Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Occupancy–abundance relationship wikipedia , lookup
Theoretical ecology wikipedia , lookup
Molecular ecology wikipedia , lookup
Latitudinal gradients in species diversity wikipedia , lookup
Reconciliation ecology wikipedia , lookup
Old-growth forest wikipedia , lookup
Habitat conservation wikipedia , lookup
Operation Wallacea wikipedia , lookup
Reforestation wikipedia , lookup
Biological Dynamics of Forest Fragments Project wikipedia , lookup
Edge effects on the regeneration of forest fragments in south Brazil A thesis presented by Efraim Rodrigues to The Department of Organismic and Evolutionary Biology in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the subject of Biology Harvard University Cambridge, Massachusetts July, 1998 by Efraim Rodrigues All rights reserved II Abstract Forested environments are being cleared all over the world. From 1981 to 1990, the loss of tropical forests was estimated at 154 million ha. As a result of that, the interface between forest and open environment has increased greatly. General patterns of species distribution and dynamics in these forest edges must be determined, if we are to recommend guidelines for the management of forest edges. This work addresses three main questions: 1) Do different types of species occur at different distances from the edge in the tropics? 2) Are non-monotonic patterns of plant density common in edges? 3) If so, which mechanism is generating it, and how do fragment size and edge orientation affects nonmonotonic patterns? In order to answer these questions, I established 48 transects 4 m wide, 25100m long. I identified 192 species, on a sample of 16,674 saplings higher than 1m, and smaller than 5 cm DBH. Transects were located in 19 forest fragments, ranging from 0.4 ha, up to 650 ha. Two additional data sets were used. Fourteen transects, 10 m wide, 45-100m long, were surveyed for density of trees (above 10cm DBH), and fifty-five transects 50m or 100m long were surveyed for light and temperature measurements. The study site is located in North Paraná, Brasil, at 50oW lon 23oS lat. The annual average temperature is 22.5o C. Monthly averages vary from 16o C up to 24o C. The average annual rainfall in the region is 1615 mm. III A historical survey of the region showed that human impact on the forests of North Paraná was very small prior to the construction of the main transportation network linking the region to the coast in 1934. The rapid deforestation process that occurred in North Paraná soon after 1934, caused edges to be even aged. The elongated shape of farms caused fragments to have straight edges, and the interspersion of large and small farms, caused small and large fragments to be interspersed. These four landscape aspects conditioned by history, plus the fact that North Paraná is devoid of major altitudinal and soil gradients, indicates that at this study site, historical and environmental variation is low, making comparisons among edges possible. The occurrence of pioneers and canopy species closer to edges than the average individual, indicates that the light enhancement at these sites, is selecting species adapted to this condition. Furthermore, where light penetrates further into the forest, (north edges), the species composition of sites deeper into the forest, resembles the species composition of sites closer to the edge of protected (south) edges. Sapling density showed to have a non-monotonic pattern. It is associated with species composition when all transects are pooled, and when different transect subsets are considered. The association sapling density/species composition ceases to exist around 35 m from the edge, in all different groups considered. Sapling density is related to (either controlling or being controlled by) lateral light on the edge, but trees control light deeper on. The point at which IV one condition switches to the other, coincides with the point at which Vapor Pressure Deficit stabilizes (35m). Further evidence that the edge is 35 m wide, is that plant diversity peaks at 35 m. This probably indicates that two different communities meet at this point. Species composition on the seven subsets (large-small, north-south, and three sizes of saplings) confirm that species composition is related to distance from the edge up to around 35 meters. After that, no pattern is shown. Species composition of whole transects in small fragments somewhat resemble the typical species composition found in edges in this region. Therefore, in addition to the fact that small fragments are more exposed to edge effects because of their higher perimeter/area relation, small fragments are, as a whole, more affected by edge effects. This fact is coupled with the enhanced amplitude of the crest and hollow pattern of density on small fragments, in relation to large fragments. The most likely mechanism enhancing edge effects in small fragments, is the additional effect of secondary edges over the one being considered. The diameter of small fragments is many times smaller than 200 m, so that entire transects in small fragments are closer than 100 m to a secondary edge. In those cases, a weak secondary edge effect is added to the entire transect, amplifying the magnitude of the edge effect. V Table of Contents Chapter 1: Life in the edge Effects on the margin of forested ecosystems ................. 1 A) Edge effects as an outcome of deforestation.......................................................1 B) Concepts of edge effects.......................................................................................2 B1) Two Approaches to the investigation of edge effects 2 Edge as ecotones (the ecological approach) 2 A general theory about boundaries (the landscape approach) 3 B2) Mathematical functions used to study Edge Effects 4 "All or nothing" function 4 "Wall" function 5 Monotonic function 5 Non-monotonic functions 6 C) Edge Effects and their causes...............................................................................6 C1) Human and landscape causes of edge effects 7 Characteristics of the forest at the moment of deforestation 7 Deforestation methods 8 Fire 9 Fragment Size 10 Matrix 12 Facing orientation 14 Slope 15 C2) Abiotic effects 15 Solar radiation 15 Humidity 16 Wind 17 C3) Direct Biological Effects 18 Regeneration 19 VI Tree Density 19 Death rates 21 Mid-distance peaks of plant density. 22 C4) Biological interactive effects 23 Herbivory 24 Egg Predation 24 Seed predation 25 Pollination 25 Plant Species Composition 27 D) How do edges change in time?...........................................................................27 E) Methodological Issues .........................................................................................29 F) Raising Questions about Edge Effects ...............................................................33 Literature Cited .........................................................................................................36 Chapter 2 The History of Forest Fragmentation in North Paraná, Brazil....................54 A) Introduction ..........................................................................................................54 B) Ancient transportation phase..............................................................................55 C) Modern transportation phase ..............................................................................58 C1) The Old North: A transportation infra structure in its beginnings 58 C2) The construction of a railway opens North Paraná 59 C3) The plan of the “Companhia de Terras 60 D) Outcomes of the colonization projects...............................................................61 Reduction of the native cover 64 E) Conclusions..........................................................................................................65 Literature Cited .........................................................................................................66 Chapter 3 Materials and Methods .................................................................................76 A) Site location and description ..............................................................................76 VII B) Data Collection .....................................................................................................78 B 1)Sapling Data set 79 The choice of plant size 79 The choice of forest fragments 81 Plant Identification 83 B 2) Tree Data set 84 B 3) Abiotic data set 84 Appendix 1 Leaf clearing protocol...........................................................................86 Literature Cited .........................................................................................................87 Chapter 4 The effect of forest edges on different groups of species......................... 97 A) Introduction .......................................................................................................... 97 B) Material and Methods........................................................................................... 99 B 1) Criteria for Species Classification 99 B 1 1) Pioneers and Climax species 100 B 1 2) Rare and Abundant species 101 B 1 3) Understorey and Canopy Species 102 B 1 4) Wind-dispersed and animal-dispersed species 103 B 1 5) Exotic species 103 B 2) Analysis 103 C) Results................................................................................................................ 105 D) Discussion.......................................................................................................... 107 Question 1) Is enhanced light causing the increase of pioneers and canopy species on the edge? 107 Question 2) Does the absence of seed predation at the edge translate into reduced occurred of animal-dispersed species? 109 Question 3) Is the edge an entrance of species to the fragment? 109 VIII Question 4) Do species that occur close to the edge, also occur in small fragments? 111 E) Conclusions........................................................................................................ 113 F) Literature Cited ................................................................................................... 114 Chapter 5 Non-monotonic patterns on edges of tropical forests………….…………..130 A) Introduction ........................................................................................................ 130 B) Materials and Methods....................................................................................... 132 Analysis 133 C) Results................................................................................................................ 136 The existence of non-monotonic patterning 136 The influence of landscape on non-monotonic patterns 140 North and South edges 140 Large and Small fragments 141 D) Discussion.......................................................................................................... 143 D1) Non-monotonic patterns and edge width estimates 143 D2) Mechanisms generating non-monotonic patterns 144 Wind disturbance 144 Niche Partitioning 144 Asymmetrical competition 145 The influence of landscape on non-monotonic patterns 146 Edge heterogeneity 148 E) Conclusions........................................................................................................ 149 F) Literature Cited ................................................................................................... 151 Chapter 6 Closing Remarks ........................................................................................ 166 A) Patterns on edge effects .................................................................................... 166 B) Answers to the Questions on chapter 1 ........................................................... 166 C) Suggestions for future studies.......................................................................... 170 IX Literature Cited ....................................................................................................... 171 X Acknowledgments This work was made possible by the contribution of many people and Institutions. Starting this thesis by showing my gratitude to them, is more than a formality to me. I look forward to having a chance to return their most valued help. First and foremost, I would like to thank my parents Lydia and Efraim for not watching TV on that warm night, back in 1964, and for putting up with all the long-term consequences of such act. Much of this work results from the perfectionism inherited from my mother, and the pragmatism of my father. My older sister Joyce also has been a constant source of support, particularly during the long Bostonian winters. Fabia, there must be a less formal way of showing my appreciation for your continuous effort during the elaboration of this work. I hope to make up for the many moments I could not enjoy your presence, and I promise that this was the last Ph.D. thesis I will ever write. Working with Otto Solbrig was a hard and enriching experience. I was benefited by his large experience, many interests, and by his great capacity of criticism. Otto was the person who most seriously took the job of training me. I thank him for the many hard times he gave me, and for exposing my weaknesses. Professors Peter Ashton, Richard Forman and William Bossert had a major influence on this work since the beginning. I thank them for putting me XI back on the track of science based on real, field data, and for the comments on earlier versions of this study. Working with a bunch of field assistants was a headache. The most fun, productive, and enriching one I have ever had. This work would have been impossible without their contribution: Ann Holland, an intuitive taxonomist, that had never worked in Brazil, and managed to teach me how to identify some species. Bruce Moreira, a computer hacker that put our GPS up and running. The nightclubs in Londrina greatly miss his presence. John Waterman, who always asked “what for?”, and in so doing, managed to change the whole field design, after asking his most preferred question a number of times. Waminda Parker, a sturdy and well-humored Australian, who traveled around the globe to participate in the data collection. Douglas Goldemberg, an experienced USDA officer, who taught me a number of tricks about how to deal with field assistants under stressing conditions. I hope Douglas will continue enlarging his collection of bot-flies. He managed to have twelve at a time, all on his head. Trevor Pattison, a tough guy, who got rid of bot-flies by painfully squeezing them out. Laila and Londa Knoll. You gals are the most productive crew ever! As promised, that large fragment in Paiquere is named here as Paiquere-Twins. XII Waldely Silva has an endless passion for field work. We always had to drag him out of the forest after 10 hours of field work. I hold no hopes for him. He will finish his days buried under piles of specimens, in dark herbariums, identifying just “one more species”. Betania Christiane Herrmann is a hard working Agronomy student. Judging by the appreciation ticks have for her, she will have a bright career as an Entomologist. Adriana Tamie Otutumi is a shy and observant student of Agronomy. I was much benefited by having her eagle eyes around. Four local Institutions offered their help, as they continuously did, over the last seven years: The Universidade Estadual de Londrina provided the labs, a field car, assistants and various materials, including 21,000 handcrafted numbered tags. The completion of this work owns a great deal to the continuous effort of this Institution to train and assist its professors. I am particularly indebted to Maria de Fátima Guimarães, Christiane Medina and Ana Maria Arruda, who took care of my teaching load when I was in Cambridge, and to the officer Dorival, and the University shop, who handcrafted 21,000 metal numbered tags. The Instituto Ambiental do Paraná, through the officers Tarcisio and Rachel, offered access to aerial photographs and hours of enriching conversation about the forests of the region. The Instituto Agronômico do Paraná efficiently provided meteorological data through the Net, after I was back in USA. XIII The Federal Ministery of Agriculture donated a large set of aerial photographs, that were used in this work, and that are serving the new-born Laboratory of Landscape Ecology of the University of Londrina I definitely look forward to going back to Londrina and to continue working with them. I am extremely grateful to Capes (Coordenadoria de Aperfeiçoamento de Pessoal de Ensino Superior). My presence in Cambridge and my field trip to Brazil would not had been possible without their aid. Their organization and punctuality were remarkable, particularly in face of all the infrastructure problems. The David Rockefeller Center for Latin American Studies (DRCLAS) and the Department of Organismic and Evolutionary Biology also provided funds that made the data collection possible. Several statisticians provided continuous, patient and valuable help during the data analysis of this thesis. Professor Robin Chazdon provided real-time assistance through the net. The analysis of diversity and of spatial patterns was much facilitated by her assistance. Professor Michael Palmer was extremely helpful on putting CANOCO to work. He must be an innate professor, who spends much of his time assisting people. I look forward to meet him personally. XIV Professor Robert Rosenthal rebuilt my faith in ANOVAS. That was not an easy task, but I am sure it will have long term effects on my future professional life. The soon to be Professor William Hoffmann also helped to calm my hard feelings about ANOVAS. His statistical assistance dates from much before the data analysis of this study. Professor Eugene Gallagher was instrumental with the CNESS analysis of diversity. He spent a whole day taking me for a walk in multidimensional space. Weber Antonio Amaral was a key person on helping me adapt to Cambridge. I profited of his company both personally and scientifically. I do hope that our friendship and partnership continue after our return to Brazil. I also thank John and Ann Fraser, Anna Herspberger, Swee Peck Quek and Pedro Laterra for many suggestions on previous versions of this thesis. I do look forward to return their long effort of reading my manuscripts. Renee Gonzalez brought a component of Mexican happiness to the early phases of this work. The 13 billion volumes included in the Harvard Libraries would be useless without the patient, kind and effective assistance of their librarians. I hope that in a next life, Judy Warnement, Lisa Ann DeCesare, Gretchen Wade and Dorothy Solbrig return as grad students, while I will be the librarian. Many taxonomists assisted the identification of the plant specimens. I am life indebted to them. The choice of asking their help was not solely based in competence, but also in kindness and friendship. Working with them was not XV only effective, but also pleasant. They were: João Batista Baitello, Marilda Dias, Maria Lucia Kawasaki, João Aurélio Pastore, Manoel Paiva, Charles Schnell, Lucia Helena da Silva and Roseli Torres. XVI Chapter 1 Life in the edge: Effects on the margin of forested ecosystems A) Edge effects as an outcome of deforestation Tropical forests are being cleared all over the world. From 1981 to 1990, their loss was estimated at 154 million ha (FAO 1993). Deforestation is a many layered process. From the social point of view, it means losing ancient culture and knowledge about the forest that will take many years to be re-acquired, if ever. From the perspective of natural resources, deforestation means the misuse of a resource that could otherwise provide clean water, new medicines, food and many forestry products at low costs. The damage of deforestation is not restricted to the amount of area cleared. When a patch of forest is cleared, the portion of the forest contiguous to the open area (hereafter called edge) experiences a new environmental condition that results from exposure to increased solar radiation, wind, rain, as well as the effect of herbicides, insecticides and fertilizers. The phenomenon is somewhat similar to a natural forest gap (Denslow 1987), but the intensity of the processes differ. Gaps are the result of one or more tree falls, whereas forest margins are exposed to kilometers of agricultural fields or urbanized areas (Rodrigues 1993). Gaps in the forest will heal themselves after some years of intense tree growth Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 1 (Brokaw 1985), but the agricultural/urbanized matrix is maintained by frequent plowing or road maintenance. Species loss is therefore not restricted to the direct action of tree cutting. The exposure of the remnant forest to the open environment means further disturbance both in time and space. Skole and Tucker (1993) estimated that even if just 6% of the Amazon forest was cleared by 1988, 15% was actually affected by fragmentation and edge effects. (“edge effects” is used here as generic term referring to the change in abiotic and biotic factors, in response to proximity to a forest edge). B) Concepts of edge effects B1) Two Approaches to the investigation of edge effects Two main approaches have been used to study edge effects. One arose from the study of transitions between ecosystems, and originated in ecological studies at the beginning of this century. The other originated in landscape design, and as such, emphasizes shapes. Both approaches differ on the scale they use. While the ecological approach focuses on a small scale (less than hundred meters), the landscape approach focuses on a scale of tens of kilometers. This thesis uses a number of concepts and definitions taken from both approaches. Edge as ecotones (the ecological approach) The earliest mention of the term “ecotone” appears in Clements (1905). The author uses ecotone to refer generically to the transition between two Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 2 communities. While Clements believed that succession leads to a unique climax state, Gleason believed in multiple possible climax states, defined by history (Gleason and Cronquist 1964). Despite strong disagreements between Clements and Gleason, both acknowledge the importance of ecotones. Odum (1971) was the first author to point out that human-made boundaries were a type of ecotone, leading to a broader definition of the term ecotone. Hansen et al. (1992) suggest no strict use of the words ecotone, edge, boundary and transition zone, widely used as synonyms. Murcia (1995) and Lovejoy et al. (1986) make a clear distinction between ecosystem boundaries (termed ecotones), and human-made boundaries (here termed edges). However, the search of common patterns between both situations seems more promising for the developing of a theory on boundaries. A general theory about boundaries (the landscape approach) The study of landscapes has developed as a way to converge different concepts of environment. It involves both social and natural sciences (Forman and Godron 1986). The search for general principles, arising from such contrasting subjects, has driven landscape ecologists over the last decade. Forman and Moore (1992) proposed principles that would include humanmade edges into a general theory of boundaries. Forman and Moore (1992) classified the structure and function of boundaries into several groups, presented in table 1. In spite of Forman and Moore's (1992) attempt, the scientific production on the area continues separated into natural and man-made Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 3 boundaries, and into small scale (ecotone approach), and large scale (landscape approach). B2) Mathematical functions used to study Edge Effects Edge effects are frequently described in terms of a function relating factors and distances from the edge. More than simply a method, the function used to describe how a factor varies in relation to distance from the edge influences the way data are collected, analyzed and the conclusions drawn from them. Edge functions have been defined with an increasing number of parameters. However, the more parameters a function has, the higher the number of replications needed for their adequate estimation. This probably explains the abundance of papers using the more exploratory functions ("point" and "wall" functions), and the lack of papers using more sophisticated non-monotonic models. "All or nothing" function The "all or nothing" function has the only purpose of refuting or confirming the existence of an edge effect. Points are taken at the edge and on the interior, and their differences tells whether an edge effect exists or not. Exploratory papers like Williams-Linnera (1990b); Leopold (1933); Holl and Lulow (1997) and Morato (1994) used "all or nothing" functions to describe edge effects. To describe an edge as an “all or nothing” process, facilitates field work by reducing the sampling and by concentrating all the statistical power on asking whether a certain pattern occurs or not. However, once the existence of a pattern is proven, more detailed experiments are needed to correlate the Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 4 intensity of the pattern to distance from the edge, and then to understand the underlying mechanism. "Wall" function Similarly to the "all or nothing" function, the "wall" function assumes that edge effects either occurs or not, without intermediate steps. In addition to that, wall functions assume that a major discontinuity, analogous to a wall, separates the edge from the interior. The “wall” model represents an improvement over the point model, since it raises the question of where the "wall" is located, or how wide the edge is. Among all the edge aspects, the determination of edge width is of major importance to conservationists. Indeed, its existence would provide estimators of how much area is lost to forest degradation at landscape level, after some basic geometry involving fragment area and shape (Leopold 1933; Laurance and Yensen 1990; and Paton 1975). In spite of the importance that such width could have for forest management, it is quite unlikely that a unique measurement will ever be found, considering the many different abiotic and biotic factors involved (Murcia 1993 and Laurance et al. ,1997). Monotonic function Monotonic functions are characterized by having a first derivative that does not change its sign, being either positive or negative. They are ever increasing or ever decreasing functions. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 5 The underlying assumption of monotonic functions when applied to edge effects is that the closer to the edge, the more intense an effect is. Much more information can be obtained by using non-monotonic functions than the previous two functions. Patterns can be characterized in terms of slope of decay as one gets away from the edge, and the distance in which a factor reaches values similar to forest interior can be assessed, even if no major discontinuity exists. Non-monotonic functions As opposed to monotonic functions, the first derivative of non-monotonic functions may change its sign from positive to negative. Non-monotonic functions have crests and hollows. The non-monotonic function assumes that factors may not only increase or decrease with increasing distances from the edge, but also may oscillate. Quadratic functions, as well as sine-cosine functions, can be non-monotonic. The literature has occasionally indicated that some edge effects may have non-monotonic trends (Murcia 1995), but to my knowledge, non-monotonic models were never used in describing forest edges. There is, however, substantial evidence of their existence. The following works contain graphs showing non-monotonic patterns: Monro (1992); Camargo and Kapos (1995); Turton and Freiburger (1997) and Viana et al. (1997). The authors however, either considered the trend as random noise, or did not offer any explanation for their existence. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 6 C) Edge Effects and their causes For the complexity of their many interacting causes, forest edges have captured the attention of many ecologists for several decades. Several attempts have been made to disentangle these effects (Rodrigues 1993, Murcia 1995, Turton and Freiburger 1997, Laurance 1997), and they all have reached, with different words, the same conclusion: Effects are divided into physical or abiotic effects, direct biological effects and interactive biological effects. Abiotic effects are among others, solar radiation, humidity, and wind. Direct biological effects are composed by the direct action of the abiotic effects over the community, like the increase in plant density due to increased solar radiation. Interactive biological effects refer to processes involving two or more species, like competition, predation, herbivory and pollination. C1) Human and landscape causes of edge effects Characteristics of the forest at the moment of deforestation When edges are created, plants are totally wiped out or exposed to high levels of light, depending on where they are in relation to the edge. Depending on the species, response to the pulse of light may vary from vigorous growth to withering to death. Recently created edges show high mortality rates during the first years (Lovejoy et al. 1984). Canham and Marks (1985) speculate that disturbance kills tree species differentially, in face of the many traits that reduce their susceptibility, like Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 7 butresses, and root sprouting. However, no data are available on which species or groups of species are more susceptible to death from exposure to the edge. Distribution of tree species also varies in time (Lieberman and Lieberman 1987). Trees die, gaps are opened, and new individuals occupy these gaps. This mechanism is so important to the occurrence of tree species, that forests were once defined as a mosaic of former gaps (Aubréville 1938). Edges created in large gaps will supposedly have different dynamics than other edges. Large gaps are the exclusive site of occurrence of juvenile pioneers (Brokaw 1985). Pioneers are adapted to make the best use of the pulse of light and nutrients available in gaps by growing fast and producing large amounts of seeds. These traits suggest that pioneer species will dominate the edge after some generations. It is likely that the dispersal of pioneer tree species along the edges of a fragment start from those edges created in existing gaps. Deforestation, however, is rarely an all or nothing event. It is normally preceded by or associated with selective logging, hunting, grazing, and opening of roads. All these factors interact to further impoverish the forest. Deforestation methods The methods used for deforesting an area influence the forest fragments more heavily during the first years after fragmentation. Describing how deforestation methods affect the edge of remaining forests is a difficult task, since in most cases, very few records are left from older deforestation episodes. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 8 Chainsaws were invented around the end of the 1950’s. Before that time, tractors in the tropics were rare and restricted to agriculture. Both facts support the notion that before 1950, the forest was felled with axes and two-person saws. It is said that a large fig tree could take days to be felled (Companhia Melhoramentos do Norte do Paraná 1975). Fire was extensively used (and still is, in many regions) to eliminate the debris, and to get rid of unmarketable wood. In the specific case of north Paraná-Brasil, local farmers claim that not all wood was sold. Some lots were too far from the sawmill to allow transportation. It is possible that trees close to the edge were felled towards the forest fragment interior, in those cases where the wood was not marketed. This is the easiest way of disposing trunks. On the other hand, many current forest owners in the region claim that the forest was left because of the environmental awareness of the landowners. In that case, it is reasonable to think that the landowners were sufficiently involved with the process, to prevent damaging the remaining forest. Tree felling onto edges is not a problem in regions with shifting cultivation. Farmers fell the trees in the field, in order to burn them, and enhance soil fertility (Williams-Linnera 1990b). Recent deforestation methods involve large tractors, and chain saws. Most of the litter is now pilled on leveled contours. Modern deforestation practices are more likely to cause soil disturbance on forest edges, given the large size of the machinery used. According to Williams-Linera (1990b), soil disturbance along edges, triggers germination of pioneer seeds, accelerating their occupation of forest edges. The industrial pace and scale of deforestation as it is done today, Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 9 precludes detailed attention, as to the occurrence of poor soils, stones, and soggy areas. Those areas could be left as forest remnants with no reduction of agricultural production. Fire Fire is another threat to the recently created fragment. In most cases, fire is used for removing the debris. It is likely that in many instances, it spread into forest fragments. As in tree felling, it is possible that the small fraction of landowners willing to have a forest fragment in their farms would also be also the most protective about them. According to Primack (1995), when a forest is fragmented, increased wind, lower humidity, and higher temperature at the forest edges make fires more likely. Fires may spread into habitat fragments from nearby agricultural fields that are burned regularly, as in sugarcane harvesting. Viana et al. (1997) and Lovejoy et al. (1984) describe such situations in detail, with sugarcane and pasture, respectively. Fire is more frequent in regions where forest-clearing is taking place Nepstad et al. (1997). In those regions, large areas of degraded, fire-prone forest are coupled with frequent use of fire. Large fires are then inevitable, as happened in north Paraná-Brasil, in 1963, Indonesia in 1982/3 1997 1998 (Cifor 1997) and Roraima-Brasil, in 1998 (Folha de São Paulo 1998). It is likely that many of the mechanisms related to edge dynamics like tree death, are promoted by fire. It is also very likely that many fragments undergo an Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 10 event of fire during their early existence, and also after that, depending on the land use of the region. Fragment Size Edges and fragment size are linked by a simple geometrical relation: The larger the fragment, the further a patch of forest can be away from the edge. This constitutes the basis of the core area model of Laurance and Yensen (1990). Therefore, fragments with diameter less than twice the width of an edge effect, are considered to have no interior (area not affected by edge effects). Malcolm (1994) shows the importance of other edges, apart from the one being studied, on edge effects. According to that author, an additional effect is added to the main one when a transect is close to other edges. In the case of a small square fragment, the effect of the two other edges, close to the one that is being studied, will reduce differences of factors along the transects, by affecting all the transect evenly. Following Malcolm’s (1994) hypothesis, Kapos (1989) proved that the average temperature within 60 m from the edge in small fragments is higher than that within the same distance in large fragments. Temperature reaches lower values within the first 60m in large transects than in small fragments. If other abiotic variables follow these temperature measurements, then edges of small fragments are drier and warmer than edges of large fragments. Large fragments, however, have a higher boundary contrast (sensu Hansen et al. 1992) and boundary width (sensu Forman and Moore 1992), in relation to abiotic variables. It is unknown whether plant populations Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 11 will respond more intensively to a wider, higher contrasting gradient (large fragments) or to a narrow gradient with extreme values (small fragments). Yahner (1988) proposes an alternate hypothesis. The author mentions that edge length is more important than edge width. Assuming that the edge environment is different from the forest, and that forest species will have to be substituted by edge species, then the longer the edge, the higher the probability of being invaded by edge species. Therefore, the edge of large fragments would be more intensively taken by edge species than small fragments. Ashton (pers.comm.) confirms that fragment area and edge effect may interact in such a way. Nee and May (1992) propose that soon after fragmentation, fragments are dominated by species with outstanding competitive performance (K-selected species). These species, however, cannot effectively disperse into other fragments. As the K-selected species go extinct, they are substituted by species with higher dispersal performance. This was named the extinction-debt theory (Nee and May 1992). The debt to which the theory refers is the increase in species extinction expected, by considering dispersal differences among species. Therefore, following this hypothesis, small fragments soon become dominated by dispersors, so that the boundaries of small fragments should have a larger width and lower contrast in terms of species composition, than large fragments The three aforementioned hypotheses point to two contrasting scenarios: One claims that small fragments are more affected by edge effects, and the Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 12 other that large fragments will have a stronger occupation of edge species. The final answer to this question can only come from field data, which, excepting that presented by Kapos (1989), are totally absent. Matrix The open environment between forest fragments is referred to as the matrix. The contact between the matrix and the forest is the primary cause of edge effects. Considering the importance of the matrix for edge effects, it would be expected that many ecologists would have worked on comparing the effects of different matrices on forest edges. However, the matrix has rarely been the object of specific investigation, particularly in the tropics. Metzger (1997) compared the importance of fragment size and isolation, forest connectivity and matrix heterogeneity on tree species diversity on fragments in São Paulo State, Brasil. The author concluded that the last two factors (both connected to the matrix) were more closely related to tree species diversity than fragment size and isolation. Bierregaard and Stouffer (1997) studied six species of birds in forest fragments in two types of matrices in the Amazon. The authors concluded that six frugivorous bird species showed no change in density whether the matrix was dominated by Vismia sp or Cecropia sp. The apparent contradiction between Metzger (1997) and Bierregaard and Stouffer (1997) illustrates the different concepts of “contrasting ” matrices. The landscape studied by Metzger (1997) is formed by a mosaic of urban areas, Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 13 agricultural areas, reforested areas, meadows and semideciduous forests. Contrasting matrices in this context means different combinations of these elements. The matrix of Bierregaard and Stouffer (1997) is compounded of continuous forests and secondary growth forests, in this context, “contrasting” whether the matrix is dominated by Vismia sp or Cecropia sp. Apart from the differences in the matrices themselves, both articles suggest that the effect of the matrix on mobile species (frugivorous birds) is higher than on sessile species (trees). Facing orientation The orientation of the edge determines whether its not an effect per se, but it controls the edge exposure to anisotropic or directionally dependent elements. Solar radiation and wind are the two principal elements. Edges facing the equator are more exposed to sunlight, than those facing poleward. Differences in solar radiation between north and south facing edges get more important further from the equator (Forman and Godron 1986). In the temperate region, edges facing the equator show a deeper or sharper gradient than those facing poleward, in terms of both biotic and abiotic effects (Wales 1972; Fraver 1994, Matlack 1993 and 1994; Palik and Murphy 1990; Cadenasso et al. 1997). Little can be concluded from the two papers (Viana et al. 1997, Williams-Linnera 1993) discussing the effect of orientation in the tropics, for lack of replications, or historical differences between the edges analyzed. Turton and Freiburger (1997) working on a fragment in Australia found Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 14 no differences in seedling density between north and south edges. The present works are not sufficient to conclude whether edge orientation is an important factor in the tropics or not. Besides north-south differences, it is also possible that east (morning) edges are more prone to occupation by sun-loving species, given that photosynthesis is more intense during the morning (John Grace, pers comm.). Brothers and Spingarn (1992), however, claim that the frequency of alien species is higher on the west and south (equator-facing) edges. The authors did not contrast eastwest edges alone. The higher dominance of alien species in south-west edges than on north-east edges may result only from differences between north and south edges. Two papers (Turton and Freiburger 1997) and Viana et al. (1997) suggest that plant density was higher on edges facing west and east than north and south. Both authors, however, do not present explanations for the difference. Orientation has also great importance in regions where the wind has a constant direction. Furthermore, variation of wind pattern throughout the year may cause edges that are protected from wind disturbance in one season to be exposed in another seasons. Slope The effect of slopes on forest edges is poorly known. To my knowledge, no article about the matter was ever published. I believe that the slope may act as an orientation enhancer i.e., south edges on a south-facing slope will be the Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 15 most shaded (south hemisphere), whereas those in the same condition, but in a north facing slope would be more exposed to the sun. C2) Abiotic effects Solar radiation Light is a vital resource for plants, since it is the provider of energy. Light also affects the water content of the forest by increasing evaporation. In addition, species respond differently to both resources (light and water). The amount of solar radiation a patch of forest receives exerts influence over germination, survival and death of trees. The increase of solar radiation along edges has been described many times in the literature (Brothers and Spingarn 1992; Cadenasso et al. 1997; Matlack 1993; Kapos 1989). Decreasing values of light radiation normally reach forest interiors values within less than 15m from the edge. (table 2) The effects of lateral solar radiation are not limited to a narrow band along the edge. Solar radiation heats up and dries a large mass of air in the surrounding matrix. Edge characteristics, like the amount of biomass and edge orientation to prevailing winds, determine how far the mass of dry air coming from the matrix will penetrate the forest. Humidity Humidity is produced by plants as evapotranspiration. Water vapor constantly flows from plants to the atmosphere at a rate determined mostly by air temperature and atmospheric humidity. Wind intensity and its incidence angle Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 16 on the forest edge determines the patterns of air circulation at the edge, and how water vapor moves among forest edge, forest interior, and matrix. The interaction between air temperature, wind and humidity makes estimates of edge width for humidity very variable. Edge width for humidity was wider than the solar radiation edge in most cases presented in table 2. VPD (Vapor Pressure Deficit) width was also wider than light width whenever both were measured, as in Kapos (1989) and Matlack (1993). The interaction between light and humidity will therefore create three zones from the edge towards the forest interior. Close to the edge, light is high and humidity is low. After that, comes a mid zone, where light is as low as in the forest interior, and humidity is still increasing. Finally comes the zone where both light and humidity values are similar to the forest interior. The width of these zones are likely to vary throughout the year, as the meteorological conditions change. The lower the solar track is (during the winter), the more profound the penetration of lateral light will be. However, air temperature and humidity in the matrix are lower during winter (in south America). Wind blows differently in different seasons of the year. Even if all these factors do not seem to affect the existence of the zoning (given the various locations and seasons represented in table 2), it certainly affects its size. As shown by Turton and Freiburger (1997), VPD widths vary between 10m and 30m in the same site, with measurements taken only 1 month apart. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 17 Wind Trees bar the wind in forest edges. In doing so, they dissipate its kinetic energy with two major consequences for the forest community: Trees are subjected to wind shear forces, and the forest interior is sheltered from wind action. Measurements above the Amazon Forest have shown that wind speed above a forest can be 16 times higher than interior values (Robert et al. 1990). The sudden reduction of wind speed causes turbulence. Air will flow freely in the matrix until it reaches a forest edge, when it forms vortices. A steady flow of air is more effective as a mean of transportation of energy, propagules and water, than a vortex (Forman 1995). The flow of air into the forest edge and the formation of vortices around it suggest that the edge acts as a sink, receiving dry or humid air, and propagules from either the matrix or the forest interior, depending on angle of incidence of the wind on the edge. The area around the edge in which wind circulation is affected can be very large. Measurements on scale models determined that wind reduction occurs 5.5 tree heights to the windward side of a forest, and is reestablished 11 heights to the leeward side of a scale model of a pine forest (Meroney 1970). Values are supposed to be higher in tropical forests, given their higher biomass and coefficient of attrition. Indeed Laurance (1997) affirms that wind shear forces may act much stronger than Meroney (1970) suggests. According to Laurance (1997), the community can be disturbed as deeply as 500 m into the forest. The orientation of the edge in relation to prevailing winds is crucial for determining wind-shearing forces on trees, and their implications for free Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 18 mortality. As shown by Seginer (1975), wind impact is correlated with the cosine of the angle of impact. According to the cosine function, small departures from perpendicularity do not affect wind impact heavily, but a sudden drop in impact force occurs, as the angle of impact grows larger than 45 degrees. C3) Direct Biological Effects The increase of light, Vapor Pressure Deficit and wind in forest edges has direct consequences for the plant community. Despite the differential response of species to these three factors, wind and desiccation increase the chance of death, and enhanced light normally causes enhancement in plant density. Direct biological effects refers to non-specific effects to the community. Regeneration Five out of six papers presented in table 3, show a trend of decreasing density of regeneration further from the edge, into the forest. Edge conditions (enhanced solar radiation, wind, and lower humidity) seem to be the most likely cause for this. These five different papers (Ranney 1981; Malcolm 1994; Camargo and Kapos 1995; Willson and Crome 1989 and this work, on a later chapter) come from five different forests, both temperate and tropical, from Australia, United States, Amazon Forest and Atlantic Rainforest. Turton et al. (1997) present an opposite trend. The study was performed on an old fragment (70 years) of seasonal mesophyll vine forest in Australia. It is likely that under drier situations and after a long period (70 years), the water stress is so severe that even the numbers of saplings of sun-loving species will Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 19 be reduced close to the edge. Another possibility is that the single 20 ha fragment used by the author was not a representative one. Methodological issues will be discussed later in this work. Willson and Crome (1989) showed that wind dispersed and vertebrate dispersed species produce larger amounts of seeds on edges, relative to the forest interior. Increased amounts of light are responsible for both the increase in reproduction rates and the increase in tree growth (Chen et al. 1992 and Mayaka et al. 1995). Tree Density The reduction of tree density towards the interior of the forest is a very visible effect in edges. Most examples in table 3 indicate that edges have more trees when compared to the forest interior. The three exceptions to the rule are worth commenting on: A) Young et al. (1995) studied a nutrient gradient in which the scattered canopy of an impoverished area, intensely grazed in the past, meets the better-conserved forest. Therefore, the described increase in plant density with distance from the edge is solely a result of a nutrient gradient, different from the maintained edges discussed in the rest of table 3. B) Camargo and Kapos (1995) studied edges still under the intense change experienced soon after deforestation. As shown by Lovejoy et al. (1984), Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 20 edges experienced high rates of death within two years after deforestation. Therefore, soon after deforestation, tree density may be smaller in the edge, than in the forest interior. Camargo and Kapos (1995) and Lovejoy et al. (1984) worked in the same region, if not in the same fragment. C) The site of Chen et al. (1992) supposedly was in the same initial stage as Camargo and Kapos (1995), even if chronologically delayed, due to a colder climate. The case presented by Williams-Linera 1990a best illustrates the most illustrative about time series. The author shows tree density data on edges aged from 10 months up to 12 years in which density at the edge gets higher with increased age of the edges. In all cases considering trees, a reduction in plant density with distance from the edge was noticed whenever sufficient time since fragmentation was allowed. Death rates Increased tree death along edges is a consensus in the literature. Besides the seven works presented in table 3, Williams-Linera (1990a), Ruth and Yoder (1953), Gratkowsky (1956), Alexander (1964), Ranney (1977), all present data confirming that rates of tree death increase along the edge. A combination of desiccation and wind throw seems to enhance death along edges. Esseen (1994) managed to isolate wind-induced death from other Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 21 factors by showing that in small fragments death and tree diameter are positively correlated since the impact of wind is stronger on larger trees. In larger fragments, death and tree diameter are independent, because trees die from causes other than wind. The increase of tree density in edges discussed in the previous section seems to be at variance with the increase of tree death. Apparently, if more trees die at the edge than in the forest interior, the edge should have fewer trees than the forest interior, not more. It is possible that higher rates of recruitment at the edge are creating the higher density, which is supported by the higher density of regeneration at the edge. Actual data on plant recruitment at edges, comprising sequential collections, are absent. Mid-distance peaks of plant density. The current paradigm of edge effects states that the closer a patch is to the edge, the more intense the edge effect is on that patch. However, plant density has been shown to peak at intermediate distances from the edge in several instances (Monro 1992; Camargo and Kapos 1995; Turton and Freiburger 1997; Viana et al. 1997). Little attention has been paid to the fact that a peak of plant density at an intermediate distance from the edge contradicts that paradigm. Crest and hollow patterns may be generated by three different mechanisms: 1) asymmetrical competition 2) niche-partitioning 3) wind-related death. It is very likely that they all work concurrently in edges. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 22 1) Lateral incidence of light causes an increase in plant density and biomass in the border of forest fragments, as shown in table 3. According to Chen et al. (1992) and Mayaka et al. (1995), individuals along the border grow faster than interior ones. They have more access to light and they shade an area close to them, towards the interior of the forest fragment. Recruitment is reduced in this area. Deeper in the forest, plants again have more light because they are close to a patch with fewer plants. The alternation of suppressed/non-suppressed patches creates a crest and hollow pattern of density, also termed by Franco and Harper (1988) as competition-effect wave. 2) Crest and hollow patterns of plant density may also arise as a result of a combination of different environmental gradients. Light gradients are much steeper than VPD gradients. Edges are characterized by much light and low humidity. Towards the forest interior, humidity rises, and light declines. Humidity values will take a longer distance to achieve forest interior values than light. So, an intermediate region is formed with interior values of light and still reduced humidity. Many different environments occur within a short space of tens of meters, and they are fairly constant with time. This constancy supposedly allows plant populations to find their optimal site and thrive there, reaching high density values. This mechanism is similar to gap niche-partitioning, (Ricklefs 1977; Hartshorn 1978; Denslow 1980; and Orians 1982). I believe that niche partitioning on edges is stronger than in Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 23 gaps since the environmental gradients are more intense and persist longer. 3) Crest and hollow patterns can also be created by wind-related death, as proposed by Sprugel (1976), in temperate forests. It is possible that a middistance peak is simply a result of increasing the probability of wind death of trees, the closer they are to the edge. As a result, the whole border is maintained in an early stages of succession with a large number of small plants occupying it. The trend of increasing density does not extend till the edge because of extreme dryness. Reduction of density at the edge creates the peak. C4) Biological interactive effects Biological interactive processes refer to those arising from interaction between two or more species, like predation, herbivory, and competition. Such processes are not yet fully understood even in well-conserved tracts of forest. Human impact on these interactions contributes an additional layer of complexity, making it even harder to understand general mechanisms. Herbivory A large number of studies has established that increases in plant biomass at the edge result in increases in herbivore densities (Leopold 1933 Alverson et al. 1988; Owen 1971; Brown and Hutchings 1997; Terborgh et al. 1997). These papers cover a large array of species and environments: Alverson et al. (1988) described the enhancement of deer populations in forest edges in Wisconsin, Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 24 USA. Brown and Hutchings (1997) and Owen (1971) worked with butterflies in south America and Africa, respectively. Terborgh et al. (1997) described the increase of leaf-cutter ant density by an order of magnitude between small islands on Guri Lake (Venezuela) and control areas. In addition to increased plant biomass, several edge effects (reduced humidity or increased wind), can force predators out of the edge. Some of the cases of increased herbivory mentioned here may also be influenced by a lack of predators. Egg Predation Reduced population of birds in fragmented landscapes has been connected to nest predation (Gates and Gysel 1979). Two factors seem to explain this connection: Nest density is higher along edges (Gates and Gysel 1979), and predators prefer edges. According to Forman and Godron (1986), in such sites, predators can integrate resources from more than one ecosystem. Paton (1994) reviewed the works describing the effect of proximity to forest edges on bird nest predation and parasitism. According to the author, depredation rates were higher closer to the edge in 10 out of 14 artificial nest studies, and in 4 out of 7 natural nest studies. Parasitism was higher closer to the edge in 3 out of 5 studies. Seed predation Burkey (1993) affirms that seed predation increases with distance from a secondary road, amidst a tropical forest in Mexico. However, Holl and Lulow Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 25 (1997) failed to obtain any effect of distance from the edge on seed predation, between a forest and a pasture in Belize. Osunkoya (1994) also worked with forest/pasture edges in Australia. The author concluded that seed predation was higher in the forest interior than in edges. Sork (1983) reached the same conclusion, working on pignut hickory (Carya glabra). The four papers agree that seed predation is higher in the forest interior, whenever differences are found. Nonetheless, these scanty data on seed predation on edges warn against early conclusions. An interesting question about seed predation in edges is what is keeping seed predators away from edges. Is it the lack of food sources (species that produce large, nutritious seeds)? Alternatively, if the food sources are present, is some edge effect acting directly on the seed predators? Pollination It seems reasonable to believe that many pollinator species cannot cope with edge conditions so that plant species that depend upon these pollinators will have reproductive problems. However the little information available about pollination in edges seems to refute the notion that pollinators are affected by edge effects. Murcia (1993) studied pollination in the edge of three medium sized (75, 300 and 700 Ha) fragments in the Colombian Andes. None of the 16 plant species studied (including 2 that were introduced experimentally) showed consistent Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 26 changes in the number of pollen tubes reaching the base of the style, in association with exposure to the forest edge. Pollination seems to be more associated with fragment size than distance from the edge. Powell and Powell (1987) Aizen and Feisinger (1994), Jennersten (1988), and Spears (1987) found effects of fragment size on pollination, working in the Amazon, the dry forest in Argentina, Sweden, and Florida, respectively. The mechanism seems to be related to the fact that small and isolated populations in small fragments are less attractive to pollinators than large populations in large fragments. Edge exposure and its associated environmental conditions seems to play a less important role than fragment size, particularly if edge populations are close to other conspecifics further into the forest. The only work actually showing a relation between edge exposure and pollinators is by Morato (1994). It is inconclusive, for it only demonstrates that species of Euglossine bees occurring on one edge are different from those occurring in one patch of continuous forest. Since no work was done with host plants, it is not clear whether the bees are not able to cope with edge conditions, or their host plants does not occur on the edge. Plant Species Composition The occupation of edges by a different set of species than those in the interior is a recurrent theme in most works on edge effects, since Shelford (1913). Change on species composition seems to be a function of time since Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 27 edge creation (table 4). The older an edge is, the more time populations have to respond to the different environments at the edge and the more clear the differences between edge and interior become. Besides differential reproduction, it is possible that different species are more susceptible to death by disturbances like windthrow and desiccation. If that is the case, forest edges can develop a characteristic species composition in less time than would be necessary by differential reproduction. These two mechanisms imply different diversity levels in the edge. Increased reproduction of a set of species in the edge implies that the edge will be more diverse during intermediate steps through relaxation (a number of species will be included, and they will dominate the edge by excluding the others), whereas differential death suggests that diversity decreases soon after the edge is created (susceptible species become extinct locally). Data that would answer this question (assessments of species diversity at different distances from the edge) are non-existent, with one exception: Matlack (1994) showed no marked differences on species richness at different distances from the edge. D) How do edges change in time? Age is a major factor defining edges. At the very moment of deforestation, the edge is no different from the rest of the forest. Differences are established as the edge grows older. The question of whether edges take more area as they Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 28 grow older, or stabilize, protecting the forest against further degradation, has remained an unanswered question since the study of Lovejoy et al. (1984). According to Ranney et al. (1981) and Matlack (1994), edges pass through three phases: a) Pattern formation, when the community responds strongly to the recent creation of the edge; b) Reassortment of physical gradients, when closure of the side canopy blocks radiation, and c) Pattern relaxation (considering the unrealistic possibility that the edge is allowed to expand into the matrix). The data presented in table 2 and 3 permit us to improve the edge time sequence proposed by Ranney et al. (1981) and Matlack (1994). Soon after deforestation, extensive tree death occurs in the edge as show by Laurance (1997) and Kapos (1997). Density of large individuals consequently increases from the edge towards the interior of the forest (Chen 1992; Camargo and Kapos 1995 and Malcolm 1994). The establishment of VPD and light gradients (table 2) will then select the species that recolonize the edge. After the new species grow, surveys on large individuals will show a tendency of decline from the edge towards the interior of the forest (Rodrigues 1993, Willson and Crome 1989, Williams-Linnera 1990a, Palik and Murphy 1990, Ranney et al. 1981). Ranney et al. (1981) speculate that the increase in tree density close to the edge is transient, whereas the increase in basal area is permanent. According to the authors, thinning occurs after the increase in density, so that edges end up having approximately the same number of trees as before, but trees are larger than before fragmentation. This is still an untested hypothesis and depends on Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 29 how succession will proceed at the edge. If new species with smaller size, life span, and low wood density (pioneers) occupy the border, then it is not likely that the basal area will ever be the same as before fragmentation. On the other hand, if similar species to those from the original forest occupy the edge, then the edge plants may return to their original basal areas after these individuals grow, which in the tropics, will take at least several decades. A negative feedback equilibrium may then be reached, when the dry hot environment favors of the tolerant pioneer species and these pioneer species grow to large densities, as they often do, sealing the edge and preventing the penetration of the hot air in return. On the other hand, if edge "sealing" is not sufficient, dry air penetrates further into the forest, moving the edge further into the forest. Tree death (table 3) will be high, either because the tree species from the pristine community can no longer survive in the edge, or because the new light demanding species have a shorter life span. E) Methodological Issues The study of edge effects has become more quantitative over time, as in many new areas of research. To my knowledge, the first study concerning edge effects was done by Leopold (1933), where the author describes edge effects based solely on his own personal subjective experience. Gysel (1951) then made a description of one edge in Michigan. Wales (1972) did one of the first Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 30 hypothesis tests based on quantitative assessments of edges. The author tested the hypothesis that north and south edges do not differ. Forest edge communities are the result of many interacting factors. Therefore, adequate hypothesis testing involves controlling for unwanted variation. When testing, for instance, the hypothesis that edges with different orientations do not differ, it is crucial to compare edges with similar land use history. Williams-Linera (1993) and Wales (1972) attempted to compare different edge orientations that also had marked differences of management history between them. Therefore, it was impossible to separate orientation from history. Gysel (1951) facing the same problem, opted not to refer to differences between north and south edges. A second problem in comparing edges is that there is always variation between patches of forest, even after ruling out all possible factors, like soil, succession, slope, exposure etc. Two patches of forest always differ. Wales (1972) first pointed out the problem of contrasting edges when background variation is present (which is always the case). In spite of the differences in species composition between south and north transects, the author concluded that they were non-significant, because they were smaller than the differences in species composition between south and north slopes in the region. A more classical way of controlling for background variation than the one used by Wales (1972), is to use replications. Williams-Linera (1993) and Viana et al. (1997) compared edge orientations without adequate replication (some classes of orientation had replications, some did not). Therefore, it is impossible Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 31 to know whether the differences attributed accredited to orientation were not just background variation. Forest fragments differ in their history, environmental condition and in the type of forest in which they originated. By restricting the sampling to one single fragment, one is susceptible to strong error when generalizing the results (Palik and Murphy 1990). Whitney and Runkle (1981); Wales (1967); Wales (1972); Viana et al. (1997) and Tabanez (1997) results derived from one fragment. Pseudo-replication consists of considering two measurements closely located as independent when they are not. As it is widely known today, biotic and abiotic data are frequently autocorrelated (Griffith 1987; Legendre 1993). This means that measurements that are taken at close locations tend to be similar. A statistical problem results from not having a new degree of freedom with a new quadrat, but a fraction of it, because by being too close, the value of a second quadrat is not independent of the first. As a result, confidence intervals seem smaller than they are, and tests become too liberal. The following works on edges are likely to be subjected to the problem just described: Matlack (1994) placed replications 5 m apart, Cadenasso et al. (1997) 20m apart. Wales (1972) transects were either contiguous or 50m apart. Some transects in Turton and Freiburger (1997) were as close as 20m, some 40m, Turton and Duff (1992) used one single quadrat, 50m wide. ANOVAS comparing plots at different distances from the edge, along a transect, were performed in several cases (Rodrigues 1993; Wales 1972; Reichman, et al. 1993 and Viana et al. 1997). Such works are severely flawed, Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 32 for not meeting sampling independence, a basic assumption of ANOVA (Sokal and Rohlf 1995). In this case, two alternatives are available: to take random sample locations that requires additional work in the field, as done by Ferreira and Laurance (1997), or to use a nested ANOVA, in which each different distance from the edge consists of a repeated measure of the subject (transects, in this case). Repeated measure ANOVAS must meet the assumption of circularity, i.e., the variance of the difference between any two subsequent levels (distances form the edge) must have the same value. (von Ende 1993). Nicotra et al. (1998) suggest a creative way of avoiding spatial autocorrelation in transects: to measure how far apart two samples must be in order to be independent (by using semivariograms), and then perform a series of ANOVAS among samples that are sufficiently apart one from the other, so that they are not correlated. Transect length has also been a methodological problem in the study of edge effects. Long transects tend to overestimate the edge for lack of points along them, as occurred in Laurance (1991), or to fail to find a threshold, as mentioned by Paton (1994). Short transects, on the other hand, are not capable of encompassing the whole effect. Nevertheless, there is no way of avoiding small transects when fragments themselves are so small that the end of the transect is closer to another edge. Laurance et al. (1997) affirm that abiotic effects are likely to be contained within 100m, excluding wind disturbance which can extend for a longer distance. The authors present data suggesting that 100m is enough to include most of the interactive biological processes. Paton Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 33 (1994) suggests that most changes in avian nest predation occur within 200 m from the edge. Several papers discussed here came from transects shorter than 50m: Viana (1997); Matlack (1993); Cadenasso (1997); Mayaka (1995); Williams-Linera (1990a) and Palik and Murphy (1990), therefore, they are not likely to encompass the whole width of the edge effect. F) Raising Questions about Edge Effects A) Is the edge more diverse than the forest interior? I expect that edges have more species than forest interior, given that there is a higher chance of immigration in the edges. On the other hand, if edges have consistently less species than the forest interior, then it is likely that species are becoming extinct as a result of edge effects. B) How do species composition edge effects change over time? If edge-adapted species are reproducing more than other species, then younger plants should show increasing differences in species composition (the more time passes, the more the edge-adapted species will dominate the edge). However, if the pattern of change in species composition with distance from the edge is less clear in juveniles than in older plants, then the edge effects on species composition is getting weaker with time, indicating the occurrence of succession in the edge. C) Do edges of large fragments versus small fragments differ ? This question is associated with the relative importance of four factors: Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 34 1)The more intense edge effects for abiotic factors on small fragments (Malcolm 1994 and Kapos 1989). 2) The wider, higher contrasting edge effects for abiotic factors on large fragments (Malcolm 1994 and Kapos 1989). 3) The more intense immigration into a longer area of edge in large fragments (Yahner 1988). 4) The faster extinction of K-selected species in small fragments (Nee and May 1992). D) Do edges facing north and south differ in relation to edge width of biotic and abiotic effects? There is evidence that the differences in light exposure are sufficient to cause differences in edge width in biotic and abiotic effects in the temperate region. However, in the tropics, differences in light exposure between the north and south facing edges are smaller, and there are more species involved. In addition to that, there is no basic information on south-north contrasts on the tropics. E) Are non-monotonic patterns of plant density common in edges? I suspect that the non-monotonic patterns, present in so many instances in the literature (Monro 1992; Camargo and Kapos 1995; Turton and Freiburger 1997; Viana et al. 1997) are not just noise. There are a number of prospective mechanisms that may generate crest and hollow patterns. Assuming that the answer to E is yes, then: Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 35 E1) Is niche-partitioning or asymmetrical competition creating the wave-pattern competition? The two mechanisms create different situations in the edge. A fine division of environmental conditions appear as a result of niche partitioning and a progressive change of species composition towards the forest interior. The niche-partitioning mechanism also implies enhanced diversity along the edge, as a result of the close and contrasting species pools. Asymmetrical competition implies a regularly oscillating pattern of plantdensity. Different species are associated with suppressed and non-suppressed patches, so that dense patches have similar species composition. According to this mechanism, species composition should repeat itself at a certain wave period that matches the plant density wave period. The close location of contrasting species pools supposedly enhances diversity close to the edge. E2) What is the importance of light in creating non-monotonic patterns? How are edges affected by different orientation (as a surrogate of differential exposure to light) I expect that in accord with the trend in temperate areas, edges facing the equator will be deeper than those facing the poles. However, no information is available to speculate on how this could affect the crest and hollow pattern. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 36 Literature Cited Aizen, M.A. and Feinsinger,P. 1994 Forest fragmentation, pollination, and plant reproduction in a chaco dry forest, Argentina. Ecology 75(2): 330-351 Alexander, R.R. 1964 Minimizing windfall around clear cuttings in Spruce-fir forests. Forest Science 10:130-143 Alverson, W. S. Waller, D.M. and Solheim, S.L. 1988 Forests too deer: Edge effects in northern Wisconsin Conservation Biology 2(4):348-358. Alverson, W. S., Kuhlman,W., Walker, D.M. 1993 Wild Forests Conservation Biology and Public Policy. Island Press, Washington, D.C. 300 pp. Aubréville, A. 1938 La forêt coloniale: Les forêts de l’Afrique occidentale française. Ann.Acad.Sci.Colon. Paris 9. Bierregaard, R.O. and Stouffer,P.C. 1997 Understory birds and Dynamic Habitat Mosaics in Amazonian Rainforests in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Brokaw, N.V.L. 1985 Treefals, Regrowth, and Community Structure in Tropical Forests in: Pickett, S.T.A. and White, P.S. The Ecology of Natural Disturbance and Patch Dynamics Academic Press, London 472 pp. Brothers, T. S., and Spingarn, A. 1992 Forest fragmentation and alien plant invasion of central Indiana old growth forests. Conservation Biology 6(1) 91100. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 37 Brown, K.S. and Hutchings, R.W. 1997 Disturbance, Fragmentation and the Dynamics of Diversity in Amazonian Forest Butterflies in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Burkey, T.V. 1993 Edge effects in seed and egg predation at two neotropical rainforest sites Conservation Biology 66:139-143 Cadenasso,M.L., Traynor,M.M., and Pickett, S.T.A. 1997 Functional location of forest edges: gradients of multiple physical factors Canadian Journal of Forest Research 27: 774-782 Camargo, J.L.C. 1993 Variation in soil moisture and air vapor pressure deficit realative to tropical rain forest edges near Manaus, Brazil. M. Phil. Dissertation, University of Cambridge, Cambridge, UK 112 pp. Camargo, J.L.C. and Kapos, V. 1995 Complex edge effects on soil moisture and microclimate in Central Amazonian forest Journal of Tropical Ecology 11 205221 Canham, C. D. and Marks, P.L. 1985 The Response of Woody Plants to Disturbance in: Pickett, S.T.A. and White, P.S. The Ecology of Natural Disturbance and Patch Dynamics Academic Press 472 pp. Chen, J. Franklin,J.F., and Spies, T.A. 1992 Vegetation responses to edge environments in old-growth douglas-fir forests. Ecological Applications 2 (4) 387-396. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 38 Cifor 1997 Fires in Indonesia [Cifor Homepage] Available: http://www.cgiar.org/cifor/misc/events/fire.html Clements, F.E. 1905 Research Methods in Ecology. Jacob North Company 334 pp. Companhia Melhoramentos Norte do Paraná 1975 Colonização e Desenvolvimento no Norte do Paraná Editora São Paulo, São Paulo 137 pp. Denslow, J. 1980 Gap partitioning among tropical rainforest trees. Biotropica 12:47-55. Denslow, J. S. 1987 Tropical rainforest gaps and tree species diversity. Ann Rev. Ecol. Syst. 18: 431-452 Didham,R.K. 1997 The Influence of Edge Effects and Forest Fragmentation on Leaf Litter Invertebrates in Central Amazonia. in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Esseen, P.A. 1994 Tree mortality patterns after experimental fragmentation of an old-growth conifer forest. Biological Conservation 68: 19-28 FAO 1993 Summary of the final report of the forest resource assessment 1990 for the tropical world FAO, Rome 212 pp. Ferreira, L.V. and Laurance, W.F. 1997 Effects of forest fragmentation on mortality and damage of selected trees in Central Amazonia Conservation Biology 11(3): 797-801. Folha de Sao Paulo, 1998 Incêndio em Roraima persiste July, 20. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 39 Forman, R.T.T. 1995 Land mosaics: the ecology of landscapes and regions. New York, Cambridge University Press, 632 pp. Forman, R.T.T. and Godron, M. 1986 Landscape Ecology. New York, Wiley, 619 pp. Forman, R.T.T. and Moore, P.N. 1992 Theoretical Foundations for Understanding Boundaries in Landscape Mosaics in: Hansen, A.J. and Castri,F.di Landscape Boundaries: Consequences for Biotic Diversity and Ecological Flows Springer-Verlag, New York 452 pp. Franco, M. and Harper, J. L. 1988 Competition and Formation of Spatial Pattern in Spatial Gradients: An example using Kochia Scoparia. Journal of Ecology 76:959-974. Fraver, S. 1994 Vegetation responses along edge-to-Interior gradients in the Mixed hardwood Forests of the Roanoke River Basin, north Carolina Conservation biology 8(3):822-832 Gates, J.E. and Gysel, L.W. 1979 Avian nest dispersion and fledging success in field-forest ecotones. Ecology 59: 871-883 Gleason,H.A., Cronquist,A. 1964 The Natural Geography of Plants Columbia University Press New York 420 pp. Gratkowsky, H.J. 1956. Windthrow around staggered settings in old growth Douglas Fir. Forest Science 2:60-74. Griffith, D.A. 1987 Spatial Autocorrelation Resource Publications in Geography, Washington, D.C. 86 pp. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 40 Gysel, L.W. 1951 Borders and openings of beech-maple woodlands in southern Michigan. Journal of Forestry 49: 13-19 Hansen, A.J., Risser, P.G. Castri,F.di 1992 Epilogue: Biodiversity and Ecological flows across Ecotones in: Hansen, A.J. and Castri,F.di Landscape Boundaries: Consequences for Biotic Diversity and Ecological Flows Springer-Verlag, New York 452 pp. Hartshorn, G.S. 1978 Treefalls and tropical forest dynamics in: P.B. Tomlinson and M.H. Zimmermann Tropical Trees as Living Systems Cambridge University Press, New York 675 pp. Holl, K.D. and Lulow,M.E. 1997 Effects of species, habitat and distance from edge on post dispersal seed predation in a tropical rainforest Biotropica 29:459-468 Hubbell,S.P. 1979 Tree Dispersion, Abundance, and Diversity in a tropical dry forest. Science 203:1299-1309 Janzen, D. H.1983 No park is an island: increase in interference from outside as park size decreases Oikos 41: 402-410 Jennersten, O. 1988 Pollination in Dianthus deltoides (Caryophilaceae): Effects of habitat fragmentation on visitation and seed set. Conservation Biology 2: 359-366. Kapos, V. 1989 Effects of isolation on the water status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology 5:173-185. Kapos, V., Ganade,G., Matsui, E.,Victoria R.L. 1993 ∂13C as an indicator of edge effects in tropical rainforest reserves. Journal of Ecology 81:425-432. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 41 Kapos, V., Wandelli,E., Camargo, J.L. Ganade, G. 1997 Edge-related Changes in Environment and Plant responses due to Forest fragmentation in Central Amazonia in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Kaspari, M. 1993: Body size and microclimate use in neotropical granivorous ants. Oecologia 96, 500-507. Klucking, E.P. 1988 Leaf venation Patterns. Vol. III; Myrtaceae. J.Cramer, Stuttgart, west Germany. 278 pp. Laurance, W.F. 1991 Edge effects in tropical forest fragments: application of a model for the design of nature reserves. Biological Conservation 57:205-219. Laurance W. F 1997 Hyper-disturbed Parks: Edge effects and the ecology of Isolated rainforest reserves in Australia in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Laurance, W. F. and Yensen, E. 1990 Predicting the impacts of edge effects in fragmented habitats. Conservation Biology 55: 77-92 Laurance, W., Bierregaard, R.O., Gascon, C., Didham, R.K. Smith, A.P. Lynam,A.J. Viana, V.M. Lovejoy,T.E., Sieving,K.E., Sites, J.W. Andersen, M., Tocher, M.D. Kramer, E.A. Restrepo, C. Moritz, C. 1997 Tropical Forest Fragmentation: Synthesis of a Diverse and Dynamic Discipline in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 42 Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Legendre, P. 1993 Spatial Autocorrelation: Trouble or new Paradigm? Ecology 74: 1659-1673. Leopold, A. 1933 Game Management. Charles Scribner’s Sons New York. 287p. Lieberman,D. and Lieberman,M. 1987 Forest tree growth and dynamics at La Selva, Costa Rica (1969 - 1982). J.Trop.Ecol. 3(4), 347-358. Lovejoy, T.E., Bierregaard JR, R.O., Rylands, A.B., et al. 1986. Edge and other effects of isolation on Amazon forest fragments. in: Soule, M.E., Wilcox, B.A. (eds). Conservation Biology: An Evolutionary - Ecological Approach. Sinauer Associates Inc, Sunderland, Massachussets 584 pp. Lovejoy, T.E., Rankin, J.M., Bierregaard, R.O., Brown, K.S.Jr, Emmons, L.H. Van der Voort, M. E. 1984. Ecosystem decay of Amazon forest fragments. in: Nitecki, M.H. Extinctions. University of Chicago Press, Chicago 354 pp. Maack, R. 1968 Geografia Física do Estado do Paraná. Livraria José Olympio Editora Rio de Janeiro 356 pp. Malcolm, J.R. 1994 Edge Effects in Central Amazonian Forest Fragments. Ecology 75 (8):2438-2445. Matlack, G.R. 1993. Microenvironmental variation within and among forest edge sites in the eastern United States. Biological Conservation 66:185-194. Matlack, G.R. 1994 Vegetation dynamics of the forest edge – trends in space and successional time. Journal of Ecology 82 113-123. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 43 Mayaka,T.B. Fonweban,J.N. Tchanou,Z. Lontchui,P. 1995 An assessment of edge effect on growth and timber external quality of ayous (Triplochiton scleroxylon K Schum) under Cameroon rain forest conditions. Annales des Sciences Forestieres (Paris) 52(1): 81-88. Meroney, J. 1970 Wind Tunnel studies of the air flow and gaseous plume diffusion in the leading edge and downstream regions of a model forest. Atmospheric Environment 4 597-614 Metzger, J. P. 1997 Relationships between landscape structure and tree species diversity in tropical forests of South-East Brazil. Landscape and Urban Planning 367 (in press) Morato, E.F. 1994 Abundance and diversity of Euglossini males (Hymenoptera: Apidae) in a terra firme forest and cleared areas, in the vicinity of Manaus (Brazil). Boletim do Museu Paraense Emilio Goeldi Série Zoologia 10 (1): 95105 Monro A.K. 1992. The effect of forest/clearing interfaces on sapling and understorey community dynamics. Account of first year fieldwork, Cambridge University, England 53 pp. Murcia, C. 1993 Edge Effects on the Pollination of Tropical Cloud Forest Plants PhD thesis, University of Florida 102 pp. Murcia, C. 1995. Edge Effects in fragmented forests: implications for conservation. Trends in Ecology and Evolution 10: 58-62. Nee, S. and May,R. 1992 Dynamics of metapopulations: habitat destruction and competitive coexistence Journal of Animal Ecology 61:37-40 Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 44 Nepstad, D.C., Klink, C. A, Uhl, C.; Vieira I.C, Lefebvre, P.; Pedlowski M.; Matricardi E.; Negreiros G., Brown I.F., Amaral E.; Homma A., Walker R. 1997 Land-use in Amazonia and the Cerrado of Brazil. Ciência e Cultura 49(1-2): 73-86 Nicotra, A., Chazdon, R., Iriarte, S. 1998 Spatial Heterogeneity of light and wood regeneration in tropical wet forests. paper accepted to Ecological Monographs. Odum E.P.1971 Fundamentals of Ecology WB Saunders Company 639 pp. Orians, G.H. 1982 The influence of tree-falls in tropical forests on tree species richness. Tropical Ecology 23: 255-279 Osunkoya, O.O. 1994 Post-dispersal survivorship of north Queensland rainforest seed and fruits: effects of forest, habitat and species. Aust. J. Ecol. 19:52-64. Owen, D.F. 1971 Tropical butterflies. Clarendon Press, Oxford 452 pp. Palik,B.J., Murphy,P.G. 1990. Disturbance versus edge effects in sugarmaple/beech forest fragments. Forest Ecology & Management 32: 187-202. Paton, P.W. 1994 The effect of edge on avian nest success: How Strong is the Evidence? Conservation Biology 8(1):17-26. Patton, D.R. 1975 A diversity index for quantifying habitat edge. Wildl.Soc.Bull. 3:171-173 Powell, A. H. and Powell, G.V.N. 1987 Population Dynamics of male Euglossine bees in Amazonian forest fragments Biotropica 19(2) 176-179 Primack, R. 1995 Primer of Conservation Biology. Sinauer, Sunderland -MA 287 pp. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 45 Ranney, J.W. 1977 Forest island edges their structure, development, and importance to regional forest ecosystem dynamics. Environmental Sciences Division Publication Number 1069. Oak Ridge national Laboratory, Oak Ridge, Tennessee, USA 108 pp. Ranney,J.W., Bruner,M.C.& Levenson,J.B. 1981. The importance of edge in the structure and dynamics of forest islands. in: Burgess,R.L. and Sharpe,D.M. Forest island dynamics in man dominated landscapes 309 pp. Ricklefs, R.E. 1977 Environmental heterogeneity and plant species diversity: a hypothesis. American Naturalist 111:376-381. Robert, J., Cabral, O.M.R., Aguiar, L.F. 1990 Stomatal and boundary layer conductances in an Amazonian Terra Firme rain forest Journal of Ecology 27(1): 336-353 Rodrigues, E. 1993. Ecologia de fragmentos florestais ao longo de um gradiente de urbanização em Londrina-PR. Master's Dissertation. Universidade de São Paulo, São Carlos-SP Brasil 110 pp. Ruth,R.H., Yoder, R.A. 1953 Reducing wind damage in the forests of the Oregon coast range. USDA Forest Service Pacific northwest Forest and Range Experimental State Research Paper 7 35 pp. Seginer, I. 1975 Flow around a windbreak in oblique wind. Boundary Layer Meteorology 9 133-141 Shelford, V. 1913 Animal communities in temperate America as illustrated by the Chicago Region. University of Chicago Press, Chicago 326 pp. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 46 Skole, D., Tucker,C. 1993 Tropical deforestation and habitat fragmentation in the amazon: Satellite data from 1978 to 1988 Science 260:1905-10 Sokal, R.R. and Rohlf, F.J. 1995 Biometry: The principles and practice of statistics in biological research. W.H. Freeman New York 887 pp. Sork, V.L. 1983 Distribution of pignut hickory (Carya glabra) along a forest to edge transect, and factors affecting seedling recruitment Bulletin of the Torrey Botanical Club 110(4):494-506 Spears, E.E.,Jr 1987 Island and mainland pollination ecology of Centrosema virginianum and Opuntia stricta. Jouranl of Ecology 75:351-362 Sprugel, D.G. 1976 Dynamics of structure of wave-generated Abies balsamea forests in the north-eastern United States. Journal of Ecology 64 889-910. Tabanez, A.J. and Viana, V.M. 1997 Patch arrangement and dynamics in seasonal Atlantic forest fragments and implications for conservation. manuscript submitted to Biotropica Tabanez, A.J., Viana, V.M. Dias, A.S. 1997. Consequências da fragmentação e do efeito de borda sobre a estrutura, diversidade e sustentabilidade de um fragmento de floresta de planalto de Piracicaba, SP. Revista Brasileira de Biologia 57 (1): 47-60. Terborgh,J., Lopez, L. Tello,J. Yu, D.,Bruni,A.R. 1997 Transitory states in relaxing ecosystems of land-bridge islands. in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 47 Turton, S.M. and Duff, G.A. 1992 Light environments and floristic composition across an open forest-rainforest boundary in northeastern Queensland. Australian Journal of Ecology 17: 415-423. Turton, S.M. and Freiburger,H.J. 1997 Edge and Aspect Effects on the Microclimate of a Small Tropical Forest Remnant on the Atherton Tableland, northeastern Australia in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Viana, V.M., Tabanez, A.A.J., and Batista, J.L.F. 1997 Dynamics and Restoration of Forest Fragments in the Brazilian Atlantic Moist Forest. in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. von Ende, C.N. 1993 Repeated measures Analysis: growth and Other TimeDependent measures in: Scheiner, S.M. and Gurevitch, J. Design and Analysis of Ecological Experiments Chapman and Hall, New York 445 pp. Wales, B.A. 1967 Climate, microclimate and vegetation relationships on north and south forest boundaries in New Jersey. William L. Hutchesen Mem. Forest Bull. 2(3):1-60. Wales, B.A. 1972 Vegetation analysis of northern and southern edges in a mature oak-hickory forest. Ecological Monographs 42: 451-471 Whitney, G.G. and Runkle,J.R. 1981 Edge versus age effects in the development of a beech-maple forest Oikos 37:377-381 Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 48 Williams-Linera, G. 1990a Vegetation structure and environmental conditions of forest edges in Panama. Journal of Ecology 78:356-373. Williams-Linera, G. 1990b Origin and early development of forest edge vegetation in Panama. Biotropica 22:235-241. Williams-Linera, G. 1993 Vegetación de Bordes de un bosque nublado en el parque Ecológico Clavijero, Xalapa, Veracruz, México Revista de Biologia Tropical. 41:443-453. Willson, M.F., and Crome, F.H.J. 1989 Patterns of seed rain at the edge of a tropical Queensland rain forest. Journal of Tropical Ecology 5: 301-308. Yahner, R.H. 1988 Changes in wildlife communities near edges Conservation Biology 2(4):333-339. Young,T.P., Patridge,N. and Macrae, A. 1995 Long-term glades in Acacia bushland and their edge effects in Laikipia, Kenya Ecological Applications 5(1):97-108. Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 49 Table 1 – Boundary functions (Forman and Moore 1992) Structure Explanation Verticality Total height of vegetation Form Curvilinearity of the boundary Width Edge portion with characteristic environmental conditions and species composition Contrast Difference between variables on the two extremes of the ecotone* Function Explanation Habitat Distinctive environment occupied by distinct species Conduits Mass flow along boundaries Herbivores, and seeds herbivore fur in Filter Thorny “mantel” prevents crossing the edge Source Dispersal of “edgy” species into the forest and matrix Sink Animals gather, protection from weather seeking *Hansen et al. (1992) Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 50 Table 2 – Studies on abiotic factors on edges. Studies are arranged by age of the edge within each process. Process Edge width Edge Age Frag area Transect length Nonpseudo replication s Matlack 1993 Light 10m >63 yrs Not mentioned 50m 2 Cadenasso et al. 1997 PAR 3.02m (reanalysis) > 60 yrs > 40ha 25-35m 2 This work Light 5m 60 yrs 1 ha – 650 ha 50m –100m 55 Matlack 1993 Light 20m 1 yr Not mentioned 50m 6 Kapos 1989 PAR 20m 1 month 100 ha 100m 2 PAR 2m Not mentioned 8ha- 23 ha 50m 4 Turton et al. 1997 Temp 10m >70 yrs 20ha 70m 4 Cadenasso et al. 1997 Max temp. 2.5m (reanalysis) > 60 yrs > 40ha 55m 2 Williams-Linera 1990a Max temp. 7.52 ± 5.66m 0.8-12 yrs Not mentioned This work VPD 35.25m 6.63m Turton et al. 1997 VPD Cadenasso et al. 1997 Light Brothers 1992 and Spingarn Temperature 45m 5 VPD and Moisture ± 60 yrs 1 ha- 650 ha 50m 100m 55 10-30m >70 yrs 20ha 70m 4 Relative Humidity 4.0m (reanalysis) > 60 yrs > 40ha 55m 2 Cadenasso et al. 1997 Soil moisture 2.8m (reanalysis) > 60 yrs > 40ha 25-35m 2 Camargo 1993 VPD and soil moisture 15 m (reanalysis) 4.5 yrs 100 ha 200m 3 Matlack 1993 VPD Linear decay up to 50m 1 yr Not mentioned 50m 6 Kapos 1989 VPD 60m 1 month 100 ha 100m 6 Kapos 1989 VPD 20m 1 month 1ha 60m 14 Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 51 Table 3- Studies on biotic factors on edges Process Trend towards the interior Edge width Edge Age Transect length Non pseudo repet. Frag. Area (ha) Increase 30m >70 yrs 70m 4 20 Decrease 10 – 15m >70 yrs 30m >25 Decrease 35m 70 yrs 100m 48 Decrease 34.9 ± 4.59m 4-9 yrs 130m Decrease 60m 4yrs 200m Not avail . 108m >51 30m 1 year 10 years 200 m 200 m 3 5 Decrease 10 – 15m >70 yrs 30m 76 Increase Decrease 20m 70 yrs >50 yrs 200m 45m 9 5 Decrease Increase 120m 60 yrs 10-15yrs 70m 240m 12 16 Overstory thickness Overstory density Increase 60.0m ±10.5m 4-9 yrs 130m 4 Increase 60m 4 yrs 200m Williams-Linera 1990a Stem density More Decrease with age 60m 5 Tabanez 1997 Structure and comp. 10 months12 yrs 100 yrs 140m 4 Regeneration Turton et al. 1997 Ranney et al. 1981 This work Malcolm 1994 Camargo and Kapos 1995 Willson and Crome 1989 Monro 1992 Seedling number Sapling density Sapling density Understory Thickness Understory density Wind and Vertebrate dispersed seeds Saplings Vines Palms Decrease (mid dist. Peak) 0.4-650 Ha Cont. forest 100 100 Tree Density Ranney et al. 1981 Young et al. 1995 Palik and Murphy 1990 Rodrigues 1993 Chen et al. 1992 Malcolm 1994 Camargo 1995 and Kapos Canopy density Plant density Overstory density Stem density Stem density (2) 80 m-100 m (2) no effect 2.7, 4.7 1 Not avail. Cont. Forest 100 9.5 Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 52 Continuation of table 3 Tree growth Chen et al. 1992 Mayaka et al. 1995 Growth rate of Fir and Hemlock Dbh growth Ayous plantation Decrease 60m 10-15yrs 240m 16 Not avail. Decrease 10m 60 years from planting 35 m Indiv. level 100yrs 80 years 40 m ?2000m 3 32 plots 9.5 10-15yrs 270m 14 Not avail. Death and Canopy-Opening rates Viana et al. 1997 Laurance 1991 Chen et al. 1992 Ferreira and Laurance 1997 Laurance 1997 Kapos 1997 Tree Death Canopy damage Dead standing trees Fallen boles Myrtaceae death Biomass Decrease Decrease 150 m Decrease Decrease 100m 5 yrs Plots 56 Increase 100m 5 yrs 1000 m 39-27 1-cont. forest 100 Proportion of Decrease 40m 4.5 yrs 550m 3 gap occupation Matlack 1993 Canopy Decrease 1-yr 40m 14 Not opening avail. Ferreira and Laurance. (1997), Camargo and Kapos (1995), Monro (1992), Malcolm (1994), and Kapos (1997) refer to the same region Tabanez (1997) and Viana et al. (1997) refer to the same region Rodrigues (1993), and this work refer to the same region Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 53 Table 4 Species composition change on forest edges Plant Stage Edge age Depth Trn Length Point sampling Frg Size Gysel 1951 1m high-10cm DBH 10 cm DBH-30cm DBh >30cm DBH Not available Brothers and Spingarn 1992 Not informed Not Avail. "species composition was very similar" 8m 50m 8-23 ha 7+7 Wales 1972 > 30 cm high Williams-Linera 1990b > 8cm high, <2m high Not available 10 mths Not Available No effect Point sampling 20m 26.3 ha 3 Cont. forest 8 Williams-Linera 1990a > 5 cm DBH >2m high< 5 cm DBH <2m high 10mths12yrs No effect 60m Cont. forest 5 Rodrigues 1993 >10 cm DBH 60 yrs Decrease of pioneer sp. up to 45 m 45m 1 ha 12 Matlack 1994 <2m high 1.5 yrs>82yrs 40m 40m Not avail. 14 Fraver 1994 “saplings” second growth forest >55 yrs 30m 100 Not avail 23 12 to 60 acres Non pseudo replications 9 Chapter 1 Life in the Edge: Effects on the Margin of Forested Ecosystems 54 Chapter 2 The History of Forest Fragmentation in North Paraná, Brazil A) Introduction This chapter consists of an assessment of the human impacts on North Paraná along the last three centuries, and consists on a complement to the biotic and abiotic analysis on chapters 4 and 5. The continuous forest that once covered the North of Paraná State–Brazil, was part of the inner Atlantic Rainforest (Maack 1968). North of Paraná flora is the result of the contact between the savannas of the interior, with the coastal part of the Atlantic rainforest. Therefore, its species assemblage contains elements of both ecosystems. The original forest cover of North Paraná was reduced from around 84% in 1900 (SOS Mata Atlântica/INPE 1993), to 6.86 % in 1980 (Rodrigues 1993), ending the original vegetation continuity at a regional scale, and causing major environmental changes at a local scale, like widespread fires and soil erosion. Many other sites in Latin America were also subjected to extensive deforestation (Dean 1995; Sader and Joyce 1988; Dirzo and Garcia 1992). However, unlike most of them, land development companies planned deforestation in North Paraná. The areas occupied by the development companies can be considered an experiment at the landscape level. A large area was involved; in one of the cases, a single company occupied 1,236,040 ha. There are some historical Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 54 records prior to the occupation of the areas. There are also records regarding the methods employed by the land development companies. Therefore, the fragmentation process can be evaluated based on prior and subsequent conditions. I expect that the analysis of this “experiment” may contribute to understanding and perhaps avoiding some of the social and ecological mistakes in regions that are being currently deforested, like the Amazonian Forest. Transportation infrastructure has shown to be closely associated to deforestation both in North Paraná and in the Amazonian forest (Fearnside 1987). I have used the development of transportation infrastructure to divide the history of occupation of North Paraná into two periods: 1) The slow deforestation process related to the absence of efficient means of transportation, 2) The rapid deforestation process, after an efficient transportation network was installed. B) Ancient transportation phase Northern Paraná was considered for all its history, until the 1960', as an “empty” space, devoid of people and resources (Mota 1993). The only interest showed by both the Federal and State Governments, was to maintain sovereignty over the region. As pointed by Tomazi (1997), the emptiness of North Paraná is nothing but a fallacy, for many different native groups occupied the region. Historians have discussed how much native populations impacted the forest. The groups currently living in the Amazonian region provide evidence of Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 55 the impact of native dwellers on pristine forests. They manage small forest gaps by introducing light-demanding species and using controlled fire. Areas are abandoned for many years before they are utilized once more (Anderson 1983). Assuming the native groups in North Paraná used similar techniques, it is unlikely that the forest encountered by the first European expeditions was extensively managed. However, the extension of the transformation resulting from this management is totally dependent on how large ancient native populations were, and how they were distributed in the area. The demography of ancient populations is mostly based on the reports of the first Spaniards that crossed the State of Paraná. According to Meggers (1993), Spaniards frequently exaggerated their reports, in order to boost their achievements. The author presents many inconsistencies in reports of old travelers. According to Meggers (1993), populations were indeed small and incapable of causing major environmental transformation. Dom Alvaro Nuñez Cabeza de Vaca was the first white man to use the Peabiru track. Peabiru linked the Atlantic to the Pacific shore, crossing all of South America, 200 kilometers to the South of North Paraná. The native Carijós were the group that used the Peabiru track most frequently, before the arrival of the white colonizers. Starting in São Francisco do Sul, on 1541, Dom Alvaro Nuñez Cabeza de Vaca arrived in Asunción, Paraguay 5 months later. After Cabeza de Vaca, several other explorers crossed the South portion of South America, like Don Diego de Sanabria in 1550 and Ulrich Schmidel in 1552 (starting in Asunción, and arriving in São Vicente). Those were crossings without Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 56 major consequences, since they were not expected to, and they did not, stimulate settlements of any kind (Maack 1968). The Spanish Jesuit reductions were the first occurrence of “white” settlement in the area. Reductions were meant to facilitate the contact with Indians, and their christianizing. Reductions on North Paraná were most active during the first half of the 17th century, but they were soon exterminated by the Bandeirantes (Portuguese expeditioners) by the end of the 17th century. There is consistent evidence that European agriculture was practiced on the Jesuit reductions, implying that some area of forest was cut. However, the total area deforested at this time seems negligible. According to Blummers (1992), the records from one of the reductions (Loreto) indicate a population of 5,000 inhabitants and an account of victims of the Bandeirantes counts 30,000 (including Paraná and other regions). The three lines of argument presented; a) Native populations were very sparse and therefore incapable of changing large extensions of forest. b) Their forest management is likely to have had very low impact on the vegetation. c) The impact of the Jesuit reductions on the forest was punctual. point to the fact that the forest was mostly pristine till the start of the modern transportation phase. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 57 C) Modern transportation phase C1) The Old North: A transportation infra structure in its beginnings The second half of the 19th century in North Paraná was marked by several attempts of colonization that endured very slow development because of the lack of adequate transportation means. Around 1860, some coffee growers moved to the Cinzas region, on the left side of Itararé River, and to Jacarezinho (figure 1). These settlers changed to growing corn and raising pigs, because the higher aggregated price of the end product (pork) covered its transportation costs. The absence of adequate transportation in the old north raised the costs of coffee growing, preventing its expansion (Payes 1984). In 1855 a military fort was founded in Jataí, at the Tibagi River margins (figure 1). Despite the very fertile soils, the colony did not develop till 1865, as a result of malaria and attacks by Caiuá and Tereno native groups. The fort counted 211 people (including soldiers and their families) by then. Several villages were created in the old north, close to the end of the 19th century: Tomazina in 1865, Santo Antonio da Platina and Wenceslau Brás in 1866 and São José da Boa Vista, in 1867 (figure 1). Those were a result of the expansion of the traditional São Paulo coffee estates. The agricultural elite of São Paulo state was using its accumulated capital to buy land beyond the agricultural frontier, in the middle of a dense forest, three days away from the closest railroad station (Bernardes 1953). Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 58 Around 1900, a ferryboat began the route Ourinhos-Paraná, facilitating access to Cerqueira Cesar, the last station of the railroad linking the region to the city of São Paulo at the time (figure 1). In 1908 the railroad was extended to Ourinhos, reducing transportation costs in the west portion of North Paraná. The West portion of North Paraná, towards the Tibagi River, continued inaccessible in spite of the ferry boat service. C2) The construction of a railway opens North Paraná From 1922 untill 1925 an extension of the Sorocabana Railroad was built towards Cambará, entirely paid by coffee growers of the region. This work enabled the expansion of coffee growing to the Tibagi River, 50 km ahead of Cambará. During this period, coffee was transported by dirt roads, which were so poor that an eighty kilometer trip could take up to three days (Companhia Melhoramentos Norte Do Paraná 1975). In 1924, a British Commercial Mission visited North Paraná, in order to establish a Land Development Company. The "Companhia de Terras Norte do Paraná" (Paraná Plantations Ltd.) was founded in September, 1925 (Jofflily 1985). The Company was a joint venture of several investors, including the Rottschild & Sons Bank (Faraco 1988). It had a very large start up capital, enough to build expensive infrastructure, like whole cities, roads, and to continue the railroad towards the west. The area occupied by the Companhia de Terras Norte do Paraná was 1,236,040 ha (Cancián 1977) (figure 2). This area represents 6.6% of the Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 59 surface of the State of Paraná, but is located on the small portion of the State that is covered with extremely fertile oxisols (Faraco 1988). The infrastructure built by the Companhia de Terras Norte do Paraná was used by other development companies, so that the influence of the Company over the region was larger than the area actually occupied by them (Payes 1984). In 1929, the Companhia de Terras resumed the construction of the railroad from Cambará. Freight costs were reduced, and large extensions of forest were cut to plant coffee (Payes 1984). C3) The plan of the “Companhia de Terras Landscape transformation followed strict criteria established on the plan of the Companhia de Terras and of other development Companies that worked in the area. It was based on the flow of agricultural products, the linkage with the Santos harbor, and the foundation of nuclei over the whole rural area, which supported farming activities. The plan of the Companhia de Terras involved the establishment of main urban centers at regular distances, over the whole area. The cities of Londrina, Maringá, Cianorte, and Umuarama were located at 100 km intervals one from each other. Between the main urban centers, small villages were founded at regular distances of 15 km. Their function was to be used as a supply center for the rural zone. Some of these villages grew large, such as Apucarana, Cambé, Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 60 Rolândia, Arapongas, Astorga, Mandaguari, Nova Esperança and Jandaia. Roads linked cities and small villages. By 1943, 1132 km of roads had been built, by 1943, 4000 km, and by the end of the colonization, in the 1950’s, just the dirt roads totaled 5000 km (Companhia Melhoramentos Norte do Paraná 1975). The Companhia de Terras divided the rural zone into small lots, in such a way that all of them would include a piece of lowland and a ridge. This way, all lots would have running water and road access (figure 3). The settlers built their houses on the lowland, where they planted some fruit trees and a vegetable garden. On the highest portions of the landscape, less affected by frosts, they planted coffee, which was the principal source of income (Companhia Melhoramentos Norte do Paraná 1975). This design was not restricted to the area of the Companhia de Terras. Many other development Companies used it in North Paraná, both before and after the Companhia de Terras. New owners had to pay 30% of the land before extracting any wood from the lot. Yearly installments on the remaining 70% had to be paid starting after three years of occupation. D) Outcomes of the colonization projects The occupation of the ridges for cities and roads, and riparian ecosystems for housing and gardening, caused the deforestation of parts of the landscape that would later be declared as permanent protection areas (Brazilian Forest Code 1965). Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 61 The corridor shaped lot, downhill oriented, is still discernable today (figure 4), and it has influenced the shape of the forest fragments. Most of them have straight edges because they are circumscribed to narrow lots. Narrow lots had also stimulated downhill planting, since the reduced lot width prevents the building of contour lines. The schedule of payments of the Companhia de Terras forced the new land owners to deforest immediately after getting settled in order to obtain income to pay the yearly installments. Therefore, there is a close relation between how much area was sold in a year, and how much area was deforested. Figure 5 shows a peak of area sold around 1945. The accumulated area sold by the Companhia de Terras increased sharply between the years of 1942 and 1950 Cancián (1977), indicating a sharp decrease on the forested area during that period. Bernardes (1953) divides the process of occupation in the State of Paraná into two types: The fast and organized process occurred in the North, and the slow paced process on the rest of the State. The first type created a sharp edge between forest and occupied landscape (figure 6). The second one is less organized, in the sense that is not the result of a unique plan, but the summation of many historical factors, such as settlers and natives activities, and economic constraints. This more complex process, that occurred in most parts of both States of São Paulo and Paraná, leads to a more diverse landscape, where the edges of forest fragments are not evenly aged, and the time since land clearing may vary within short distances. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 62 The deforestation process was so fast, that the wood could not always be utilized. In many cases, the low prices, and the distance to a sawmill would stimulate farmers to burn, even large trunks of the valuable Aspidosperma polyneuron (peroba) (early settler, pers comm). According to Muller (1956), the forest was cut in many cases, not by the land owner, but by a contractor, on a six-year-long rent. The pressure to grow coffee and harvest it during this period was another factor for the deforestation to be quick and to cover the largest area. The use of land for a restricted period of time is a determinant of land degradation. No farmer will invest in long term conservation techniques, like contour lines or liming, if they will not stay on the land long enough to get the long term profits. An extensive fire occurred on the region in 1963. It was the result of both a very dry winter (June, July, August), and a series of frosts (Dean 1995) The fire destroyed a wide portion of the remaining primitive rainforest, mainly on the islands of pristine rainforest, isolated on the region of secondary forests (Maack 1968). Another outcome of the colonization projects was the moving and sometimes killing of native indian populations. North Paraná was considered empty land for a long time (Mota 1993). The Company claims that its enterprise transformed a "land of nobody" into a rich region (Companhia Melhoramentos Norte do Paraná 1975). Nevertheless, it was recently discovered that the company had a guard of its own, and even that some members are still alive. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 63 These guards maintain that the area had many Indian groups, that were removed by various means (Tomazzi, pers comm). In a period of less than 40 years, North Paraná was transformed into a large agricultural region, very powerful economically. Thirty-Four percent of the population of the State lived in North Paraná by 1960, in 172 cities (Companhia Melhoramentos Norte do Paraná 1975). The original transitional flora between two forest types was destroyed, as a result of deforestation and forest fragmentation. Reduction of the native cover The native forest coverage of the State of Paraná underwent a major reduction over the last 100 years (Figure 7). However, the forest cover is not uniform over the whole State. The hills close to the Atlantic Ocean shore , and on the middle of the State, contain most of the non-arable land, and have a larger cover of native forests, than North Paraná. In North Paraná, the native forest cover was 6.86 % in 1980, with a 95% error confidence interval from 5.35% to 9.53% (Rodrigues 1993). This value is an order of magnitude higher than surrounding areas with equally fertile soils, like Piracicaba in São Paulo State (Viana et al. 1997) Deforestation is still occurring in North Paraná. In an area of 6,095,000 ha around Londrina, 914.25 ha of forests were cut between 1970 and 1980 (Rodrigues et al. 1995). Fifty-five percent came from fragments smaller than 20 Ha. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 64 In order to cut any piece of forest, landowners need to obtain a permission from the State environmental agency (Instituto Ambiental do Paraná). However, law enforcement is difficult. The forest fragments are numerous and are scattered over a large area. The Government is not equipped to detect all small damages, like plowing the edges of forest fragments. The recent increase of the penalties for deforesting has finally caused landowners to be are afraid of being caught (pers. obs.). E) Conclusions The Companhia de Terras do Norte do Paraná, and other development Companies, played a major role in shaping the landscape of North Paraná. The straight edges of forest fragments and the similar age of the edges are results of the planned occupation. These conditions are probably not positive in terms of tree species populations. A more diversified landscape would probably host a larger number of species. However, the uniformity on edges age, and the straight edges, makes of North Paraná an ideal site for forest fragmentation studies. Besides the human aspects, large areas of North Paraná are free of major altitudinal and soil gradients, what provides uniformity at environmental level, as well. The local population believes that no forest was left after the “Company”. Measurements of native forest cover indicate the opposite. Not only the North Paraná still has a relatively large forest cover, but also deforestation is an Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 65 ongoing process on the region. It is important to publicize that it is still possible to reduce the remaining forest coverage if no regulatory measures are taken. Increasing environmental awareness is putting more pressure on landowners in South Brazil, and suggests an optimistic future for the environment in North Paraná, in spite of the current deforestation events. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 66 Literature Cited Anderson, A. 1983 Summa ethnobotanica Editora Vozes Petrópolis 427 pp. Bernardes, L.M.C. 1953 O Problema das frentes pioneiras no estado do Paraná Revista Brazileira de Geografia 3: 3-49 Blumers, A. 1992 La contabilidad en las reducciones guaranies. Imprenta Salesiana, Asunción 189 pp. Cancián, N.A. Cafeicultura Paranaense: 1900-1970 1977 Estudo das conjunturas Doctoral thesis Universidade federal do Paraná 497 pp. Castro, A.B. 7 ensaios sobre a economia Brasileira. 1980 Forense Universitária, Rio de Janeiro 521 pp. Código Florestal in: Secretaria de Estado do Desenvolvimento Urbano e Meio Ambiente Coletânea de Legislação Ambiental 1990 Governo do Paraná, Curitiba 536 pp. Companhia Melhoramentos Norte Do Paraná 1975 Colonização e Desenvolvimento no Norte do Paraná. São Paulo 212 pp. Dean, W. 1995 With Broadax and Firebrand California University Press 482 pp. Dirzo,R. and Garcia,M.C. 1992 Rates of deforestation in Los Tuxtlas, a Neotropical areas in Southeast Mexico. Conservation Biology 6(1) 84-90. Faraco, J.C. 1988 Adensamento Central e Dispersão Periférica: Levantamento e Sistematização de indicadores que permitam qualificar os Desequilíbrios Intraurbanos de Londrina. Master's thesis 237 pp. Fearnside, P.M 1986 Spatial Concentration of Deforestation in the Brazilian Amazon. Ambio 15(2): 74-81. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 67 Fearnside, P. M. 1987 Causes of Deforestation in the Brazilian Amazon. In: Dickinson, R.E. The Geophysiology of Amazonia Vegetation, and climate Interactions. John Wiley and Sons, Inc. New York, 526 pp. Fundação SOS Mata Atlântica/INPE - 1993 Atlas da Evolução dos remanescentes florestais e ecossistemas associados do domínio da Mata Atlântica no Estado do Paraná no período 1985-1990 23 pp. Joffily, J. 1985 Londres - Londrina. Paz e Terra, Rio de Janeiro 260 pp. Levi-Strauss, C. 1997 Tristes tropiques Modern Library, New York 517 pp. Maack, R. 1968 Geografia Física do Estado do Paraná. Livraria José Olympio Editora, Curitiba 450 pp. Meggers, B.J. 1993 Amazonia on the eve of european contact: ethnohistorical, ecological and anthropological perpectives. Revista de Arqueologia Americana 8: 92-115 Mota, L. T. 1993 A transformação dos territórios indíginas situados no Paraná em imensos vazios demográficos Cadernos de Metodologia e Técnica de Pesquisa 4: 1-54. Muller, N.L. 1956 Contribuição ao Estudo do Norte do Paraná. Boletim Paulista de Geografia 22:75-77 Payes, M.A.M. 1984 O Norte do Paraná: Expansão Cafeeira e Apropriação da Renda Fundiária desde fins do Século XIX até 1960 Master's thesis Universidade Federal do Paraná 173 pp. Perlin,J. A 1983 Forest Journey Harvard University Press Cambridge 445 pp. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 68 Rodrigues, E. 1993 Ecologia de Fragmentos Florestais no Gradiente de Urbanização do Norte Paraná. Masters Thesis Universidade de São Paulo, 110 pp. Rodrigues, E. Camara, C.D. Dias, A.T. 1995 Comparação entre os padrões espaciais de fragmentos florestais no gradiente urbano de Londrina. Semina: Ci.Agr. Londrina 16 (1) 34-39 Romariz, D.A. 1953 Mapa da vegetação original do Estado do Paraná. Revista Brazileira de Geografia 15(4):597-611 Sader, S.A and Joyce, A.T. 1988. Deforestation rates and trends in Costa Rica, 1940 to 1983 Biotropica 20:11-19 Silva, L.H.S. 1990 Fitossociologia Arbórea da Porção Norte do Parque Estadual Mata dos Godoy, Londrina, PR. Master's thesis Universidade Federal do Paraná 196 pp. Tomazi, A.D. 1997 “Norte do Paraná": História e Fantasmagorias Doctorate thesis Universidade Federal do Paraná 305 pp. Uhl, C., Barreto, P., Veríssimo, A., Vidal, E., Amaral, P., Barros, A.C. Souza Jr., C., Johns, J. and Gerwing, J. 1997 Natural Resource Management in the Brazilian Amazon. Bioscience 47(3): 160-168. Viana, V.M., Tabanez, A.A.J., and Batista, J.L.F. 1997 Dynamics and Restoration of Forest Fragments in the Brazilian Atlantic Moist Forest. in: Laurance,W.F. and Bierregard, R.O Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities Chicago University Press, Chicago 616 pp. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 69 Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 70 Brasil Cerqueira Cesar â Ourinhos Cambaráâ â Paraná State â Jacarezinho âJataí â Londrina Santo Antonio da Platina â Tomazina â Wenceslau Brás â â São José da Boa Vista São Paulo State TIBAGI RIVER ITARARÉ RIVER N 90 0 90 180 Kilometers Figure 1 Location of some of the cities founded during the Ancient transportation # phase Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 71 Brasil São Paulo Paranapanema River Paraná Maringá (1954) Londrina (1934) Apucarana (1943) Cambará (1925) Cascavel (1973) N 100 0 100 200 Kilometers # Figure 2 Map of the North Paraná, showing the area bought by the Companhia de Terras do Norte do Paraná (two polygons), the year of arrival of the railroad (in parenthesis), and the railroad (double line) Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 72 Source: (Bernardes, 1953) Lots Hard, straight lines Creeks Hard, crooked lines Ridges Dashed lines Figure 3 Lot planning on the Colônia Sertanópolis. The same type of planning was later used by the Companhia de Terras. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 73 1 Km Figure 4 Aerial photograph of a site south of Londrina, showing the straight edges of forest fragments, and the elongated farms. Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 74 800000 60 40 400000 20 0 1974 1970 1960 1950 1940 0 Area sold (ha) 80 1930 Average lot area (ha) 1200000 100 Years Source: Cancián, 1977 Figure 5 - Average lot area (bars) and accumulated area sold by Companhia de Terras Norte do Paraná (solid line). Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 75 Paraná State 1950 1940 1920 1900 N W 200 0 200 400 Kilometers E S Source: (Bernardes, 1953) Figure 6 Deforestation front on 1900, 1920, 1940 and 1950 in Paraná, Brazil Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 76 # Native Forest Coverage 100 80 60 40 20 0 1895 1930 1937 1950 1955 1960 1965 1980 1985 1990 Source: SOS Mata Atlântica/INPE (1993) Figure 7 Percentage of native forest coverage along the years in the State of Paraná Chapter 2 - The History of Forest Fragmentation in North Paraná, Brazil 77 Chapter 3 Materials and Methods A) Site location and description The state of Paraná is one of the states comprising the Southeast region of Brasil. The study site is located in the northern portion of the State of Paraná, at approximately 50oW lon. and 23oS lat. (figure 1) Rolling hills dominate the landscape of North Paraná. Hilltops are rounded, hillsides are long, and the valleys are open (Vizintim and Queiroz Neto 1992). The altitude varies between 650 and 350 m above the sea level. Slopes within the study area were never greater than 10% (figure 2). The North of Paraná region is known for the high fertility of its soil, originated from Basaltic rocks. This was one of the main causes of its fast deforestation process. The region is covered by a suite of three different eutrophic soils: Dark Red Eutrophic Latossols on the weathered plateaus, presenting no textural B horizons; Dark Red Podzols (Terra Roxa Estruturada) on the hillsides, and small pockets of Eutrophic Litossols on the most hilly sites. These three different soils result from different intensities of soil weathering. In plateaus, the intense downward flux of rainfall water carry clays down in the soil profile, so that the whole profile presents a relatively constant amount of clay. In hillsides, runoff is more intense, and the downward flux of water on the soil profile is weaker. Some of the clay from the higher soil horizons migrates to lower horizons, originating a textural gradient, in which higher Chapter 3 - Materials and Methods 76 horizons have relatively less clay and more sand than lower horizons. In the most hilly sites, the absence of rainfall water deeper into the soil originates a very young soil, in which pieces of the parent material still remain. The soils of North Paraná are generally very deep. In many occasions, excavations as deep as 15 or 20m, still find soils that resemble the one in the surface, including the smell of organic matter (pers. obs.) Dark Red Eutrophic Latossols and Dark Red Podzols differ on their susceptibility to erosion when they are plowed. Dark Red Eutrophic Latossols are located on flatter terrain, and water infiltrate easily in it. Little runoff is produced. Dark Red Podzols are located on the hillsides and have an A horizon composed by a relatively high fraction of sand. In addition to that, water infiltration is impaired by the textural gradient. The sand on the A horizon tend to be carried by the increased runoff on Dark Red Podzols. These two soil types represent extremes, and they are rarely found on the area. However, most of the field site is covered with soils that are intermediate to both types. According to Rodrigues (1993), among 18 samples taken from 0-40 cm depth, clay % varied between 74.57% and 43.7%. The Cation Exchange Capacity was between 10.93 and 20.76, and the Saturation of Basis (V%) was between 90.66 and 25.70. According to the Koeppen classification, the climate on the region is Cfa (humid subtropical) (Correa et al 1982). The annual average temperature is 22.5o C. The lowest monthly average occurs in July, and the maximum in December (figure 3). Chapter 3 - Materials and Methods 77 The average annual rainfall in the region is 1615 mm. The rainiest month is January, with more than 200 mm on average. The driest months are July and August, when total rainfall averages less than 60 mm. From November to June, rainfall exceeds evaporation, and between June and November evaporation exceeds rainfall (figure 3A and 3B). Between 1961 and 1994, four daily minimum temperatures were below 0o C: 7/75, 7/81, 7/81 and 6/94 (Meteorological Series provided by IAPAR). Intensity and direction of the winds in North Paraná were assessed with the meteorological series provided by IAPAR (figure 4). Wind in North Paraná is mostly calm. On more than eighty percent of the days, the maximum daily peak is below 10m/s. The predominant wind direction during the 1.2% most windy days was South-West. On calmer days, wind direction shifts towards east (figure 5). The continuous forest that once covered the North of Paraná State–Brazil, was part of the inner Atlantic Rainforest (Maack 1968). North of Paraná flora is the result of the contact between the savannas of the interior, with the coastal part of the Atlantic rainforest. Therefore, its species assemblage contains elements of both ecosystems. The local vegetation was classified by Hueck and Seibert (1972) as subtropical decidual and mesophitic forest of southeastern Brasil, with a high proportion of evergreen species. B) Data Collection Apart from the saplings data set that are used in most of the analyses presented in this thesis, two other data sets are used (tree data set and abiotic Chapter 3 - Materials and Methods 78 data set). The three of them were collected within the same area presented in figure 1, but at independent locations, and at different times. The three data sets were randomly and independently located on edges within a 60 x 60 km area, between 50.94oW lon 23.53oS lat and 51.58oW lon 23.03oS lat. Collecting the three data sets at different locations minimized the influence of random local effects, like gaps and minor soil differences. If, for instance, the occurrence of a gap reduced tree density at a site, this local effect did not affect light, because light was not measured at the same site. The independent location of the three data sets guarantees an independent random effect on each one of data sets, and minimizes the chance of obtaining a correlation because of local effects. Besides that, the three data sets were replicated (abiotic-55 transects; saplings-48 transects; trees-14 transects), and scattered throughout the entire study area (figure 1). B 1)Sapling Data set Sapling Data were collected in two field seasons: 6/15/96 - 8/30/96 and 1/4/97 - 3/15/97. The choice of plant size I chose to survey all free-standing plants between 1m high and 5 cm DBH (hereafter referred as saplings). Forest surveys generally include the largest trees. Therefore, they have minimum size thresholds, but not maximum. The upper limit of plant size was set to ensure that all individuals sampled had only established following fragmentation and that the plants were not the result of prefragmentation reproduction. The presence of an individual at a certain site may Chapter 3 - Materials and Methods 79 not have any connection with distance from the edge and fragment size, if that individual was established before fragmentation. Understory species grow very slowly (Lang and Knight 1983, Lieberman et al. 1985, Welden et al. 1991 and Terborgh et al. 1997). Assuming that most of the plants between 1m and 5 cm DBH are from understory species, it would be possible that this size class comprises older individuals than larger trees. In that case, the assumption according to which saplings are the result of reproduction after fragmentation would be flawed. However, Welden et al. (1991) classified only three species, out of the 108 species studied on the 50 ha plot on Barro Colorado Island, as understory specialists. The vast majority of species (79 out of 108) were found to be generalists, with high survival and slow growth under high and low canopy conditions, what supports the assumption that larger individuals are in average older. The lower limit of 1m was set to avoid seasonal fluctuations of sapling density, which happens as a result of the phenology of the different species, and to avoid plants without mature leaves, which are often difficult to identify. Data collection was accelerated by the restriction to saplings. Individual trees may take as long as one hour to sample (particularly those covered with vines). On average, trees take 4 times longer to be sampled than saplings. The increase in speed of data collection permitted a much larger data set to be gathered. Sample size is a crucial aspect for this work, given the large variance of both biological and physical aspects between small fragments. Chapter 3 - Materials and Methods 80 The choice of forest fragments Forests fragments were first selected on aerial photos at the recently established Landscape Ecology Laboratory of the Universidade Estadual de Londrina. The main criterion was shape (as close as possible to square), and size, in order to maintain the survey balanced in relation to the range of sizes considered. Fragments of different sizes were selected across the whole study area, in order to avoid fragments of similar size being too close to each other. This would cause spatial variation to be confused with variation due to fragment size. The locations of all transects are presented in figure 1. Figure 1 also includes the centroid of all the transects, and groups of transects that were analyzed in chapter 4. The second criterion used in choosing fragments was that they had not been subjected to any major recent effect, such as fire or intense selective logging. Fragment edges were sampled with transects (figure 6) Transects were chosen for being easy to lay out and to recover for resampling at later dates, and because contiguous quadrats along a transect reduce undesired environmental variation resulting from having quadrats separated from each other. The shortcoming of transects is the lack of independent sampling. Contiguous quadrats tend to be similar (spatial autocorrelation), which constitutes a basic problem when performing ANOVAS This problem can be overcome in a number of ways, already mentioned in chapter 1. Chapter 3 - Materials and Methods 81 Before laying the transect, all fragments smaller than 200 m wide were measured with a tape. One side was then picked at random, and the transect was located at its middle. The precise measurement of the small fragments was necessary to avoid having the interior end of a transect too close to another edge. Transects were typically 4m wide, 100 m long (figure 6). Forty-eight transects were laid in 19 fragments, ranging from 0.4 ha to 650 ha. Twenty-nine transects were 100m long and 19 were shorter. The shortest transect was 25 m long (table 1). All self-supporting plants above 1m height and below 5 cm DBH were recorded. Among the 16776 saplings recorded, 102 were left without identification, since they died before a sample could be collected. The entire sapling data set comprised 197 species or morphospecies. Most transects were laid by two people. One person would lead the rope, (frequently through a dense net of thorns and vines), following a compass bearing, and the second came behind, laying marking posts at every five meters, and ensuring the rope was as straight as possible. Plant marking and recording started once the transect was laid. All plants between 1 m high and 5 cm DBH were tagged with a circular aluminum tag 4cm wide attached with a 13 cm copper wire. Bigger saplings (close to 5 cm DBH) were nailed. The post marking and plant marking were performed to facilitate getting back to the plants in case of identification problems, field mistakes and to enable reassessing them on the future. Many saplings were difficult to identify in the field. These were collected, for later identification. Chapter 3 - Materials and Methods 82 Three months after the end of the data collection, all sites were revisited. I aimed to collect herbarium samples from plants that lacked leaves during the first collection, to check inconsistencies in identification, and to have an overall estimate of error in the data set, by checking the identification of a number of plants at random. I revisited 944 individuals, comprising both individuals with high chance of having been misidentified, and others, randomly taken on the field. Plant Identification Identifying the plants was a major task in the data collection. The vast majority of identifications were performed by comparing vegetative characters with reproductive specimens on the Herbarium of the Universidade Estadual de Londrina (FUEL). The strong emphasis of this Herbarium on local species facilitated the work. We became more proficient with species identification over time after developing an artificial classification system based on Gentry (1993). Species in the Myrtaceae were particularly difficult to identify. Two leaves of every individual of this family were cleared (see leaf clearing protocol on appendix 1), following previous taxonomic work by Cardoso (1996) and Klucking (1986). Leaf clearing is not recommended as an identification technique. The nervation pattern of a sample can be observed against a light bulb, and compared to the patterns in the literature, without clearing them. Nervation pattern is one character among many used to identify a species. In Myrtaceae, leaf clearing is not better, nor a substitute for characters like leaf color and size, leaf insertion angle and bud hairiness. Myrtaceae is a group with Chapter 3 - Materials and Methods 83 very few marked differences at the vegetative level, so all the characters must be pooled, in order to identify individuals, including a careful observation of individuals in the field, where characters like bark, leaf smell and taste can be better observed. Three months after the end of the data collection, all sites were revisited. I aimed to collect herbarium samples from plants that lacked leaves during the first collection, to check inconsistencies in identification, and to have an overall estimate of error in the data set, by checking the identification of a number of plants at random. I revisited 944 individuals, comprising both individuals with high chance of having been misidentified, and others, randomly taken on the field. Sapling identification took twice as much time as the data collection. It took 467 days to have the whole data set ready for analysis. 150 days in the field, and 317 identifying the plants and typing the data. B 2) Tree Data set Tree Data were collected between 9/1/1991 and 6/1/1992, and were already published elsewhere (Rodrigues 1993). Two more transects were added between 6/15/1996 and 8/30/1996. All trees above 10 cm DBH were recorded, in fourteen 100m long, 10m wide transects perpendicular to forest edges. B 3) Abiotic data set Abiotic data were collected during the dry season, from July, through August, 1993. Chapter 3 - Materials and Methods 84 Fifty-five transects were laid out in fourteen forest fragments. Those transects were all perpendicular to the edge, measuring either 50 or 100m in length, according to the size of the fragment. Twenty transects were 50m long and thirty five were 100m long. Dry and wet air temperatures were measured at five meter intervals, at 1 meter above the ground, for a total of 955 data points. Measurements of Photon Flux Density (PFD) were obtained by averaging three 400 -700 ηm sensors. The far-red component was measured with a 725-735 ηm sensor, and the red component by a 655-665 ηm sensor. All sensors were connected to a data logger. A period of two minutes was necessary to go from one point on the transect to the next, so that the interior extreme of a 50m and 100 m transect was measured 20 and 40 minutes, respectively, after starting on the edge. A second data logger was left outside of the forest, under full sun, in order to compensate for the passage of clouds during transect measurement. Vapor Pressure Deficit (VPD) was calculated by the Tetens formula (Jones 1992): VPD = 0.611 17.27(Tw-273) Td-36 Tw Wet bulb temperature Td Dry bulb temperature Chapter 3 - Materials and Methods 85 Appendix 1 Leaf clearing protocol 1. Immerse the leaves in Sodium Hydroxide till they become transparent green (1-2 days) (The epidermis tend to retain bubbles of oxidized pigments. Punching these bubbles allows reduction of the time in Sodium Hydroxide). Take the leaves out of the Sodium Hydroxide as soon as they became transparent. Reducing the exposure to Sodium hydroxide helps to maintain leaf overall structure. 2. Rinse the leaves with water 3. Immerse in Lactic Acid for around ten minutes when they should become transparent 4. Rinse them with water 5. Dehydration series: 5h -50% alcohol, 10h - 95 %alcohol 2h - Absolute Alcohol This phase is crucial to avoid bubbles on the mount. Longer periods are necessary if leaves are bulky. 6. Immerse the leaves for 15’ in Hemo-De, when they must be looking like glass. 7. Mount them on microscope slides, with toluen resin. Alternatively, leaves may be photographed, which will save the trouble of mounting the glass slides without bubbles. Chapter 3 - Materials and Methods 86 Literature Cited Cardoso, C.M.V. 1996 Uso do padrão de nervação como caráter taxonômico em Myrtaceae A.L. Juss. Monography Universidade Estadual de Londrina 73 pp. Correa, A.R. Godoy,H. Bernardes,L.R.M. 1982 Características climáticas de Londrina Circular IAPAR 5 Gentry, A. 1993 A field guide of woody plants of Northwest South America (Colombia, Ecuador, Peru) Conservation International, Washington 895 pp. Hueck, K. and . Seibert, P. 1972 Vegetationskarte von Sudamerika. Mapa de la Vegetacion de America del Sur. Erlauterungen zur Karte 1: 8 Millionen. G. Fischer, Stuttgart 69 pp. Jones, H.G. 1992 Plants and microclimate: a quantitative approach to environmental plant physiology. Cambridge University Press, Cambridge 428 pp. Klucking, E.P. 1988 Leaf venation Patterns. Vol. III; Myrtaceae. J.Cramer, Stuttgart, West Germany. 278 pp. Lang, G.E. and Knight, D.H. 1983 Tree growth, mortality, recruitment, and canopy gap formation during a 10-year period in a tropical moist forest Ecology 64(5) 1075-1080. Lieberman, D., Lieberman, M., Hartshorn,G. and Peralta,R. 1985 Growth rates and age-size relationships of tropical wet forest trees in Costa Rica. Journal of Tropical Biology 1 97-109. Chapter 3 - Materials and Methods 87 Rodrigues, E. 1993. Ecologia de fragmentos florestais ao longo de um gradiente de urbanização em Londrina-PR. Master's Dissertation. Universidade de São Paulo, São Carlos-SP Brasil 110 pp. Terborgh,J., Flores,C., Mueller,P. Davenport,L., 1997 Estimating the ages of successional stands of tropical trees from growth increments Journal of Troipcal Ecology 14:833-856. Vizintim, M. and Queiroz Neto, J.P. de 1992 Evolução do uso do solo na bacia do ribeirão cafezal- PR entre 1980 e 1987 Semina. Ci. Agr. 13 (1):24-31. Welden, C.W., Hewett,S.W.,Hubbell,S.P., Foster,R.B. 1991 Sapling survival, growth and recruitment: relationship to canopy height in a neotropical rainforest. Ecology 72(1):35-50 Chapter 3 - Materials and Methods 88 50 W Brazil N NN N 49 47 48 46 51 23 S 50 45 N 53 NN NN 25 ð 26 sm no al 22 23 21 N 19 N NN N 20 ð 27 16 17 15 18 ð NNNN NN so NNN N 3 NN 1 ð ð 30 9 Paraná 4 5 6 28 29 8 lg 7 12 11 10 13 14 N 43 44 42 NNN N 37 0 20 40 38 36 N 60 Kilometers + centroids X transects al centroid of all transects sm centroid of fragments smaller than 10 Ha no centroid of transects facing north so centroid of transects facing south lg centroid of fragments > 10 Ha Figure 1 - Localization of the study site and the 53 sapling transects surveyed. Centroids were calculated for all transects altogether, and for different groups of transects that are studied on chapter 5. The closeness of the centroids and their location around the center of the area, indicate that the location of the transects groups are not geographically biased. Abiotic and tree transects are located within the same area. Chapter 3 - Materials and Methods 89 Figure 2 General picture of the landscape, indicating the lack of major altitudinal gradients. Chapter 3 - Materials and Methods 90 220 Rainfall Evaporation 200 mm/month 180 160 140 120 100 80 60 40 26 Monthly average temperature o Temperature ( C) 24 22 20 18 16 14 Feb April June August October December source: Meteorological series provided by IAPAR Figure 3A Monthly averages of total rainfall and evaporation between 1985 and 1994. 3B Monthly averages of temperature between 1985 and 1994. Chapter 3 - Materials and Methods 91 20 10 max. daily peak (m/s) 30 0 100 80 60 40 20 0 % days Data source: Meteorological series provided by IAPAR Figure 4 Peak of wind speed on a daily basis in Londrina, North Paraná, in relation to % of the days accounted. Measurements referring to 4254 days between 1/1/86 and 8/31/1997 Chapter 3 - Materials and Methods 92 N NW W 0,4 0,3 0,2 0,1 0 1.2% highest (>20m/s, 50days) NE 4.6% highest (>15m/s, 193 days) 9.8% highest (>13m/s, 418 days) E 100% (>1m/s, 4254 days) 9.8% lowest (<5m/s, 374 days) SW SE S Data source: (Meteorological series provided by IAPAR) Figure 5 Prevailing wind direction. Data is grouped according to the daily peaks of wind speed. Measurements referring to 4254 days between 1/1/86 and 8/31/1997 Chapter 3 - Materials and Methods 93 FOREST FRAGMENT transect orientation TRANSECT 5m 2m Forest edge 0L 5L 10L 15L ooo ooo 95L 0R 5R 10R 15R ooo ooo 95R 5 m from the edge 15 m from the edge 95 m from the edge Figure 6 – Sketch of the plot design on the field Chapter 3 - Materials and Methods 94 Table 1 Description of the 19 fragments surveyed. Transects 3, and 32 through 36 were not used on this thesis. TRN Lon (W) Lat(S) Length Edge Plants/ Fisher’ Frg. (m) facing trn sAlpha Area Orient. (ha) Fragment name 1 51.2386 23.3815 100 289 402 14.21 13.43 Zelia 2 51.2366 23.3812 100 109 326 19.01 13.43 Zelia 3 51.2365 23.3809 100 Interior 400 20.05 13.43 Zelia 4 51.2300 23.3808 80 290 226 15.74 2 Mae morta Plants/ frg sp/ frg 1128 85 5 51.2274 23.3817 80 110 288 16.11 2 Mae morta 514 66 6 51.2402 23.3782 80 19 443 21.51 2.2 Eucalyptus 443 66 7 51.2409 23.4209 100 276 297 17.24 67.82 Pastejo 8 51.2336 23.4152 100 105 160 14.08 67.82 Pastejo 693 79 1917 99 415 30 1845 89 9 51.2349 23.4148 100 20 236 23.4 67.82 Pastejo 10 51.2490 23.4446 100 106 355 17.28 650 Godoy 11 51.2491 23.4443 100 110 564 23.66 650 Godoy 12 51.2583 23.4425 100 292 473 16.6 650 Godoy 13 51.2588 23.4451 100 292 489 15.97 650 Godoy 14 51.2493 23.4521 100 0 148 14.66 650 Godoy 15 50.9912 23.3029 100 346 607 18.89 100 Doralice 16 50.9894 23.3021 100 340 333 16.84 100 Doralice 17 50.9872 23.3021 100 78 533 13.56 100 Doralice 18 50.9677 23.3025 100 260 444 10.72 100 Doralice 19 51.0143 23.2870 25 39 214 6.93 0.35 Pedra bonita 20 51.0153 23.2871 25 150 201 5.54 0.35 Pedra bonita 21 51.4916 23.2699 100 160 356 15.39 25 Perdidao 22 51.4913 23.2680 100 116 587 14.14 25 Perdidao Perdidao 23 51.4968 23.2671 100 250 570 13.92 25 24 51.4939 23.2660 100 20 332 13.6 25 Perdidao 25 51.4949 23.2563 100 16 432 14.22 9 Perdidinho 26 51.4946 23.2594 100 190 178 11.38 9 Perdidinho 610 54 27 51.4618 23.3259 70 50 110 15.36 1.7 Broken elba 110 32 28 51.5062 23.3956 45 340 222 4.97 0.8 Futebol 29 51.5062 23.3967 45 242 195 5.97 0.8 Futebol 417 31 30 51.5089 23.4000 100 58 339 17.21 4 Vagabundo 31 51.5101 23.3999 100 238 250 20.71 4 Vagabundo 589 74 4774 137 2633 111 959 66 32 51.2534 23.4451 100 Interior 605 19.27 650 Godoy 33 51.2534 23.4451 100 Interior 744 22.79 650 Godoy 34 51.2510 23.4458 100 Interior 657 21.41 650 Godoy 35 51.2510 23.4458 100 Interior 739 19.37 650 Godoy 36 51.0069 23.5145 100 70 424 14.35 56 Paiquere-twins Paiquere-twins 37 51.0142 23.5127 100 330 420 14.79 56 38 51.0123 23.5099 100 50 379 16.8 56 Paiquere-twins 39 51.0141 23.5124 100 300 436 14.57 56 Paiquere-twins 40 51.0070 23.5198 100 190 513 18.49 56 Paiquere-twins 41 51.0091 23.5113 100 60 461 16.28 56 Paiquere-twins 42 51.1342 23.4615 70 180 306 9.79 2 Placa 43 51.1334 23.4596 70 50 375 10.19 2 Placa 44 51.1346 23.4601 70 300 278 11.04 2 Placa Chapter 3 - Materials and Methods 95 45 51.0596 23.0509 70 225 452 12.83 4 46 51.0589 23.0495 70 45 416 12.82 4 Tania Tania 47 51.0643 23.0479 50 60 238 15.33 10 Paulo 48 51.0669 23.0481 50 260 379 14.59 10 Paulo 49 51.0393 23.0361 40 318 271 14.89 0.7 Clube de meninos 50 51.0404 23.0370 40 115 238 10.04 0.7 Clube de meninos 51 51.0259 23.0236 80 290 298 20.37 3.5 Extremo norte 52 51.0252 23.0235 75 110 222 15.93 3.5 53 51.0131 23.1020 100 20 440 17.89 4 868 61 617 62 509 53 Extremo norte 520 64 Saco cheio 440 58 19 20004 197 Chapter 3 - Materials and Methods 96 Chapter 4 The effect of forest edges on different groups of species A) Introduction This chapter discusses some of the patterns presented on chapter 1. It tests whether biotic and abiotic characteristics of forest edges influence the occurrence of groups of species at different distances from the edge. This chapter also investigates differences between edges of large and small fragments. Some consensus has been reached in relation to several patterns on forest edges. These include the following: 1) Among the patterns described in the edge literature, perhaps the best established is the increase of sun radiance along forest edges. (Matlack 1993; Cadenasso et al. 1997; Matlack 1993; Kapos 1989 and Brothers and Spingarn 1992). 2) Another pattern frequently referred to in the literature is that species composition is frequently different at forest edges, when compared to the forest interior (Brothers and Spingarn 1992; Matlack 1994; Rodrigues 1993 and Fraver 1994). Two different sites always present some difference in species composition, regardless of how close to each other they may be. A circular reasoning commonly used with species composition is to classify species occurring close to the edge, as edge species. Chapter 4 Effect of forest edges on different types of species 97 3) Plant density increases in the edge. This pattern seems to hold for trees (Williams-Linera 1990a; Palik and Murphy 1990 and Rodrigues 1993), and for saplings (Ranney et al. 1981; Camargo and Kapos 1995; and Willson and Crome 1989). 4) Lack of seed predation in forest edges is a fourth pattern, repeatedly reported (Sork 1983; Osunkoya 1994 and Burkey 1993). 5) Small fragments are constantly referred as more affected by edge effects than large fragments, because of their greater perimeter relative to interior area (when having the same shape, small fragments have a larger portion of their area within short distances from the edge, than large fragments). This claim is based on the assumption that edges of small and large fragments are similar, an assumption for which there is little evidence. The interaction distance from the edge x fragment size was studied only once, by Kapos (1989). According to Kapos, average temperature in the first 60 m in small fragments, was higher than in large fragments. In spite of the importance of such studies for conservation, no biological data were ever collected to compare the edges of small and large fragments. With established consensus for some of the patterns of edge effects, it is time to go beyond pattern seeking, and investigate some of the mechanisms generating those patterns (Murcia, 1995). This chapter asks a number of questions that test the link between species and environment in several ways. It also tests the assumption that edges of small and large fragments are similar. More explicitly, I am asking: Chapter 4 Effect of forest edges on different types of species 98 1) Whether enhanced light is causing the increase of pioneers and canopy species on the edge. It is assumed that pioneers and canopy species, require more light to get established than climax and understorey species. Early successional stages in forests are characterized by large levels of light (pattern 1), particular species composition (pattern 2), and high plant densities (pattern 3). This has led authors like Wales (1972) and Gysel (1951) to conclude that edges consist of an "arrested" succession condition, and that edges are analogous to gaps. Such conclusions arise after a long chain of speculations, rather than evidence that light demanding species occur at the forest edge. This chapter attempts to provide this evidence. 2) Whether the absence of seed predation close to edge translates into reduced occurrence of animal dispersed species in relation to wind-dispersed species. 3) Whether the edge is a point of entrance of species on forest fragments, such that exotic species occur closer to the edge. 4) Whether species that occur close to the edge, also tend to occur in small fragments, making edge effects more pronounced in small fragments. B) Material and Methods Refer to Chapter 3 for site description and saplings data collection. B 1) Criteria for Species Classification The purpose of classification of species into relatively consistent groups is to simplify the information contained in species lists, to reveal general patterns, and to allow predictions about forest processes (Swaine and Whitmore 1988). Chapter 4 Effect of forest edges on different types of species 99 Species were classified into successional stages, commonness, species height, and whether a species was exotic or not. B 1 1) Pioneers and Climax species The classification of species according to their place in succession is associated to whether a species grows fast under full sun and occupies gaps soon after their creation, or whether the species is tolerant to low amounts of light, and occupy old growth sites (Budowski 1965). There are four papers classifying species according to their place in succession in South Brasil (Gandolfi 1991; Durigan & Leitão-Filho 1995; Lorenzi 1992; and Tabanez and Viana 1998). However, they used different number and types of categories. Gandolfi's (1991) categories were pioneers, early secondary and late secondary. Lorenzi (1992) classified species as pioneers or climax. Durigan & Leitão-Filho (1995) used three categories: Pioneers, heliophyte nonpioneers and shade tolerant non-pioneers. Tabanez and Viana (1998) used four categories: pioneers, gap-opportunists, shade-tolerant canopy species and shade tolerant understory species. Reconciling the data from the four papers is further complicated because they largely disagree about the classification of many species. The explanation for such disagreements is that there exists a wide range of responses (growth) to the light environment in tropical succession, and the differences between shade adapted species and light adapted species are not clear in tropical forests (Bazzaz and Pickett 1980). Chapter 4 Effect of forest edges on different types of species 100 I considered as pioneers those species that were classified as such in all papers in which they were mentioned (all papers included a pioneer category). Similarly, I considered as climax species, only those species that were included in the most shade adapted category in each paper (late secondary, climax, shade tolerant non-pioneers and shade tolerant canopy species), whenever they were mentioned. By adopting these criteria, I probably made both pioneer and climax groups smaller than they actually are. However, given the restricted knowledge about the species, and their wide range of responses to light, a less strict criterion would risk including misclassified species in the groups. Considering the small size of the groups, misclassified species could exert a strong influence on the groups average distance from the edge/fragment size. Eight species comprising 386 individuals were classified as pioneers, and eighteen species, comprising 1265 individuals, were classified as climax. B 1 2) Rare and Abundant species The 16,764 individuals in the entire data set were divided into two approximately equal halves containing the 11 most abundant species, (8263 individuals) and the 186 least abundant species (8501 individuals). Species with absolute abundance from one individual, up to 421 individuals were considered as rare species, whereas those containing from 434 individuals, up to the maximum abundance (1491 individuals) were considered to be abundant species. The abundance thresholds defining abundance and rarity did not originate from an absolute definition of the terms. The limits used here are only of relative Chapter 4 Effect of forest edges on different types of species 101 value, serving to compare the characteristics of two groups of individuals, in which one is composed by individuals from the most abundant species, and the other from individuals of the least abundant species. B 1 3) Understorey and Canopy Species The classification of species into the understory and canopy categories used two criteria: one based on the size structure of the species, and another based on secondary sources. All species that were represented by 50 or more individuals, in two or more fragments, were classified according to the height of the 90th percentile of the individuals. I considered as understorey species, all species in which ninety percent of the individuals were below 300 cm. Those species for which the height of the 90th percentile was above 440 cm, were considered as canopy species. The largest plant collected was 5cm DBH. Therefore, it could be that some of the species classified as “canopy” species do not grow up to the canopy. Similarly, it is possible that some of the species classified as understorey grow up to the canopy, in other regions. A second criterion was then added. Species were excluded from the understorey class if they were present in any other tree (above 5 cm DBH) surveys in South Brasil (22 publications were consulted, including one compilation of 192 tree surveys on the Atlantic rainforest, by Siqueira (1994)). All canopy species were widely present on the other surveys. The understorey group included nine species and 1638 individuals, and the canopy group included eight species and 890 individuals. Chapter 4 Effect of forest edges on different types of species 102 B 1 4) Wind-dispersed and animal-dispersed species Species were classified as wind-dispersed or animal-dispersed by examining seeds of fertile specimens at the Harvard University Herbaria. Those species with low-density seeds and large wing-like appendages were considered wind-dispersed. Species were included in the animal-dispersed class if they had fleshy endocarp and colorful fruit. Both criteria were evaluated on a qualitative basis. No measurement was taken. The wind-dispersed group included fifteen species and 1832 individuals, and the animal dispersed group included forty-five species and 1779 individuals. B 1 5) Exotic species This survey included four exotic species: Coffea arabica (coffee), Melia azedarach (an Asian tree, with a weedy behavior, and high resistance to fire), Citrus sp.(orange), and Eugenia pitanga (native in north Brasil). 106 individuals were included. B 2) Analysis The different groups listed above were evaluated in relation to the distance from the edge, and fragment size in which they occurred. Frequencies of individuals at different distances from the edge and in fragment sizes were recorded, and averages calculated. Abundance values were subtracted from the group average, so that trends could be visually examined, according to the following formula: Chapter 4 Effect of forest edges on different types of species 103 Relative frequency at = a given distance from the edge number of saplings _ of a given group__________ total number of saplings at a given distance from the edge average frequency of saplings of a given group, across all distances from the edge The subtraction of the average frequency is necessary to nullify the influence of the species classification on the relative frequency. Certain groups are frequent just because they are easier to classify, not because they are more frequently found on the field. Contrasts between averages of both distance from the edge and fragment size were assessed by a two sample “t” test on the untransformed data, assuming unequal variances (Sokal and Rohlf 1995). A Canonical Correspondence Analysis (CCA) was used to assess the overall effect of distance from the edge, fragment size, and edge direction (northing) on species composition. CCA is an ordination method based on weighted averaging. Species are ordered according to the weighted average of environment variables in the sites they occur (frequency of individuals X distance from the edge, fragment size, or northing, in our case). CCA finds, by means of iteration, the linear combination of environmental variables that optimizes the dispersion of species weighted averages (Jongman et al. 1995). The common way of presenting the results of a CCA is a biplot. In this diagram, arrows represent environmental variables, and points represent species. Species are plotted according to their weighted averages. The arrows point towards the direction of ordination of the species weighted averages. The closer a species is to the tip of an arrow, the higher is its weighted average in relation to that environmental variable. The length of the arrow indicates the Chapter 4 Effect of forest edges on different types of species 104 correlation of that environmental variable with the ordination axis, and therefore, the correlation between that environmental variable and the pattern of variation of species in the ordination diagram (Jongman et al. 1995). C) Results The survey comprised 16,764 individuals, and 192 species or morphospecies. The thirty-six morphospecies comprised 726 individuals (3.5% of the data set). 102 individuals (0.5% of the data set) were not identified, because they lacked leaves during the first collection, and they were dead when I visited the sites a second time, a few months later. These unidentified individuals were included on the averages of distance from the edge, and fragment size, but not on any of the analyses involving species identification. The overall average of individuals' distance from the edge was 42.49m, and the overall average fragment size in which individuals occurred was 105.83 ha (table 1). Among the 994 individuals reassessed in the field, 6% were misidentified. This represents the highest estimate of error in this data set, not only because those 944 individuals comprised a biased sample (those individuals whose identification I doubted), but also because those errors were fixed. Among the three environmental factors considered on the CCA presented in figure 2 (fragment size, distance from the edge and northing), fragment size was the single factor that best separates sapling species, i.e. different species tended to occur at different fragment sizes, as indicated by the longest arrow in figure 2. As mentioned before, the length of the arrows in a CCA diagram indicates how closely related that environmental variable is to the pattern of Chapter 4 Effect of forest edges on different types of species 105 variation in species composition. Distance from the edge was the second most powerful factor on separating species, followed by northing (edge orientation). The species sequence obtained when ranking species according to their average fragment size, was rather different from the ranking of species according to their average distance from the edge, as indicated by the approximately perpendicular angles between both environmental axes on the CCA. The angle between both environmental axis is 58 degrees, and their correlation is r=0.52. This indicates an intermediate situation between situation A: species ranking according to fragment size and distance from the edge are not correlated, in which case both axis are perpendicular; and situation B: species ranking according to fragment size and distance from the edge are positively correlated, in which case the arrow point to the same direction. This answers question 4, whether species that occur in small fragments also occur on edges, or not. The case of species occurring further from the edge, and on larger fragments, was more frequent than the opposite. However, this trend was weak, not only because the correlation between both axis is weak, but also because both CCA axes explained only 23.8% percent of the variance of species composition in samples. In the CCA, edge orientation towards North was negatively correlated with fragment size. Therefore, in general, species that were frequent in large fragments were rare at north edges. The eigenvalue of the edge orientation effect was much smaller than the other two, indicating that edge orientation had a Chapter 4 Effect of forest edges on different types of species 106 weaker effect on species composition than distance from the edge, or fragment size. The average distance from the edge for pioneer, canopy, rare, exotic and wind dispersed groups was significantly smaller than that that for their opposite groups (climax, understorey, abundant, non-exotic and animal-dispersed), as indicated by the P values in figure 3. The exotics group was heavily influenced by 86 individuals of Coffea arabica that represented 81% of the entire exotics group. Coffea arabica once covered most of the cultivated area in the region. Individuals from rare species showed a trend of decreasing abundance from the edge towards the interior (figure 3). The average fragment size of pioneer, understorey, exotic, and animaldispersed groups was significantly higher than that for their opposite groups: climax, canopy, and wind dispersed, as indicated by the P values on figure 4. D) Discussion Question 1) Is enhanced light causing the increase of pioneers and canopy species on the edge? Individuals associated with high light regimes (pioneer and canopy groups) occurred significantly closer to the edge than did those associated with low light regimes (climax and understory). This suggests that the light enhancement at the edge (described by Matlack (1993); Cadenasso et al. (1997); Matlack (1993); Kapos (1989) and Brothers and Spingarn (1992)) is strong enough to enhance the dominance of individuals associated with high light regimes. Chapter 4 Effect of forest edges on different types of species 107 Species composition change on forest edges was already described for the temperate region (Brothers and Spingarn 1992; Matlack 1994 and Fraver 1994 ). However, no link was shown between the edge environment and the environmental requirement of those species occurring close to the edge. Previous papers on the tropics have failed to establish differences in species composition between the edge and the forest interior (Williams-Linera 1990a; and Williams-Linera 1990b), probably due to the young age of the edges (less than twelve years). The older age of the landscape studied (60 years) suggest that differences in species composition take longer to get established. Furthermore, this study surveyed saplings, whereas Williams-Linera (1990a) and Williams-Linera (1990b) surveyed trees. Species composition differences are likely to develop sooner among younger plants than among trees. The preferential occurrence of saplings of light demanding species closer to the edge and on seems to agree with an often repeated (Gysel 1951; Whitney and Runkle 1981, Kapos 1989 and Wales 1972), although untested, idea that forest edges are similar to gaps. Furthermore, as in gaps, plant density is higher at the edge than in the forest interior (pattern 3), (Ranney et al. 1981; Camargo and Kapos 1995; and Willson and Crome 1989). Rare species, however, occurred significantly closer to the edge, which was unexpected, since gaps are normally dominated by few abundant pioneer species. The presence of rare species at the edge, coupled with high plant densities, suggests that edges have high species diversity, which is not characteristic of early successional stages, such as gaps. This is possibly a result of higher persistence of edge Chapter 4 Effect of forest edges on different types of species 108 environmental characteristics. Therefore, species may adapt to a series of environments that occur at different distances from the edge. The close location of contrasting species assemblages supposedly causes sapling diversity to be higher at the edge. Question 2) Does the absence of seed predation at the edge translate into reduced occurred of animal-dispersed species? The group of individuals pertaining to animal-dispersed species not only presented an average distance from the edge significantly higher than the wind dispersed individuals, but also presented the highest average distance from the edge among all groups. A positive feedback is possibly established between the lack of seed predation and animal-dispersed species. The less animal-dispersed plants at the edge, the less the edge is visited by dispersors. Less dispersors bring less seeds, and less animal-dispersed plants at the edge. The absence of animal-dispersed individuals closer to the edge is at odds with Willson and Crome (1989). The authors showed no effect of animal or wind dispersion on seed movement at different distances from the edge. Question 3) Is the edge an entrance of species to the fragment? The preferential occurrence of exotic plant species closer to the edge, has been repeatedly described for the temperate region (Fraver 1994; Ranney 1981; Ambrose & Bratton 1990; Brothers and Spingarn 1992), but not for the tropics. Individuals of exotic species occurred closer to the edge than non-exotic, but the small number of exotic individuals found on edges here may suggest just Chapter 4 Effect of forest edges on different types of species 109 the opposite: Exotic species, on the vast majority of cases, fail to invade forest fragments. Brothers and Spingarn (1992) came to the same conclusion studying temperate forest fragments in Indiana. Coffea arabica, the most abundant species among the exotic (81%), once covered most of the cultivated soil of the region. It has fleshy fruits that are possibly dispersed by birds. However, its reduced population on fragment edges (86 indiv.) indicates that it failed (when it was common in the region) to colonize the edges of forest fragments. The average size of fragment occupied by exotic species was 11.7 ha. It is possible that exotic species need large amounts of light to become established, and that this condition is only found on small fragments. The answer to question 3 is based on the generalization of immigration patterns of exotic species to all species. Therefore, it is necessary to exclude the possibility that these exotic individuals are occurring close to the edge because of other reasons, than having recently arrived to the fragment. That is not possible. The four exotic species overlay with rare species, and pioneers (even if they were not considered as pioneers on this study, because of the criteria used here). The answer to this question demands obtaining actual estimates of immigration, what is only possible with sequential assessments. Chapter 4 Effect of forest edges on different types of species 110 Question 4) Do species that occur close to the edge, also occur in small fragments? The Canonical Correspondence Analysis indicated that species composition of plots on large fragments resemble the species composition of plots that are distant from the edge. This trend, however, was weak. The separate observation of groups explains why distance from the edge and fragment size had a small correlation on the Canonical Correspondence Analysis. Canopy and wind-dispersed species occurred preferably closer to edges, and in large fragments. Individuals on edges of large fragments are supposedly part of a large population. The continuous arrival of seeds from mother plants on the forest interior, coupled with increased light at the edge, is likely to create a different condition than the edge of small fragments. The relative frequency of pioneer individuals was higher on fragments smaller than 1ha, and within 5m from the edge. On a 1 ha square fragment, 19% of its surface is within 5m from the edge. It is possible that this represents a size limit, below which pioneer species dominate the fragment. The relative frequency of saplings from rare species was higher at edges of both small and large fragments. A simple account of species richness on both conditions could lead to the conclusion that they are similar. However, pioneer and exotic individuals are more frequent on fragments larger than 10 ha. The small correlation between edge orientation (northing) and sapling species composition seems to be coupled with a latitude effect. While edges with different orientations show marked differences on their species composition in Chapter 4 Effect of forest edges on different types of species 111 the temperate region (Wales 1972; Fraver 1994, Matlack 1993 and 1994 and Palik and Murphy 1990), tropical edges show little or no differences (Viana 1997; Williams-Linnera 1993). Edges may differ more in the temperate region because of the lower angle of the sun. According to Kapos (1989), edges of small fragments are warmer than edges of large fragments, and regarding air temperature, large fragments have higher boundary contrast (sensu Hansen et al. 1992), and boundary width (sensu Forman and Moore 1992). If the results of Kapos (1989) can be generalized, small fragments have an intense edge effect over their whole area, and large fragments have a wider edge effect, in which environmental variables reestablish their original values deeper into the forest. Distance from the edge, and fragment size were positively correlated with their effect on species composition, which indicates that species composition of plots on large fragments resemble the species composition of plots that are distant from the edge. Therefore, the higher exposure to the open environment in small fragments (higher temperature, as pointed by Kapos 1989) has a more powerful effect on shaping species composition than the larger width of edges of large fragments. Also, edges of large fragments are contiguous with a large area of forest interior, which may prevent changes in species composition by acting as a seed source for the forest edge. When generalizing edge width estimates, one needs to keep in mind that at the same distance from the edge, a plot in a small fragment tends to have a species composition more characteristic of edges than a plot in a large fragment. Chapter 4 Effect of forest edges on different types of species 112 Even if edges and small fragments have a similar species composition in general, canopy and wind-dispersed species occurred preferably closer to edges, and in large fragments. E) Conclusions The relation established between increase in pioneer and canopy species, and light enhancement at the edge, represents a step forward in relation to the single description of species composition change on edges, described in previous papers. An ecological mechanism of species selection is now described. The indication of diversity enhancement at forest edges should not be taken out of context. Besides being the preferred site for rare species, edges are also the preferred site for exotic and pioneer species. The diversity enhancement on edges, may be only an intermediate phase in this 60-year-old landscape, before the edge is totally taken over by few “weedy” species. Chapter 4 Effect of forest edges on different types of species 113 F) Literature Cited Ambrose, J.P. and Bratton, S.P. 1990 Trends in landscape heterogeneity along the borders of Great Smokey Mountains National Park. Conservation Biology 4 135-143. Bazzaz,F.A. and Pickett,S.T.A. 1980 Physiological ecology of tropical succession: A comparative review. Annual Review of Ecology and Systematics 11:287-310. Brokaw, N.V.L. 1985 Treefals, Regrowth, and Community Structure in Tropical Forests in: Pickett, S.T.A. and White, P.S. The Ecology of Natural Disturbance and Patch Dynamics Academic Press, New York 472 pp. Brothers, T. S., and Spingarn, A. 1992 Forest fragmentation and alien plant invasion of central Indiana old growth forests. Conservation Biology 6(1) 91100. Budowski, G. 1965 Distribution of tropical American rainforest species in the light of successional processes Turrialba 15 (1) 40-42. Burkey, T.V. 1993 Edge effects in seed and egg predation at two neotropical rainforest sites Conservation Biology 66:139-143 Cadenasso,M.L., Traynor,M.M., and Pickett, S.T.A. 1997 Functional location of forest edges: gradients of multiple physical factors Canadian Journal of Forest Research 27: 774-782. Durigan, G. & Leitão-Filho, H.F. 1995. Florística e fitossociologia de matas ciliares do oeste paulista. Rev.Inst.Flor., São Paulo, 7(1):197-239 Chapter 4 Effect of forest edges on different types of species 114 Forman, R.T.T. and Moore, P.N. 1992 Theoretical Foundations for Understanding Boundaries in Landscape Mosaics in: Hansen, A.J. and Castri,F.di Landscape Boundaries: Consequences for Biotic Diversity and Ecological Flows Springer-Verlag, New York 452 pp. Fraver, S. 1994 Vegetation responses along edge-to-Interior gradients in the Mixed hardwood Forests of the Roanoke River Basin, North Carolina Conservation biology 8(3):822-832 Gandolfi, S. 1991. Estudo florístico e fitossociologia de uma mata residual na área do Aeroporto Internacional de São Paulo, município de Guarulhos - SP. Tese de mestrado, Instituto de Biologia, Universidade Estadual de Campinas, Campinas, SP 232 pp. Gysel, L.W. 1951 Borders and openings of Beech-Maple woodlands in Southern Michigan. Journal of Forestry 49: 13-19. Hansen, A.J., Risser, P.G. Castri,F.di 1992 Epilogue: Biodiversity and Ecological flows across Ecotones in: Hansen, A.J. and Castri,F.di Landscape Boundaries: Consequences for Biotic Diversity and Ecological Flows Springer-Verlag, New York 452 pp. Jongman, R.H.G. , Ter Braak,C.J.F. and Van Tongeren, O.F.R. 1995 Data Analysis in Community and Landscape Ecology. Cambridge University Press, Cambridge 297 pp. Kapos, V. 1989 Effects of isolation on the water status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology 5:173-185. Chapter 4 Effect of forest edges on different types of species 115 Lorenzi,H. 1992 Árvores Brasileiras. Editora Plantarum Ltda, Piracicaba-Brasil 387 pp. Martins, A., Semir, J. Goldemberg, R. and Martins, E. 1996 O Gênero Miconia Ruiz & Pav. (Melastomataceae) no estado de São Paulo [Homepage of the Base de Dados Tropical],[Online]. Available: http://www.bdt.org.br/bdt/miconia/ Matlack, G.R. 1993 Microenvironmental variation within and among forest edge sites in the eastern United States. Biological Conservation 66:185-194. Matlack, G.R. 1994 Vegetation dynamics of the forest edge – trends in space and successional time. Journal of Ecology 82 113-123. Mayaka,T.B. Fonweban,J.N. Tchanou,Z. Lontchui,P. 1995 An assessment of edge effect on growth and timber external quality of ayous (Triplochiton scleroxylon K Schum) under Cameroon rain forest conditions. Annales des Sciences Forestieres (Paris) 52(1): 81-88. Murcia, C. 1995. Edge Effects in fragmented forests: implications for conservation. Trends in Ecology and Evolution 10: 58-62. Osunkoya, O.O. 1994 Postdispersal survivorship of North Queensland rainforest seed and fruits: effects of forest, habitat and species. Aust. J. Ecol. 19:52-64. Palik,B.J., Murphy,P.G. 1990. Disturbance versus edge effects in sugarmaple/beech forest fragments. Forest Ecology & Management 32: 187-202. Ranney,J.W., Bruner,M.C.& Levenson,J.B. 1981. The importance of edge in the structure and dynamics of forest islands. in: Burgess,R.L. and Sharpe,D.M. Chapter 4 Effect of forest edges on different types of species 116 Forest island dynamics in man dominated landscapes Springer Verlag, New York 309 pp. Rodrigues, E. 1993. Ecologia de fragmentos florestais ao longo de um gradiente de urbanização em Londrina-PR. Master's Dissertation. Universidade de São Paulo, São Carlos-SP Brasil 110 pp. Siqueira, M.F. 1994 Análise Florística e Ordenação de Espécies arbóreas da Mata Atlântica através de dados binários. Master thesis, UNICAMP. 100pp Sokal, R.R. and Rohlf, F.J. 1995 Biometry: The principles and practice of statistics in biological research. W.H. Freeman New York 887 pp. Sork, V.L. 1983 Distribution of pignut hickory (Carya glabra) along a forest to edge transect, and factors affecting seedling recruitment Bulletin of the Torrey Botanical Club 110(4):494-506 Swaine, M.D. and Whitmore, T.C. 1988 On the definition of ecological species groups in tropical rain forests Vegetatio 75 81-86. Tabanez, A. A. J. and Viana, V. M. 1998 Patch arrangement and dynamics in seasonal Atlantic forest fragments and implications for conservation. Paper submitted to Biotropica Williams-Linera, G. 1990a. Vegetation structure and environmental conditions of forest edges in Panama. Journal of Ecology 78:356-373. Williams-Linera, G. 1990b. Origin and early development of forest edge vegetation in Panama. Biotropica 22:235-241. Whitney, G.G. and Runkle,J.R. 1981 Edge versus age effects in the development of a beech-maple forest Oikos 37:377-381 Chapter 4 Effect of forest edges on different types of species 117 Willson, M.F., and Crome, F.H.J. 1989. Patterns of seed rain at the edge of a tropical Queensland rain forest. Journal of Tropical Ecology 5: 301-308. Chapter 4 Effect of forest edges on different types of species 118 Table 1 Species list, including abundances, averages of distance from the edge, and fragment size, and five grouping criteria ABUN DIST SIZE HEIG udst DISP SUCC EXOT RARE Acanthaceae Geissomeria pubescens Nees 63 31.3 80.2 Jacobinia carnea (Lindl.) Nicholson 10 61.5 63.2 rare 661 44.1 136.6 abun 197 32.0 173.1 Justicia brasiliana Roth. pion rare Anacardiaceae Astronium graveolens Jacq cano wind rare Annonaceae Annona cacans Warm. 1 55.0 2.3 anim rare Rollinia sylvatica (St.Hill.) Mart. 22 43.6 139.3 anim rare Aspidosperma polyneuron Muell. Arg 915 49.1 327.8 wind Peschiera australis (Muell. Arg.) Miers 51 7.2 28.6 113 70.6 54.3 2 57.5 2.0 19 50.5 70.9 wind rare 1 15.0 0.7 wind rare Apocynaceae clim abun rare Arecaceae Euterpe edulis Mart. Syagrus romanzoffiana (Cham.) Glassm. anim rare rare Bignoniaceae Jacaranda puberula Cham. Bombacaceae Chorisia speciosa St Hill. Boraginaceae Cordia ecalyculata Vell. 34 49.0 199.2 rare Cordia superba Cham. 1 90.0 650.0 rare Cordia trichotoma (Vell.) Arrab. ex Steud. 2 60.0 35.0 rare 5 36.0 246.6 anim 3 38.3 51.8 anim 18 20.8 41.8 5 53.0 520.1 Acalipha gracilis A.Gray 1110 46.5 95.7 Actinostemon concolor (Spr.) Muell. Arg. 1491 46.4 163.8 abun Alchornea glandulosa Poepp. 16 45.0 238.0 rare Alchornea triplinervia (Spreng.) Muell. Arg. 5 63.0 2.3 112 27.2 339.5 47 45.9 174.5 rare Sebastiania commersoniana (Baill.) Smith & Downs 434 49.3 55.9 abun Indeterminate Sebastiania 32 155 59.9 94.3 rare rare Caricaceae Jaracatia spinosa (Aubl.) DC. rare Cecropiaceae Cecropia glazioui Sneath. pion rare clim rare Celastraceae Maytenus aquifolium Mart. Elaeocarpaceae Sloanea monosperma Vell. rare Euphorbiaceae Croton floribundus Spr. Sapium glandulatum (Vel.) Pax udst abun rare pion rare Flacourtiaceae Banara parviflora (A.Gray) Bentham 47 20.1 37.4 Casearia decandra Jacq. 8 31.3 5.7 Casearia gossypiosperma Briq. 27 24.3 259.6 Casearia obliqua Spr. 2 42.5 353.0 rare Casearia sylvestris Sw. 70 33.9 50.5 rare Prockia crucis P. Bromme ex Linn. 24 25.2 156.9 rare rare wind rare Chapter 4 Effect of forest edges on different types of species 119 Xylosma ciliatifolium (Clos) Eichl. 7 40.7 112.1 rare 8 55.0 155.2 rare Icacinaceae Citronella megaphylla (Miers) Howard Lauraceae Endlicheria paniculata (Spr.) Macbr. 41 42.0 170.5 300 48.5 108.2 rare Ocotea elegans Mez 2 15.0 100.0 rare Ocotea indecora Schott 56 46.1 198.2 rare Ocotea puberula (Rich.) Nees 5 6.0 131.5 rare Nectandra megapotamica (Spr.) Mez. clim rare Ocotea silvestris Vatt 44 36.4 158.0 rare indeterminate Lauraceae 19 4 23.8 155.2 rare indeterminate Ocotea 24 1 75.0 650.0 rare indeterminate Ocotea 25 3 33.3 204.5 rare indeterminate Ocotea 26 1 15.0 56.0 rare 6 40.0 127.3 Acacia lacerans Benth. 96 33.0 76.8 Acacia polyphilla DC. 78 14.0 41.1 Apuleia leiocarpa (Vog.) Macbr. 3 21.7 240.6 Bauhinia forficata Link 21 19.5 21.1 pion rare Calliandra foliolosa Benth 34 38.2 22.2 clim rare Enterolobium contortisiliquum Morong 6 49.2 27.8 Holocalix balansae Mich. 421 41.3 96.5 Inga marginata Willd. 470 54.5 276.8 anim abun Inga sessilis (Vell.) Mart. 21 58.8 401.0 anim rare Inga striata Benth 23 81.1 589.5 anim Lonchocarpus campestris Mart ex Bentham 120 27.5 37.6 cano Lonchocarpus muehlbergianus Hassl. 215 30.5 73.9 cano 2 27.5 56.0 28 39.8 133.7 rare 111 26.6 71.7 rare Lecythidaceae Cariniana estrellensis (Raddi) O. Kuntze wind rare Leguminosae Machaerium aculeatum Raddi Machaerium hatschbachii Rudd. Machaerium minutiflorum Tul. udst rare rare rare rare rare rare rare rare wind rare Machaerium nictitans (Vel.) Benth. 9 26.1 225.3 wind rare Machaerium paraguariense Hassl. 47 29.0 168.2 wind rare Machaerium stiptatum Vog. 4 12.5 11.5 rare Myroxilum peruiferum L. 27 44.6 24.0 rare Parapiptadenia rigida (Benth.) Bren. 235 15.2 9.4 Peltophorum dubium (Spreng.) Taub. 1 65.0 3.5 132 21.2 5.9 rare indeterminate Albizia 2 1 30.0 3.5 rare indeterminate Leguminosae 20 5 48.0 100.0 rare indeterminate Leguminosae 21 15 31.3 100.0 rare indeterminate Leguminosae 22 5 44.0 3.6 rare indeterminate Leguminosae 23 11 40.0 132.3 rare 15 33.3 41.3 rare 8 42.5 67.4 Leandra scabra DC. 5 77.0 586.3 rare Miconia discolor DC 162 47.9 524.5 rare 5 55.0 319.3 rare Piptadenia gonoacantha (Mart.) Macbr. rare wind rare Loganiaceae Strychnos brasiliensis (Spr.) mart. Malpighiaceae Bunchosia pallescens Skottsberg clim rare Melastomataceae Miconia hymenonervia (Raddi) Cogn Chapter 4 Effect of forest edges on different types of species 120 Miconia minutiflora (Bonpl.) DC 4 25.0 613.2 rare Miconia tristis Spring 6 60.8 239.9 rare Cabralea canjerana (Vell.) Mart. 63 53.3 264.5 Cedrela fissilis Vell. 11 33.6 36.9 Guarea kuntiana A. Juss 219 50.3 312.1 Guarea macrophilla Vahl. 93 43.4 113.8 Melia azedarach Blanco 13 1.9 12.3 Trichilia casaretti C. DC. 233 50.0 334.8 Trichilia catigua A. Juss. 316 44.0 227.1 rare Trichilia clausenii C. DC. 465 49.4 243.2 abun Trichilia elegans A.Juss. 715 35.8 112.4 abun 71 48.5 223.9 rare 176 47.3 274.7 rare Mollinedia clavigera Tul. 43 54.7 506.0 rare Mollinedia widgrenii A.DC. 2 82.5 100.0 rare Meliaceae Trichilia pallens C.DC. Trichilia pallida Sw. rare wind clim rare rare rare exotic rare rare Monimiaceae Moraceae Ficus guaranitica Chodat & Visher 4 41.3 32.1 Sorocea bonplandii (Baillon) Burg. 564 46.5 160.8 anim rare abun Myrsinaceae Rapanea lanceolata Mez. 13 30.0 58.3 rare Rapanea umbellata (Mart. ex. A. DC. Mez 59 36.9 110.4 rare 11 39.1 81.0 anim Calypthranthes concinna DC. 1 35.0 25.0 anim clim rare Campomanesia guazumifolia (Camb.) Berg. 31 50.2 219.0 anim clim rare Campomanesia xanthocarpa Berg 51 41.0 98.5 anim clim rare Eugenia blastantha (Berg) Legr. 15 52.7 59.6 anim clim Eugenia burkartiana Berg 19 36.3 13.8 anim Eugenia florida DC. 12 34.6 122.7 anim clim Eugenia hiemalis Camb. 10 46.0 18.4 anim clim Eugenia involucrata DC. 37 56.4 52.0 anim Eugenia moraviana Berg 64 47.7 63.9 anim clim Eugenia pitanga (Berg.) Kiaerskou 3 41.7 7.3 anim clim 166 46.6 246.4 anim Myrtaceae Calycoretes psidiflorus (Berg) Sobral Eugenia ramboi Legr. Eugenia uniflora Linn. Eugenia verrucosa Legr. 23 24.3 59.7 111 42.6 256.7 cano rare rare rare rare rare rare rare exotic rare rare anim rare anim rare rare Hexachlamys itatiaiae Mattos 65 33.0 63.8 anim Myrcia grandiflora (Berg) Legr. 3 68.3 650.0 anim Myrcia laruotteana Cambess. 19 12.6 4.3 anim Myrcia rostrata DC. 10 57.0 247.9 anim rare Neomitranthes glomerata (Legr.) Legr. 59 39.0 607.3 anim rare Plinia trunciflora Berg 1 40.0 559.1 anim rare indeterminate Myrciaria 22 12 53.8 488.7 anim rare Boungainvillea spectabilis Willd. 19 28.4 48.9 wind rare Guapira opposita (Vell.) Reitz 53 39.8 313.5 Pisonia aculeata Linn. 10 57.0 188.9 68 22.4 75.7 cano rare clim rare Nyctaginaceae cano rare rare Phytolaccaceae Gallesia integrifolia (Spreng.) Harms. wind rare Chapter 4 Effect of forest edges on different types of species 121 Seguieria aculeata Jacq. 81 40.1 176.1 rare Ottonia martiana Miq. 7 27.9 293.4 rare Piper aduncum Linn. 177 56.0 211.5 rare Piper amalago (Jacq.) Yunck. 411 28.4 28.3 rare Piperaceae Piper arboreum Aubl. 16 49.1 399.9 rare Piper crassinervium H.B.K. 21 51.4 106.6 rare Piper glabratum Kunth. 67 61.3 148.2 Piper xylosteoides Steud. 46 38.4 357.2 indeterminate Ottonia 27 3 85.0 283.3 rare indeterminate Ottonia 28 1 95.0 100.0 rare indeterminate Piper 29 3 38.3 164.2 rare 6 29.2 68.7 rare 32 33.4 117.7 rare Coffea arabica Benth. 86 35.3 497.3 anim Psychotria barbiflora DC. 2 72.5 650.0 anim rare Psychotria carthaginensis Jacq. 19 41.8 139.0 anim rare Psychotria deflexa DC. 11 52.3 535.3 anim rare Psychotria kleinii Smith & Downs 20 74.3 650.0 udst anim rare udst rare udst rare Rhamnaceae Colubrina glandulosa Perk. Rosaceae Prunus sellowi Koehne Rubiaceae exotic rare Psychotria leiocarpa Cham. & Schlecht. 83 53.9 404.0 Psychotria macrobotrys Ruiz & Pav. 20 69.8 340.5 Psychotria myriantha Muell. Arg. 81 51.3 404.6 Rudgea jasminioides Muell. Arg. 11 51.8 467.1 anim rare indeterminate Randia 31 4 43.8 35.4 anim rare wind udst anim rare anim rare anim rare Rutaceae Balfourodendron riedelianum (Engl.) Engl. 489 40.4 81.9 Esenbeckia febrifuga (St. Hill.) A. Juss. 99 24.7 4.4 pion abun rare Esenbeckia grandiflora Mart. 20 28.0 650.0 clim rare Pilocarpus pennatifolius Lem 368 38.9 23.4 Zanthoxylum hyemale A. St. Hil. 10 41.0 98.7 rare Zanthoxylum rugosum A.St.Hil. 9 45.0 150.5 rare indeterminate Citrus Linn. 5 4 16.3 57.0 indeterminate zanthoxylum 36 2 5.0 6.8 rare exotic rare rare Sapindaceae Allophylus edulis (St. Hill.) Radlk. 100 30.7 70.0 Cupania vernalis Camb. 9 43.9 28.2 anim rare rare Matayba elaegnoides Radlk. 10 43.0 438.1 rare Paullinia meliaefolia Juss. 29 46.6 153.3 rare Sapotaceae Chrysophyllum gonocarpum (Mart. & Eich.) Engl. 115 42.6 233.3 rare Chrysophyllum marginatum (H. & A.) Radlk. 7 42.1 236.8 rare Pouteria beaurepairei (Glaziou & Raunk.) Boehni 3 10.0 0.4 rare 134 42.2 235.0 Cestrum calycinum Willd. 12 12.5 42.4 Cestrum intermedium Sendtn. 35 42.9 95.5 Solanum argenteum Dun. 55 60.5 291.3 anim rare Solanum caavurana Vell. 8 41.3 5.8 anim rare Simaroubaceae Picramnia ramiflora Planch. cano rare Solanaceae clim rare rare Chapter 4 Effect of forest edges on different types of species 122 Solanum mauritianum Blanco 7 35.7 42.5 anim rare Solanum sanctae-catharinae Dun. 1 5.0 2.0 anim rare indeterminate Solanum 33 5 36.0 551.0 rare indeterminate Solanum 34 3 58.3 38.7 rare 1 65.0 326.1 rare 21 33.3 174.6 Styracaceae Styrax pohlii A. DC. Tiliaceae Heliocarpus americanus Linn. wind rare Ulmaceae Celtis iguanae (Jacquin.) Sargent 11 35.5 20.9 Trema micrantha (Linn.) Blume 2 27.5 218.5 pion rare 81 31.8 44.5 pion rare Aegiphila mediterranea Vell. 61 39.3 26.1 Aegiphilla sellowiana Cham. 5 36.0 336.9 Indeterminate Petraea 30 1 10.0 100.0 949 38.4 130.8 Indeterminate 35 1 45.0 100.0 indeterminate alternate 10 3 45.0 172.6 rare Indeterminate alternate 4 20 31.0 35.6 rare indeterminate alternate 8 2 40.0 3.5 rare indeterminate alternate 9 4 26.3 35.3 rare indeterminate compound 12 1 5.0 326.0 rare indeterminate compound 13 5 69.0 115.5 rare indeterminate compound 14 3 48.3 220.8 rare indeterminate compound 15 2 30.0 488.1 indeterminate opposite 1 396 41.7 69.4 indeterminate opposite 16 3 33.3 11.9 rare indeterminate opposite 17 1 0.0 56.0 rare indeterminate opposite 18 6 10.0 136.3 rare indeterminate opposite 6 15 53.3 185.2 rare rare Urticaceae Urera baccifera (Linn.) Gaud. Verbenaceae rare pion rare rare Violaceae Hybanthus biggibosus (H. H.) Hassl. udst abun Indeterminate Unindentified Total (excluding unindentified) rare rare udst rare 102 16764 ABUN: Absolute species abundances DIST: Average of the distance from the edge, among all occurrences of the species SIZE: Average of the fragment size, among all occurrences of the species HEIG: udst Understory species cano Canopy species DISP: wind – Wind dispersed species anim Animal dispersed species SUCC: pion Pioneers species clim Climax species EXOT: Exotic species RARE: rare – Rare species com – Common species Chapter 4 Effect of forest edges on different types of species 123 Fig 1 Box plots of heights of the sixty eight species with more than 50 individuals, and present in at least two fragments. Horizontal lines refer to median, boxes indicate 50 percentile, and bars indicate 90 percentile. Data above and below the 90 percentile are plotted individually, as circles. Chapter 4 Effect of forest edges on different types of species 124 1000 Plant height (cm) 800 600 400 200 0 5 10 15 20 25 30 35 40 45 50 55 60 65 Chapter 4 Effect of forest edges on different types of species 125 Legend to figures 1 and 2 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 pipxyl psylei psymyr acagra indop1 sapgla geipub ingstr hybbig psykle eugram cofara pipgla acapol acalac sebsp1 eutedu jusbra pesaus actcon pipadu necmeg micdis ocosil eugmor asppol triele tricat segacu parrig ingmar cassyl sorbon guakun Piper xylosteoides Steud. Psychotria leiocarpa Cham. & Schlecht. Psychotria myriantha Muell. Arg. Acalipha gracilis A.Gray Indeterminate opposite 1 Sapium glandulatum (Vel.) Pax Geissomeria pubescens Nees Inga striata Benth Hybanthus biggibosus (H. H.) Hassl. Psychotria kleinii Smith & Downs Eugenia ramboi Legr. Coffea arabica Benth. Piper glabratum Kunth. Acacia polyphilla DC. Acacia lacerans Benth. Indeterminate Sebastiania 32 Euterpe edulis Mart. Justicia brasiliana Roth. Peschiera australis (Muell. Arg.) Miers Actinostemon concolor (Spr.) Muell. Arg. Piper aduncum Linn. Nectandra megapotamica (Spr.) Mez. Miconia discolor DC Ocotea silvestris Vatt Eugenia moraviana Berg Aspidosperma polyneuron Muell. Arg Trichilia elegans A.Juss. Trichilia catigua A. Juss. Seguieria aculeata Jacq. Parapiptadenia rigida (Benth.) Bren. Inga marginata Willd. Casearia sylvestris Sw. Sorocea bonplandii (Baillon) Burg. Guarea kuntiana A. Juss 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 Galint Urebac Hexita Pipama Macpar Molcla Rapumb Alledu Croflo Pipgon Guamac Esefeb Cabcan Tripal Tricas Sebcom Balrie Tricla Holbal Pilpen Camxam Chrgon Aegmed Ocomed Tripal Macmin Picram Lonmue Astgra Eugver Loncam Guaopp Neoglo Solarg Gallesia integrifolia (Spreng.) Harms. Urera baccifera (Linn.) Gaud. Hexachlamys itatiaiae Mattos Piper amalago (Jacq.) Yunck. Machaerium paraguariense Hassl. Mollinedia clavigera Tul. Rapanea umbellata (Mart. ex. A. DC. Mez Allophylus edulis (St. Hill.) Radlk. Croton floribundus Spr. Piptadenia gonoacantha (Mart.) Macbr. Guarea macrophilla Vahl. Esenbeckia febrifuga (St. Hill.) A. Juss. Cabralea canjerana (Vell.) Mart. Trichilia pallida Sw. Trichilia casaretti C. DC. Sebastiania commersoniana (Baill.) Smith & Downs Balfourodendron riedelianum (Engl.) Engl. Trichilia clausenii C. DC. Holocalix balansae Mich. Pilocarpus pennatifolius Lem Campomanesia xanthocarpa Berg Chrysophyllum gonocarpum (Mart. & Eich.) Engl. Aegiphila mediterranea Vell. Ocotea indecora Schott Trichilia pallens C.DC. Machaerium minutiflorum Tul. Picramnia ramiflora Planch. Lonchocarpus muehlbergianus Hassl. Astronium graveolens Jacq Eugenia verrucosa Legr. Lonchocarpus campestris Mart ex Bentham Guapira opposita (Vell.) Reitz Neomitranthes glomerata (Legr.) Legr. Solanum argenteum Dun. Chapter 4 Effect of forest edges on different types of species 126 pesaus acapol parrig croflo geipub galint 2 Axis 2 (7.1%) 1 macpar esefeb macmin astgra pipama alledu lonmue hexita acalac pipgon guaopo eugver cofara indop1 trielerapumb hybbig ocosil loncam neoglo segacu urebac pipxyl picram micdis chrgon rudpar pilpen triida asppol cassil balriejusbra aegmedcamxan tricat sebcom frag size psymir holbal guakun psykle tricla guamac sorboneugram ingstr eugmor northing cabcan ocoind tricas acagra ingmar actcon triens psylei pipadu 0 -1 edge dist necmeg -2 pipgla sebsp1 eutedu -2 solarg 0 2 Axis 1 (16.7%) Figure 2 Canonical correspondence analysis diagram. Arrows indicate axis of maximum dispersion of species in relation to the environmental variables they represent. Species with higher averages of an environmental variable are located closer to the tip of the arrow of that environmental variable. Therefore, a species like asppol (Aspidosperma polyneuron) occurred in larger fragments than a species like aegmed (Aegiphilla mediterranea), as presented in table 1. Chapter 4 Effect of forest edges on different types of species 127 0.08 Climax Pioneers P < 0.001 0.04 0.00 -0.04 Understorey Canopy Relative frequency (Standardized Units) 0.04 P < 0.001 0.00 -0.04 0.2 Abundant Rare 0.1 P < 0.001 0.0 -0.1 -0.2 Non-Exotic Exotic 0.008 P < 0.001 0.004 0.000 -0.004 -0.008 0.08 Animal Wind 0.04 0.00 -0.04 P < 0.001 0 20 40 60 80 100 Distance from the edge (m) Figure 3 - Group frequency at different distances from the edge. Frequency values were subtracted from the group average across all distances from the edge. Chapter 4 Effect of forest edges on different types of species 128 80 Climax (197.5 ha) Pioneer (93.8 ha) 40 P < 0.001 0 -40 40 Relative frequency (Standardized Units) 20 0 P< 0.001 -20 Canopy (168.3 ha) Understory (93.2 ha) -40 -60 Abundant (105.2 ha) Rare (106.5 ha) 80 40 0 P = 0.676 -40 -80 80 Exotic (11.7 ha) Non-Exotic (107.0 ha) 40 P< 0.001 0 -40 40 Animal (124.3 ha) Wind (171.9 ha) 20 P< 0.001 0 -20 0-1 1 - 10 10 - 100 100 - 1000 Fragment Size (ha) Figure 4 - Group frequency at different fragment sizes. Frequency values were subtracted from the group average across all distances from the edge. Chapter 4 Effect of forest edges on different types of species 129 Chapter 4 Effect of forest edges on different types of species 130 Chapter 5 Non-monotonic patterns on edges of tropical forests A) Introduction The recent literature on forest edges assumes that, when entering a forest, factors like plant density, light, and species composition change linearly or monotonically with distance from the edge (Cadenasso et al. 1997; WilliamsLinnera 1990a). Little attention has been given to wave (non-monotonic) patterns, in spite of their appearance in a number of works: Monro (1992); Camargo and Kapos (1995); Turton and Freiburger (1997) and Viana et al. (1997). Sprugel (1976) first described a mechanism generating non-monotonic patterning within temperate forests. The author describes a disturbance mechanism in which trees are more prone to be wind-blown, if they are on the wind-exposed side of a gap. As a tree is broken, it increases the exposure of the area to the windward of it, generating a wave-shaped pattern. If wind is a major force acting on tropical edges, then the edge community must be similar to other communities submitted to frequent disturbances. Therefore, it should be dominated by pioneer species, and have low diversity (similar to other initial phases of succession in the tropics). Franco and Harper (1988) described a non-monotonic pattern as a result of asymmetrical competition among neighbors. Individuals planted at the edge of pots had more light availability than their neighbors did in the interior. These neighbors were suppressed, and their small size provided an advantage to the Chapter 5 Non-monotonic patterns on edges of tropical forests-130 next row of individuals. A wave shaped pattern arose as a result of the alternation of suppressed-not suppressed areas. Murcia (1995) describes a third mechanism generating non-monotonic patterns at edges. According to her, waves may result from opposing trends of environmental variables along borders (light and water, mostly). In this case, both extremes are detrimental for plant growth (too much light and dryness, or humidity without light). High environmental predictability at the edge allows plant populations to find their optimal site and thrive there, reaching high density values. This mechanism is similar to niche partitioning in gaps (Ricklefs 1977; Hartshorn 1978; Denslow 1980; Orians 1982). The three different mechanisms stated above (figure 1) have different effects on the edge community. Wind-related death (Sprugel 1976) implies reduced diversity at the edge, and increased abundance of shorter species with shorter life-span (pioneers). Asymmetrical competition (Franco and Harper 1988) implies a regularly oscillating pattern of plant-density. If asymmetrical competition is an important driving force for the edge community, then different species are associated with suppressed and non-suppressed patches (understood here as those having low and high sapling density), so that dense patches have similar species composition. According to this mechanism, species composition should repeat itself at a wave period, matching the plant density wave period. The proximity of contrasting species pools supposedly enhances species richness close to the edge. The niche partition mechanism (Murcia 1995) implies a fine division of environmental conditions, and a progressive change of species Chapter 5 Non-monotonic patterns on edges of tropical forests-131 composition towards the forest interior. The niche partition mechanism also implies enhanced diversity along the edge, as a result of the close and contrasting species pools. This work will attempt to test the existence of non-monotonic patterns on forest edges, and in so doing, it will investigate the mechanisms that may be generating non-monotonic patterns. Subsequently, the influence of landscape features (fragment size and edge orientation) on non-monotonic patterns will be investigated. B) Materials and Methods Refer to Chapter 3 for site description and data collection of tree data, abiotic data and sapling data. The three data sets were randomly and independently located on edges within a 60 x 60 km area. The measurement of factors at different locations may appear to be a problem for their correlation. However, it consists of a more strict condition than if they were collected on the same sites. Plant density and light at the ground level are linked, so that random factors increasing plant density would also reduce light if they were both collected on the same sites. By using different sites to collect light and plant density, random factors (gap occurrence, proximity to large trees, etc) affect sites differently, so that any correlation obtained between sites can be attributed to the only aspect they have in common, that is distance from the edge. Chapter 5 Non-monotonic patterns on edges of tropical forests-132 Analysis The presence of a non-monotonic (crest and hollow) pattern of sapling density was tested with a Monte Carlo permutation test (Manly 1997). Sapling density values were randomized within transects, and then averaged according to their randomized distances from the edge. After each of the 10,000 permutations, the summation of departures from the sapling density overall average was recorded. Therefore, the observed summation of departures from the sapling density overall average (arrows in figure 2A), were compared to a distribution of the same value, built under the null hypothesis that distance from the edge has no influence on the pattern of sapling density. Another permutation test was devised to assess how pervasive a nonmonotonic (crest and hollow pattern) was among the 48 sapling transects. A number of transects (from 1 up to 48) were taken at random, and the plots were averaged according to distance from the edge. After each averaging, I recorded whether the sapling density of plots 5-20m was higher than both 0m, and 2535m. Ten thousand permutations were performed. Results are presented in terms of how many of these permutations presented the pattern, as a function of how many transects were averaged. Vapour Pressure Deficit, Species richness, Light, Tree density and Sapling Density were averaged for each distance from the edge. Confidence intervals (95%) are presented for each average. A piecewise regression (Neter et al. 1985; Williams-Linnera 1990a) was fit to each one of the fifty-five VPD transects. This regression allows fitting two lines to a data set. One stands for a linear Chapter 5 Non-monotonic patterns on edges of tropical forests-133 change of VPD towards the interior, and the second is a horizontal line, standing for the VPD value on the forest interior. This regression informs the point where a certain measure ceases to change, i.e.: the edge width. Sapling species composition and Sapling species diversity were analyzed on 20 pooled samples (hereafter referred to as PS), obtained by pooling all plots at the same distance from the edge, across all 48 sapling transects. This method is also known as transect composing, and has the advantage of averaging out the variation on individual transects, providing clearer trends (Whittaker 1978). Partial PS were obtained by pooling north and south samples separately. North PS were obtained by pooling 19 transects, and South PS by pooling 29 transects. The same was done for large and small fragments. Small and Large PS were obtained by pooling 24 transects. Besides the diversity estimates for PS, species richness was assessed in individual plots. The first measurement estimates heterogeneity among all plots at a given distance from the edge, and the former estimates heterogeneity within individual plots. A linear regression was performed to evaluate trends of species richness and species diversity on a qualitative basis, since the spatial autocorrelation of this data set prevents hypothesis testing. A cluster analysis was used to assess similarity of species composition among the 20 PS. The number of samplings over the PS varied from 568 (80m), to 1155 (15m). The analysis of species composition of groups with a varying number of individuals suffer from density bias, that is the increase of species in groups with more individuals, solely because of a sampling effect. Chapter 5 Non-monotonic patterns on edges of tropical forests-134 The problem of different sized PS was solved by CNESS (Chord Normalized Species Shared), an index created by Trueblood et. al. (1994). The index is based on the probability of species to appear on a random sample of a given size, in fact a hypergeometrical standardization of the data set, followed by plot normalization (by Trueblood et al. 1994). The Euclidean distances between these values is CNESS. PS were clustered by group averaging (UPGMA), based on CNESS. Seven diversity indexes were calculated for each pooled sample (PS): Singletons (species with one individual), Doubletons (species with two individuals), Ace (Chao et al. 1993), Chao1 (Chao 1984), Fisher's Alpha (Magurran 1988), Shannon-Weaver (Magurran 1988), Simpson (Magurran 1988). A Principal Component Analysis (PCA) was employed to analyze differences between the indices. The 48 transects were grouped into two equal halves, comprising 24 transects located on the smallest fragments (0-10 ha), and 24 located on the largest fragments (10.1-650 ha). A Monte Carlo permutation test was employed to test the hypothesis that the variance of sapling density along the transects had no influence of fragment size. Transects were allocated randomly to one of the two groups (smallest and largest fragments). After each randomization, the variance of plant density along the transect was calculated for both groups. Subsequently, the difference between both variances was calculated, and recorded. This process was repeated 10,000 times, in order to build a distribution Chapter 5 Non-monotonic patterns on edges of tropical forests-135 of variance differences, under the null hypothesis that the fragment size has no influence on plant density variance. The 48 transects were located on 19 edges facing the north quadrant, and 29 transects facing the south quadrant. Averages and confidence intervals were calculated separately for south and north facing transects. Differences in species composition on large-small fragments and north-south edges were assessed with the same index (CNESS) used for the entire data set. A PCA was used on the standardized matrix, because it provides the best estimate least square fit, two dimensional metric scaling of CNESS among PS (Trueblood et al. 1994). This method is referred to as PCA-H by Trueblood et al. (1994). The null hypothesis is that no structure should be found when quadrats originated from independent transects are pooled. On the other hand, if some structure is found on the PCA-H, then quadrats are correlated with species composition. C) Results The existence of non-monotonic patterning The null hypothesis according to which distance from the edge has no influence creating a non-monotonic (crest and hollow) pattern was refuted. The summation of departures from average, shown in figure 2A, was equal or higher than the observed, in only 3.4% of the permutations. The non-monotonic pattern (whether sapling density at plots 5-20m were higher than plots 0m and 25-35m) was pervasive over the entire data set. 31 transects out of 48 presented the nonmonotonic pattern indicated in figure 2A. The crest and hollow pattern of plant Chapter 5 Non-monotonic patterns on edges of tropical forests-136 density appeared more than 95% of the times 9 transects taken at random were averaged (figure 2B). Non-monotonic patterns appeared when transects were divided into those in north and south edges (Figure 5), small and large fragments (Figure 7) and into three different plant sizes (Figure 4B). Figure 3 presents the transects in which the non-monotonic trends of sapling density were among the most and the least clear. Figure 3A shows the transect with the highest linear coefficient, and figure 3B 3A shows the transect with the highest Spearman rank correlation with the average sapling density presented on figure 2. Not surprisingly, saplings species richness followed the same pattern as saplings density (Figure 4A). A reduced number of individuals fit into the 20m2 plot used to record the plants in the field. With a reduced number of individuals, plant density and species richness are generally linearly related, which leads to a similar edge profile of species richness, and sapling density. Both crest and hollow pattern of sapling density and sapling species richness seem to end at 35m from the edge. Species richness on individual plots also showed a general trend of increasing deeper into the forest, whereas species diversity of P.S. showed a general trend of decreasing into the forest. Confidence errors for the curves are presented, but the autocorrelation of this data set prevents a test of hypothesis. The separate fitting of the 55 VPD transects onto a piecewise regression yielded an average stabilizing point (edge width) of 35.25m ± 6.63m from the Chapter 5 Non-monotonic patterns on edges of tropical forests-137 edge (Confidence Interval, P =0.95). The vertical dashed line in figures 3A and 3B, indicates this point. Species composition clustered PS into five groups at the following distances 0m, 5m-35m, 40m-70m, 75m and 80-90m (Figure 4A). This clustering was consistent with the crest and hollow pattern of sapling density. 0m (before the crest) has a different species composition from the other PS. The crest of sapling density (5-15m) had a dissimilarity below 0.45, as did the hollow (30-35m). Between them, there were two intermediate pooled plots (20-25m) that are also similar (CNESS below 0.45). Beyond 40m from the edge, PS cluster, indicating differences in species composition, in relation to 0-35m. This does not indicate that edge effect of species composition is limited to the first 35m. If this were the case, species composition would not correspond to distance from the edge beyond 35m, as they do. The correspondence between distance from the edge and species composition is lost only after 75m from the edge, on the fourth cluster. Species composition clustering was also consistent with PS species diversity (fig 3A). Before 35m, PS diversity fluctuates around a higher average (141.8) than after 35m (120.9). Sapling diversity peaks at 35m, where the region before 35m meets the region after 35m. Diversity rises after 80m, after the correspondence with distance (and the edge effect) weakens. Species composition also relates with light and VPD profiles (fig.3A). At 0m, where light and VPD are very different from interior, species composition was also very different. From 5 m up to 35m from the edge, where only VPD is higher Chapter 5 Non-monotonic patterns on edges of tropical forests-138 than on the forest interior, a cluster of plots with similar species composition is formed. From 35m, up to 70 m, species composition is rather similar, and so are VPD and light. At 70m, light again peaks where two species composition clusters meet. Sapling diversity indices of PS seemed to agree with one each other, to a large extent. Excluding Doubleton, they all showed high loadings with the first PCA axis. Among the seven diversity indices, Chao 1 was the one that most agreed with the six other indexes, and, therefore, it was chosen to summarize the trend of PS diversity towards the interior of the forest. The peak of diversity shown by Chao 1 at 35 meters (Figure 4A), also occurred with ACE, Singletons and Fisher's Alpha. Simpson and ACE, similarly to Chao 1, showed higher values of diversity before 35m, than after that distance. Tree density decreased from 0m, up to 70 m from the edge (fig 3b). Most of this reduction occurred within 0 and 20 m. Despite the non-significant correlation between tree density, sapling density and light, there was a peak of tree density at 30 meters that coincided with a hollow of sapling density and a peak of light. Also, at 70m, there was a hollow in tree density, that coincided with a peak in sapling density and a peak of light. Sapling and light profiles mirror one another from 0m up to 35m, and tree profiles seems to mirror light, from 40 up to 90m from the edge. The three data sets (biotic, tree and saplings) seem to coincide on the location of their peaks. The first, stronger one is located at 35 m, and the second, Chapter 5 Non-monotonic patterns on edges of tropical forests-139 weaker one, is located at 70m from the edge (indicated by vertical dashed lines on graphs 3A and 3B). The influence of landscape on non-monotonic patterns North and South edges The crest and hollow pattern of sapling density is present in both North and South edges (Figure 5), but the patterns differ in profile. While density peaks at 5 m on south edges, it peaks only at 15m in north transects. These two peaks appear on saplings smaller than 2 meters (Figure 4B), but not on those higher than 2 meters. Plant density increases sharply between 35 and 40m on south edges. On north transects, it stays low until 55-60m, and does not rise much above the average, even after 100m from the edge. This suggests that the crest and hollow pattern is contracted on south transects, and extended in north transects. The contraction of the crest and hollow pattern seems to be associated with less penetration of dry air into the forest. The average point at which VPD stabilized on North edges was 42.2 ± 11.6 m (confidence interval, 95%), and on south edges it was 29.7 ± 7.0 m (confidence interval, 95%). The pattern of contraction and extension is also found for species composition. The two axis explaining more variance on the PCA-H divide all PS into North and South PS, and correlate with distance from the edge, until approximately 30m from the edge. This indicates that North and South PS have different species composition, and that species composition is related to distance Chapter 5 Non-monotonic patterns on edges of tropical forests-140 from the edge, until 30m from the edge. Furthermore, at the same distance, north PS are more typical of edges (have a higher score on PCA-H axis one), than are south PS. Species composition of a south PS was more typical of edges in only one case (30m). Species composition was distinct on north and south PS, in spite of the closeness among many south and north transects on the field. Species like Lonchocarpus muhelbergianus and Sebastiania commersoniana were more frequent on South edges than on North edges (Figure 6B), and Sebastiania sp32, were more frequent on North edges than on South edges. Species composition change was consistent on North and South PS, until 30m from the edge. Species like Inga marginata and Trichilia clausenii were more frequent further from both North and South edges, whereas Lonchocarpus campestris and Parapiptadenia rigida were more frequent close to North and South edges. The PCA-H diagram on Figure 6A also informs whether the difference between North and South edges is larger than the difference between edge and interior, in terms of species composition. The variance explained by PCA-H axis 2 was 14.5%, and PCA-H axis 1 was 16.3%, informing that both contrasts were as large. Large and Small fragments Sapling density varied more along edges of small fragments than on edges of large fragments (Figure 7). Maximum average of sapling density on edges of small fragments was 28.4 saplings/plot, at 15m from the edge, and the minimum was 12.8 saplings/plot, at 80 meters from the edge. Average density on large Chapter 5 Non-monotonic patterns on edges of tropical forests-141 fragments varied from 16.8 saplings/plot, up to 23.3 saplings/plot. The Monte Carlo permutation test showed that the variance along transects in large fragments is significantly smaller (P=0.0064) than the variance along transects on small fragments. Despite the higher variance of sapling density on smaller fragments, crests and hollows seemed to be located at similar positions on both small and large fragments. Similarly to the North and South contrast, species composition on small fragments PS, and large fragments PS were distinct and consistent (Figure 8). Species like Piper aduncum, Aspidosperma polyneuron, Miconia discolor were more frequent on edges of large fragments, than on edges of small fragments, and species like Piptadenia gonoacantha and Balfourodendron riedelianum were more frequent on the edges of small fragments than on edges of large fragments. Differences between edge and interior were consistent on small and large fragments, until 20-25m. Species like Inga marginata and Actinostemum concolor were frequent on the interior of both large and small fragments, whereas Acacia polyphilla was rare. Species composition change was related to distance from the edge on small fragments until 25 or 30m from the edge, and until 15 or 20m from the edge on large fragments. The contrast between Large and Small fragments PS was larger than the contrast between edge and interior PS, in terms of species composition (Figure 8). The first PCA-H axis that separates large and south fragments PS explained 27.1% of the variation, whereas the second PCA-H axis, that separates edge and interior PS, explained 10.9% of the variation. Chapter 5 Non-monotonic patterns on edges of tropical forests-142 D) Discussion D1) Non-monotonic patterns and edge width estimates The coincidence of peaks for tree, saplings, and light transects indicates that saplings may be controlling lateral light incidence on forest fragments within 35m from the edge. After that, when lateral incidence of light is no longer significant, taller trees exert a stronger influence over light at ground level. The coincidence of peaks for light, tree and sapling density in independently located transects indicate that the open environment has a strong influence on edges of this landscape. Different factors showed singularities (stabilization points, peaks, clusters, or the period of the crest and hollow pattern) at 35 m from the edge. This seems to indicate that at 35 meters, factors are similar to the edge interior, so that this is the estimate of edge width on the landscape studied. In addition to that, another singularities occurred at 70m from the edge, indicating the existence of a second edge width, with less contrast (sensu Hansen et al. 1992) than the one at 35m. This second edge consists of one of the strong evidences in favor of nonmonotonic patterns on edges. The existence of non-monotonic patterns on edges, suggests that in many instances on the literature, edge widths may have been underestimated. Studies based on small transects, like Viana (1997), Matlack (1993), Cadenasso (1997), Mayaka (1995), Williams-Linera (1990) and Palik and Murphy (1990), and Rodrigues (1993) are the most likely candidates for such underestimate. In those cases, one side of the crest and hollow pattern may have been interpreted as a Chapter 5 Non-monotonic patterns on edges of tropical forests-143 linear increase/decrease. Therefore, it is likely that the actual edge width, in many instances, was larger than estimated. Murcia (1995) and Laurance et al. (1997) suggested that unique edge width estimates are impossible to be found, given the multitude of factors acting on edges. However, edge width estimates of abiotic factors, trees and saplings seemed to coincide in this study, reinforcing the idea of a unique edge width, or at least that few edge width estimates are able to summarize a large number of factors. D2) Mechanisms generating non-monotonic patterns Wind disturbance The implications of the wind disturbance process to the edge community did not seem to occur when all edges are pooled. (partial PS presented a different figure). Tree density is higher at the edge, where trees are most exposed to wind. This is incompatible with high turnover rates in frequently disturbed environments. Wind, however, has been reported to cause extensive mortality on younger edges (Lovejoy et al. 1984). It is possible that, within the sixty years since fragmentation, wind resistant species occupied the edge (not considering other possible differences between north and south Brazilian forests). Niche Partitioning The notion that different environments are created at different distances from the edge, and that different species composition are associated with each one of these environments was repeatedly described for temperate edges (Gysel Chapter 5 Non-monotonic patterns on edges of tropical forests-144 1951; Wales 1972 and Brothers and Spingarn 1992), but not for the tropics. The two previous studies on the tropics (Williams-Linnera 1990a and WilliamsLinnera 1990b) failed to find differences of species composition at different distances from the edge, perhaps due to the young age of the edges (less than 12 years). Species composition was associated with the different environments formed at different distances from the edge, what supports that niche partitioning is a key mechanism shaping saplings species composition on the edge. Asymmetrical competition An important aspect associated with asymmetrical competition is the regular alternation of suppressed and non-suppressed patches, with reducing intensities at each cycle, at further distances from the edge (Franco and Harper 1988). The distance of 35 meters seemed to be a recurrent theme in many of the factors: VPD stabilized at 35 meters from the edge, sapling species composition clustered at 5-35m, and at 40-70m. Sapling density showed a crest and hollow pattern within the first 35 m, and a peak at 70m, as did tree density. This seems to suggest that suppressed and non-suppressed patches alternate at regular 35m intervals. Contrary to the expectation, different species were not associated with suppressed and non-suppressed patches. PS 0m and PS 30m have both low sapling density, but their species compositions were very different. Similarly, 15m and 70m had both high sapling densities, but different species compositions, as well. Niche partitioning and asymmetrical competition seems to interact. Plant density varies at regular intervals as a result of asymmetrical competition. Chapter 5 Non-monotonic patterns on edges of tropical forests-145 However, species compositions are not the same at each cycle, because of the different environment formed at different distances from the edge. The influence of landscape on non-monotonic patterns The displacement of edge zones towards the forest interior, on sun-exposed edges, has been already described for the temperate region by Wales (1972), Fraver (1994), Matlack (1993), Matlack (1994), Palik and Murphy (1990) and Cadenasso et al. (1997). Much less data is available on the tropics. Two papers discussing the effect of edge orientation in the tropics (Viana 1997; WilliamsLinnera 1993) are inconclusive, for lack of replications, or historical differences between the edges analyzed. Also, Turton and Freiburger (1997) found no differences in seedling density between North and South edges, working on a single fragment in Australia. Data presented here indicate that the same displacement described previously in the temperate region, also occurs in the tropics. In fact, the pattern could be better described as a contraction, where changes on plant density, VPD and species composition occur within shorter distances on sun-protected edges, than on sun-exposed edges. Average sapling density varied more on edges of smaller fragments, than on edges of larger fragments, without major changes on location of the crest and hollow pattern. According to Malcolm (1994), the intensity of edge effects on a patch of forest is a function of distance to all close edges, not only the closest one, that is being studied. On small fragments, edges are closer to one each other, than on large fragments (All fragments smaller than 10 ha used on this study were square). This suggests that edges on small fragments are subject to Chapter 5 Non-monotonic patterns on edges of tropical forests-146 an additional edge effect, coming from other close edges, that enhances the original one, causing the increase on density variance. The increased variance of plant density on edges of smaller fragments seems to be at odds with both niche partitioning and asymmetrical competition. Asymmetrical competition creates crest and hollow patterns through shading. On small fragments, the shaded patches count with an additional source of light, from other close edges. Therefore the variance of sapling density on small fragments should be smaller, and not larger. Niche partitioning creates crest and hollow patterns through the environmental differences that enhance survivor of some species, and reduce survivor of others. Environmental differences are supposedly smaller in small fragments, because of the closeness of other edges. Therefore, a crest and hollow pattern created as a result of niche partitioning would be less marked in small fragments than in larger fragments. Among the three mechanisms proposed to explain the existence of crest and hollow patterns on edges, wind disturbance is the only one expected to be more intense on small fragments. Esseen (1994) showed that wind caused an increase of mortality in small fragments of conifer forest, in relation to larger fragments. If edges of small fragments lack trees, then the density of saplings supposedly is higher, generating the higher variance on edges of small fragments. Wind disturbance may also be the cause of the marked differences in species composition of large and small fragments PS. Patches at different distances from the edge are not differentially affected by wind disturbance, or at least not enough to reduce tree density or to cause a marked reduction of sapling Chapter 5 Non-monotonic patterns on edges of tropical forests-147 species richness, as expected. However, whole fragments seem to be affected, so that the differences in species composition between large and small fragments are enhanced. There is no consensus on the literature, whether edges of large or small fragments are expected to have a larger width and contrast (sensu Hansen et al. 1992). Kapos (1989) and Malcolm (1994) suggest that edges of smaller fragments are larger and have more contrast, what agrees with the predictions of the extinction-debt theory of Nee and May (1992). However, Yahner (1988) proposes that larger fragments will have more intense edge effects because of the larger extension of edges they have. The relation between distance from the edge and species composition was maintained until deeper into the forest on the smaller fragments, than on the larger fragments, indicating that the edge effect has a larger width on small fragments. This result is an important contribution to implement models that estimate edge area at landscape level, like the one presented by Laurance and Yensen (1990). Up to now, fragments of different sizes were assumed to have the same edge width. Edge heterogeneity The regressions of species richness of individual plots, and species diversity of PS provided qualitative evidence that edges have less species than forest interior, and that they differ more among them than the forest interior does. If later studies confirm this trend, it would indicate a change on the diversity spatial pattern of the forest, reducing diversity at the scale of 20m2, and increasing the Chapter 5 Non-monotonic patterns on edges of tropical forests-148 diversity at the scale of tens of kilometers. This increased patchiness implies more isolation among individuals of the same species. The consequences of such isolation for the community are unpredictable at this point. E) Conclusions Non monotonic patterns of sapling density were described and related to light, VPD, tree density, edge orientation, and fragment size. This does not imply, however, that all factors on edges should be expressed as a non-monotonic function of distance from the edge. VPD decreased linearly with distance from the edge, and PS diversity of saplings presented a discontinuity, typical of the "wall" function described on chapter 1. Species composition showed a more complex behavior, forming clusters at 35 m intervals. Each of the four functions described in chapter 1 are related to different conceptions of edge effects. The conception of edge width implies that factors either have a major discontinuity at a certain distance from the edge (wall function), or that they decrease monotonically. The existence of non-monotonic patterns on edges, implies the existence of sequential edge widths, with decreasing contrasts at increasing distances from the edge. On this study, a strong edge was formed at 35 meters from the edge, and a second one was formed at 70 m from the edge The similar edge width presented by several factors suggests that few edge width estimates may summarize a large number of factors. Niche partitioning, asymmetrical competition and wind disturbance interact to create non-monotonic patterns on edges. Contrary to our initial expectation, no Chapter 5 Non-monotonic patterns on edges of tropical forests-149 single mechanism can be accredited, since different aspects of the nonmonotonic pattern can be accredited to each of the three mechanisms. The extended crest and hollow pattern found on north (sun-exposed) edges suggests that they should concentrate the efforts of conservation, as well as small fragments. The protection of small fragments must include mechanisms to reduce the impact of wind on their edges. The close relation between the crest and hollow pattern, and species composition, indicate the development of Rapid Ecological Assessment techniques of forest edges, involving sapling density. Chapter 5 Non-monotonic patterns on edges of tropical forests-150 F) Literature Cited Cadenasso,M.L., Traynor,M.M., and Pickett, S.T.A. 1997 Functional location of forest edges: gradients of multiple physical factors Canadian Journal of Forest Research 27: 774-782 Camargo, J.L.C. and Kapos, V. 1995 Complex edge effects on soil moisture and microclimate in Central Amazonian forest Journal of Tropical Ecology 11 205221 Chao, A. 1984 Non-parametric estimation of the number of classes in a population. Scandinavian Journal of Statistics 11: 265-270 Chao, A. Ma,M.C. and Yang, M.C.K. 1993 Stopping rules and estimation for recapture debugging with unequal failure rates Biometrika 80: 193-201 Correa, A.R. Godoy,H. Bernardes,L.R.M. 1982 Características climáticas de Londrina Circular IAPAR 5 Denslow, J. 1980 Gap partitioning among tropical rainforest trees. Biotropica 12:47-55. Esseen, P.A. 1994 Tree mortality patterns after experimental fragmentation of an old-growth conifer forest. Biological Conservation 68: 19-28 Franco, M. and Harper, J. L. 1988 Competition and Formation of Spatial Pattern in Spatial Gradients: An example using Kochia Scoparia. Journal of Ecology 76:959-974. Fraver, S. 1994 Vegetation responses along edge-to-Interior gradients in the Mixed hardwood Forests of the Roanoke River Basin, North Carolina Conservation biology 8(3):822-832 Chapter 5 Non-monotonic patterns on edges of tropical forests-151 Hansen, A.J., Risser, P.G. Castri,F.di 1992 Epilogue: Biodiversity and Ecological flows across Ecotones in: Hansen, A.J. and Castri,F.di Landscape Boundaries: Consequences for Biotic Diversity and Ecological Flows Springer-Verlag, New York 452 pp. Hartshorn, G.S. 1978 Treefalls and tropical forest dynamics in Tropical Trees as Living Systems P.B. Tomlinson and M.H. Zimmermann eds. Cambridge University Press, New York 675 pp. Kapos, V. 1989. Effects of isolation on the water status of forest patches in the Brazilian Amazon. Journal of Tropical Ecology 5:173-185. Laurance, W. F. and Yensen, E. 1990 Predicting the impacts of edge effects in fragmented habitats. Conservation Biology 55: 77-92 Laurance, W., Bierregaard, R.O., Gascon, C., Didham, R.K. Smith, A.P. Lynam,A.J. Viana, V.M. Lovejoy,T.E., Sieving,K.E., Sites, J.W. Andersen, M., Tocher, M.D. Kramer, E.A. Restrepo, C. and Moritz, C. 1997 Tropical Forest Fragmentation: Synthesis of a Diverse and Dynamic Discipline in: Laurance, W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Lovejoy, T.E., Rankin, J.M., Bierregaard, R.O., Brown, K.S.Jr, Emmons, L.H. Van der Voort, M. E. 1984. Ecosystem decay of Amazon forest fragments. in: NITECKI, M.H. (eds). Extinctions. University of Chicago Press, Chicago 354 pp. Chapter 5 Non-monotonic patterns on edges of tropical forests-152 Magurran, A. 1988 Ecological diversity and its measurement. Princeton University Press, Princeton 179 pp. Malcolm, J.R. 1994 Edge Effects in Central Amazonian Forest Fragments. Ecology 75 (8):2438-2445. Manly, B. 1997 Randomization, Bootstrap and Monte Carlo methods in biology. Chapman & Hall, New York 399 pp. Matlack, G.R. 1993. Microenvironmental variation within and among forest edge sites in the eastern United States. Biological Conservation 66:185-194. Matlack, G.R. 1994 Vegetation dynamics of the forest edge – trends in space and successional time. Journal of Ecology 82 113-123. Mayaka,T.B. Fonweban,J.N. Tchanou,Z. Lontchui,P. 1995 An assessment of edge effect on growth and timber external quality of ayous (Triplochiton scleroxylon K Schum) under Cameroon rain forest conditions. Annales des Sciences Forestieres (Paris) 52(1): 81-88. Monro A.K. 1992. The effect of forest/clearing interfaces on sapling and understorey community dynamics. Account of first year fieldwork, Cambridge University, England. Murcia, C. 1995. Edge Effects in fragmented forests: implications for conservation. Trends in Ecology and Evolution 10: 58-62. Nee, S. and May,R. 1992 Dynamics of metapopulations: habitat destruction and competitive coexistence Journal of Animal Ecology 61:37-40 Neter,J., Wassernan, W., and Kutner, M. 1985. Applied linear statistical models, 2nd : Richard E. Irwin, Inc., Illinois 569 pp. Chapter 5 Non-monotonic patterns on edges of tropical forests-153 Orians, G.H. 1982 The influence of tree-falls in tropical forests on tree species richness. Tropical Ecology 23: 255-279 Palik,B.J., Murphy,P.G. 1990. Disturbance versus edge effects in sugarmaple/beech forest fragments. Forest Ecology & Management 32: 187-202. Ranney,J.W., Bruner,M.C.& Levenson,J.B. 1981. The importance of edge in the structure and dynamics of forest islands, in: Burgess,R.L. and Sharpe,D.M. Forest island dynamics in man dominated landscapes Springer-Verlag, New York 309 pp. Ricklefs, R.E. 1977 Environmental heterogeneity and plant species diversity: a hypothesis. American Naturalist 111:376-381. Rodrigues, E. 1993. Ecologia de fragmentos florestais ao longo de um gradiente de urbanização em Londrina-PR. Master's Dissertation. Universidade de São Paulo, São Carlos-SP Brasil 110 pp. Serviço Nacional de Levantamento e Conservação de Solos 1984 Levantamento de Reconhecimento dos Solos do Estado do Paraná Tomos I e II Boletim Técnico do IAPAR/EMBRAPA, Londrina, 57 pp. Sprugel, D.G. 1976 Dynamics of structure of wave-generated Abies balsamea forests in the North-Eastern United States. Journal of Ecology 64 889-910. Trueblood, D. D, E. D. Gallagher, and D. M. Gould. 1994. The three stages of seasonal succession on the Savin Hill Cove mudflat, Boston Harbor. Limnol. Oceanogr. 39: 1440-1454. Turton, S.M. and Freiburger,H.J. 1997 Edge and Aspect Effects on the Microclimate of a Small Tropical Forest Remnant on the Atherton Tableland, Chapter 5 Non-monotonic patterns on edges of tropical forests-154 Northeastern Australia in: Laurance, W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Viana, V.M., Tabanez, A.A.J., and Batista, J.L.F. 1997 Dynamics and Restoration of Forest Fragments in the Brazilian Atlantic Moist Forest. in: Laurance, W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Wales, B.A. 1972 Vegetation analysis of northern and southern edges in a mature oak-hickory forest. Ecological Monographs 42: 451-471 Whittaker, R.H. 1978 Ordination of Plant Communities Dr W. Junk bv Publishers, Boston 388 pp. Williams-Linera, G. 1993. Vegetación de Bordes de un bosque nublado en el parque Ecológico Clavijero, Xalapa, Veracruz, México Revista de Biologia Tropical. 41:443-453. Williams-Linera, G. 1990a Vegetation structure and environmental conditions of forest edges in Panama. Journal of Ecology 78:356-373. Williams-Linera, G. 1990b Origin and early development of forest edge vegetation in Panama. Biotropica 22:235-241. Yahner, R.H. 1988 Changes in wildlife communities near edges Conservation Biology 2(4):333-339. Chapter 5 Non-monotonic patterns on edges of tropical forests-155 Table 1 PCA loadings of different diversity indexes % explained 1st axis 53.785 2nd axis 22.694 SINGLETON DOUBL ACE CHAO1 ALPHA SHANNON SIMPSON 0.818 -0.273 0.897 0.904 0.77 0.667 0.601 0.352 0.703 0.369 -0.019 0.402 -0.479 -0.665 PCA Chapter 5 Non-monotonic patterns on edges of tropical forests-156 PROCESSES DIVERSITY AND SPECIES COMPOSITION AS IMPLIED BY THE PROCESSES Wind disturbance (Sprugel 1976) Diversity towards interior Wind exposure creates a gradient of disturbance closer to the edge. Plant density increases at intermediate levels of disturbance. Increase Speciescomposition towards interior Monotonic change Asymmetrical Competition (Franco and Harper, 1988) Trees at the edge have access to more light, so that they end up suppressing plants close to them. Plants close to suppressed patches also have access to more light. Decrease Non Monotonic, and associated with plant density Niche Partitioning (Murcia 1995) Distinct species occupy different environments created in result of proximity to the edge. Plant density is enhanced in certain environments, but not in others. Decrease Monotonic change Figure 1 Mechanisms generating nonmonotonic patterns on forest edges Chapter 5 Non-monotonic patterns on edges of tropical forests-157 30 cases presenting the pattern 1.0 Saplings/20m 2 25 20 15 P = 0.034 0 20 40 60 A 80 0.9 0.8 0.7 0.6 100 distance from the edge (meters) B 0 10 20 30 40 50 transects averaged Figure 2A-Sapling density as a function of distance from the edge. Circles represent averages, and bars represent 95% confidence intervals. P value refer to the significance of the crest and hollow pattern, as obtained from a Monte Carlo permutation test. Solid horizontal line refers to the overall average of sapling density (20.85 saplings/20m2 plot). 2B – Presence of the crest and hollow pattern on a number of transects taken at random and averaged. Chapter 5 Non-monotonic patterns on edges of tropical forests-158 25 trn 14 2 saplings/20m plot 20 15 10 5 A 0 35 trn 38 25 2 saplings/20m plot 30 20 15 10 5 B 0 0 20 40 60 80 100 distance from the edge (m) Figure 3 Sapling density profiles on the most monotonic, and non-monotonic transects A Sapling density on the transect with highest linear coefficient (transect 14). B Sapling density on the transect with highest Spearmann rank correlation with the average density on figure 2A (transect 38) Chapter 5 Non-monotonic patterns on edges of tropical forests-159 0 20 40 60 80 100 2.0 C (n=55) VPD kPa 1.6 1.2 Species composition 0.60 0.55 0.50 0.45 0.40 5 10 15 20 25 45 50 55 60 65 70 80 95 85 90 75 A (n=48) 14 12 10 8 6 4 30 35 40 Species richness 180 Species diversity of P.S. A (n=48) Diversity index (Chao 1) Sp/20m2 0 A (n=48) Similarity Index (CNESS) 0.8 160 140 120 100 0 20 40 60 80 100 Distance from the edge (m) A - saplings data-set C - abiotic data set Figure 4A Edge profiles as a function of distance from the edge. Bars represent 95% confidence intervals. The regressions on species richness and species diversity diagrams are accompanied by their 95% confidence intervals. Chapter 5 Non-monotonic patterns on edges of tropical forests-160 C (n=55) Light 10 1 20 B (n=14) Stems/100m 2 Light (% outside) 100 Tree density 15 10 5 A (n=48) 20 15 10 9 8 7 6 5 4 3 Sapling size classes saplings >2m 1.27m < saplings <2m 1m< saplings <1.26m 0 20 40 60 80 A (n=48) Plants/20m 2 Plants/20m 2 Sapling density 25 100 Distance from the edge (m) A - saplings data-set B - tree data set C - abiotic data set Figure 4B Edge profiles as a function of distance from the edge (bars represent 95% confidence intervals) Chapter 5 Non-monotonic patterns on edges of tropical forests-161 30 North Density n=19 VPD n=26 Saplings/20 m2 28 26 24 22 20 18 16 14 30 South Density n=29 VPD n=29 Saplings/20 m2 28 26 24 22 20 18 16 14 0 20 40 60 80 100 Distance from the edge (m) Figure 5 Contrast among North and South transects, in relation to sapling density (circles), and VPD stabilizing point (horizontal 95% confidence errors). Chapter 5 Non-monotonic patterns on edges of tropical forests-162 PCA-H Axis 2 (14.5%) 2 1 0 -1 1 95N 85N90N 25N 40N45N 30N 80N 50N 10N 75N 70N 55N 35N 60N 20N 65N 15N ing mar pip adu 0 5N 80S 55S 75S 70S 65S 35S 90S 40S 95S 85S50S 60S 45S20S 25S 30S15S 10S 5S B 32 seba nec meg pip gon cro flo tri cla 0N pil pen lon cam ese feb eut edu mic dis par rig pip ama 1op 0S lon mue seb com -2 -1 -2 -1 0 1 2 3 -1 PCA-H Axis 1 (16.3%) 1op 32s eba aca lac aca pol act con asp pol bal rie ese feb eut edu ing mar species Indet. 1 Sebastiania sp32 Acacia lacerans Acacia polyphilla Actinostemum concolor Aspidosperma polyneuron Balfourodendron riedelianum Esembeckia febrifuga Euterpe edulis Inga marginata 0 1 PCA-H Axis 1 (16.3%) lon mue mic dis nec meg par rig pil pen pip adu pip ama pip gon seb com tri cla Lonchocarpus muhelbergianus Miconia discolor Nectandra negapotamica Parapitadenia rigida Pillocarpus pennatifolius Piper aduncum Piper amalago Piptadenia gonoacantha Sebastiania commersoniana Trichilia clausenii Figure 6A Pca-H ordination of species composition on pooled samples (PS) of North and South edges. Figure 6B Covariance biplot of species, showing only the species that explained 3% or more of species composition variance. Chapter 5 Non-monotonic patterns on edges of tropical forests-163 large frags (24 trns) small frags (24 trns) saplings/20m2 25 20 15 P = .0064 0 20 40 60 80 100 distance from the edge (m) Figure 7 Sapling density on large (>10 ha) and small (<10 ha) fragments. P value refers to the contrast between variance on small and large fragments (solid and dashed arrows). Chapter 5 Non-monotonic patterns on edges of tropical forests-164 4 0L PCA-H Axis 2 (10.9%) 3 1 A B aca pol 5L lon mue 0S 2 10L 15L 30L 20L 25L 45L 50L 35L 90L 40L 55L 95L 70L 60L 80L 85L 75L 65L 1 0 -1 -2 pippar ama rig seb com 1op nec meg mic32s dis eba 5S 10S 15S 20S 40S 25S 45S30S 90S 60S 55S 95S 35S 70S 50S 75S 85S65S 80S 0 bal lac rie aca asppip poladu pip gon ese feb tri ele pil pen eut edu act con tri cla ing mar -3 -1 -2 -1 0 1 PCA H Axis 1 (27.1%) 1op 32s eba aca lac aca pol act con asp pol bal rie ese feb eut edu ing mar lon mue 2 -1 0 1 PCA-H Axis 1 (27.1%) species Indet. 1 mic dis Miconia discolor Sebastiania sp32 nec meg Nectandra negapotamica Acacia lacerans par rig Parapitadenia rigida Acacia polyphilla pil pen Pillocarpus pennatifolius Actinostemum concolor pip adu Piper aduncum Aspidosperma polyneuron pip ama Piper amalago Balfourodendron riedelianum pip gon Piptadenia gonoacantha Esenbeckia febrifuga seb com Sebastiania commersoniana Euterpe edulis tri cla Trichilia clausenii Inga marginata tri ele Trichilia elegans Lonchocarpus muhelbergianus Figure 8A Pca-H ordination of species composition on pooled samples (PS) of Small and Large fragments. Figure 8B Covariance biplot of species, showing only the species that explained 3% or more of species composition variance. Chapter 5 Non-monotonic patterns on edges of tropical forests-165 Chapter 6 Closing Remarks This chapter enumerates the most important contributions of this thesis, and relates it with the current knowledge about edge effects. A) Patterns on edge effects The last comprehensive revision about edge effects (Murcia 1995) emphasizes that edges are so complex that few general patterns can be noticed. However, a close examination of the literature presented a number of patterns: From the edge towards the forest interior: 1) Light and Vapor Pressure Deficit decreases (air humidity edge width is frequently larger than light edge width). 2) Density of trees and regeneration decrease 3) Tree death decreases 4) Seed predation increases 5) Species composition of plants change B) Answers to the Questions on chapter 1 1) Is the edge more diverse than forest interior? Rare species occurred closer to the edge, as shown in chapter 4, and species richness showed a weak tendency to be lower in the edge. In addition, edge heterogeneity was higher than the heterogeneity among forest interiors (a restriction applies to the former two claims, see chapter 5). A consistent trend of change in species composition was found from the edge towards the interior (chapter 5). Chapter 6 Closing Remarks 166 These four aspects seem to indicate that species achieve high abundances at the scale of a plot (20m2), but different species occupy different transects (or sites), increasing heterogeneity. Therefore, diversity is reduced at the small scale, and increased at the larger scale. It is likely that the change on the spatial pattern of species may alter their gene flow, and lead them to inbreeding problems. However, part of the species occurring in edges are pioneers, which normally occur in gaps that are isolated from one each other. On the first chapter, higher plant diversity in edges was said to indicate intense immigration in edges. Not only this did not happen, but a small number of exotic individuals was found in this study, indicating that edges are not likely to be a preferential “entrance” to the fragments. 2) Do edges of large fragments and small fragments differ? Yes. Their sapling species composition differ, and the relation between sapling species composition and distance from the edge penetrates deeper into small fragments than on large fragments, meaning that the edge width of sapling species composition is larger on fragments smaller than 10 ha, than those larger than 10 ha. According to the Canonical Correspondence Analysis, species composition on small fragments resembled the species composition on edges. This resemblance, and the larger edge width on small fragments seems to be enough evidence to refute the hypothesis proposed by Yahner (1988) according to which large fragments would have a stronger edge effect than small fragments, because of their larger edge extension. Chapter 6 Closing Remarks 167 Sapling density on edges of small fragments varied more than on large fragments. This seems to be explained by two mechanisms: The secondary edges effects created by the proximity of other edges on small fragments (Malcolm 1994), and the increased death by wind-throw on small fragments (Esseen 1994). Both mechanisms also explain the marked differences on species composition between large and small fragments. 3) Do edges facing north and south differ in relation to edge width of biotic and abiotic effects on the tropics? Yes. See question 4b 4) Are non-monotonic patterns of plant density common on edges? Yes. A non-monotonic pattern appeared 95% of the times 9 transects were taken at random from the pool of 48 transects, and averaged. Non-monotonic patterns appeared when transects were divided into small and large fragments, north and south transects, and into three different plant sizes. This result suggests that many of the edge width estimates based on short transects (smaller than 50m) (Palik and Murphy 1990; Williams-Linera 1990; Matlack 1993; Rodrigues 1993; Mayaka; 1995; Cadenasso et al. 1997 and Viana et al.1997) may be underestimated, because they considered the area after the first crest to be the forest interior. The existence of non-monotonic patterns on edges does not imply that all factors on edges should be expressed as a non-monotonic function of distance from the edge. VPD decreased linearly with distance from the edge, and species diversity Chapter 6 Closing Remarks 168 of saplings presented a discontinuity, typical of the "wall" function described on chapter 1. 4a) Assuming that the answer to 4 is yes, then which mechanism is creating the crest and hollow pattern? The original attempt of explaining non-monotonic patterns by a single mechanism failed, for different mechanisms explained different aspects of the crest and hollow pattern. Sapling density presented a marked crest at 15m, and a weaker one at 70m from the edge. Asymmetrical competition, as proposed by Franco and Harper (1988) is the only mechanism among the three proposed, that may create repeated patterns, at different distances from the edge. Species composition of saplings did not repeat itself at regular intervals, as did sapling density. Species composition changed regularly at increasing distances from the edge, up to 70m from the edge. After that, no pattern was shown. This trend was expected from the niche partitioning mechanism, operating on a sequence of different environments located at different distances from the edge, as described by Murcia (1995). Wind disturbance seems to generate the differences on the crest and hollow pattern, between large and small fragments. Enhanced tree death by wind make crests and hollows to be more salient on small fragments. 4b) What is the importance of light on creating non-monotonic patterns? How are edges affected by different orientation (as a surrogate of differential exposure to light) Chapter 6 Closing Remarks 169 On the tropical region studied, edges facing the equator were deeper than those facing the poles, similarly to the edges of temperate forests. Edge width of Vapor Pressure Deficit was larger on north edges than in south edges; the crest and hollow pattern of sapling density was expanded on north edges, and species composition of plots deeper into north edges was similar to the species composition of plots closer to the south edges. North Paraná is close to the tropic of Capricorn. It is unknown how far this different is maintained at sites closer to the Equator. C) Suggestions for future studies An attempt was made to study immigration on forest edges, by diversity measurements, and by the occurrence of exotic saplings. A more effective way of studying immigration is to reassess the plots, and obtain actual estimates of immigration and local extinction, by comparison with the results presented on this study. The moving of the deforestation front over North Paraná (figure 6, chapter 2) suggests that the change of forest edges over time could be studied if plots are located on different portions of the region. This will allow better control of unwanted variation, than comparing edges of different regions. Chapter 6 Closing Remarks 170 Literature Cited Cadenasso,M.L., Traynor,M.M., and Pickett, S.T.A. 1997 Functional location of forest edges: gradients of multiple physical factors Canadian Journal of Forest Research 27: 774-782 Esseen, P.A. 1994 Tree mortality patterns after experimental fragmentation of an old-growth conifer forest. Biological Conservation 68: 19-28 Franco, M. and Harper, J. L. 1988 Competition and Formation of Spatial Pattern in Spatial Gradients: An example using Kochia Scoparia. Journal of Ecology 76:959-974. Malcolm, J.R. 1994 Edge Effects in Central Amazonian Forest Fragments. Ecology 75 (8):2438-2445. Matlack, G.R. 1993. Microenvironmental variation within and among forest edge sites in the eastern United States. Biological Conservation 66:185-194. Mayaka,T.B. Fonweban,J.N. Tchanou,Z. Lontchui,P. 1995 An assessment of edge effect on growth and timber external quality of ayous (Triplochiton scleroxylon K Schum) under Cameroon rain forest conditions. Annales des Sciences Forestieres (Paris) 52(1): 81-88. Murcia, C. 1995. Edge Effects in fragmented forests: implications for conservation. Trends in Ecology and Evolution 10: 58-62. Palik,B.J., Murphy,P.G. 1990. Disturbance versus edge effects in sugarmaple/beech forest fragments. Forest Ecology & Management 32: 187-202. Chapter 6 Closing Remarks 171 Rodrigues, E. 1993. Ecologia de fragmentos florestais ao longo de um gradiente de urbanização em Londrina-PR. Master's Dissertation. Universidade de São Paulo, São Carlos-SP Brasil 110pp. Viana, V.M., Tabanez, A.A.J., and Batista, J.L.F. 1997 Dynamics and Restoration of Forest Fragments in the Brazilian Atlantic Moist Forest. in: Laurance,W.F. and Bierregard, R.O. Tropical Forest Remnants Ecology, Management, and Conservation of Fragmented Communities, Chicago University Press, Chicago 616 pp. Williams-Linera, G. 1990a Vegetation structure and environmental conditions of forest edges in Panama. Journal of Ecology 78:356-373. Yahner, R.H. 1988 Changes in wildlife communities near edges Conservation Biology 2(4):333-339. Chapter 6 Closing Remarks 172