Download Edge effects on the regeneration of forest fragments in south Brazil

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