* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Chapter 2
Reforestation wikipedia , lookup
Introduced species wikipedia , lookup
Ecological fitting wikipedia , lookup
Theoretical ecology wikipedia , lookup
Renewable resource wikipedia , lookup
Plant breeding wikipedia , lookup
Biological Dynamics of Forest Fragments Project wikipedia , lookup
Storage effect wikipedia , lookup
1 The nature of the plant community: a reductionist view 2 3 J. Bastow Wilson Botany Department, University of Otago, P.O. Box 56, Dunedin, New Zealand. 4 5 Andrew D.Q. Agnew Institute of Biological Sciences, University of Wales Aberystwyth, SY23 3DA, U.K. 6 Chapter 2: Interactions between species 7 1 Introduction ................................................................................................................... 2 8 9 10 11 12 13 14 15 16 17 18 2 Interference: negative effects between plants ............................................................... 3 2.1 Competition ........................................................................................................... 4 2.2 Competition and individual ramets ....................................................................... 5 2.3 Factors for which competition occurs ................................................................... 6 2.4 Plasticity and competition ..................................................................................... 8 2.5 Cumulative effects of competition ........................................................................ 9 2.6 Rarity and competitive ability ............................................................................. 12 2.7 Allelopathy .......................................................................................................... 13 2.8 Parasitism ............................................................................................................ 14 2.9 Spectral Interference (red/far-red) ....................................................................... 15 2.10 Pest carriers ......................................................................................................... 16 19 20 21 22 23 24 25 26 27 28 3 Subvention: positive effects between plants ............................................................... 18 3.1 Water flow and nutrient redistribution ................................................................ 18 3.2 Ambient conditions (temperature, evapotranspiration and wind) ....................... 19 3.3 Aerial leakage of nutrients .................................................................................. 20 3.4 Salt ....................................................................................................................... 20 3.5 Shelter from ultra-violet ...................................................................................... 21 3.6 Protection from soil frost-heave .......................................................................... 21 3.7 Hydraulic lift ....................................................................................................... 21 3.8 ‘Talking trees’ ..................................................................................................... 22 3.9 Subvention conclusion ........................................................................................ 22 29 30 31 32 33 34 35 36 37 38 4 Litter: the necessary product ....................................................................................... 22 4.1 The timing and type of litter production ............................................................. 23 4.2 Litter decomposition ........................................................................................... 27 4.3 Effects of litter ..................................................................................................... 28 4.4 The direct mechanical effect of litter .................................................................. 29 4.5 Alteration in environment by litter ...................................................................... 29 4.6 Leachates from litter............................................................................................ 30 4.7 Control of nutrient cycling .................................................................................. 30 4.8 Peat formation ..................................................................................................... 31 4.9 Can reaction via litter evolve? ............................................................................. 33 39 40 41 42 43 44 5 Autogenic disturbance: plants as disturbers ................................................................ 34 5.1 Movement and contact ........................................................................................ 35 5.2 Growth thrust....................................................................................................... 36 5.3 Crown shyness..................................................................................................... 37 5.4 Growth and gravity.............................................................................................. 37 5.5 Lianas .................................................................................................................. 38 45 46 6 Plant-plant interactions mediated by other trophic levels ........................................... 39 6.1 Below-ground benefaction .................................................................................. 40 Wilson & Agnew, chapter 2, Interactions, page 2 of 54 47 48 49 6.2 Pollination ........................................................................................................... 41 6.3 Herbivory ............................................................................................................ 42 6.4 Indirect impacts of diseases................................................................................. 46 50 51 52 53 7 Interactions .................................................................................................................. 47 7.1 Litter-herbivore interactions ................................................................................ 47 7.2 Litter-fire interactions ......................................................................................... 49 7.3 Litter-herbivore-fire interactions ......................................................................... 49 54 55 8 Conclusion ................................................................................................................... 51 56 1 Introduction 57 Plants interact. In chapter 1, we described how plants arrive at a locale and start to 58 develop. In later chapters we shall seek assembly rules – generalised patterns that are the 59 imprint of restrictions in the community assembly process. In this chapter we follow our 60 reductionist approach by considering the plant-plant interactions that lead to those restrictions. 61 Such interactions are often via reactions: modifications of the plant’s local environment. We 62 must beware of assuming that any or certain interactions necessarily occur, but any set of 63 organisms that are in close proximity must affect each other and these interactions can be 64 mediated through many processes. 65 There are classifications of interactions between organisms. Occasionally their labels 66 are easy to apply, though ‘commensalism’ and ‘amensalism’ are found much more frequently 67 in textbooks than in real scientific work. In other cases, application is problematic. Take an 68 experiment on interference. This can be set up to examine a replacement situation, or an 69 additive situation, or both (Fig. 2.1). Monoculture of A A A A B Replacement Comparison (de Wit) Additive Comparison B Mixture A B B A Mixture B Monoculture of B Fig. 2.1. Replacement and additive comparisons in experiments that examine the interaction between species. Wilson & Agnew, chapter 2, Interactions, page 3 of 54 70 Suppose species A has a somewhat stronger competitive ability than species B. If we 71 ask about gains and losses (+ and –) as textbook tables do, the question is gain or loss relative 72 to what? In the replacement comparison (i.e. comparison with monocultures at the same 73 overall density, the de Wit replacement design(define): Fig. 2.1), A will gain and B will lose: 74 a +– situation. Yet in an additive comparison (i.e. comparison in which each species has the 75 same density in monoculture as in mixture), both will suffer competition to some extent: a – – 76 situation. This looks at first sight to be an artefactual distinction of experimental design, but it 77 applies equally to questions in the field. This leads to misunderstandings between animal and 78 plant ecologists, who are really quite different species (though they have been known to 79 interbreed). Animal ecologists think in terms of adding one species to another. Weed 80 ecologists do normally use an additive design, because they are interested in the effect of a 81 weed on a particular density of a crop. However, the de Wit replacement design (Harper 82 1977) is so entrenched in the mind of most plant ecologists that they will expect that in 83 competition one plant will gain and the other lose. So really you are distinguishing between 2 84 types of ecologists, animal ecologists and weed ecologists (who use the same additive design) 85 and plant ecologists who use a replacement design. Of course these are grand generalizations 86 that don’t always hold up, but what bothers me the most is the insinuation that plant 87 ecologists see competition as resulting in one plant gaining and one loosing b/c of their 88 experimental design. Now that’s a truly grand assumption without any backing. 89 2 Interference: negative effects between plants 90 The literature is confused on interference. Generally plant ecologists regard 91 competition as one kind of interference, whilst animal ecologists regard interference as one 92 kind of competition. We follow the precedence of Clements et al. (1929), and the textbook 93 definition of Begon et al. (1996) to give: 94 interference (negative effects) 95 competition = via removal of resources from the environment 96 allelopathy = via toxic substances 97 parasitism = removal of resources directly from the plant 98 spectral interference 99 pest carriers. 100 [Other negative effects are discussed elsewhere, via: litter (sect. 4), autogenic disturbance 101 (sect. 5), other trophic levels (sect. 6) and other types of reaction such as pH modification 102 (chapt. 3).] Interference can occur extremely early in the development of the plant 103 community, as seed-seed interference before the seeds emerge. This could be through Wilson & Agnew, chapter 2, Interactions, page 4 of 54 104 competition for water for imbibition and anchorage by mucilaginous coat, but allelopathy is 105 the more likely cause (Harries and Norrington-Davies 1977; Murray 1998). 106 2.1 Competition We use the definition of Begon et al. (1996): “Competition is an interaction between 107 108 individuals, brought about by a shared requirement for a resource in limited supply, leading to 109 a reduction in the survivorship, growth and/or reproduction of at least some of the competing 110 individuals concerned”. Grime’s (2001) definition very close: “The tendency of neighbouring 111 plants to utilise the same quantum of light, ion of mineral nutrient [or] molecule of water …”. 112 (Grime added competition for space; we have omitted it because we do not believe that it 113 occurs.) Don’t rooted plants need some soil surface on which to exist? Please provide more 114 115 explanation here, or at least state that the topic will be discussed in more detail later in the 116 chapter. 117 You need to either give references or say that you will explain later, or both. 118 Why not? Isn’t space a resource, independent of other resources that plants compete 119 for? 120 121 The subject of competition has occupied most plant ecologists from time to time. The 122 result is a weighty literature and a plethora of ideas, models and theories. We could not do it 123 justice to all this here (though you could cite select references of good synthesis or histories), 124 and we deal only with selected topics. In all species of plant there are more seeds and juvenile 125 plants than adults, which indicates mortality. Some have proposed that there is no observable 126 competition in existing communities, perhaps only the residuum of past competition. We shall 127 discuss this in chapter 6 (sect. 9.2). 128 Competition must be for some limiting resource, such as light, water, CO2, nutrients, 129 radiant heat and, some claim, space. As Clements et al. (1929 pp. 316-317) expressed it: 130 “Competition is a question of the reaction of the plant upon the physical factors that 131 encompass it and of the effect of these modified factors upon the adjacent plants. In the exact 132 sense, two plants do not compete as long as the water-content and nutrients, the heat and light 133 are in excess of the needs of both. The moment, however, that the roots of one enter the area 134 from which the other draws its water supply, or the foliage of one begins to overshade the 135 leaves of the other, the reaction of the former modifies unfavorably the factors controlling the 136 latter, and competition is at once initiated”, and later: “When the immediate supply of a single 137 necessary factor falls below the combined demands of the plants, competition begins”. Note Wilson & Agnew, chapter 2, Interactions, page 5 of 54 138 that he emphasises that all competition is via reaction. No intrinsic competitive 139 characteristics? 140 141 Transitivity in interference networks is discussed in chapter 5. Give more here, at least a definition or leave it out till later. 142 143 144 Transitivity? Definitions of terms not yet dealt with would be helpful 2.2 Competition and individual ramets Competition must occur between individuals, even if our interest is in interspecific 145 interactions. If we follow the Gausian principle of competitive exclusion, two similar species 146 cannot occupy the same niche (Krebs 1972). It follows that the greatest potential for 147 competition should be between individuals with the same resource requirements, and there 148 can be no greater similarity than between individuals of the same species. There is no 149 fundamental difference in the competitive process between motile individual animals and our 150 plants. One owl eats a mouse, so another cannot, and perhaps dies. One plant takes up water 151 from the soil, so another cannot, and perhaps dies. The competition is sometimes ultimate 152 rather than proximate: a lion defends its territory rather than competing directly, but the 153 existence of territorial behaviour is set by evolution in order to defend resources, and the size 154 of territories are probably adjusted in ecological time to the density of resources. Similarly, a 155 plant may not compete so strongly with its neighbours because they are suppressed by its 156 allelopathy or by its harbouring pests/s, and these behaviours may be adaptations (or may not: 157 Newman 1978). One usual difference is that since animals are quite aplastic, competition 158 normally affects densities rather than individual size, but plants are plastic, so that 159 competition affects plant size before density. Thus, the effects of plant competition are best 160 expressed as total biomass (often above-ground, for practical reasons) per area. What 161 happened here? There has been a change in gears with minimal transition 162 Often, some decrease in density does result from competition and this has been 163 generalised as an approach to a straight line on a plot of log (mean individual plant weight) 164 against log (density). The slope of such a line has been justified on theoretical (especially 165 geometric) grounds and on the basis of observations as being any of -3/2, -1/2 or -3/4 (Harper 166 1977; Torres et al. 2001). When competition is for light and is cumulative (sect. 1.1.6 below), 167 there is a particular reason to expect mortality: a dominance hierarchy will result in which the 168 small/short plants lag further and further behind the large/tall ones, until they die (Black 169 1958). When competition is for nutrients and water no such effect will be present (J.B. Wilson 170 1988 %019) so self-thinning should be much less prominent and the -3/2 (etc.) relation may 171 be absent. This has not been investigated, probably because it would be difficult to impose Wilson & Agnew, chapter 2, Interactions, page 6 of 54 172 realistic competition for nutrients or water without there being competition for light. Nutrient 173 availability can moderate the relation caused by competition for light, not changing the -3/2 174 (etc.) slope but changing the density and mean weight per plant at which self thinning starts to 175 take place (Bi 2004, Morris and Myerscough 1991). 176 Shoots generated by vegetative reproduction via rhizomes or stolons are produced at 177 lower densities than those sometimes seen in seed-established populations. Moreover, mutual 178 support through ramets (Marshall and Sagar 1968) should reduce mortality, perhaps 179 eliminating self-thinning. However, self-thinning mortality has been seen in these situations: 180 in the culms of the Gynerium sagitatum (uva grass; Dekroon and Kalliola 1995), in the tufted 181 regeneration of two Sasa spp. (bamboo; Makita 1996) and in shoots of Urtica dioica (nettle; 182 Hara and Srutek 1995). Developmental controls are certainly capable of suppressing bud 183 development, as they do during tree growth, so that we can only presume that its failure in 184 clonal herbs has value, such as in the maintenance of stands in the monospecific condition in 185 spite of accidental death of individual shoots. The above is an awkward paragraph. It’s intent 186 is unclear- what point are you trying to get across.? 187 2.3 188 Factors for which competition occurs Competition by definition occurs for resources, but what are resources? They are 189 molecules or types of energy necessary for plant growth/maintenance, that can be absorbed by 190 one plant or another. This clearly includes light, C (CO2), water, macronutrients (NPK) and 191 micronutrients (Ca, S, Mg, etc.). Even though Clements claimed radiant heat, it seems more 192 like pH, an environmental factor, but how is it different from light? Is soil O2 a resource, 193 because a molecule absorbed by one plant means it is not available for another? But it is 194 related to redox potential, which seems more like an environmental factor. These matters are 195 difficult at the edges. 196 Light is the resource for which competition has most often been analysed, and we 197 shall refer to it several times. It differs from most other resources in that it is available 198 instantaneously and disappears if not used; it is also directional in source (e.g. sect. 2.5). 199 Competition for mineral nutrients, mainly N, P and K, is well established by experiments in 200 pots and in the field (Clements et al. 1929; Wilson and Newman 1987). The main issue is that 201 because N is more mobile in soils than P it is possible for two plants to come into competition 202 for N before they compete for P (Bray 1954). The mobility of K is intermediate 203 204 205 I’m not seeing the hierarchical arrangement of interference versus competition. It seems that they could easily be flip flopped, according to the earlier definition of each. The rôle of competition for water between trees and their understorey is well known, Wilson & Agnew, chapter 2, Interactions, page 7 of 54 206 particularly in savannah. This can be two-way, the trees reducing the growth of the 207 understorey and vice-versa (Knoop and Walker 1985). Anderson et al. (2001) found that 208 although the canopy of Prosopis glandulosa (mesquite) subvents the survival of saplings 209 below the canopy, competition for water develops as they age. Robberecht et al. (1983) found 210 that in the Sonoran Desert removing all other plants surrounding a tussock of Hilaria rigida 211 increased the water potential which led to the plants remaining green for longer into the dry 212 reason and growing more. Perez-Salicrup and Barker (2000) found that after lianas had been 213 removed from seasonally-dry tropical forest in Brazil the water potential of trees increased 214 and they grew faster but the authors did not discuss how much of this was due to the obvious 215 increase in light due to removal of the liana canopy. Like mineral nutrients but in contrast to 216 light soil water remains available for some time if not absorbed, though not indefinitely. 217 Competition could theoretically occur for the carbon source, CO2. Plants can reduce 218 the CO2 levels around themselves through photosynthesis, and this could affect the growth of 219 neighbours (Oliver and Schreiber 1974). However, there is also production of CO2 by 220 respiration of soil and litter (Bazzaz and McConnaughay 1992) and considerable air mixing 221 (Reicosky 1989). This means depletion of CO2 within a canopy may be uncommon, but it has 222 been recorded from dense canopies. Bazzaz and McConnaughay (1992) recorded a decrease 223 by 9 ppmv (parts per million by volume) compared to that above the canopy in Abutilon 224 theoprasti (velvet-leaf) and Buchmann et al. (1996) recorded a decrease of up to 26 ppmv in 225 daytime in the understorey of temperate forests with an open tree canopy and a dense field 226 layer. Species do differ greatly in their effects on the CO2 regime, e.g. Reicosky (1989) found 227 daily minimum CO2 levels differed between crops Zea mays (maize) and Glycine max 228 (soybean) by 5 ppmv. Most species show growth responses to CO2 concentration over this 229 range (Hunt et al. 1991). Competition for CO2 is probably more important than has usually 230 been assumed. Grime (2001) did not list CO2 as one of the resources for which competition 231 would occur, though Clements et al. (1929) did. 232 Temperature is a rate-regulator, not a resource, although Clements et al. (1929) 233 suggested that plants may compete for radiant heat. They also suggested that plants could 234 compete for soil air in waterlogged conditions. 235 Ecologists often write of competition for physical space. When they use ‘space’ as 236 shorthand for all resources this is justifiable, though it misleads the young and impressionable 237 and obscures the differences between resources in their mechanisms of competition. 238 However, sometimes it is made clear that space is seen as a resource separate from light, 239 water, NPK, etc., e.g. “competition is defined as ‘the tendency of neighbouring plants to 240 utilize the same quantum of light, ion of a mineral nutrient, molecule of water, or volume of Wilson & Agnew, chapter 2, Interactions, page 8 of 54 241 space’” (Grime 1979, 2001); “weeds competing for light and space in the first year of growth, 242 rather than moisture or nutrient stress” (Sage 1999). Grist (1999) attempted to model “plant 243 competition for light and space”. Yodzis (1986) envisaged that: “competition for space is so 244 different from what we normally think of as consumptive competition that it makes more 245 sense … to think of it as a completely different category of competition. Certainly space is 246 quite different from any other resource”. In these cases space should be equivalent to volume, 247 though we suspect that most authors have soil surface area in mind. 248 As far as spatial volume is concerned, Chiarucci et al. (2002) measured the percentage 249 volume occupancy of eight plant communities (four in each in each of Italy and NZ, 250 comprising four grasslands and four shrublands). In spite of all the assumptions and 251 speculation in the literature, this had never before been measured. They found that only 252 0.44% to 2.89% of the available volume within the canopy was occupied by plant tissue and 253 concluded: “physical space is probably never limiting by itself in terrestrial higher-plant 254 communities, so that competition for space, distinct from competition for resources such as 255 light, water and nutrients, is not likely to exist”. Clements (1916, 72) understood this: “In a 256 few cases, such as occur when radish seeds are planted too closely, it is possible to speak of 257 mechanical competition or competition for room. … However [this] seems to have no 258 counterpart in nature. There is no experimental proof of mechanical competition between 259 root-stocks in the soil, and no evidence that their relation is due to anything other than 260 competition for the usual soil factors – water, air and nutrients.” Actually, there is no evidence 261 that it occurs between close radishes. Even Tilman (1982), though devoting a chapter to 262 competition for space, is careful to note that physical space “may be irrelevant”. He notes that 263 disturbance can create open space, but that “it would seem better to study explicitly the 264 resources supplied by disturbance”. We agree. The attempt by neighbouring plants to use the 265 same space can result in contact and damage, but this is not competition, it is autogenic 266 disturbance (sect. 5 below). Isn’t it true that plants REACT to each other in competition for 267 space? The example of vines keeps coming to mind, where different species wrap around each 268 other, indirectly searching for resources. However, the growth itself is a competitive activity, 269 in search of space. 270 2.4 Plasticity and competition 271 We must remember that the competitive abilities of the plant are determined not by its 272 characters in monoculture but by those changed by plasticity in competition. This can be seen, 273 and is well-known, above ground, for example etiolation, an increase in height as a result of 274 shade (Bradshaw 2006). Collins and Wein (2000) found that two Polygonum species Wilson & Agnew, chapter 2, Interactions, page 9 of 54 275 (arrowleaf tearthumb and smartweed) increased in internode length when crowded by plants 276 of the same two species. Plasticity occurs in less visible characters too. For example, Huber- 277 Sannwald et al. (1996) found that two steppe grasses differed markedly in their plastic change 278 of specific root length. Bookman and Mack (1982) found by double-labelling of shoots that 279 when Bromus tectorum (downy brome) was growing in competition with Poa pratensis 280 (meadow grass) its root system was constricted laterally, but extended slightly deeper. 281 2.5 Cumulative effects of competition 282 The term ‘asymmetric competition’ (with the more general ‘asymmetric interference’) 283 is used in two senses. Firstly, it can mean that the interference effect of Species A on Species 284 B is different from that of B on A. Perhaps A suppresses B, but B has no effect on A, or the 285 difference can be quantitative: more effect of A on B than of B on A (e.g. Roxburgh and 286 Wilson 2000 %395; Ives and Hughes 2002 %388). This would be expected since species by 287 definition differ in characters and their interference effect is therefore very likely to be 288 different, but the degree of A/B asymmetry can be of interest as a feature of the community 289 matrix and hence of the community (chapt. 3, sect. 7.3 below). This concept can be applied to 290 individual genotypes too. I’d like to see more on plasticity (phenotypes being expressed or 291 repressed, mechanisms of plasticity, etc.) 292 The second meaning of ‘asymmetric competition’ is not so easy to define. Wedin and 293 Tilman (1993 %199) synonymise it with ‘resource preemption’, apparently implying an 294 advantage to the first plant to arrive. For Schwinning and Fox (1995 %432) it implies that 295 “large plants greatly suppress the growth of smaller neighbors”, and more precisely for 296 Weiner et al. (1997 %85) that “larger plants are able to obtain a disproportionate share of the 297 resources (for their relative size) and suppress the growth of smaller individuals”. It has 298 therefore been termed ‘size asymmetry’ to distinguish it from A/B asymmetry. The problem is 299 what is a plant, so what is size? If a large grass tiller divides into two small, physiologically- 300 independent tillers, is it no longer large? (that depends on the question you are asking) Many 301 authors have reached the conclusion that initial differences are magnified into a large effect 302 on the long-term outcome of competition, so that the process favours competitive exclusion. 303 There is truth in this, but it is too superficial because it ignores the mechanism. 304 The classic study is that of Black (1958). He established swards of Trifolium 305 subterraneum (subterranean clover), using large seeds (mean mass 10.0 mg) and small seeds 306 (mean mass 4.0 mg), all of cultivar Bacchus Marsh and therefore of near-identical genotype. 307 In monoculture, the two sizes of seed and a 50:50 mixture of them all produced essentially 308 identical sward mass throughout the experiment. However, in the mixture the initial seed Wilson & Agnew, chapter 2, Interactions, page 10 of 54 309 mass was crucial: plants from large seeds suffered no mortality and came to be 93% of the 310 mixture biomass, whereas more than half of the plants from small seeds died and those 311 remaining contributed only 7% of the biomass. The distribution of leaf area showed the cause: 312 plants from small seeds held their leaf area a little lower than those from large seeds. In 313 monoculture, this did not matter, but in mixture they captured a small share of the resources 314 (disproportionately small for their relative leaf area), so that by the end of the experiment, 315 although they contributed 10% of the leaf area it was held so low that they captured only 2% 316 of the light. It is essential to notice that this result was obtained because of a specific 317 mechanism (Fig. 2.2): (a) in T. subterraneum greater growth results in the leaf area being held 318 higher, and (b) in competition for light the height of the leaf area is crucial. This gives 319 cumulative competition: the results of competition change the competitive abilities. In this 320 context, ‘height asymmetry’ would be more appropriate than ‘size asymmetry’. Larger individual leaves Longer petioles Greater biomass Leaf area held higher Faster growth (RGR) 321 322 323 Disproportionate percentage of the light captured Fig. 2.2. In Trifolium subteranneum, the results of competition affect competitive ability. It cannot be assumed that competition for light will always be cumulative. Hirose and 324 Werger (1995 %466) found that in competition for light the taller species of a wet meadow 325 community did not have as much advantage as expected from the higher position of their 326 leaves in the canopy because in order to achieve that height they had less of their biomass in 327 leaf area and more in stems, and Bernston and Wayne (2000 %1072) to failed to find biomass 328 size-asymmetry in competition for light between Betula alleghaniensis (yellow birch) 329 seedlings, apparently perhaps because of height plasticity in small, shaded plants. 330 This effect would not be expected in below-ground competition, because there is no 331 common equivalent to the growth/height/light/growth positive feedback. A plant that is larger 332 with longer roots, and therefore able to access deep resources generally has no advantage?!!. 333 The deep roots are hardly likely to ‘shade’ roots nearer the surface from NPK rising up Wilson & Agnew, chapter 2, Interactions, page 11 of 54 334 through the soil. In any case, nutrients are generally much more available near the surface, 335 where small plants can easily root. Uptake of water is sometimes from aquifers, but it is hard 336 to envisage a root shield comparable to the canopy aboveground. In terms of allometric 337 growth, if a plant is larger by ×2 in each dimension, it will be ×23 = ×8 larger in volume and 338 in biomass. Therefore, its NPK requirements will be ×8. However, its root system will cover 339 an area only ×22 = ×4 larger to cope with the ×8 demand. The large plant will be the one 340 deficient in nutrients. Why are you looking at the above ground plant as a 3D entity and the 341 below ground plant as a 2D entity? Seems inconsistent given that plants occupy a soil volume 342 not a soil plane.Thus, it seems reasonable that Rajaniemi (2003) could find only very 343 ambivalent evidence of size-asymmetry in below-ground competition, and all other 344 investigations have found none. In fact the larger plants have usually been at a disadvantage, 345 due to greater within-plant competition (Wilson 1988 %019; Weiner et al. 1997 %085). So if 346 I were to stop here, I could maybe extrapolate as to how some of these characteristics may 347 extend to plant community characteristics… 348 Size asymmetry seems very subjective. 349 We could think of possible exceptions. If nutrients are available patchily, small plants 350 can grow in high-NPK areas, whereas a large plant might need to include in its root system 351 the nutrient-poor matrix, with actual disadvantage to the large plant. However, if nutrients 352 were available intermittently in time and space, such as animal urination patches, perhaps a 353 large plant could take up enough to satisfy the whole of itself from a few roots in the patch, 354 whereas a small plant outside the patch could not reach in (Fig. 2.3). This argument applies 355 only when the patches are transient, so it seems reasonable that Blair (2001 %199) could find 356 no evidence that spatial heterogeneity in nutrients led to size-asymmetric competition. 357 If nutrients are patchy, then all plants have a roughly – chance of landing 358 (germinating) in a high nutrient patch. Larger plants though have an increased chance of 359 capturing a high nutrient patch within their root-soil-volume. Therefore the large plant will 360 be more likely to have access to a high nutrient area than a small plant. The large plant does 361 have to support roots in low nutrient areas, but once a high nutrient area has been 362 encountered, the plant can allocated increased root growth into that soil volume. Wilson & Agnew, chapter 2, Interactions, page 12 of 54 This plant has some roots in the sheep-urine patch, and can access nutrients from there for the whole plant A sheep micturated here This plant has access This plant has no access Fig. 2.3: A possible mechanism for size-asymmetric competition belowground, when nutrient supplies become available patchily in time and space, e.g. from sheep micturition. 363 364 We conclude that cumulative competition, due to size-asymmetry, i.e. the effect of 365 initial advantage, is not a necessary feature of plant competition. It will occur only when there 366 is a specific feedback mechanism, such as that via height/light, that results in competitive 367 success increasing competitive ability per unit biomass. The conditions under which it occurs 368 have huge implications for community structure, because if competition be cumulative 369 competitive exclusion will occur much more readily, and faster, and coexistence less readily. 370 2.6 Rarity and competitive ability 371 It is often suggested that plant species become rare because they are poor competitors, 372 but the evidence here is mixed. Walck et al. (1999) found that low competitive ability may be 373 one factor contributing to the narrow endemism of the dryland species Solidago shortii 374 (goldenrod) in comparison with the widespread S. altissima, which can overtop S. shortii in 375 more mesic sites but it is simplistic to interpret a two-species comparison in terms of general 376 trends in rarity. Lloyd et al. (2002) working in New Zealand found that there was no general 377 relation between competitive ability and rarity at either high or low levels of soil fertility for 378 ten species in the rosaceous genus Acaena (biddy bid), but in the grass genus Chionochloa 379 (snow tussock) common species tended to be stronger competitors than rare species, and this 380 was significant at the low level of fertility—in some sense, either via initial 381 advantage/competition or through some sort of competitive mechanism, isn’t it inevitable that 382 common species have competed their way to their place?. Several of the common 383 Chionochloa species have expanded their geographic ranges since anthropogenic forest 384 removal, consistent with their greater competitive ability. Conversely, rare Chionochloa Wilson & Agnew, chapter 2, Interactions, page 13 of 54 385 species occupy specialized habitats that may be interpreted as refuges from competition—is 386 this the competition v. ruderal v. disturbance idea?. However, it is too simplistic to speak of a 387 plant being a generally strong or weak competitor. Apart from ruderals, every species is a 388 good competitor in the sites that it inhabits, at least a good enough one to persist. Hubbell et 389 al. (2001 %759) reported a curious relationship between rarity and neighbourhood effects in 390 neotropical forest, such that the survival by rare trees was more adversely affected by 391 conspecific neighbours than that of common species. One explanation is that the rare species 392 are using rare resources, or that they are limited by specific pests (chapt. 4, sect. 3 below) 393 This subject was covered to quickly. It would be useful to explain how rarity could increase 394 the fitness of those individuals left, and why this is useful in defining a plant community. 395 2.7 396 Allelopathy Allelopathy is another type of interference. Unlike competition, it does not involve a 397 “shared resource in limited supply” (sect. 2.1 above). Rather, a toxin produced by plants is 398 either leached or volatilised from their living parts, above- (Inderjit et al. 1999) or below- 399 ground (Kato-Nogudi 2004), from standing dead plants (Gliessman and Muller 1978), or from 400 the breakdown of their litter (e.g. Kuiters 1991; Sharma et al. 2000), and then reduces the 401 growth of neighbouring plants. Some ecologists believe that allelopathy is a widespread and 402 important process (Rice 1983; Inderjit et al. 1994). Others are sceptical (Harper 1977). Most 403 plants contain chemicals that are toxic to other plants, at least in the right circumstances. It is 404 very unlikely that the strong effects that can readily be demonstrated by leaf elutants in petri 405 dishes will happen in nature, but it is equally unlikely that soil bacteria and colloids will 406 completely neutralise the effect, and field experiments are difficult to perform (Williamson 407 1990). 408 Most work has been with shoot elutants, but Mahall and Callaway (1991 %2145; 1992 409 %874) conducted greenhouse experiments between Larrea tridentata (creosote bush) and 410 Ambrosia dumosa (white bursage). The two species occur together in the Mojave Desert. 411 Roots of A. dumosa were inhibited when they came near to roots of L. tridentata. This effect 412 was reduced by activated carbon, so it was presumably via some chemical. The two species 413 apparently have similar water extraction capabilities (Yoder and Nowak 1999b), so this 414 signalling would enable to the plants to avoid each other and postpone the onset of 415 competition. This may be a more common phenomenon than we have knowledge of at 416 present. More direct evidence for the production of toxins from roots was obtained by 417 Welbank (1963), but only after they had started to decompose in anaerobic conditions. But 418 there are always decomposing roots around a living plant and they may indeed create Wilson & Agnew, chapter 2, Interactions, page 14 of 54 419 anaerobic conditions in the immediate vicinity of the decomposing root through their own 420 decomposition or the uptake of O2 by neighbouring roots. Perhaps then it isn’t unreasonable 421 to look at this as a source of these compounds. 422 We are inclined to think in teleological terms, of plants deliberately inhibiting their 423 neighbours. Although we know this is not true, it leads to an assumption that allelopathic 424 ability has evolved as an adaptation: species that make their locale unfavourable for the 425 growth of other species will gain in selection. But can this evolve? If a mutant produces an 426 allelochemical against other species this will be at some cost to itself (many of the secondary 427 compounds involved are expensive for the plant to make), yet unless the effect is very local it 428 will benefit non-mutant plants of the same species as much as itself—but isn’t it often very 429 local? Especially in ecosystems where, e.g. the Janzen scatter hypothesis holds, species’ 430 individuals tend to spread out anyway?, and the trait will not evolve. Why shouldn’t the 431 allelopathic chemical affect non-mutants of the same species? 432 We then have to invoke group selection, is a controversial and probably weak force 433 (Wilson 1987 %493). This problem would not occur with selection for a plant to protect itself, 434 and thus eventually its species, against its own allelochemical. Williamson (1990) commented 435 that there would be equal selective pressure on the species’ neighbours to develop defences, 436 but this ignores the issue that a species always grows with itself, but has a variety of 437 neighbours against which it would have to protect itself. Newman (1978) questioned whether 438 selection was involved at all, finding in a literature survey that auto-toxicity is as common as 439 toxicity to other species. He argued that this indicates that allelopathy is an accidental result 440 of the production of secondary compounds. 441 442 We discuss the particularly complex allelopathic interactions with litter in section 3.5 443 below. Allelopathy seems more of a competitive trait than simply an interfering trait . 444 Allelopathy contributes to the competitive ability of an organism, right? So, how are 445 interference and allelopathy not functions of competition? 446 447 448 2.8 Parasitism Plant parasites are commoner in the dry tropics than in temperate climates. There are 449 partial shoot parasites (Loranthaceae, Viscaceae) and total shoot parasites (e.g. Cuscuta, 450 Cassytha) as well as partial (e.g. Rhinanthus spp.) and total root parasites (e.g. 451 Balanophoraceae, Orobanchaceae, Rafflesiaceae, Striga gesnerioides). The immediate effect 452 of a parasite is to depress the growth of the hosts. Infestation by some root parasites can Wilson & Agnew, chapter 2, Interactions, page 15 of 54 453 reduce crops by up to 70% (Graves et al. 1992; Press 1998). There is often an increase in 454 species diversity (Press 1998). This may be because the parasite’s favourite hosts are the 455 dominant species, and a reduction in their productivity allows subordinate species to flourish, 456 possibly also reducing cumulative competition (sect. 1.1.4 above). It is not clear why parasites 457 should favour the dominants, but perhaps we see the effect more when they do. On the other 458 hand, the parasite could reduce the productivity of the whole community from high to 459 medium, allowing more species to coexist if the humped-back theory of Grime (1973; 2001) 460 is correct. Situations, however, can be very complex when plant hosts, their parasites and their 461 herbivores are examined. Marvier (1996; 1998) compared aphid performance on the parasite 462 Castilleja wightii (paintbrush) in host plants differing in N content, which was highest in a 463 legume (Fabaceae) and lowest in a composite (Asteraceae). Aphid populations did best on the 464 parasite when it depended on the legume host and worst when the host was the composite, 465 leaving the C. wightii worse off on the legume, in spite of the latter’s higher nitrogen content. 466 These interactions differentially affect the host plant species and their parasite, modifying 467 competitive ability, but are complicated by the parasite’s ability to better its performance by 468 mixing its hosts. 469 2.9 Spectral Interference (red/far-red) 470 Plant canopies change the spectral quality of light that passes through them. Of 471 particular importance for plant-plant interaction is that the red/far-red ratio (R/FR) is lower in 472 light that has passed through a leaf or been reflected off one. It is therefore lower under 473 canopies and somewhat lower in canopy gaps than in the open (Turnbull and Yates 1993; 474 Davis and Simmons 1994). This can decrease the leaf chlorophyll content of plants in the 475 lower strata and hence reduce their photosynthetic capacity, leaf life, leaf area ratio, tiller 476 production and relativeh growth rate (Barreiro et al. 1992; Dale and Causton 1992; Skinner 477 and Simmons 1993; Rousseaux et al. 1996), and decrease seed germination (Batlla et al. 478 2000, Kasperbauer 2000). 479 These responses are mediated by phytochrome (Gilbert et al. 1995). However, the 480 response depends on the species. Leeflang (1999) found that only one species out of the six 481 that they examined responded to R/FR ratio in terms of biomass accumulation. Dale and 482 Causton (1992) found differences between three Veronica species (speedwell): the species 483 that grew in deeper shade responded more to R/FR. It can be difficult to separate effects of 484 light intensity and spectrum, but sometimes plants respond morphologically to the presence of 485 neighbours even before they are directly shaded, by sensing the change in R/FR in light that 486 has been reflected off neighbours (Schmitt and Wulff 1993). Wilson & Agnew, chapter 2, Interactions, page 16 of 54 487 R/FR may be less important for its direct effect on the growth of the recipient than as a 488 signal to the plant that competition is imminent, allowing it to respond by either competition 489 tolerance or competition avoidance. (Murphy and Briske 1994; Stuefer and Huber 1998). For 490 avoidance, R/FR can induce plastic changes which would allow the plant to reduce shading 491 by its neighbours, e.g. enhancing leaf elongation, making shoots and leaves more upright, and 492 increasing shoot:root ratio (Skinner and Simmons 1993; Vanhinsberg and Vantienderen 493 1997). The R/FR signal can also promote plant foraging for light gaps away from competitors, 494 through altered stem angle and longer internodes (Ballaré et al. 1995). Some plants sense and 495 grow away from each other, even at some distance. Manipulation of R/FR ratios, and 496 experimentation with pieces of plastic with the same spectral characteristics as plant leaves, 497 show this to be due to R/FR effects (Fig. 2.4; Novoplansky 1991). Germination can be 498 reduced under a canopy, but faster germination is also possible, and these R/FR responses can 499 be seen as competition avoidance and greater competitive effort, respectively (Dyer et al. 500 2000). R/FR is not everything: Muth and Bazzaz (2002) found that visible (PAR) light was 501 more important than R/FR in the gap-seeking of Betula papyrifera (birch) seedling. Experimental setup The relative frequency of shoots developing in different directions Fig. 2.4: The effect of R/FR light, as seen in surrounds of different colours. After Novoplansky (1991). 502 503 504 2.10 Pest carriers Plants can interact at a local scale using pests as weapons. Van den Bergh and Elberse 505 (1962) found that under a high nutrient regime Lolium perenne at first outcompeted 506 Anthoxanthum odoratum, but eventually A. odoratum transmitted a virus to L. perenne and 507 suppressed it. There is evidence that the invasion of the exotic Avena fatua (wild oat) into 508 Californian grasslands has been due to its increasing the incidence of a virus in the native 509 tussock grasses, severely reducing their growth, though the effect may be indirect, due to its 510 increasing the abundance of the aphids that are vectors of the virus (Malmstrom et al. 2005). Wilson & Agnew, chapter 2, Interactions, page 17 of 54 511 A similar situation may exist in California where preliminary evidence suggests that 512 Umbellularia californica (Bay laurel) is infected with the fungus Phytophthora ramorum, but 513 is little affected. However, many spores are produced and these cause the death of nearby 514 Quercus spp. (oaks) and Lithocafvrpus densiflora (tannoak) (Mitchell and Power %Collinge 515 58). We list these examples of species as pest carriers separately, though they grade into 516 plant-plant interactions moderated by diseases (sect. 4.4, below). 517 Wilson & Agnew, chapter 2, Interactions, page 18 of 54 518 519 3 Subvention: positive effects between plants 520 Just as the plant community provides a diversity of habitats for heterotrophic 521 organisms, so it augments the diversity of sites available for plant species by increasing 522 environmental heterogeneity both horizontally and vertically. One logical result of this 523 reaction (sensu Clements) should be subvention: positive effects by some species on the 524 survival and/or growth of others. Gleason (1936) gave subvention equal weight with negative 525 interference, but whilst the interference between species has been a focus of attention 526 subvention has been neglected until recently (Lortie et al. 2004). Subvention changes during 527 ecesis as plants grow and change shape, so that repeatable patterns of interaction are difficult 528 to find. Usually two species are not absolutely dependent on each other for survival. 529 Moreover, the existence of a subvention by species A on species B does not imply that they 530 have co-evolved. It is simply a corollary of reaction. The terminology of subvention is 531 confusing. We use the scheme: 532 subvention 533 mutualism (= synergy) = both species benefit relative to their being at the same 534 density on their own (i.e. an additive comparison). 535 benefaction = one species benefits as above, with no known advantage or 536 disadvantage to itself 537 facilitation = one species benefits another, to its disadvantage 538 The term ‘facilitation’ was used thus by Connell and Slatyer (1977) for Clements’ (1916) 539 concept of the mechanism of succession, but it is appropriate here since the mechanisms are 540 the same. We consider here the various environmental features which are modified by plants 541 to the benefit of their own, or other species. The list is a long one, and we are aware that many 542 potential effects have been left out. The review by Callaway (1995) discusses more examples 543 than are cited here. 544 3.1 Water flow and nutrient redistribution 545 Plants intercept rainfall. Up to half of the input can be lost by evaporation from the 546 canopy (Cape et al. 1991), but input is occasionally increased due to fog capture (chapt. 3, 547 sect. 5.4.B below). The distribution of soil water input is made far more patchy, for example 548 through stemflow amongst many interactions. Nutrients can be made even more patchy 549 because of leaf leaching. Stemflow can be an important source of nutrients for epiphytes, 550 especially of nitrogen for which stemflow concentrations can be more than ten times those in Wilson & Agnew, chapter 2, Interactions, page 19 of 54 551 rainwater (Awasthi et al. 1995; Whitford et al. 1997). This may explain the trunk-base habitat 552 of a few mosses such as Isothecium myosuroides in Western Europe. This induced patchiness 553 in nutrients is a good example of benefaction without there being any question of co- 554 evolution. Benefaction, by its nature, would not seem to involve co-evolution. It is confusing 555 to bring up the possibility of co-evolution in this way when talking about benefaction. 556 It might be better to leave it out or firm up the statement. In the clustered or striped vegetation 557 (‘tiger bush’) typical of arid and semi-arid areas (chapt. 3, sect. 5.4.B below), there can be 558 similar benefaction by shrubs and trees of their understorey. For example, litter can increase 559 infiltration into the soil by preventing runoff in semiarid areas (Tongway et al. 2001). 560 However care must be taken in interpreting these patterns. Walker et al. (2001) demonstrated 561 by plant manipulations that whilst islands of soil fertility around woody plants can increase 562 water and nutrient availability, the net effect of nurse plants can be negative due to shading 563 and/or root competition. 564 3.2 565 Ambient conditions (temperature, evapotranspiration and wind) Plants alter the physical environment around and below them in obvious ways: 566 ambient temperature, evapotranspiration and air movement. This is reaction. Canopy trees 567 dampen diurnal and seasonal fluctuations in subcanopy air temperature (Stoutjesdijk and 568 Barkman 1992; Chen et al. 1993) and this is significant because temperature stresses of heat 569 load or frost can kill cells. Daytime temperature in the field layer of a forest is generally 1- 570 3 oC less than above the canopy. Windspeed is lower, reducing transpirational load and the 571 abrasion of photosynehetic organs (sect. 3.3 below). Humidity is higher, reducing water 572 stress. Within one grass stratum, Liancourt et al. (2005) gave evidence of benefaction between 573 species in a calcareous grassland due to amelioration of water stress. Plant species differ in 574 their reaction on the microclimate (Castro et al. 1991) and in their response to microclimate 575 (Stoutjesdijk and Barkman 1992; Tretiach 1993). In extreme environments the presence of 576 one plant can allow another to survive. 577 In alpine and arctic habitats krummholz trees can shelter a shorter species from wind 578 desiccation, storm damage and wind-born ice particles (Marr 1977, Callaway 1998). Similar 579 subvention can occur amongst alpine herbs, possibly as a mutualism (e.g. Choler et al. 2001). 580 In arid and semi-arid areas tree shade has an obvious associated flora. Extensive 581 woody vegetation with isolated canopies can be an essential part of system function in warm 582 semi-arid climates, and often fine soil and organic matter accumulate below the canopy, 583 bearing a herbaceous flora. The association between nurse plants and their beneficients is 584 often so clear that no attempt is made to determine the precise environmental variable that is Wilson & Agnew, chapter 2, Interactions, page 20 of 54 585 responsible, but it is generally assumed to be temperature combined with humidity (e.g. 586 Brittingham and Walker 2000). Temperature differentials are easy to demonstrate and very 587 well documented (e.g. Suzán 1996; Godínez-Alvarez et al. 1999), and in hot deserts with 588 summer rainfall cell death by high temperatures can be a critical factor in survival. This is 589 particularly important for succulents, which can suffer from heat load in full sun, being unable 590 to lose as much heat by convection as thin-leaved plants. Water loss is also a problem for 591 plants. Under shrub canopies, the litter ameliorates soil processes, increasing water-holding 592 capacity. Shrub nurse effects are sometimes attributed to nutrient buildup but Gómez- 593 Aparicio et al. (2005) found that in dry Spanish shrublands most of the effect was due to 594 canopy shade. Belsky et al. (1993 %143) have meticulously described the complex of 595 interactions in East Africa, while Holzapfel and Mahall (1999) carefully separated subvention 596 from interference between Ambrosia dumosa (white bursage) shrubs and annuals in the 597 Mojave. Amiotti et al. (2000) show that isolated Pinus radiata trees in Argentina have similar 598 effects. In a fascinating series of subventions in western Texas, Prosopis velutina (mesquite) 599 canopy allows the establishment of a Juniperus pinchotii (juniper), while J. pinchotii helps 600 the establishment of three other woody species with a more mesic distribution (Armentrout 601 and Pieper 1988; McPherson et al. 1988). Presumably there are more mesic conditions of 602 temperature and water relations under these canopies. 603 These are good examples of subvention that do not seem to be the result of co- 604 evolution. 605 3.3 Aerial leakage of nutrients 606 Nutrients such as nitrogen can be leached from canopy leaves and absorbed by the 607 leaves of other plants below (McCune and Boyce 1992; Cappellato and Peters 1995). This 608 interaction is difficult to categorise. Presumably one organism loses and the other gains, 609 almost by definition, as in predator/prey and parasite/host relations, except that there would 610 be leakage from the loser whether there was a recipient below or not. It seems best 611 categorised as benefaction. 612 3.4 613 614 Salt A plant could shelter another from salt spray deposition (Malloch 1997). It could also increase its salt spray load. Neither effect seems to have been demonstrated. 615 Bertness and Hacker (1994 %363) showed, by removing Juncus gerardii (a rush) from 616 a New England, USA, saltmarsh showed that its presence had raised the redox potential of the 617 soil and, by reducing evapotranspiration, kept down the soil salinity. Without this benefit, the Wilson & Agnew, chapter 2, Interactions, page 21 of 54 618 associated Iva frutescens (marsh elder) died. Transplants of J. gerardii to the low and mid 619 zones of the marsh also benefited from the presence of J. gerardii and/or I. frutescens 620 neighbours because of lower salinities. However, transplants into higher zones in the marsh 621 were suppressed by the I. frutescens there, so this is a one-way subvention, benefaction rather 622 than mutualism. 623 3.5 624 Shelter from ultra-violet In some species, ultra violet light (UV-B) damages cells and their photosynthetic 625 apparatus, so we might expect that plants could exhibit benefaction by UV-B shielding. This 626 could be stronger in alpine communities, with higher UV-B. The transmittance of UV-B to 627 lower strata differs with the canopy structure of different species (Shulski et al. 2004). Species 628 and ecotypes react differently to UV-B in growth rates (Dixon et al. 2001; Robson et al. 629 2003). All the components for a benefaction seem to exist, but we cannot find any evidence 630 that it exists in nature. 631 3.6 632 Protection from soil frost-heave Ryser (1993) found that, of six species examined in limestone grassland in 633 Switzerland, two appeared to require shelter from neighbouring plants against frost heave as 634 well as drought. 635 3.7 636 Hydraulic lift This is the process by which one species takes up water from deep layers in the soil 637 and releases it in the surface layer. In arid climates, woody rheatophytes with deep roots in an 638 aquifer could benefit species with shallow roots by this mechanism. Indeed, Facelli and 639 Temby (2002) show that Atriplex vesicaria and Maireana sedifolia (both Chenopodiaceae) 640 enhance the water relations of their surrounding annuals. However, they can also compete for 641 light, leading to a complex interaction. In a nice variant of the story, Yucca schidigera in the 642 Mojave Desert, with CAM photosynthesis, apparently redistributes soil water by creating a 643 higher upper-soil water potential during the day, rather than at night when it has open stomata 644 and a higher water demand (Yoder and Nowak 1999 %093). Surrounding non-CAM plants 645 with a high water demand must benefit. The phenomenon is also known in temperate forest 646 where Dawson (1993) and Emerman and Dawson (1996) have shown that Acer saccharum 647 (sugar maple) uplifts considerable quantities of water, much of which may be taken up by 648 understory plants from near the exuding tree roots. This phenomenon is benefaction or 649 possibly facilitation, and may be much more common that formerly supposed (Caldwell et al. 650 1998), but most authors stress that soil particle size and structure can be critical in the process. Wilson & Agnew, chapter 2, Interactions, page 22 of 54 651 Roots can also distribute water to deeper soil layers (Ryel et al. 2003). 652 3.8 653 ‘Talking trees’ Plants can often increase their content of defence compounds when grazed (see 654 below). It has been suggested that they can do this upon receiving a chemical signal such as 655 methyl jasmonate released by nearby grazed plants: the ‘talking trees’ mechanism. Although 656 early work was plagued by problems such as pseudoreplication, careful work has since 657 validated the effect (Karban et al. 2000). The remaining puzzle is whether it can be a product 658 of natural selection since it benefits the neighbours of a plant (genotype) producing the signal, 659 not the plant itself (Strauss et al. 2002). It could have evolved by kin selection (Wilson 1987 660 %493) or between-plant signals could be an accidental result of within-plant signalling 661 (Karbon and Maron 2002). 662 3.9 663 Subvention conclusion One of the categories above is mutualism, and examples are well known between 664 plants and heterotrophs, for example in mycorrhizae. However, we have been able to discover 665 any examples of demonstrated mutualisms between two embryophytes. Awk sentence We 666 also note that the list above contains no clear example of facilitation. This suggests that 667 subventions may be accidents of evolution. We would expect this anyway. Plants do not rise 668 to the top of the interaction league table by being nice to others. Probably facilitation during 669 succession is a combination of mechanisms, for example an N-fixing and litter-accumulating 670 pioneer that is shaded out by subsequent species. 671 4 Litter: the necessary product 672 As we described in chapter 1 (sect. 1.1) the fundamental pattern in the natural history 673 of plants is that growth comprises the production of modules. The functional lifespan of these 674 organs, such as leaves and translocatory (vascular) tissue, is programmed and curiously short. 675 New organs and tissue must be produced, and discard of the old ones results in the production 676 of litter. The organs are mainly shed when dead, but some material is lost to the living plant 677 through damage while it is still living. All types of plant organ are included in litter, 678 especially leaves but also twigs, coarse woody debris (CWD, i.e. branches and whole trees) 679 and reproductive parts. Litter from roots and underground stems is deposited below ground 680 but remains in position and there is little evidence of its mediating plant interactions. The 681 proportion of total litterfall that is from reproductive parts varies, e.g. 16-24% in an NZ 682 podocarp/angiosperm forest (Enright 1999), 23% in a Mediterranean shrubland (Arianoutsou Wilson & Agnew, chapter 2, Interactions, page 23 of 54 683 1989), versus a much smaller proportion in a tropical forest in India (Sundarapandian and 684 Swamy 1999). 685 686 687 If litter breaks down more slowly than it is deposited and is not redistributed within the area by wind, animals, etc., climatic and local regimes determine its ultimate fate: In a dry climate and/or on a dry substrate, litter can disappear by: burial in habitats on which material such as gravel (in braided river systems), sand 688 689 or loess is being deposited, 690 transport in surface floodwater, carried into rivers, 691 fire. Even in the desert decomposition eventually happens 692 The two former fates are allogenic and do not impact on the plant communities that 693 delivered the litter. However, fire does so and is discussed below. 694 When the climate and substrate are moist enough, litter can become peat. We deal with 695 this below. What about the range of conditions between these 2 extremes which would 696 include the majority ecosystems on the planet? 697 Litter plays a major part in the functioning of most ecosystems. We start with a description of 698 litter production and then deal with its effect on its neighbours. Litter production is most often 699 examined in ecological studies for its rôle in nutrient cycling (e.g. Orians et al. 1996). Here, 700 we focus on its other impacts on plant communities. 701 4.1 702 Litter can be of three kinds: Seems unnecessarily simple for the intended audience 703 704 705 The timing and type of litter production Standing dead litter is retained for some time on the plant, decaying. It comprises leaves and stems in many herbs and grasses as well as branches and trunks from shrubs/trees. Programmed litter shed via the physiological process of senescence and abscission. 706 Such litter is almost always produced in pulses at predictable seasons. It comprises 707 mainly leaves, reproductive parts and root hairs (Larcher 1980; Osborne 1989). 708 Stochastic litter is shed by an allogenic disturbance (i.e. accidental damage) or by 709 pathogenic dieback, not initiated by the plant. Some types are produced irregularly, 710 though others are seasonal. 711 These types may differ in the ability that selection has to lead to adaptations in both the litter- 712 producing species, and the species on which the litter falls. 713 Standing dead litter 714 715 Exemplifying the effects of this type of litter, Bergelson (1990) demonstrated in a field experiment that the seedling emergence of the annuals Senecio vulgaris (groundsel) and Wilson & Agnew, chapter 2, Interactions, page 24 of 54 716 Capsella bursa-pastoris (shepherd’s purse) was very sparse where there was a high density of 717 dead Poa annua plants. Gleissman and Muller (1972) reported evidence that the dead fronds 718 of at least some ecotypes of Pteridium aquilinum (bracken) have suppressive effects on 719 underlying vegetation. On the acid forest floor in the uplands of northern Britain the grass 720 Deschampsia flexuosa is often dominant, with the litter consisting largely of D. flexuosa 721 remains (Ovington 1953). Regeneration of Quercus petraea (sessile oak) can be almost absent 722 in a wood with a developed D. flexuosa leaf-and-litter mat, in comparison to good 723 regeneration in other, nearby woods (Scurfield 1953). In this case the litter is mostly of dead 724 D. flexuousa leaves still attached to the plant, and these leaves form a thatch over the grass 725 tufts, shedding the oak leaf litter should this read “shading the oak seedlings and decreasing 726 regeneration”? Other litter effects on these acid forest floors are described below. However, 727 little has been published on the importance of abscission versus retention of dead material in 728 the functional ecology of plant communities. In many perennial herbs the vegetative shoots 729 collapse in winter but the dead flowering stems remain erect. The latter may have effects such 730 as channelling rainfall or acting as bird perches, with consequent input of nutrients and 731 disseminules. 732 Programmed litter 733 In this type, the organ being shed is usually disconnected by abscission, the result of 734 physiological processes endogenous to the plant and initiated some time before actual litter 735 fall occurs. Programmed litter is almost always pulsed by season (e.g. Arianoutsou 1989; 736 Enright 1999), exceptions being tropical figs (Nason et al. 1998) and ruderals in oceanic 737 climates. There can be several peaks of programmed litterfall during the year, often of 738 different types of a species’ replaceable plant organs (leaves, twigs, flowers, fruit, etc.) all 739 with their special phenology. Reproductive litter is always programmed. In masting species it 740 can be pulsed by year (Buck-Sorlin and Bell 1998) though not necessarily in a simple way: 741 the latter workers found that copious shoot shedding correlated with masting the following 742 year. As well as being temporally patchy, reproductive litter is also spatially patchy, for 743 example in some New Zealand forests it is greater near trees of Agathis australis (kauri; 744 Enright 1999). 745 The fall of leaf litter and other organs of photosynthesis such as cladodes and 746 phyllodes enough vocab already. Use the appropriate words, but there are places (like here) 747 where “big” words seemed be used to show that they can be. What is wrong with the words 748 leaves and petioles. is a pulse in almost all forests whether dominated by deciduous or 749 evergreen species. The number and amplitude of pulses each year depend on the phenology of Wilson & Agnew, chapter 2, Interactions, page 25 of 54 750 the vegetation and its response to the environment. The pulse of falling leaves disturbs the 751 environment of the ground flora, and each species’ leaf litter has a characteristic pH, phenolic 752 content, breakdown rate and pattern of fall, depending upon its chemistry, specific leaf area, 753 leaf size and geometry. The effects of a litter pulse can be to change light and humidity at the 754 soil surface or to start nutrient cycling. 755 Programmed litter is usually lightweight and therefore easily dispersed. Sometimes it 756 is redistributed by wind and rain into deep patches, having a great local effect (Fahnestock et 757 al. 2000). This means that its time of impact at a point does not necessarily coincide with the 758 time of it falls (this applies to lightweight stochastic litter too). It can also be redistributed by 759 animals (Theimer and Gehring 1999), for example bird scratching can remove the litter load 760 locally, and permit germination of surface seeds. Branch fall (= branch shedding = limbfall = 761 cladoptosis) is programmed in some species. The process has been well described (Millington 762 and Chaney 1973) but its potential wider evolutionary and synecological significance appears 763 to have been overlooked. It involves well-defined cleavage zones, for example with a cork 764 band produced from a phellogen, usually in twigs and small branches but also in large (e.g. up 765 to 6 cm in diameter) branches of several years of growth. Branch fall can be seasonal, usually 766 in autumn but in a few species in spring (Millington and Chaney 1973). The fall of heavy 767 litter such as palm leaves and the branches of some conifers (Agathis australis: W.R. Wilson 768 et al. 1998) can be programmed to occur at a particular season and there can be in 769 considerable amounts (e.g. c. 800 kg ha-1 yr-1: Christensen 1975; Buck-Sorlin and Bell 1998). 770 Stochastic litterfall 771 Stochastic litterfall is that which is not programmed. Rather, it is caused or triggered 772 by storm damage, droughts, floods, geomorphological change, gravitational overload, 773 herbivore destruction etc. 774 Stochastic litter can be dead or alive. It includes leaves: 68% in the upland forest 775 studied by Shure and Phillips (1987). However, the lignified plant parts (twigs and CWD) are 776 often prominent, e.g. 46% of the litter recorded in an Amazonian rain forest (Chambers et al. 777 2000). Wardle (1984), in evergreen Nothofagus in New Zealand, found that twig fall 778 comprised about 30% of the litter. It can be very sizable in total: Buck-Sorlin and Bell (1998) 779 recorded 37,000 plant fragments from one oak tree in a year; the shoots were generally short 780 shoots not older than two years. Chambers et al. (2000) recorded 3.6 tonnes ha-1 yr-1 of CWD 781 (>10 cm diameter) in an Amazonian rain forest. Amounts are somewhat less in more open 782 forests in drier areas, for example in the order of 1-2 tonnes ha-1 yr-1, mostly leaves, for two 783 Eucalyptus savannahs in Queensland, Australia (Grigg and Mulligan 1999). Trees are Wilson & Agnew, chapter 2, Interactions, page 26 of 54 784 continually shedding overtopped branches (Buck-Sorlin and Bell 1998). Broken dead trees 785 can also make a contribution, e.g. 7% in a boreal forest in Finland (Siitonen et al. 2000). In 786 many climates, epiphytes add appreciably to the litter fall, being 5-10% in old tropical cloud 787 forest in Costa Rica, where epiphyte fall is temporally sporadic (Nadkarni and Matelson 788 1992), and c. 5 % in Picea abies (Norway spruce) old-growth forests in Sweden, consisting of 789 lichens (Esseen 1985). Fig. 2.5: The seasonal pattern of litterfall in an oakwood. After Christensen (1975). 790 791 By definition stochastic litterfall occurs less predictably than programmed litter, especially in 792 the case of CWD which falls mostly during storms. However, stochastic litterfall can be 793 pulsed by phenology, falling especially at the end of the life of a modular organ. It can also be 794 pulsed by a pest epidemic or short-term weather conditions. Such pulses occur sporadically 795 between years (Harmon et al. 1986) but sometimes with seasonal peaks related to storm 796 frequency (Fig. 2.5; Gentry and Whitford 1982; Brokaw 1982; Buck-Sorlin and Bell 1998). 797 Wardle (1984) found that in NZ Nothofagus it was seasonal, though less consistently so than 798 leaf fall. The time of the peak differed between Nothofagus species and also between sites, 799 usually but not always coinciding with the peak of litter fall. In Nothofagus menziesii (silver 800 beech) there was a double peak. Both Nadkarni and Matelson (1992) and Esseen (1985) found 801 the epiphyte fall to occur mainly in storms. There is a gradation from seasonal events to those 802 separated by decades or centuries. Even programmed branch fall can be partly stochastic, i.e. 803 influenced by the environment, such as that resulting from dieback after drought-induced 804 cavitation (Rood et al. 2000). 805 806 807 Because it is less phased, stochastic litter is unlikely to disturb as much and certainly not as predictably as programmed litter. In boreal forests CWD is an important constituent of forest function, covering over 5% 808 of the forest floor (Zielonka and Niklasson 2001), the larger tree trunks taking up to a century 809 to decompose. Wilson & Agnew, chapter 2, Interactions, page 27 of 54 810 811 4.2 Litter decomposition The litter decomposition rate is obviously important because it sets the period during 812 which litter can have a mechanical and chemical effect, as well as driving nutrient cycling 813 (sect. 3.13 below). Decomposition depends on moisture, temperature, and the litter’s nutrient 814 and phenolic content. These are influenced by climate, soils and the identity of the littering 815 species. On a global scale climate affects decomposition rates more than do the characteristics 816 of the litter, with the fastest decomposition in warm, wet climates (Meentemeyer 1978), but 817 on a local scale litter nutrient and polyphenol content, and to some extent morphology, control 818 breakdown rates in conjunction with the water regime in each habitat (Osono and Takeda 819 2004). For example, the frequent occurrence of tannin-rich litter in heavily leached soils is 820 correlated with an increase in its polyphenol content which controls the release of nitrogen, so 821 important for decomposers (Northup et al. 1995). Further, Moorhead et al. (1998) 822 experimented with mycorrhizal nutrition in grassland of Bouteloua gracilis (grama grass) / 823 Schizachyrium scoparium (bluestem), and interestingly concluded that this community could 824 retard litter decay by reducing decomposer action in a nutrient-limited soil, since the grasses 825 could outcompete the decomposers for scarce nutrients. 826 Litter is broken down by a succession of microbial communities, involving 827 invertebrate shredders, commutators, fungi and mycotrophs, and bacteria (Cadisch and Giller 828 1997). If the litter is large and woody the process involves macrofauna. However, the 829 suberised exfoliating bark of Eucalyptus spp. trees in Australia is slow to break down, and 830 often needs destruction by fire (Canhoto and Graca 1999). Breakdown rates differ between 831 species and taxonomic groups, for example Cornelissen (1996) found within British regional 832 flora a rough correlation between “evolutionary advancement” and leaf decomposition rate, 833 But the slow breakdown of grasses (Poaceae) was anomalous, explained by their low base 834 status compared with the dicotyledons (except the Ericaceae) from that region. However, in 835 the study of Preston and Trofymow (2000) comparing species within one vegetation type 836 rather than an entire flora, grass litter decomposed the fastest and Fagus grandifolia 837 (American beech) the slowest. The rate is undoubtedly correlated with cellulose:lignin ratios 838 and mechanical characteristics of the leaves (Cornelissen and Thompson 1997). 839 There is some evidence that leaf litter combined from several species tend to 840 decompose faster than single-species litter (McTiernan et al. 1997; Wardle et al. 2002 %585), 841 reducing litter residence times and hence the disturbing effect of litter. Gartner and Cardon 842 (2004), reviewing the literature, found that enhanced decomposition in mixtures has been 843 found in 47% of cases reported. For example, Kaneko and Salamanca (1999) found that in 844 litter bags of species collected from a Quercus serrata (oak) / Pinus densiflora (red pine) Wilson & Agnew, chapter 2, Interactions, page 28 of 54 845 forest in Japan, mixtures containing Sasa densiflora lost weight faster than expected from the 846 decomposition rate of their components. They suggested the effect may operate via the faunal 847 commutators. However, Wardle et al. (2006 %1052) conclude that effects of litter mixing on 848 decomposition and on fauna are not related. Perhaps this synergy holds only for certain types 849 of litter: Hoorens et al. (2002) suggested that the effect was due to the high N content of some 850 species which would speed the decomposition of their lower-N associates. However, they 851 found that while broad-leaved herb litter mixes (of legumes) decayed faster, there was no 852 effect when Sphagnum recurvum (=S. fallax) was mixed with Carex rostrata (beaked sedge). 853 4.3 854 Effects of litter Litter must have a chemical effect as well as its physical one. In the case of CWD, the 855 physical effect overwhelms the chemical, but finer reproductive parts may have mainly a 856 chemical effect. Decomposition and leaching are both central to the nutrient cycling regime in 857 the plant community. Litter fall can affect community succession, facilitating it or redirecting 858 it. Deep persistent leaf litter can inhibit the invasion of later-successional species (Crawley 859 1997). In many communities, an increase in nutrient supply leads to dominance by a few fast- 860 growing tall species, and lower species richness (Willems et al. 1993). This has usually been 861 seen as due to competition, but it could equally well be due to the effect of litter (Berendse 862 1999). Litter can have differential effects on particular species and thus on their populations 863 within a plant community. Documentation has been particularly thorough in the case of leaf 864 litter (Table 1). For some of the effects listed in Table 1, opposite effects are seen in different 865 situations, depending on the depth of the litter, the affected species and the growth medium 866 (Hamrick and Lee 1987; Nilsson et al. 1999; Facelli et al. 1999). The effects of litter can be 867 classified as: 868 1. Direct mechanical effects 869 a. physical damage, 870 b. alteration in environment. 871 2. Chemical effects 872 a. leachates from litter, 873 c. control of nutrient cycling. 874 3. Complex environmental interactions 875 a. peat formation, 876 b. litter-herbivore interactions (sect. 6.5 below), 877 c. litter-fire interactions (sect. 6.5 below), 878 d. higher order interactions (sect. 6.5 below). Wilson & Agnew, chapter 2, Interactions, page 29 of 54 879 All of these depend on the characteristics of the litter material and its phenology of production 880 as described above. In addition many of them require the litter to decompose slowly enough 881 to maintain a constant presence in the A0 top soil layer. 882 4.4 883 The direct mechanical effect of litter The physical impact of litter, particularly on the forest floor, is very conspicuous, and 884 we discuss it below (sect. 5.4 below) as an autogenic disturbance. 885 4.5 886 Alteration in environment by litter Litter also causes general environmental changes at the soil surface. Plant growth, 887 especially of herbaceous plants, must be suppressed by the low light under litter, but we have 888 not found any study on this. Red/far red (R/FR) ratios are reduced by a live canopy, then 889 sometimes by half again by the litter. This can suppress germination (Vazquez-Yanes et al. 890 1990) or increase it (Fowler 1986 %131). Litter might intercept and shortly evaporate rain, 891 reducing soil water content (Myers and Talsma 1992), but it can also reduce evaporation from 892 the soil itself (Eckstein and Donath 2005). The temperature regime at the soil surface can also 893 be affected (Fowler 1986 %131). 894 Whenever species of stiff, erect form, such as the grass Holcus mollis and Vaccinium 895 myrtillus, trap and accumulate tree leaves (e.g. of Quercus spp., oaks; or Fagus sylvatica, 896 beech), other plants including Deschampsia flexuosa are killed, as demonstrated by dead D. 897 flexuosa seen below such litter (Watt 1931; Ovington 1953). Watt (op. cit.) mimicked this by 898 laying a layer of beech leaves 5 cm thick; the D. flexuosa was killed. Watt suggested this 899 could not be due to allelopathy, because where taller D. flexuosa leaves occasionally projected 900 above the litter layer they grew well, and moreover Holcus mollis growth was much improved 901 by similar treatment. This can allow H. mollis to invade areas that had been dominated by D. 902 flexuosa (Ovington 1953; Scurfield 1953). No litter is built up by H. mollis, as its dead organs 903 rot too fast (Ovington 1953), so Quercus can establish in the tree litter, or amongst H. mollis, 904 once the D. flexuosa litter mat is removed (Watt 1931; Ovington 1953). Chapin et al. (1994) 905 suggest that in Glacier Bay successions Alnus sinuata (alder) suppresses Dryas drummondii 906 partly by the allelopathic effects of its litter, but as part of complex replacement interactions. 907 In forests in the mid west USA (Michigan and Wisconsin), hardwoods (e.g. Quercus rubra, 908 red oak) can dominate the mid-successional forest. In these stands, Tsuga canadensis 909 (hemlock) cannot invade, because its seedlings are unable to penetrate the hardwood litter, 910 whereas Acer saccharum (sugar maple) seedlings can, and it therefore invades (Rejmanek 911 1999). Wilson & Agnew, chapter 2, Interactions, page 30 of 54 912 Subvention via litter can be very significant for particular species and, in some 913 temperate rain forests the seedlings of many trees depend on fallen trunks and CWD for their 914 establishment and hence regeneration (McKee et al. 1982; Agnew et al. 1993; Duncan 1993). 915 CWD has a characteristic development in old growth stands, and harbours a specialized, often 916 endangered, invertebrate fauna and bryophyte flora. Bryophytes are particularly vulnerable to 917 litter cover (Barkman et al. 1977; Oechel and Van Cleve 1986), especially in forest, where 918 litter is deposited by the herbaceous field layer as well as by woody vegetation. However, 919 During and Willems (1986) named five bryophyte species that they suggested “may thrive on 920 litter”. One possibility is that nutrients released from litter is taken up by bryophytes (Tamm 921 1964). In Scandanavian forests many species are dependent CWD for regeneration or growth, 922 including many bryophytes (Siitonen 2001). 923 4.6 924 Leachates from litter It has been long suspected that leachates from litter affected germination and growth 925 of certain plant species. Early reports of this were concerned with crop residues (Kimber 926 1973) but then evidence came from natural plant assemblages. The case of the tree Celtis 927 laevigata (sugarberry) is the clearest (Turner and Rice 1975; Lodhi 1978). Bosy and Reader 928 (1995) investigated suppression of germination of four forbs by grass litter in an oldfield. 929 Germination of seeds of at least three of the species was markedly reduced under litter, and 930 for two species an appreciable component of the suppression could be explained by the 931 inhibitory effect of leachate. However, it is also possible for litter to subvent germination 932 (Xiong and Nilsson 1999), or to act as part of a cue for germination from the seed bank 933 (Preston and Baldwin 1999). 934 Allelopathy was discussed above. The allelopathic effect of litter leachates has long 935 been suspected and repeatedly tested. For example, Rutherford and Powrie (1993) showed 936 that leaf and litter leachates from the Australian Acacia cyclops (coastal wattle) affected the 937 small-leaved shrub Anthospermum spathulatum below it. Apparently leachates from both the 938 canopy and litter are implicated in many allelopathic effects. However, experiments are easier 939 in vitro than in the field, and only a few clearly demonstrate that litter allelopathic agents have 940 real effects in nature. 941 4.7 942 Control of nutrient cycling Litter returns nutrients to the soil as mineral compounds or as organic compounds 943 which are then broken down to mineral ones. This is the well-known process of nutrient 944 cycling, but different litter types will differ. Berendse (1994a) found that in fast-growing Wilson & Agnew, chapter 2, Interactions, page 31 of 54 945 species (i.e. with high RGRmax) less N was re-translocated to the living parts before the leaf 946 fell, and the resulting litter broke down quickly. Such species would be expected to 947 predominate in N-rich environments. In contrast, species with a low RGRmax re-translocated a 948 greater proportion of their leaf N before it fell. Such N-conserving species, with a low carbon 949 turnover, would be expected early in primary succession, when N is in short supply. 950 Berendse et al. (1987) later compared the deciduous Molinia caerulea (purple moor 951 grass) with the evergreen microphyllous woody shrub Erica tetralix (heath). M. caerulea lost, 952 during one year, 63% of the N present in the above-ground biomass standing at the end of the 953 growing season and 34% of its P. Comparable losses in E. tetralix were 27% and 31%, 954 respectively. This suggests that under nutrient-poor conditions the low-nutrient-loss E. tetralix 955 will be the dominant species, and that if nutrient availability increases M. caerulea will 956 replace it. There is some empirical support for this: in a similar system, Berendse et al. 957 (1994b) found that addition of nutrient solution caused Calluna vulgaris (a species 958 ecologically similar to Erica tetralix) to decline and cover of four grass species to increase. 959 The litterbag decomposition rates given for various species by Cornelisson et al. (1999) make 960 it clear that plants of later successional stages will have litter that is slower to decompose and 961 thus with a greater potential disturbance effect. These authors report from a wide ranging 962 survey an almost universal negative correlation between decomposition rate and herbivore 963 defence. 964 4.8 965 Peat formation On wet substrates, if litter breaks down at a slower rate than it is produced peat will 966 build up. Peat is composed of plant remains in various stages of humification by a partial 967 organic hydrolysis of plant material mediated by micro-organisms (Fig. 2.6). Some types of 968 litter especially impede drainage to cause waterlogging. This yields humic colloids with 969 properties that restrict drainage and oxygenation and, being complex molecules bounded by 970 weakly acidic OH radicals, lower the pH of the soil solution. Nutrient availability and 971 oxygenation are reduced, which decrease litter production but decrease its decomposition 972 even more, so that peat accumulation increases yet more. 973 974 975 Wilson & Agnew, chapter 2, Interactions, page 32 of 54 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 Sphagnum grows and builds a mound Empodisma grows associated Ericaceae cell walls with cellulose hydrolyse to uronic acid high concentrations of polyphenols in litter NPK scavenged from low concentrations because of high base exchange capacity for NH4, Ca and K substrate surface is isolated from ground water input of rain and CO2 acid, low NPK input + leaching low pH low NPK cell walls decompose only partially, to humic acids decomposition slows anoxia brown water from leachate of semidecomposed plant material methane produced peat buildup (litter>decomposition) porosity of peat decreases Key = Empodisma paths = the main switch feedback cycle Fig. 2.7. The process of peat accumulation in mires dominated by Empodisma (wire rush) 1006 or Sphagnum moss species. 1007 There are three routes to peat formation. In eutrophic or mesotrophic conditions litter 1008 production is high but decomposition is fast too, so peat will form only in still lakes, infilling 1009 them. This is a classic succession story (Walker 1970). Paludification often follows from this 1010 point. Peat can also form on the wet, nutrient-poor soils often caused by the leaching effect of 1011 high rainfall (Crawford 2000), a process known as paludification. There are two ways in 1012 which paludification can happen. Firstly, the species adapted to these conditions often bear 1013 leaves with a high polyphenol content, which slows decomposition and thus potential for 1014 paludification (Inderjit and Mallik 1997; Northup et al. 1998). This is a widespread 1015 phenomenon, having important effects on community structure. For example Ehrenfeld et al. Wilson & Agnew, chapter 2, Interactions, page 33 of 54 1016 (1995) propose that small ericaceous shrubs trap litter in New Jersey pine savannahs, 1017 inhibiting decomposition and maintaining openings in the canopy, while Inderjit and Mallik 1018 (1997) provide experimental evidence for litter of Ledum groenlandicum (Labrador tea) 1019 limiting black spruce regeneration in Canada. 1020 The ultimate peat-forming process is the formation of an ombrogenous bog by 1021 isolation of the peat surface from its surroundings in water and nutrient supply (Klinger 1996, 1022 Anderson et al. 2003). This is effected by a guild of species which can absorb nutrients 1023 directly from rainfall, mostly bryophytes and particularly Sphagnum spp. and the restiad 1024 Empodisma minus (wire rush, Agnew et al. 1993), an odd flowering plant which has the same 1025 effect though its negatively geotropic root weft. This guild has, in common, cell walls with a 1026 cellulose-derived weak acid (uronic) that can scavenge ions from precipitation, yielding 1027 sufficient nutrients for annual growth and maintaining a high hydrogen ion concentration in 1028 the interstitial water. The effect is very well documented for Sphagnum spp. in temperate 1029 mires (Clymo 1963; Clymo and Hayward 1982; Koojiman and Bakker 1994). The rate of 1030 percolation, as seen in Darcy’s constant, is slowed by the peat formed. The waterlogging, low 1031 pH and low-nutrient conditions favour this guild. Malmer et al. (1994; 2003) suggest that since 1032 Sphagnum spp. can outcompete vascular plants under ombrotrophic conditions these 1033 conditions may arise rather suddenly. This is the state change expected when a switch is 1034 operating (chapt. 3, sect. 5 below). There are indeed records of such rapid state changes where 1035 Sphagnum spp. expansion has caused a brown moss fen to be replaced by bog (Kuhry et al. 1036 1993), destroyed forest (Anderson et al. 2003) and invaded forest around the edges of basin 1037 mires (Klinger 1996). The process may be initiated by climatic change to colder, wetter 1038 conditions (Korhola 1996). 1039 4.9 1040 Can reaction via litter evolve? It is not enough to describe the processes that occur within plant communities; we 1041 should gauge whether these processes are accidental, or whether they could be the result of 1042 evolution. This is hardly in doubt for the other types of interaction discussed in this chapter – 1043 interference, subvention and autogenic disturbance—wasn’t it in question just a little before? 1044 – because the characters of the plant involved affect its fitness and hence the transmission of 1045 its genes to the next generation. The problem with litter is that it is by definition deceased 1046 plant tissue. As Clements (1904) pointed out, all plants react on their environment, but the 1047 extent and type of reaction will be genetically controlled. A particular reaction will evolve if it 1048 enhances the fitness of the individual plant (i.e. genet) more than it enhances the fitness of its 1049 conspecific neighbours, after taking into account any cost of reaction. However, litter affects Wilson & Agnew, chapter 2, Interactions, page 34 of 54 1050 all the individuals and species in a community. In some cases it is produced when the whole 1051 plant dies, so the characters are expressed too late to affect whether the genes that caused 1052 them will increase or decrease. This is clearest in cases when the litter is produced when the 1053 plant dies or has its effect after the plant’s death, for example by paludification, by control of 1054 nutrient cycling, and as fuel for fire. There is then no possibility of plants enhancing their own 1055 fitness, in survival or fecundity, only that of their offspring (which will carry their genes. I 1056 don’t see why evolution can’t therefore act on litter). 1057 For effects that occur after the death of the plant we need to call on kin selection, with 1058 the mechanism that a plant benefits especially its neighbours, which are especially likely to 1059 carry the same alleles (Wilson 1987 %493). Another way of expressing this is selection 1060 through the extended phenotype. Northrup et al. (1998) commented that “decomposing leaf 1061 litter is not generally considered to be part of the plant’s phenotype” so that litter characters 1062 must evolve through selection on the extended phenotype. It has been remarked that just as 1063 selection in different areas with comparable environments but different species often produces 1064 similar solutions in terms of leaf characters, so similar solution in terms of reaction on the 1065 environment would be expected in different species and communities (Lewontin 1985; 1066 Mooney et al. 1977). The concept is that this reaction can lead to the evolution of secondary 1067 characters if an organism modifies its own environment, and the same induced environmental 1068 change is reimposed for sufficient generations to serve as a significant source of selection 1069 (Olding-Smee et al. 1996; Laland et al. 1996 1999)-community selection?. A different 1070 solution was proposed by Whitham et al. (2003), that traits of the extended phenotype could 1071 evolve because they affect the whole community and there could be selection between 1072 communities, a concept that most biologists find hard to swallow (but see D.S. Wilson 1997). 1073 However, most plants produce litter during their life and have an opportunity for that litter to 1074 affect their fitness, in which case Northup et al.’s argument does not apply. Still, there are 1075 problems with the idea that reaction via litter can evolve; litter effects seem to be a necessary 1076 by-product of growth. 1077 5 Autogenic disturbance: plants as disturbers 1078 Disturbances, marked changes in the environment for a limited period, often with the 1079 removal of plant material, are more common than was once believed and important for plant 1080 communities (White 1979; Rundel 1998; Grime 2001). The concept ‘autogenic disturbance’ 1081 was coined by Attiwill (1994), though it is close to the ‘endogenous disturbance’ of White 1082 (1979). Most consideration has been of allogenic disturbance, i.e. that caused by external 1083 factors such as animals or extreme weather events. For example, Bazzaz (1983) describes Wilson & Agnew, chapter 2, Interactions, page 35 of 54 1084 disturbance as continuous and necessary in all ecosystems, but discusses only allogenic 1085 causes. No less important, we suggest, are the minor disturbances within the community that 1086 inevitably result from the growth and death of plant parts, for example by plant contact and by 1087 litter (there seems little opportunity for autogenic disturbance by subterranean organs). Thus, 1088 vegetation, by its nature, disturbs itself. As we pointed out in chapter 1: “Plants move, 1089 animals don’t”. Autogenic disturbance has potentially positive and negative effects on 1090 population fitness and density, yet stands apart from the general considerations of interference 1091 and subvention described above. There is a wide range in the intensity of autogenic 1092 disturbances, from delicate adjustments resulting from leaf growth to the fall of a group of 1093 trees, but plants must tolerate all these effects. It may have profound bearing on vegetation 1094 dynamics. Autogenic disturbance may be more predictable than allogenic disturbances such 1095 as storms and earthquakes perhaps they are more predictable on a day to day basis, but over 1096 the course of an organism’s life, the probability of experiencing many allogenic disturbances 1097 is very predictable. I might not know whether or not I’m going to be hit be a storm tomorrow, 1098 but I do know that I will be hit by a storm in the course of my life. As the damage sustained 1099 by a storm or other allogenic dist can be much more severe shouldn’t all this (when combined 1100 with the plasticity issue commented on below) lead to a high level of selection on allogenic 1101 dist. related traits?, and if so selection for adaptation to it is more likely to occur. If plants can 1102 respond with plasticity to autogenic disturbance (as you suggest below) but not with allogenic 1103 disturbances, wouldn’t selective pressure be greater on allogenic dist. adaptations? However, 1104 autogenic and allogenic disturbance can interact. 1105 5.1 1106 Movement and contact Even gentle contact between plants can affect their growth, producing effects such as 1107 reduced growth, reduced soluble carbohydrates and increased transpiration (van Gardingen 1108 and Grace 1991; Keller and Steffen 1995). For example, when Biddington and Dearman 1109 (1985) brushed Brassica oleracea var. botrytis (cauliflower) and Apium graveolens (celery) 1110 seedlings with typing paper for 1.5 min day-1 it reduced their shoot dry weight (Fig. 2.7). Wilson & Agnew, chapter 2, Interactions, page 36 of 54 Control Brushed Fig. 2.5: The effect of brushing seedlings of Brassica oleracea for 1.5 min day-1After Biddington and Dearman (1985). 1111 1112 Experimentally shaking the plant can have an even greater effect, e.g. reduced shoot 1113 extension, shorter and wider stems, lower shoot:root ratios, altered root development, shorter 1114 petioles, shorter flowering culms and reduced reproductive effort (Neel and Harris 1971; 1115 Braam and Davis 1990; Gartner 1994; Niklas 1998; Goodman and Ennos 1998). Whilst 1116 brushing with typing paper and shaking by hand are artificial treatments, they are very close 1117 to the leaves of adjacent plants brushing each other, or close trees rocking each other. The 1118 changes are mediated by the up-regulation of a large number of genes (Braam and Davis 1119 1990). Plants can also change position as a result of their own growth (Evans and Barkham 1120 1992), for example Salix (willow) trunks collapse due to their own weight in wet carr forest 1121 (Agnew, pers. obs), and shrub stems can be bent by their own weight and fall to the ground 1122 (B.F. Wilson 1997). 1123 5.2 1124 Growth thrust Shoots can thrust each other aside physically as they grow (Campbell et al. 1992). For 1125 example, the leaves of Juncus squarrosus are vertical when they first emerge, but they then 1126 become horizontal, form a rosette and press down surrounding plants (authors pers. obs.) as 1127 do leaves of Plantago species (Campbell et al. 1992). Surprisingly little is known of this 1128 process. 1129 Secondary growth of tree trunks, branches and roots continually creates new habitat, 1130 initiating succession by providing opportunities for colonisation. Circumferential branch 1131 growth produces new sites for the establishment of epiphytes. Some bryophyte species are 1132 tree-base specialists (Ashton and McCrae 1970). Their specialisation is perhaps explicable by 1133 their greater shade tolerance (Hosokawa and Odani 1957) and the peculiar characteristics of 1134 trunk water flow creating a damper, more nutrient-rich environment, but this is also the zone 1135 where the fastest circumferential growth occurs. Circumferential growth of tree roots can Wilson & Agnew, chapter 2, Interactions, page 37 of 54 1136 affect the environment above the surface: woody roots near the soil surface create shallow 1137 soils and diversify the forest floor habitat (Agnew pers. obs.). The reader may like to take the 1138 ‘pers. obs’ in this section as confirmation that autogenic disturbance has not received due 1139 attention in ecological research. 1140 5.3 Crown shyness 1141 Crown shyness comprises gaps between the canopies of adjacent trees (Jacobs 1955), 1142 probably caused by abrasion in the canopy (Putz et al. 1984 %24). Franco (1986) observed 1143 the process, reporting that when branches of Picea sitchensis (Sitka spruce) and Larix 1144 kaempferi (Japanese larch) met physical damage occurred by abrasion, causing the death of 1145 the leading shoots. New shoots arising from lateral buds and were also killed. Sounds like 1146 competition for space to meLong and Smith (1992) demonstrated the mechanism by fixing 1147 wooden pickets into the canopy as artificial branches: they were broken back to the edge of 1148 the crown, due to wind-rock and consequential abrasion. Tall slender trees were subject to 1149 greater wind-rock, and thus greater frequency of impact with neighbouring crowns, leading to 1150 greater crown shyness. Putz et al. (1984 %24) measured swaying of Avicenna germinans 1151 (mangrove) and found that branches bordering crown shyness gaps generally had broken 1152 twigs and few leaves. Flexible crowns were more widely spaced than still crowns, confirming 1153 the observation of Long and Smith (op. cit.). These effects are the result of the compulsory 1154 growth and movement imposed by the modular construction of plants (chapt. 1, sect. 1.1 1155 above). 1156 5.4 1157 Growth and gravity CWD clearly has the potential to be very destructive on the forest floor. The most 1158 destructive type of stochastic litter is tree fall, which is the autogenic disturbance par 1159 excellance and implicated in all demographic processes in forests of mixed canopy species. It 1160 appears to be a normal, predictable event in many undisturbed stands, which can be measured 1161 and built into models of species and biomass turnover in boreal (e.g. Hofgaard 1993), tropical 1162 evergreen (e.g. Chandrashekara and Ramakrishnan 1994) and tropical rain forests (e.g. 1163 Kellman and Tackaberry 1993). 1164 The weight in a tree’s canopy can eventually result in its fall, obviously disturbing 1165 surrounding plants. In a forest the upper canopy trees are more likely to fall, having reached 1166 their greatest potential and being too heavy for their root support or too loaded with 1167 lianas/epiphytes (Strong 1977; Putz 1995). Tree fall is associated with storm damage, 1168 particularly in mid-latitudes (e.g. Busing 1996, Allen et al. 1997). Understorey trees are more Wilson & Agnew, chapter 2, Interactions, page 38 of 54 1169 likely to become standing dead, since pathogens or light reduction by neighbours (self- 1170 thinning) can kill, whether or not the tree still has the potential to grow further, and they are 1171 not as exposed to storm damage as are the canopy species. An exception occurs in the dry 1172 Kysna (South Africa) forest, where most canopy trees die in situ due to “competition, 1173 senescence and secondary pathogens” and gradually break up (Midgley et al. 1995). There is 1174 not enough evidence to suggest this as a general feature of any forest types. 1175 In old-growth forest, tree fall is the usual process that forms canopy gaps. In gap 1176 formation, several sudden environmental changes take place. The most important is that the 1177 forest floor is brightly lit with light of a very different spectral quality. The effects of a gap 1178 are not all positive: Lovelock et al. (1998) found photoinhibition of shade-tolerant juveniles in 1179 gaps, especially in species with short-lived leaves. Another, often overlooked, factor is 1180 rainfall/hydrology. A full canopied forest evaporates up to 40% of incoming rainfall and the 1181 soil is seldom at field capacity, but gaps receive the rain uninterrupted. There is also soil 1182 disturbance, with soil exposed in a root pit and adjacent root plate mound. The bare soil and 1183 higher light allow secondary species to invade, as observed by Ellison et al. (1993) in 1184 neotropical forest, and Narukawa and Yamamoto (2001) for Abies spp. (fir) seedlings in 1185 boreal and subalpine Japan. 1186 Branches can also cause severe impact when they fall (sect. 3.9 above). For example, 1187 in the tropical liana Connarus turczaninowii in Panama, 20-45% of the mortality of young 1188 plants is caused by branch fall (Aide 1987); in a tropical rain forest in Amazonia, Uhl (1982) 1189 found the most common (38%) cause of death in small (1-10 cm dbh) trees to be a branch fall 1190 or treefall; in Costa Rican rain forest, Vandermeer (1977) found that 46% of seedling deaths 1191 were due to falling leaves or branches, about half of these were due to falling palm fronds. 1192 Gillman and Ogden (2001 %671) found in Podocarp/angiosperm forest in northern North 1193 Island, NZ, that 10-20 % of the annual seedling mortality was caused by litterfall. 1194 5.5 1195 Lianas Lianas can be so abundant in the canopy of many tropical and sub-tropical forests that 1196 they compete with the emergent trees (Lowe and Walker 1977). Treefall, with its associated 1197 disturbance to the forest floor, can be generated by either adherent lianas (e.g. Hedera helix, 1198 ivy) or free-swinging lianas (e.g. Ripogonum scandens, supplejack). Phillips et al. (2005 1199 %1250) found that in an upper-Amazon tropical rain forest large trees were three times as 1200 likely to die if they were invested by lianas. The resulting canopy gap can also benefit lianas. 1201 Needing less investment in support they can extend fast and keep pace with the fast-growing 1202 secondary species filling the gap. Wilson & Agnew, chapter 2, Interactions, page 39 of 54 1203 Twining or tendril lianas can constrict the host stem (Lutz 1943; Clark and Clark 1204 1991; Matista and Silk 1997). Usually there is little active strangulation, but the liana stays 1205 put and the tree grows, which can produce deformations in the trunk and lead to changes in 1206 the vascular tissue and the wood fibres ((Falconer 1986; Reuschel et al. 1988; Putz 1991). 1207 Sometimes new conducting tissue is formed parallel to the constriction, but constriction can 1208 inhibit or even stop downward translocation of organic solutes (Lutz 1943; Putz 1991, 1209 Hegarty 1991). The constriction can also be a point of weakness, at which a branch or young 1210 stem is more likely to break (Lutz 1943; Uhl 1982; Putz 1991). Deformed trees are also more 1211 susceptible to pathogens (Putz 1991). Not infrequently the tree dies (Lutz 1943; Clark and 1212 Clark 1991; Uhl 1982). 1213 In the tropics, epiphytes, lianas and trees are often not distinct categories. Many 1214 genera in the Araceae (Monsteroideae) start as lianas but can maintain themselves as 1215 epiphytes without contact with the forest floor. We have found no record of an epiphyte being 1216 associated with tree or branch fall, although their contribution to litter fall has been well 1217 documented (see above). Strangler epiphytes certainly damage trees (Richards 1996). The 1218 epiphyte species involved are evergreen; many are in the genus Ficus (fig) and curiously 1219 revered by the indigenous people (F. benghalensis in Hindustan, F. natalensis in Kenya). 1220 Such stranglers begin life as epiphytes, gradually extend aerial roots downwards, eventually 1221 root in the soil, then grow around their host as a tree. The host usually dies, and does so 1222 standing, though the cause of the host’s death is not clear (Richards 1996). 1223 The interactions between support and liana are complex. Lianas are often supported by 1224 more than one tree (with a record of 27 trees for one liana on Barro Colorado Island, Panama: 1225 Putz 1984 %1713). This causes interdependence between trees that may stabilise canopy trees 1226 adjacent to treefalls (Putz 1984 %1713; Hegarty 1991). However, a corollary is that if such a 1227 connected tree falls, it damages more trees. For example, Putz (1984 %1713) found on Barro 1228 Colorado Island that 2.3 trees fell with each gapmaker, often ones that shared lianas. Liana- 1229 laden trees can also pull branches off neighbouring trees as they fall (Appanah and Putz 1230 1984). 1231 6 Plant-plant interactions mediated by other trophic levels 1232 While a macrophyte remains sedentaryi, its environment is changing, especially with 1233 respect to the various sets of organisms utilising the carbon energy source that it represents. 1234 These include herbivores, parasites and pathogens, all capable of modifying a plant’s fitness 1235 and therefore its competitive ability. There is also the potential for various combinations of 1236 heterotrophs and neighbouring plants to build complex nexus of inter-relationships. In most Wilson & Agnew, chapter 2, Interactions, page 40 of 54 1237 cases these interactions are associated with a "patch effect", which works only if one species 1238 is sufficiently abundant in a local vegetation patch to change the behaviour of the herbivore or 1239 pollinator. 1240 6.1 1241 Below-ground benefaction It is generally assumed that species with root nodules that contain nitrogen-fixing 1242 micro-organisms enhance the nitrogen economy of associated non-nodulating plants. These 1243 N-fixing plants are therefore held to be important in ecosystems and particularly in rangelands 1244 (Paschke 1997). However likely the transfer is, it is rarely proven. Kohls et al. (2003) set out 1245 to investigate this link in the well-studied site of Glacier Bay, Alaska. Using Δ15N values they 1246 concluded that nodulated plants accounted for most of the fixed N in soils and plants for up to 1247 40 years of succession; most of this being donated to the ecosystem by Alnus viridis (green 1248 alder). The assumption that fixed nitrogen is indirectly passed from nodulated to un-nodulated 1249 plants is probably justified in many cases. 1250 Common mycorrhizal networks (CMNs) link individuals below ground so that 1251 nutrient and carbon transfers have the potential to enhance or depress plant growth. In all 1252 three major types of mycorrhiza – vesicular-arbuscular, ectotrophic and ericoid/epacrid – 1253 some of the fungal species involved can infect more than one plant species (Simard and 1254 Durall 2004). The vesicular-arbuscular and ectotrophic ones are known to establish fungal 1255 networks that connect the roots of different species, and this may be true for the 1256 ericoid/epacrid type too (Cairney and Ashford 2002). It has been suggested that CMNs affect 1257 the performance of plants (e.g. Booth 2004). However, it is unclear whether plant-tissue to 1258 plant-tissue transfer of carbon actually occurs (Simard and Durall 2004). There has never 1259 been a convincing demonstration of it, certainly not to a degree sufficient to benefit the plant. 1260 In cases where carbon has been traced by radioactivity to the recipient it is likely that it is 1261 retained in the fungal hyphae, not in the plant tissues (Robinson and Fitter 1999). 1262 Similar transfers could operate in mineral nutrients. However, Newman and Ritz 1263 (1986) and Newman and Eason (1993) carefully monitored the transfer of 32P and concluded 1264 that any direct transfer via hyphal links was very minor, and too slow to substantially 1265 influence the nutrient status of the plants. There is evidence for greater nitrogen transfer when 1266 mycorrhizae are present but it is not clear whether the transfer is through the CMN or whether 1267 the mycelia increase release of N from one plant into the soil and/or uptake by the other plant. 1268 Moreover, ideas that this would even out concentrations between plants do not stand up: He et 1269 al. (2005 %897) found that there was net transfer from Eucalyptus maculata (gum) to 1270 Casuarina cunninghamiana (sheoke). This represents transfer from a plant that is not N- Wilson & Agnew, chapter 2, Interactions, page 41 of 54 1271 fixing, to one that is N-fixing and had close to double the N concentration in its tissues. This 1272 transfer was at very low rates, but the benefaction was in the opposite direction from that 1273 usually assumed between fixing and non-fixing plants. 1274 Roots and microbial activity in the soil are intimately linked, since roots provide the 1275 most of the carbon source below ground. Chen et al. (2004) used microcosms to show that 1276 Pinus radiata roots increased microbial activity and the mineralisation of organic P. There 1277 was evidence that root exudates were responsible for these effects. It is possible that the plant 1278 can obtain N from free-living N-fixing bacteria in the same way. That these effects are species 1279 specific was demonstrated by Innes et al. (2004), who found that the grasses Holcus lanatus 1280 (Yorkshire fog) and Anthoxanthum odoratum (sweet vernal) could stimulate microbial 1281 activity, whereas dicotyledonous forbs depressed it. The effect was only seen in soil of higher 1282 fertility. 1283 6.2 1284 Pollination When fruit set is limited by pollen, plants of the same or different species can interfere 1285 with each others’ reproduction by competing for the service of pollinating animals (Grabas 1286 and Laverty 1999). When pollen from another species lands on a stigma it may interfere with 1287 the germination or fertilisation of the species own pollen by occlusion or allelopathy (Murphy 1288 and Aarssen 1995a). This can reduce seed set and increase self-fertilisation (Bell et al. 2005). 1289 Plants can also compete for animal dispersers (Wheelwright 1985). 1290 There can be subvention too if one species attracts pollinators to the vicinity of 1291 another by being more attractive (a ‘magnet species’) or by mimicry (Laverty 1992). Moeller 1292 (2004) found that among Clarkia xantiana (an annual in the Onagraceae) populations in 1293 California there were more pollinating bees and less pollination limitation in populations with 1294 a greater number of congeners, implying that the mass of plants of the same pollination type 1295 attracted pollinators. An interaction that we can count as a genuine mutualism between 1296 embryophytes was demonstrated by Waser and Real (1979). If species have staggered 1297 flowering times with shared pollinators and some overlap in flowering time, there will 1298 probably be some competition for pollinators. However, the continuous availability of flower 1299 rewards over the period may maintain pollinator populations. Waser and Real gave evidence 1300 from flower numbers, fecundities and pollinating hummingbird numbers across four years at a 1301 site in the Rocky Mountains, Colorado, that the earlier-flowering Delphinium nelsonii was 1302 benefiting the later-flowering Ipomopsis aggregata thus. If we consider the maintenance of 1303 hummingbird populations from one year to the next, I. aggregata could be benefiting D. 1304 nelsonii in a similar way, forming a genuine mutualism. Wilson & Agnew, chapter 2, Interactions, page 42 of 54 In an interesting study assessing the decoy function of sterile teratological “flowers” 1305 1306 on Arabis holboellii (rock cress) caused by the rust Puccinia monoica, Roy (1996) showed 1307 that the mass flowering of mixtures of A. holboellii and Anemone patens attracted more bee 1308 visits to the latter. During bee visits the sticky pseudoflowers of A. holboellii removed pollen 1309 derived from the A. patens stamens and replaced it with fungal spermatia which reduced seed 1310 set in the A. patens. This is a complex interactive system that shows the potential for fertility 1311 interference effects, whether via pathogens or pollen vectors. 1312 6.3 1313 Herbivory Herbivores can remove large amounts of plant material and this must affect the plant 1314 community. The word herbivory suggests grazing, with mammals biting off whole leaves, but 1315 most herbivores are invertebrates, and these maintain their trophic pressure more consistently 1316 and more specifically than the vertebrates. Contrary to the situation with vertebrates, 1317 invertebrate herbivore pressure can be as great below-ground as above-ground (Brown 1993; 1318 Wardle 2002). They can be very much smaller than their substrate, giving large populations. 1319 With fast reproduction this can allow efficient selection to overcome plant defences and 1320 specialise on particular plant species. Invertebrate herbivores have parasites and carnivores 1321 that control them. There are therefore complex systems affecting plant population fitness 1322 through herbivory, but since the observation of both the herbivory and the effect on the plant 1323 are difficult, we know little about the effects of invertebrates on the balance between co- 1324 occurring plant species and hence on community structure. 1325 Herbivores are always selective to some degree. However, the degree of 1326 discrimination varies widely. Many Lepidopteran larvae feed on only one species of plant, 1327 e.g. Tyria jacobaeae (the cinnabar moth) on Sencio jacobaea (ragwort). At the other extreme, 1328 many large non-ruminant animals such as Loxodonta africana (African elephant), even 1329 though they have preferences, will readily eat a wide range of species. More importantly, 1330 some such as Equus spp. (horses) often graze finely-patterned vegetation at a relatively coarse 1331 scale, necessarily taking in species with a range of palatability. We can see the causes of 1332 differential palatability of plants most clearly within species, e.g. cyanogenesis for 1333 Gastropoda (slugs; Jones 1966), smell and visual appearance for Atherigona soccata (a shoot 1334 fly; Nwanze et al. 1998), concentrations of nonstructural carbohydrates, sodium, iron and 1335 mercury for Sciurus aberti (a squirrel; Snyder (1992) and cineole in the leaf terpenoids for 1336 Anoplognathus spp. (Christmas beetles; Edwards et al. 1993). The relative palatability of 1337 species will be affected by a complex of such characters. 1338 Differences in palatability can affect the type of interaction between species. A plant Wilson & Agnew, chapter 2, Interactions, page 43 of 54 1339 of target species S can be positively or negatively affected if conspecifics are replaced by 1340 plants of another species (analogously to a replacement comparison in interference, R below), 1341 or they can be affected positively or negatively by the presence of a neighbouring plant 1342 species when a herbivore moves into the patch (comparison A below), both by several 1343 mechanisms: 1344 positive: 1345 (1) the neighbour repels the herbivore, reducing the herbivory load on S (R) 1346 (2) the neighbour attracts the herbivore’s enemy, reducing the herbivory load on S 1347 1348 1349 1350 1351 (R) (3) defences against herbivory are induced in the neighbour, reducing its interference against S (A) (4) the neighbour attracts the herbivore, which creates open ground, which is the micro-beta niche of species S (R) 1352 (5) the neighbour attracts the herbivore, diverting the herbivory load from S (A) 1353 (6) the herbivore removes a more palatable, interfering, neighbour of S (A) 1354 (7) The herbivore defends S against a an interfering neighbour (A) 1355 negative: 1356 (8) the neighbour attracts the herbivore, increasing the herbivory load on S (A or R) 1357 (9) the herbivore removes a more palatable beneficient neighbour of S (A) 1358 (10) induced defences against herbivory in the neighbour divert the herbivore to S (R) 1359 (1) +ve: The neighbour repels the herbivore, reducing the herbivory load 1360 The repelling of herbivores by a neighbour (Augner 1994) affects plant communities 1361 at the scale which the repulsion occurs. For example, a spiny shrub deters the herbivore from 1362 consuming any other species within its canopy. A recognisably unpalatable species in a grass 1363 sward deters herbivore from taking its bite at that point, sparing any palatable species growing 1364 closely with it. As examples among mammalian browsers/grazers, Hjältén et al. (1993 %125) 1365 found that when Betula pubescens was associated with the less palatable Alnus incana (alder) 1366 the herbivores Microtus agrestis (voles) and Lepus timidus (hares) avoided those patches and 1367 B. pubescens suffered less herbivory damage, and Oesterheld and Oyarzábal (2004) found 1368 that plants of the palatable grass Bromus pictus tended to occur within one of the unpalatable 1369 tussocks. The avoidance of a less preferred patch would be part of an “extended phenotype” 1370 of the neighbour species, but confer benefit on all the plants in that community, as we have 1371 discussed above. Tuomi et al. (1994) suggested that the evolutionarily stable strategy would 1372 be for such benefit to conspecific neighbours to limited, or an undefended (‘selfish’) genotype Wilson & Agnew, chapter 2, Interactions, page 44 of 54 1373 would too readily invade, and they suggested that plant defences should therefore deter the 1374 herbivore without killing it. 1375 Chemicals from a species might repel insects not only from that species but also from 1376 their neighbours. This is widely discussed as a goal of organic gardeners, but good evidence is 1377 hard to find. Tahvanainen and Root (1972) found that planting Lycopersicon esculentum 1378 (tomato) amid Brassica oleracea (collards) plants decreased flea beetle populations on B. 1379 oleracea and increased its plant weight. In a choice experiment, adult flea beetles (Phyllotreta 1380 cruciferae) preferred excised B. oleraceus leaves with no tomato mixed in with them, which 1381 is evidence of the involvement of a repellent chemical. It is widely suggested that some 1382 Tagetes species (African marigold), as companion plants, repel insect pests and thus benefit 1383 other plants in the crop. There is a little firm evidence for this (McSorley and Frederick 1994; 1384 McSorley and Dickson 1995), but nematodes, below-ground, may be mediating the 1385 interaction, apparently not due to any chemical coming from the Tagetes sp. plants (Ploeg and 1386 Maris 1999). 1387 (2) +ve: The neighbour attracts the herbivore’s enemy, reducing herbivory load 1388 White et al. (1995 %1171) found that planting Phacelia tanacetifolia, which was 1389 attractive to hover flies probably for its pollen, around Brassica oleracea (cabbage) patches 1390 (the control was bare ground around the patches) reduced the density of Brevicoryne 1391 brassicae and Myzus persicae (both aphids) on the B. oleracea. 1392 (3) +ve: Defences against herbivory are induced in the neighbour 1393 It is well established that herbivore attack, or leaf damage mimicking herbivory, can 1394 induce in a plant chemical defences such as alkaloids, nicotine, phenolics and tannins (e.g. 1395 Boege 2004, Vandarn et al. 2000), and even morphological defences (e.g. Young et al. 2003). 1396 In most cases, there is a fitness cost of this induced defence, for example in growth rate (Strauss 1397 et al. 2002). One would expect a reduced interference ability this is not the same as inducing 1398 defences in the neighbor, benefiting neighbours, and Baldwin and Hamilton (2000 %915) 1399 produced some evidence of this, but only intra-specifically. In order for this to operate, the 1400 herbivore has to select between species at a small scale: to be exact, inside the neighbourhood 1401 within which interference occurs. 1402 (4) +ve: The neighbour attracts the herbivore, which creates open ground 1403 Hoof marks are the preferred microenvironment of some species (Csotonyi and 1404 Addicott 2004). This does not depend on the herbivore’s having any selection between 1405 species or selecting at any particular spatial scale. 1406 (5) +ve: The neighbour attracts the herbivore, diverting the herbivory load Wilson & Agnew, chapter 2, Interactions, page 45 of 54 1407 This is theoretically possible, and Atsatt and O’Dowd (1976) mention some possible 1408 examples. Mensah and Khan (1997) found that Medicago sativa (lucerne), which has a higher 1409 palatability in choice experiments to divert Creontiades dilutus (a mirid) pests from a crop of 1410 cotton. There was a dramatic effect over the four-month season. However, we can find no 1411 evidence from natural systems. The effect would be only temporary, because the herbivore 1412 population would build up on the neighbour, and probably then increase the herbivory load on 1413 the target. 1414 (6) +ve: The herbivore removes a more palatable, interfering neighbour 1415 If an interfering neighbour is palatable, herbivory could reduce interference from it, 1416 allowing increased growth of the target. This is the principle: “my enemy’s enemy is my 1417 friend”. It is remarkably difficult to find evidence for this. Part of the reason may be the 1418 difficulty of obtaining estimates of relative palatability, but confounding effects of the 1419 herbivore such as creating gaps and species-to-species tradeoffs such as between palatability 1420 and regrowth ability are other reasons (Bullock et al. 2001 %253). However, Cottam (1986) 1421 grew the grass Dactylis glomerata (cocksfoot) with the more palatable Trifolium repens 1422 (white clover). When there was no grazing D. glomerata suffered in mixture, dying out after 1423 120 days of grazing, but when there was grazing by Deroceras reticulatum (slug), 1424 D. glomerataii, grew as well as, or better than, in monoculture. Carson and Root (2000), in a 1425 very careful experiment, applied insecticide experimentally in an oldfield dominated by 1426 Solidago altissima and S. rugosa (goldenrods). It became clear that in the untreated 1427 community specialist Microrhopala spp and Trirhabda spp (chrysomelid beetles) were 1428 suppressing the Solidago spp and thus allowing increased growth of other forbs and of woody 1429 seedlings. As with other mechanisms involving release from interference, the herbivore must 1430 select between species at a fine scale. 1431 (7) +ve: The herbivore defends against an interfering neighbour 1432 Ants are often associated with plants. For example, Macaranga is an old world genus 1433 in the Euphorbiaceae (often conspicuously part of secondary forest) some of whose species 1434 bear extra-floral nectaries which attract ants and domatia hollows which shelter them. The 1435 relationship is close and complex. Fiala et al. (1989) found that on trees of Macaranga spp. in 1436 Malaya ants considerably reduced the amount of herbivory damage, mainly by forcing 1437 lepidopteran larvae (caterpillars) off leaves, and possibly by harassing Coleoptera (beetles) 1438 and Acrididae (grasshoppers). The ants involved are generalist predators often in the genus 1439 Crematogaster, yet they also seemed to protect against lianas: Macaranga species that were 1440 myrmecophytic had far fewer lianas attached, and this difference was seen only with plants Wilson & Agnew, chapter 2, Interactions, page 46 of 54 1441 occupied by ants. The effect was caused by the ants biting the tips of invading liana, and 1442 Federle, Maschwitz and Holldobler (2002) showed that they also pruned adjacent tree species 1443 reducing canopy contact and presumably also competition. 1444 (8) -ve: The neighbour attracts the herbivore, increasing the herbivory load White and Whitham (2000) found that Populus angustifolium fremontii 1445 1446 (cottonwood) plants growing under the highly palatable Acer negundo (box elder) suffered 1447 much more herbivory from the insect Alsophila pometaria (fall cankerworm) than those under 1448 their own species or in the open. Sessions and Kelly (2002) suggested that the grass Agrostis 1449 capillaris (bent) created a moist habitat suitable for Derocerus reticulatum (a slug) that then 1450 attacked a nearby fern (Botrychium australe). Mechanism ‘8’ can operate at any spatial scale 1451 and the herbivore does not need to be selective so long as it is attracted to the patch by one 1452 species. It seems to be the situation to which the strange term ‘apparent competition’ has been 1453 applied. The term ‘magnet species’ for the neighbour is more helpful. 1454 (9) -ve: The herbivore removes a more palatable, subventing neighbour 1455 This is a theoretical possibility, but we know of no example. As with mechanisms 1456 involving competitive release, the herbivore must select at a fine scale. 1457 (10) -ve: Induced defences in the neighbour divert the herbivore to the target 1458 Induced defences are now known in many species, which could protect that species 1459 but not a neighbour. We do not know of any explicit examples of this. 1460 6.4 1461 Indirect impacts of diseases For some of the mechanisms of positive plant-plant interactions via herbivores that we 1462 discussed above, such as ‘2’, ’4’ and ‘7’,it is hard to envisage analogues for fungal or viral 1463 diseases. However, others seem possible. The neighbour could repel viral vectors such as 1464 aphids (Birkett et al. 2000), reducing the disease load on S, the equivalent of herbivory 1465 interaction ‘1 +ve’ above. Defences against fungal diseases are often induced in the host, and 1466 these might carry a cost (though the evidence for this is weak), reducing the host’s 1467 competitive ability against S (cf. herbivory ‘3 +ve’). The effect (translating to disease terms) 1468 “5 +ve: The neighbour attracts the fungus, diverting the load from S”, could be applied when 1469 a neighbour had a strong ‘fly-paper effect’, trapping fungal spores (chapt. 4, sect. 4.3, below). 1470 Disease could remove a more susceptible neighbour that normally interferes with S 1471 (cf. herbivory ‘6 +ve’). For example, the introduction of the Asian pathogen Cryphonectria 1472 parasitica (= Endothia parasitica; chestnut blight fungus) to forests of northeastern USA led 1473 to almost complete death of the dominant Castanea dentata (chestnut) trees. In one area this Wilson & Agnew, chapter 2, Interactions, page 47 of 54 1474 resulted within c. 14 years to a doubling of the basal area of Quercus rubra (red oak) and a 1475 threefold increase in Quercus montana (Chestnut oak) (Korstian and Stickel 1927). Almost all 1476 the other tree species increased slightly. In other areas different species took advantage of the 1477 released resources. 1478 A neighbour could be very susceptible to the disease, and, whilst not quite being a 1479 magnet, it could spread spores or vectors and increase the disease load on S. For example, 1480 Power and Mitchell (2004, %79) found in a field experiment with plots of 1-6 grass species 1481 that plots containing Avena fatua (wild oats), which was highly susceptible to a particular 1482 generalist aphid-vectored virus, were over 10 × more heavily infected than communities 1483 lacking A. fatua. They termed this ‘pathogen spillover’, but it is a close equivalent of our 1484 herbivore situation “8 -ve: The neighbour attracts the [disease], increasing the [disease] load”. 1485 For those wishing to use the term ‘apparent competition’, it is the disease equivalent. 1486 Possibly, the disease could removes a more susceptible neighbour that a facilitator of S 1487 (herbivory ‘9 -ve’). An equivalent to herbivory mechanism ’10 -ve’ seems just possible. 1488 The impact of diseases is seen most obviously when they appear suddenly, as in the 1489 case of chestnut blight. We can be fairly sure that diseases in the past have had a major hand 1490 in shaping the plant communities that we see today and continue to. (Janzen scatter 1491 hypothesis?) 1492 7 Interactions 1493 7.1 1494 1495 Litter-herbivore interactions Grazing, especially by large grazers, can result in: 1. less natural litter production, both because plant vigour may be reduced by the 1496 removal of photosynthetic organs and because leaves have a lower probability of 1497 reaching senescence, 1498 1499 1500 2. redistribution of lying litter by trampling and habitat use, often speeding decomposition (Eldridge and Rath 2002), 3. an increase in the production of leaf litter induced by animal movement damage, the 1501 tearing of edges of leaves that have been bitten, etc., and more woody litter including 1502 CWD from collateral damage by tree canopy browsers (notably by elephants: 1503 Plumptre 1994), 1504 4. material shed around the point of damage, e.g. Buck-Sorlin and Bell (2000), found 1505 that spring defoliation and shoot tip removal in Quercus spp. (oak) trees in Wales led 1506 to increased shoot shedding; Wilson & Agnew, chapter 2, Interactions, page 48 of 54 1507 5. reduced suppression of sub-canopy species, potentially adding more of their litter. 1508 The complexity of such processes and systems are well illustrated by Long et al. (2003), who 1509 examined the interaction between the beetle herbivore Trirhabda virgata and its food plant 1510 Solidago altissima (goldenrod). Beetle populations were highest in the densest S. altissima 1511 patches, leading to reduced plant vigour and lower litter production. Litter, when present in 1512 quantity, increased aboveground biomass and reduced species richness. Thus the net effect of 1513 the beetle and S. altissima concentrations was to enhance plant species diversity by reducing 1514 litter disturbance. 1515 In heavily grazed vegetation, litter is reduced and can have little or no effect, so that 1516 more of the system of interspecific interactions must depend on growth disturbance, as well as 1517 interference and subvention. This conclusion conflicts with the usual assumption that 1518 competition will be less intense under heavy grazing. There is nothing in the first sentence 1519 that proves that just because litter has no effect, that interference (neg. interactions., aka 1520 competition) and subvention (pos. interactions) are necessarily the forces that are driving 1521 community structure- it is merely stated as a fact without backing. Heavy grazing might give 1522 rise to a community structured by the ability to survive the grazing or to recolonize areas 1523 opened by disturbance, in other words a community structured not be plant-plant interactions, 1524 but by plant-herbivore interactions. It might favour rare species (escape from grazing) or 1525 unpalatable species. Indeed as you state that both negative and positive plant interactions are 1526 more important in these systems, even that statement doesn’t lead to the conclusion that 1527 negative interactions (competition) will increase or even dominate the system interactions. 1528 If only living plant presence is recorded and litter is ignored, only growth interactions 1529 can be assessed. This may explain why the strongest evidence for assembly rules has been 1530 found in uniformly and heavily grazed (i.e. mown) lawns (Wilson and Roxburgh 1994), in a 1531 heavily grazed saltmarsh (Wilson and Whittaker 1995) and a sand dune (Stubbs and Wilson 1532 2004), in all of which there is very little accumulation of litter and interactions due to growth 1533 are not obscured by the litter effect. This suggests that assembly rules should be sought which 1534 include not only the living components but also the litter, as implied by Grime (2001) in his 1535 interpretation of the productivity-richness humpback curve. 1536 CWD can also interact with herbivory. Long et al. (1998) suggest that treefall mounds 1537 provide a refuge from herbivory for Tsuga canadensis (hemlock) regeneration, which is 1538 important in spite of its being a less favourable environment for establishment of that species 1539 than the forest floor. Wilson & Agnew, chapter 2, Interactions, page 49 of 54 1540 7.2 1541 Litter-fire interactions Sometimes the autogenic effect of litter interacts with allogenic disturbance. Fire is 1542 perhaps the best example. If litter accumulates faster than it decomposes, and local conditions 1543 of climate and soil are too dry to allow peat accumulation, its ultimate fate on a landscape 1544 scale is destruction by fire. This is basically an autogenic effect because there can be no fire 1545 without fuel, and the initiating fuel of a fire is usually a buildup of standing dead, 1546 programmed and stochastic litter. However, an allogenic process has to start the fire. After 1547 fire, the litter accumulation cycle restarts. This is a process intrinsic to heathlands and dry 1548 forests over large geographical areas. 1549 Various factors affect the probability and the type of fire. In a flash fire burning 1550 mainly volatiles either in the field layer or the canopy, litter will play little part. However a 1551 fire that consumes the litter will burn for longer. All fire will impact more on the less fire- 1552 resistant and more fire-intolerant species (Whelan 1995). In many vegetation types, branch 1553 fall provides the fuel build up because CWD is long-lived, much of it persisting until burnt. 1554 Treefall also provides fuel (Whelan 1995). The spread of fire into the canopy is facilitated by 1555 the retention of dead limbs (Johnson 1992), so branch fall will help to confine fire to the 1556 ground flora, sparing the canopy stratum. Lianas can have a similar effect: Putz (1991) 1557 observed flames travelling up lianas, with the result that Pinus spp. (pine) trees with lianas 1558 (especially Smilax spp., greenbriar etc.) were more likely to suffer crown scorch than liana- 1559 free trees. Fire is especially prevalent in communities of conifers (particularly Pinus: 1560 Ne’eman and Trabaud 2000) and of Eucalyptus (gum) apparently because of their flammable 1561 resin/oil content (Whelan 1995; Williams et al. 1999). At the other extreme, some types of 1562 vegetation are very fire resistant. For example, in arid central Australia, where most 1563 vegetation is fire-prone, Acacia harpophylla (wattle) does not burn (Pyne 1991) because of 1564 several features: their chemistry makes the leaves barely flammable; the simple leaf shape 1565 results in flat, poorly-aerated litter fuelbeds; leaf fall is keyed to intermittent rains, when fire 1566 is less likely; bark is not shed; and finally the microclimate of the forest floor discourages 1567 drying and promotes decomposition. 1568 7.3 1569 Litter-herbivore-fire interactions There are several examples of good evidence for higher-order interactions between 1570 grazing, litter and fire. For example, in North American Pseudotsuga menziesii (Douglas fir) / 1571 Physocarpus malvaceus (mallow ninebark) forest, grazing by livestock (including cattle, 1572 sheep and deer) reduces the herb layer and hastens the decay of litter by trampling 1573 (Zimmerman and Neuenschwander 1984), and the unpalatable herbs left tend to be less Wilson & Agnew, chapter 2, Interactions, page 50 of 54 1574 flammable. This reduces fire frequency, but allow the buildup of coarse woody debris so that 1575 when fires occur they are more severe and may reach the canopy. A similar effect of grazing 1576 reducing the fuel load and hence the frequency of fires occurs in the African savannah (Fig. 1577 2.8; Roques et al. 2001). In habitats with both a grass and a woody component (savannah or 1578 ecotone) a number of processes can occur (Fig. 2.8; Brown and Sieg 1999; Touchan et al. 1579 1995). 1580 This process includes many complexities. It could converge to an equilibrium, it could 1581 operate as a cyclic succession, or a switch could also operate in the system, giving alternative 1582 stable states (Fig. 2.8; Dublin 1995). In the 1960s and 70s, fire caused a decline in Serengeti 1583 woodland, because higher rainfall then led to greater grass growth (Dublin et al. 1990). 1584 Loxodonta africana (African elephant) inhibited the recovery and kept it as grassland, and 1585 because of their grazing the grass crop is not enough to support hot fires. On the other hand, 1586 elephants avoid the fire-resistant thickets, which are therefore stable as an alternative stable 1587 state. There are both negative and positive feedback processes here. 1588 Wilson & Agnew, chapter 2, Interactions, page 51 of 54 1589 1590 Fire repelled 1591 1592 1593 Woody thicket 1594 remains Elephant browsing 1595 repelled 1596 Woody patch escapes 1597 the fire 1598 Catastrophic fire, 1599 triggered by an 1600 exceptional event1 1601 Trees Woody plants killed, re-establish 1602 at least above ground 1603 1604 Fire frequency Less palatable 1605 reduced forbs dominate 1606 More light is allowed 1607 into the field layer 1608 Litter buildup 1609 reduced 1610 1611 1612 Invasion of Grass growth promoted woody species Fewer old 1613 Litter decays suppressed leaves 1614 faster 1615 1616 1617 Regular Trampling Grazing 1618 fires 1619 Low 1620 High rain rain The grazing Low 1621 resource increases rain Herbivore density 1622 increases 1623 1624 1 1625 A rainy season giving a heavy crop of grass followed by a dry season; or a dry season 1626 making savannah burnable; or high browser density causing CWD accumulation 1627 = switch 1628 Fig. 2.8: Pathways of community change in dry savannah. 1629 8 Conclusion 1630 We have described the many processes that can be involved in the development of Wilson & Agnew, chapter 2, Interactions, page 52 of 54 1631 mixed species stands. Most types of interference, subvention, litter effects and autogenic 1632 disturbance are based on the universal principle of reaction, that organisms in general, and 1633 plants in particular, modify their physical environment. All these reactions have effects on 1634 neighbours. The wide variety of modifications and the wide variety of responses by 1635 neighbours are the reasons for the long list of interactions we gave, a list that we tried to make 1636 exhaustive but surely have not. 1637 Plants must be seen within the framework of the ecosystem. Autotrophs make up most 1638 of the stored living carbon framework of terrestrial ecosystems, i.e. its biomass, and a plant’s 1639 biomass is also a necessary part of its work in foraging for light and soil resources. But 1640 phytomass is never alone. Operators from other trophic levels are always present and can 1641 have profound influences on plant species presence, adding to the complexity. Clements 1642 envisaged reaction as being on the physical environment. We are reluctant to expand this to 1643 the biotic environment because of the ensuing complications, but we have listed interactions 1644 “mediated by other trophic levels”. There are complex interactions amongst heterotrophic 1645 levels too, which will also impact on our dear plants. Paine’s original usage of ‘Keystone 1646 species’ (chapt. 5, sect. 11 below) is a clear expression of these. 1647 The plant community comprises continual renewal: disturbance, death, establishment, 1648 growth and both sexual and vegetative reproduction of great numbers of offspring – greater 1649 than the habitat can support. The negative and positive interactions that we have listed 1650 moderate those processes. In the next chapter we consider how those processes operate at the 1651 whole-community level, to determine successional changes after allogenic or autogenic 1652 disturbance, and to affect the spatial pattern of vegetation. 1653 Wilson & Agnew, chapter 2, Interactions, page 53 of 54 1654 1655 Table 1. Effects of litter on the population dynamics of species in the community, as reported in the literature. 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 Feature Change References Germination (or emergence) reduced or delayed Hamrick and Lee 1987 Peterson and Facelli 1992 Bosy and Reader 1995 Facelli and Kerrigan 1996 Molofsky et al. 2000 Barrett 1931 Facelli and Kerrigan 1996 Myster and Pickett 1993 Cintra 1997 Hamrick and Lee 1987 Myster 1994 Berendse et al. 1994 Xiong and Nilson 1997 Facelli et al. 1999 Facelli and Kerrigan 1996 Facelli et al. 1999 Theimer and Gehring 1999 Barrett 1931 Hamrick and Lee 1987 Facelli and Pickett 1991a Moore 1917 Facelli and Facelli 1993 Facelli 1994 Nilsson et al. 1999 Facelli et al. 1999 Monk and Gabrielson 1985 Peterson and Facelli 1992 Facelli and Facelli 1993 Molofsky et al. 2000 Peterson and Facelli 1992 Facelli et al. 1999 Watt 1931 Scurfield 1953 Beatty and Sholes Facelli and Pickett 1991b Beatty and Sholes 1988 Helvey and Patric 1965 1694 ILLUSTRATIONS 1695 Fig. 2.1. Replacement and additive comparisons in experiments that examine the interaction 1696 increased Seed predation reduced Establishment reduced increased Seedling mortality Biomass increased increased reduced Shoot:root ratio increased Species composition changed Species habitat range Water interception changed increased between species. 1697 Fig. 2.2. In Trifolium subteranneum, the results of competition affect competitive ability. 1698 Fig. 2.3: A possible mechanism for size-asymmetric competition belowground, when nutrient 1699 supplies become available patchily in time and space, e.g. from sheep micturition. Wilson & Agnew, chapter 2, Interactions, page 54 of 54 1700 Fig. 2.4: After Novoplansky et al. (1990). 1701 Fig. 2.5: After Biddington and Dearman (1985). 1702 Fig. 2.6: The seasonal pattern of litterfall in an oakwood. After Christensen (1975). 1703 Fig. 2.7. The process of peat accumulation in mires dominated by Empodisma (wire rush) or 1704 1705 Sphagnum moss species. Fig. 2.8: Pathways of community change in dry savannah. i ii Microscopic algae do move. when planted as 66% of the mixture