Postprint of: Ecology Letters 21(6): 865-874 (2018)
1 Opposing effects of floral visitors and soil conditions on 2 the determinants of competitive outcomes maintain 3 species diversity in heterogeneous landscapes
4
5 Jose B. Lanuza, 1 Ignasi Bartomeus, 2 Oscar Godoy 3*
6 1School of Environmental and Rural Science, University of New England, Armidale, 2350, NSW 7 Australia
8 2 Estacion Biologica de Doñana (EBD-CSIC), C/Americo Vespucio 26, E-41092 Sevilla, Spain.
9 3Instituto de Recursos Naturales y de Agrobiología de Sevilla (IRNAS-CSIC), Avda. Reina 10 Mercedes 10, E -41012 Sevilla, Spain.
11 * corresponding author: [email protected] +34 954 624 711
12 13 Statement of authorship : I.B. and O.G. designed the study. J.B.L and O.G. conducted 14 fieldwork. All authors analysed the results, and J.B.L and O.G. wrote the manuscript 15 with substantial contributions from I.B. 16 Data accesibility statement : Raw data will be uploaded in Dryad upon acceptance. 17 Abbreviated form of title : Species coexistence and multitrophic interactions . 18 Keywords : demography, coexistence, community assembly, fitness, multitrophic 19 interactions, mutualism, niche, pollinators, salinity, spatial structure. 20 Type of article : Letters. 21 Number of words : abstract (150); main text considering equations (4985). 22 Nuber of references : 50. 23 Number of figures, tables and text boxes : figures (4); tables (1); no text boxes.
1 24
25 Abstract 26 Theory argues that both soil conditions and aboveground trophic interactions have 27 equivalent potential to limit or promote plant diversity. However, it remains 28 unexplored how they jointly modify the niche differences stabilising species 29 coexistence and the average fitness differences driving competitive dominance. We 30 conducted a field study in Mediterranean annual grasslands to parameterise 31 population models of six competing plant species. Spatially explicit floral visitor 32 assemblages and soil salinity variation were characterized for each species. Both 33 floral visitors and soil salinity modified species population dynamics via direct 34 changes in seed production and indirect changes in competitive responses. Although 35 the magnitude and sign of these changes were species-specific, floral visitors 36 promoted coexistence at neighbourhood scales while soil salinity did so over larger 37 scales by changing the superior competito rs’identity.Ourresultsshowhowbelow 38 and aboveground interactions maintain diversity in heterogeneous landscapes 39 through their opposing effects on the determinants of competitive outcomes. 40
41 Introduction 42 One central aim in ecology is understanding how plant species diversity is 43 maintained. Extensive empirical work has documented that variation in soil 44 conditions and multitrophic interactions modulate key processes of plant 45 population dynamics. For instance, plant offspring and the strength of competition 46 depend on the combined species' ability to deplete shared limiting soil resources 47 (Tilman 1982) and to cope with stressful soil conditions such as the amount of salt 48 (Bertness & Shumway 1993; Crain et al. 2004). Likewise, mutualistic and 49 antagonistic biotic interactions with herbivores (Hulme 1996; Olff & Ritchie 1998), 50 soil biota (Bever 2003; Bennett et al. 2017; Teste et al. 2017), pathogens (Mordecai 51 2015; Parker et al. 2015), and floral visitors (Bastolla et al. 2009; Carvalheiro et al. 52 2014; Weber & Strauss 2016) can profoundly impact plant performance. It is 53 obvious that outcrossing plants directly depend on their mutualistic floral visitors to 54 maximize their reproductive success (Morris et al. 2010; Ollerton et al. 2011). 55 Subtler is the fact that floral visitors indirectly mediate competition among plants 56 through a wide variety of density-dependent processes including variation in the 57 number and diversity of floral visitors as well as heterospecific pollen deposition 58 (Moeller 2004; Arceo-Gómez & Ashman 2011; Runquist & Stanton 2013). 59 Despite the fact that variation in soil conditions and multitrophic biotic interactions 60 occurs simultaneously in nature, evaluations of their effects on modulating the 61 strength and sign of plant competition have been explored individually. It remains
2 62 unknown how these drivers can jointly maintain plant diversity. Theoretical work 63 has advocated that both types of interactions should be viewed symmetrically as 64 they have equivalent potential to limit or promote diversity (Chesson & Kuang 65 2008). Although empirical tests of this prediction have remained so far elusive, we 66 can progress by framing our research within recent advances of coexistence theory 67 (Chesson 2000), and by learning from prior work applying these theoretical 68 advances to multitrophic antagonistic interactions, mainly predators and pathogens 69 (Chesson & Kuang 2008; Kuang & Chesson 2010; Stump & Chesson 2017). 70 ccordingtoChesson’s(2000)framework,bothsoilconditionsandfloralvisitors 71 can promote the stabilising niche differences that favour plant coexistence, which 72 occur when intraspecific competitive interactions exceed interspecific competition, 73 and the average fitness differences that favour competitive exclusion and determine 74 the competitive winner in the absence of niche differences. Ecologists have paid 75 much more attention to the relationship of these two factors with fitness differences 76 (e.g. soil conditions (Tilman 1982; Casper & Jackson 1997; Rees 2013); floral 77 visitors (Herrera 2000; Waites & Ågren 2004; Arceo-Gómez & Ashman 2011)) than 78 with niche differences (Silvertown 2004; Levine & HilleRisLambers 2009) and this 79 is particularly evident for floral visitors (Pauw 2013). Most likely both drivers 80 modify niche and fitness differences simultaneously, yet the extent of such 81 modifications is poorly understood. Therefore, a rigorous evaluation of the 82 equivalent potential of soil conditions and floral visitors on maintaining plant 83 diversity can only be done by a mechanistic understanding of how these two types 84 of interactions relatively modify the determinants of competitive outcomes. 85 When relating theory to field experiments, it is important to consider two critical 86 aspects. One is selecting ecological systems that are relatively easy to observe. For 87 example, recent work in Mediterranean annual grassland (Godoy & Levine 2014) 88 has described how niche and fitness differences influence species' population 89 dynamics. Some of these grasslands are subjected to strong variation in soil salinity, 90 which negatively correlates with soil fertility (Olff & Ritchie 1998; Hu & 91 Schmidhalter 2005). Moreover, floral visitor assemblages in these Mediterranean 92 environments are particularly interesting because they are composed of an array of 93 insects including solitary bees, hover flies, beetles, and butterflies. While some of 94 these floral visitor types act as true mutualisms (Pauw 2013), others rob plant 95 nectar or pollen or damage flowers (Morris et al. 2003). The second aspect is that 96 these drivers of plant competition tend to show spatial structure (Tilman 1994; 97 Weber & Strauss 2016). Coexistence theory predicts that plant diversity can be 98 maintained at the neighbourhood scale when species niche differences overcome 99 fitness differences. This can occur when either plant competitor equalises its fitness 100 differences, increases its niche differences, or a combination of both. If not, the 101 superior competitor excludes the inferior species (Chesson 2000). However, 102 competitive exclusion outcomes can also maintain plant diversity at larger scales if 103 the spatial structure of variation in soil salinity and floral visitors change the 104 identity of the superior competitor across locations. This latter process might be the
3 105 cause of reported turnover patterns of species and functional attributes across soil 106 salinity gradients (Bertness 1991; Pavoine et al. 2011). 107 Here, we considered three layers to test how the belowground environmental 108 conditions (i.e. soil conditions) and the aboveground trophic interactions (i.e. floral 109 visitors) influence coexistence of the middle layers (i.e. plant species). We 110 specifically focus on three questions: (1) How do soil salinity and floral visitors 111 modify species' population dynamics via direct changes in per capita seed 112 production and indirect changes in species' responses to competitive interactions? 113 (2) Do these direct and indirect effects modify niche and fitness differences between 114 plant species?, and finally, (3) At which spatial scale are these modifications on the 115 determinants of competitive outcomes limiting or promoting diversity? 116 We answered these three questions by first parameterizing a general plant 117 competition model from which the stabilising niche differences and average fitness 118 differences were quantified. To parameterise these models of pairwise competition 119 between six annual native grassland species, we quantified their vital rates and 120 competition coefficients in field plots relating seed production of focal individuals to 121 a density gradient of numerous different competitors. We then assessed how 122 seasonal and spatial variation in the number of floral visits and in soil salinity 123 changes species fecundity and their responses to competition (Question 1). Once, 124 the model was parameterised, we estimated niche and fitness differences with and 125 without considering the effect of soil salinity and floral visitors on species fecundity 126 (Question 2), and compared how strong niche differences offset fitness differences 127 between scenarios (Question 3). Our work is novel in quantifying the effects that 128 distinct environmental conditions and trophic interactions have on modifying niche 129 and fitness differences, and showing under field conditions that they maintain 130 diversity in heterogeneous landscapes through their opposing effects on the 131 determinants of competitive outcomes.
132 Methods 133 Study system 134 Our study was conducted in Caracoles Ranch (2,680 ha), an annual grassland system 135 located in Doñana NP, southwest Spain (37º04'01.5"N 6°19'16.2"W). The climate is 136 Mediterranean with mild winters and average 50-y annual rainfall of 550-570 mm 137 with high interannual oscillations (Muñoz-Reinoso & García Novo 2000). Soils are 138 sodic saline (electric conductivity > 4dS/m and pH < 8.5) and annual vegetation 139 dominates the grassland with no perennial species present. The study site has a 140 subtle micro topographic gradient (slope 0.16%) enough to create vernal pools at 141 lower parts from winter (November-January) to spring (March-May) while upper 142 parts do not get flooded except in exceptionally wet years. A strong salinity- 143 humidity gradient is structured along this topographic gradient. Additionally, salt 144 can reach upper parts of the soil by capillarity resulting overall in heterogeneous 145 soil salinity patterns at the local and at the landscape scale (Appendix S1). This
4 146 salinity gradient is strongly correlated with soil nutrient availability at our study 147 location, and more saline conditions correlate with less fertile soils (Clemente et al. 148 2004). 149 We recorded 19 annual plants at the study site. Of this initial species set, three were 150 not further considered due to their low abundance (only recorded in 5 out of the 151 324 subplots evaluated). The 16 species finally selected represent a broad range of 152 taxonomic families, plant morphology and flowering phenology co-occurring at the 153 scale of the entire study system. All species were considered for estimating 154 competitive interactions, but we only observed enough visits of insects to the 155 flowers of six species. Hence, we further focus on this particular set of species to 156 compare the effect of soil salinity and floral visitors on niche and fitness differences 157 (Table 1). 158 Modeling approach to quantify the niche and fitness differences between 159 species pairs 160 Our observational study was designed to field-parameterise a mathematical model 161 describing annual plant population dynamics (Levine & HilleRisLambers 2009). This 162 model allows quantifying stabilising niche differences and average fitness 163 differences between species within a trophic level (Godoy & Levine 2014). 164 Importantly, there have not been previous attempts to quantify how soil condition 165 or multitrophic interactions change the strength of niche and fitness differences 166 between species within a single trophic level, and here we show how these effects 167 can be incorporated into this model. The model is described as follows:
168 (1)
169 where is the per capita population rate, and is the number of seeds of 170 species i in the soil prior to germination in winter of year t. The germination rate of 171 species i, , can be viewed as a weighting term for an average of two different 172 growth rates: the annual survival of ungerminated seed in the soil ( ), and the 173 viable seeds produced per germinated individual ( ). In past work, , was 174 expanded into a function describing how the average fecundity of each germinated 175 seed that becomes an adult (i.e. per germinant fecundity) declines with the density 176 of competing number of individuals in the system (Godoy & Levine 2014). Now, we 177 slightly modify this function to include the additional effect of soil conditions and 178 floral visitors on the per germinant fecundity as follows:
179 (2) 180
5 181 where and control the effect of soil salinity ( ) and floral visitors ( ) 182 respectively on the per germinant fecundity of species i in the absence of 183 competition ( ). In addition, is modified by the germinated densities of other 184 species including its own ( ). To describe the per capita effect that species j is 185 mediating on species i, we multiplied these germinated densities by a sum of three 186 interaction coefficients ( + + ), which describes the additional direct 187 effect of soil salinity and the apparent effect of floral visitors on the competitive 188 interactions between species. Notice that we considered explicitly in our study the 189 effect that soil salinity and floral visitors have on species' fecundity ( ), but not on 190 the other two vital rates, seed germination ( ) and seed soil survival ( ). This is 191 because we did not follow seed performance over several years not because the 192 model cannot be extended to account for multitrophic effects on these two vital 193 rates. 194 With the direct and apparent dynamics of competition described by this population 195 model, we followed the approach of Chesson (2013) to determine fitness and niche 196 differences between species pairs. Our procedure here parallels previous work 197 described in Godoy & Levine (2014), and allows us to define stabilising niche 198 differences and fitness differences without and with considering the effect of floral 199 visitors and soil salinity on plant coexistence. For the model described by eqns (1) 200 and (2), we define niche overlap ( ) as follows:
201 (3) 202 203 If multitrophic interactions are not considered by setting their values to zero (i.e. 204 =0, =0), then collapses into an equation that reflects the average degree to 205 which species limit individuals of their own species relative to heterospecific 206 competitors based on their interaction coefficients ( 's) (Godoy & Levine 2014). 207 Conversely, if multitrophic interactions are considered as present (i.e. = = 208 1) and are the terms controlling changes in average niche differences between a 209 pair of species. For example, two species with different sets of floral visitors could 210 increase niche overlap by having positive apparent competitive effects of each 211 species on the other (i.e >0). With ( ) defining niche overlap between a pair of 212 species, their stabilising niche difference is expressed as 1- . 213 As an opposing force to stabilising niche differences, average fitness differences 214 drive competitive dominance, and in the absence of niche differences, determine the 215 competitive superiority between a pair of species. Addressing the modifications 216 done in the annual population model described by eqns (1) and (2) to include the
6 217 effect of floral visitors and soil conditions, we define average fitness differences 218 between the competitors ( ) as:
219 (4) 220 221 and
222 When the ratio >1 this indicates that species j has a fitness advantage over species 223 i. Both soil salinity and floral visitors can be seen as equalising mechanisms 224 promoting coexistence because they can reduce fitness differences between a 225 species pair by two contrasted pathways. They can modify the 'demographic ratio' 226 ( ) which describes the degree to which species j produces more seeds 227 ( ) per seed loss due to death or germination (1-(1- ) ) 228 than species i, and they can also modify the 'competitive response ratio' 229 ( ) which describes the degree to which species j is less 230 sensitive to competition than species i (eqn (4)). Notice that these modifications can 231 produce the opposing effect and promote species' competitive dominance by a 232 combination of high demographic rates and low sensitivity to competition. 233 Competitors can coexist when niche differences overcome fitness differences, 234 allowing both species to invade (i.e. increase its populations) when rare (Chesson 235 2013). This condition for mutual invasibility is satisfied when:
236 (5) 237 238 Therefore, coexistence at the neighbourhood scale occurs when niche differences 239 between species pairs overcome their differences in fitness. If not the species with 240 higher fitness exclude the inferior competitor from the neighbourhood. We used this 241 condition to evaluate how strongly soil salinity and floral visitors increase or 242 decrease the likelihood of coexistence between competitors across scales. 243 Specifically, we computed for each species pairs the differences between observed 244 and predicted niche differences needed to determine stable coexistence at the
7 245 neighbourhood scale according to the observed asymmetry in fitness between 246 species. 247 Field observations used to parameterise the model 248 In September 2015, we established nine plots of 8.5m x 8.5m along a 1km x 200m 249 area. Three of these nine plots where located in the upper part of the topographic 250 gradient, three at the middle, and the last three at the lower part. Average distance 251 between these three locations was 300m and average distance between plots within 252 each location was 15m (minimum distance 10m). In addition, each plot was divided 253 in 36 subplots of 1m x 1m with aisles of 0.5m in between to allow access to subplots 254 where measurements were taken (total of 324 subplots) (Appendix S2). 255 This spatial design was established to parameterise for each focal species the annual 256 plant population model (eqns. 1 and 2). Specifically, the core of the observations 257 involved measuring per germinant viable seed production as a function of the 258 number and identity of neighbours within a radius of 7.5cm including individuals of 259 the same species (see analyses below). This radius is a standard distance used in 260 previous studies to measure competitive interactions among annual plant species 261 (Levine & HilleRisLambers 2009; Mayfield & Stouffer 2017), and has been validated 262 to capture the outcome of competition interactions at larger scales ( ) under 263 homogeneous environmental conditions (Godoy & Levine 2014). We measured one 264 individual per subplot for widespread species and several individuals per subplot 265 when species were rare (max. 324 individuals/species). To additionally incorporate 266 the effect of soil salinity, from November 2015 to June 2016 we measured soil 267 humidity (%) and soil salinity (dS/m) bimonthly at the subplot center with a TDR 268 (Time Domain Reflectometer) incorporating a 5cm probe specially designed and 269 calibrated for these sodic saline soils (EasyTest, Poland). We summarised the 270 amount of soil salinity experienced by each germinant, which was highly correlated 271 with soil moisture (r=0.77), as the sum over their lifetime of the soil salinity 272 measured at the subplot scale. 273 Moreover, floral visitors were measured during the complete phenological period of 274 all species (from January to June 2016). We surveyed weekly the number of floral 275 visitors for all species within each subplot. Visits were only considered when the 276 floral visitor touched the reproductive organs of the plant, and insects were 277 collected with a hand net. All subplots within a plot were simultaneously surveyed 278 during 30 minutes each week. Plot survey was randomized between weeks to avoid 279 sampling effects. Overall, this procedure rendered approximately 90 hours of 280 overall floral visitors sampling. Floral visitors to each species and subplot were 281 grouped in four main taxonomic groups (bees, beetles, butterflies, and flies). We 282 summarised the number of floral visits by insects to each germinant as the total sum 283 of visits at the subplot scale. 284 Finally, we quantified the germination of viable seeds ( ) by counting the number 285 of germinants in 18 quadrats of 1m x 1m placed close to the plots (2 quadrats per 286 plot) from seeds collected the previous year and sown on the ground prior to the
8 287 first major storm event after summer (September 2015). Similarly, we quantified 288 seed bank survival ( ) with the same seed material by burying seeds from 289 September 2015 to September 2016 following the methods of (Godoy & Levine 290 2014). 291 Analysis 292 To determine the role of soil salinity separated from the role of floral visitors in 293 plant fecundity (eqn. 2), we used maximum likelihood methods to fit empirical 294 observations in the following way. We fit changes in and (both bounded to be 295 positive) as a function of the total number of floral visitors ( ) and the accumulated 296 soil salinity ( ) experienced by each germinant at the subplot scale over their 297 lifetime (optim method="L-BFGS-B"). The effect of floral visitors on fecundity was 298 calculated by either considering all visits together or separating visits from each of 299 the four taxonomic groups. Soil salinity ( , ) and floral visitors ( , ) 300 parameters were not bounded to any specific range as we hypothesized that they 301 can have both positive and negative effects on per germinant fecundities. We further 302 distinguished using AIC (Akaike Information Criterion) whether soil salinity and 303 floral visitor effects on competition were specific to each pairwise interaction 304 (model 1), common to all interactions (model 2), or had no effect (model 3) 305 (Appendix S4 for details). For all three models, individuals of the other 10 species 306 surveyed apart from our six focal species were separated into two groups 307 depending on their degree of relatedness, either confamilial or heterofamilial, and 308 their competitive effects on the focal species were summarized as independent 309 parameters. Estimates of mean and standard error for each parameter of the best 310 model selected by AIC across species are included in Appendix S3. Note that model 311 estimations of values were not validated experimentally so we do not know the 312 species' selfing rates. All analyses were conducted in R (version 3.3.1) (R Core Team 313 2016).
314 Results 315 The six focal species experienced a great variation in soil salinity and the type and 316 number of floral visitors. Along the salinity gradient, Beta macrocarpa and Pulicaria 317 paludosa grew mainly in high soil salinity concentrations, in contrast, Melilotus 318 elegans and Leontodon maroccanus grew in relatively low saline soils, while 319 Chamaemelum fuscatum and Melilotus sulcatus showed a more tolerant behaviour 320 growing in a wider range of salt concentrations (Fig. 1). Number of floral visits by 321 insects also varied greatly among plant species. Overall, the main groups of floral 322 visitors in our system were flies (581 visits) and beetles (496 visits), followed by 323 bees (161 visits) and butterflies (87 visits). The three Asteraceae species were the 324 most visited species. Among them, L. maroccanus received 636 visits followed by C. 325 fuscatum 293 visits, and P. paludosa 291 visits. The rest of the species B. macrocarpa 326 (64), and the Fabaceae congeners M. sulcatus (35) and M. elegans (6) had in 327 comparison a much lower number of visits. Moreover, species also showed variation
9 328 in the assemblage of floral visitors. Of the three plant species with higher number of 329 visits, flies were the most abundant insects visiting C. fuscatum , and P. paludosa 330 while beetles were the most abundant for L. maroccanus (Fig. 1). 331 The wide variation in soil salinity concentrations and the number of floral visits 332 observed in our study modified the seed production in the absence of neighbours 333 ( )andthestrengthofthespecies’responsesto competitive interactions ( ) of 334 the three Asteraceae species plus M. sulcatus (model 2, lowest AIC values, Appendix 335 S4), though AIC values did not help to distinguish between models 2 and 3 for P. 336 paludosa (AIC < 10). Interestingly, the sign of the f loralvisitors’effectson and 337 varied among these species. While higher number of visits to C. fuscatum increased 338 its potential fecundity, and reduced the negative effect of both intra and interspecific 339 competition on seed production, the opposite pattern was observed for L. 340 maroccanus and P. paludosa (Fig. 2). The floral visitor groups that contribute greater 341 to these positive and negative effects were those that visited each focal plant species 342 more frequently, bees and flies in the case of C. fuscatum , beetles in the case of L. 343 maroccanus , and flies in the case of P. paludosa (Appendix S5 and S6). Soil salinity, in 344 contrast, had a similar effect across species increasing seed production in the 345 absence of neighbours and promoting weaker competitive interactions. For the 346 other two non-Asteraceae species, AIC values suggest that soil salinity and floral 347 visitors did not have a strong effect on and . In neither case, did model 348 selection support the view that floral visitors and soil salinity separately modified 349 each pairwise competitive interaction (i.e. model 1 showed consistently higher AIC 350 values) (Appendix S4). 351 Soil salinity and floral visitors exerted positive, negative or no effect on plant 352 fecundity, yet they modified the determinants of competitive outcomes in opposite 353 and specific directions (Fig. 3). While floral visitors tend to maintain stable 354 coexistence at the neighbourhood scale (3 out of 15 species pairs), or to promote 355 coexistence by equalising fitness differences (8 out of 15 pairs moved from the 356 exclusion region into, or closer to, the coexistence region), soil salinity tended to 357 promote competitive exclusion by increasing competitive asymmetries between 358 species pairs (4 pairs moved out of the coexistence region) (Fig. 3). As a result, floral 359 visitors significantly reduced the niche differences needed for coexistence at the 360 neighbourhood scale (estimated from the mutual invasibility, eqn.5) across species 361 pairs (paired t-test, t = 2.45, P = 0.046) (for separated effects of each floral visitor 362 guild, see Appendix S7), while soil salinity increased on average the niche 363 differences needed for coexistence (paired t-test, t = 5.72, P < 0.001). 364 When comparing how strongly soil salinity and floral visitors modify the 365 determinants of competitive outcomes, we observed that they do not have 366 equivalent potential to limit or promote diversity at neighbourhood scales. The 367 effect of soil salinity on increasing competitive asymmetries between species 368 overwhelmed in most cases the equalising effect of floral visitors (Fig. 4). 369 Nevertheless, the identity of competitive winners (estimated from eqn. 4) changed 370 in one third of the species pairs (5 out of 15). Overall these results suggest that plant
10 371 diversity in our system is primarily maintained by the effect of soil salinity on 372 changing the dominant competitor across contrasting soil conditions.
373 Discussion 374 Recent work has increased awareness among ecologists of the equivalent potential 375 of soil conditions and aboveground multitrophic interactions on promoting or 376 impeding diversity maintenance (Chase et al. 2002; Chesson & Kuang 2008). 377 However, empirical tests of this prediction have remained elusive due to the 378 difficulties in connecting theory with detailed field observations. Our ability to 379 combine coexistence theory advances with plant population models and spatially 380 explicit observations provides direct evidence that variation in soil salinity content 381 and floral visitor frequency modify the likelihood of plant coexistence, yet they do so 382 in opposite directions. While floral visitors' effects on plant fecundity promoted 383 plant diversity maintenance at the neighbourhood scale, soil salinity effects drove 384 competitive exclusion. These modifications occurred via direct changes in per capita 385 seed production and indirect changes in competitive responses. Nevertheless, the 386 strength of these modifications differed between these two drivers. Variation in soil 387 salinity overrode in many species pairs the effect of variation in floral visitors on 388 promoting plant coexistence at the neighbourhood scale but it changed the identity 389 of the superior competitor across contrasted saline conditions (Figs. 3 and 4). 390 Overall, these results show that both soil conditions and aboveground multitrophic 391 interactions are important for maintaining plant diversity, but suggest that, at least 392 in our system, soil conditions were more determinant at neighbourhood scales, and 393 that regional diversity is primarily maintained by spatial changes in soil salinity 394 conditions occurring over larger scales. 395 Floral visitors consistently promoted species coexistence at the neighbourhood 396 scale by reducing the niche differences needed to overcome fitness differences 397 between species pairs (Fig. 3a). The positive effect of floral visitors on diversity did 398 not only occur due to a positive effect expected from mutualistic interactions. 399 Rather, we observed both positive and negative effects on plant fecundity. For 400 instance, floral visitors strongly increased the seed production in the absence of 401 competition and reduced to a lesser extent the negative effect of competition on the 402 seed production of C. fuscatum individuals (Fig. 2a). At the other extreme, floral 403 visitors reduced the fecundity of species such as L. marocanum and P. paludosa by 404 both reducing seed production in the absence of competition and increasing their 405 sensitivity to competition (Fig. 2b and 2c). Detailed analyses of the effect of each 406 particular group of floral visitors on plant fecundity showed that the strongest 407 effects on each focal plant species were exerted by their main group of floral visitors 408 (Fig.1)(Appendix S5, S6 and S7). The most common groups visiting C. fuscatum 409 individuals were bee and fly pollinators (mostly Syrphidae species), which are 410 considered mutualistic species. In contrast, beetles visiting primarily L. marocanum 411 flowers were pollen feeders belonging to the families Chrysomelidae and Melyridae 412 (Wäckers et al. 2007) and the principal visitors of the late-flowering species P.
11 413 paludosa were mainly Bombyliidae species, which are poor pollinators (Polidori et 414 al. 2005). 415 Critically, the equalising effect of floral visitors on plant coexistence likely happened 416 becausepositiveandnegativeeffectswereinfluencedbythespecies’competitive 417 ability. The negative effect of floral visitors occurred for those species that were, on 418 average, superior competitors, whereas positive effects occurred for the inferior 419 competitors. This process arises from the fact that our system was dominated by 420 non-specialist interactions and may be a common scenario in this type of system. 421 For instance, beetles acted as herbivores that tend to focus on the most abundant 422 resource, and therefore target the most abundant species (Table 1). Meanwhile, 423 species with high pollinator dependence (i.e. self-incompatible mating system) tend 424 to be subdominant and the ones that benefit substantially from pollinator visits (Tur 425 et al. 2013). Although we did not observe that floral visitors increased niche 426 differences between plant species in our system (Fig. 3a), this does not mean that 427 this stabilising effect can occur in more specialised systems. It is reasonable to argue 428 that equalising and stabilising effects occur in combination, as many plant species 429 trade-off between being sufficiently specialised to differentiate in their pollination 430 niche, while being able to attract a sufficient number of mutualistic partners 431 (Vamosi et al. 2014; Coux et al. 2016). 432 Conversely to floral visitors, soil salinity promoted competitive exclusion at the 433 neighbourhood scales of species interactions by reducing niche differences while 434 increasing fitness differences among species pairs (Fig. 3b). Nevertheless, the 435 identity of the competitive winner changed across contrasting soil salinity 436 conditions. For instance, B. macrocarpa and L. maroccanus were competitive 437 winners against P. paludosa under low soil salinity concentrations but losers under 438 high soil salinity concentrations. For the particular case of P. paludosa , competitive 439 superiority came mostly from the strong positive effect of salinity in reducing its 440 sensitivity to competitive interactions rather than from an increase in the species' 441 ability to produce seeds in the absence of neighbours (Fig. 2f). The consistent effect 442 of soil salinity in determining competitive exclusion across species pairs predicts 443 reduction of species diversity in homogeneous landscapes under constant soil 444 salinity conditions, favouring species that either prefer or refuse salt. But in 445 heterogeneous landscapes like our system, diversity is maintained because of the 446 species' inability to be competitive superiors across all soil salinity conditions. 447 Indeed, these results align with the well-known effect of environmental 448 heterogeneity on promoting diversity (Chesson 2000), and agree also with spatial 449 patterns of species turnover found for very similar salty grasslands in other 450 Mediterranean areas (Pavoine et al. 2011). Yet, our results highlight that 451 competitive interactions rather than niche partition (see Allouche et al. 2012) is 452 likely the main mechanism driving documented patterns of species turnover. 453 Our methodological approach is novel in showing how to incorporate the effect of 454 different abiotic and biotic variables into the estimation of niche and fitness 455 differences between species pairs from models that describe species population 456 dynamicsviaspecies’vitalratesandinteractioncoefficients.Thisapproachallows
12 457 experimental testing of the prediction that soil conditions and aboveground 458 multitrophic interactions have equivalent potential for promoting or limiting 459 diversity maintenance. Our methodology is readily available to be extended to 460 consider other soil conditions such as nutrient content or other kinds of interactions 461 beyond the scope of this study such as herbivores, leaf pathogens and root 462 mutualisms. These soil conditions and interactions could potentially explain 463 changes in fecundity of those species for which soil salinity and floral visitors did 464 not have a significant effect (Landwehr et al. 2002; Pan et al. 2015), or could be 465 indirectly influencing observed patterns. 466 Another important step when studying the effect of multitrophic interactions on 467 plant coexistence is to move from di rectpairwiseeffectstoinclude“higherorder 468 effects”amongspecies(Mayfield&Stouffer2017).Higherordereffectsoccurwhen 469 the presence of a third competitor changes per capita competitive interactions 470 within a species pair. One main challenge to this is to achieve sufficient sampling 471 size to capture the variability in species composition and multitrophic interactions 472 (Levine et al. 2017). Yet, our study was not able to capture this complexity as model 473 selection by AIC highlighted a common effect of floral visitors and soil salinity across 474 species. However, this could be caused because we measured interactions in a 475 relatively dry year and the abundance of some floral visitor groups such as bees and 476 butterflies were relatively low. This last point also makes us aware that climatic 477 variability across years is another layer of complexity that we do not include in our 478 study. Variation between years in the amount of rainfall can change the spatial 479 configuration of soil salinity conditions. Also it can change the abundance, the 480 strength and the specificity of the effect of floral visitors on plant fecundity. 481 Together, our study shows that soil conditions and multitrophic interactions 482 represented by floral visitors have contrasting outcomes in determining coexistence 483 at the neighbourhood scale of plant species interactions. While variation in soil 484 salinity promotes competitive exclusion, variation in floral visitors promotes 485 coexistence. These differences were mostly explained by equalising processes 486 rather than by stabilising processes. Nevertheless, soil salinity variation was the 487 primary driver of plant diversity in our systems and promoted plant coexistence 488 over larger scales by changing the identity of the competitive winner under 489 contrasting soil salinity conditions. Our results highlight that the spatial structure of 490 soil conditions and multitrophic interactions needs to be considered explicitly when 491 evaluating their effects on maintaining species diversity.
492 Acknowledgments 493 We thank three anonymous referees for their helpful comments on the manuscript. 494 O.G. acknowledges postdoctoral financial support provided by the European Union 495 Horizon 2020 research and innovation program under the Marie Sklodowska-Curie 496 grant agreement No 661118-BioFUNC. I.B and O.G. acknowledge also financial 497 support provided by the Spanish Ministry of Economy and Competitiveness, Explora 498 Program (CGL2014-61590-EXP, LINCX). Francisco Rodriguez-Sanchez provided
13 499 useful tips and templates to develop all this work in R markdown. Curro Molina and 500 Oscar Aguado helped with field work and identifying insect species. We finally thank 501 the Doñana NP staff for granting access to Caracoles Ranch, and Manu Saunders for 502 reviewing the manuscript.
503 References 504 1. 505 Allouche, O., Kalyuzhny, M., Moreno-Rueda, G., Pizarro, M. & Kadmon, R. (2012). 506 Area –heterogeneity tradeoff and the diversity of ecological communities. Proc. Natl. 507 Acad. Sci. USA. , 109, 17495 –17500. 508 2. 509 Arceo-Gómez, G. & Ashman, T.-L. (2011). Heterospecific pollen deposition: Does 510 diversity alter the consequences? New Phytol. , 192, 738 –746. 511 3. 512 Bastolla, U., Fortuna, M.A., Pascual-García, A., Ferrera, A., Luque, B. & Bascompte, J. 513 (2009). The architecture of mutualistic networks minimizes competition and 514 increases biodiversity. Nature , 458, 1018 –1020. 515 4. 516 Bennett, J.A., Maherali, H., Reinhart, K.O., Lekberg, Y., Hart, M.M. & Klironomos, J. 517 (2017). Plant-soil feedbacks and mycorrhizal type influence temperate forest 518 population dynamics. Science , 355, 181 –184. 519 5. 520 Bertness, M.D. (1991). Interspecific interactions among high marsh perennials in a 521 new england salt marsh. Ecology , 72, 125 –137. 522 6. 523 Bertness, M.D. & Shumway, S.W. (1993). Competition and facilitation in marsh 524 plants. Amer. Nat. , 142, 718 –724. 525 7. 526 Bever, J.D. (2003). Soil community feedback and the coexistence of competitors: 527 Conceptual frameworks and empirical tests. New Phytol. , 157, 465 –473. 528 8. 529 Carvalheiro, L.G., Biesmeijer, J.C., Benadi, G., Fründ, J., Stang, M. & Bartomeus, I. et al. 530 (2014). The potential for indirect effects between co-flowering plants via shared 531 pollinators depends on resource abundance, accessibility and relatedness. Ecol. 532 Lett. , 17, 1389 –1399. 533 9. 534 Casper, B.B. & Jackson, R.B. (1997). Plant competition underground. Annu. Rev. Ecol. 535 Syst. , 28, 545 –570.
14 536 10. 537 Chase, J.M., Abrams, P.A., Grover, J.P., Diehl, S., Chesson, P. & Holt, R.D. et al. (2002). 538 The interaction between predation and competition: A review and synthesis. Ecol. 539 Lett. , 5, 302 –315. 540 11. 541 Chesson, P. (2000). Mechanisms of maintenance of species diversity. Annu. Rev. Ecol. 542 Syst. , 31, 343 –366. 543 12. 544 Chesson, P. (2012). Species competition and predation. In: Encyclopedia of 545 Sustainability Science and Technology . (ed. Meyers, R.A.). Springer, New York, pp. 546 10061 –10085. 547 13. 548 Chesson, P. & Kuang, J.J. (2008). The interaction between predation and competition. 549 Nature , 456, 235 –238. 550 14. 551 Clemente, L., García, L.V., Espinar, J.L., Cara, J.S. & Moreno, A. (2004). Las marismas 552 del Parque Nacional de Doñana. Investigación y Ciencia , 332, 72 –83. 553 15. 554 Coux, C., Rader, R., Bartomeus, I. & Tylianakis, J.M. (2016). Linking species functional 555 roles to their network roles. Ecol. Lett. , 19, 762 –770. 556 16. 557 Crain, C.M., Silliman, B.R., Bertness, S.L. & Bertness, M.D. (2004). Physical and biotic 558 drivers of plant distribution across estuarine salinity gradients. Ecology , 85, 2539 – 559 2549. 560 17. 561 Godoy, O. & Levine, J.M. (2014). Phenology effects on invasion success: Insights from 562 coupling field experiments to coexistence theory. Ecology , 95, 726 –736. 563 18. 564 Herrera, C.M. (2000). Flower-to-seedling consequences of different pollination 565 regimes in an insect-pollinated shrub. Ecology , 81, 15 –29. 566 19. 567 Hu, Y. & Schmidhalter, U. (2005). Drought and salinity: A comparison of their effects 568 on mineral nutrition of plants. J. Plant Nutr. Soil Sci. , 168, 541 –549. 569 20. 570 Hulme, P.E. (1996). Herbivory, plant regeneration, and species coexistence. J. Ecol. , 571 84, 609 –615. 572 21. 573 Kuang, J.J. & Chesson, P. (2010). Interacting coexistence mechanisms in annual plant
15 574 communities: Frequency-dependent predation and the storage effect. Theor. Popul. 575 Biol. , 77, 56 –70. 576 22. 577 Landwehr, M., Hildebrandt, U., Wilde, P., Nawrath, K., Tóth, T. & Biró, B. et al. (2002). 578 The arbuscular mycorrhizal fungus Glomusgeosporum in European saline, sodic and 579 gypsum soils. Mycorrhiza , 12, 199 –211. 580 23. 581 Levine, J.M. & HilleRisLambers, J. (2009). The importance of niches for the 582 maintenance of species diversity. Nature , 461, 254 –257. 583 24. 584 Levine, J.M., Bascompte, J., Adler, P.B. & Allesina, S. (2017). Beyond pairwise 585 mechanisms of species coexistence in complex communities. Nature , 546, 56 –64. 586 25. 587 Mayfield, M.M. & Stouffer, D.B. (2017). Higher-order interactions capture 588 unexplained complexity in diverse communities. Nat. Ecol. Evol. , 1, 0062. 589 26. 590 Moeller, D.A. (2004). Facilitative interactions among plants via shared pollinators. 591 Ecology , 85, 3289 –3301. 592 27. 593 Mordecai, E.A. (2015). Pathogen impacts on plant diversity in variable 594 environments. Oikos , 124, 414 –420. 595 28. 596 Morris, W.F., Bronstein, J.L. & Wilson, W.G. (2003). Three-way coexistence in 597 obligate mutualist-exploiter interactions: The potential role of competition. Am. 598 Nat. , 161, 860 –875. 599 29. 600 Morris, W.F., Vázquez, D.P. & Chacoff, N.P. (2010). Benefit and cost curves for typical 601 pollination mutualisms. Ecology , 91, 1276 –1285. 602 30. 603 Muñoz-Reinoso, J.C. & García Novo, F. (2000). Vegetation patterns on the stabilized 604 sands of Doñana Biological Reserve. In: Proceedings IAVS symposium . pp. 162 –165. 605 31. 606 Olff, H. & Ritchie, M.E. (1998). Effects of herbivores on grassland plant diversity. 607 Trends Ecol. Evol. , 13, 261 –265. 608 32. 609 Ollerton, J., Winfree, R. & Tarrant, S. (2011). How many flowering plants are 610 pollinated by animals? Oikos , 120, 321 –326.
16 611 33. 612 Pan, C., Zhao, H., Feng, Q., Liu, J., Liu, L. & Cai, Y. et al. (2015). Temporal variations of 613 ground-dwelling arthropods in relation to grassland salinization. Europ. J. Soil Biol. , 614 68, 25 –32. 615 34. 616 Parker, I.M., Saunders, M., Bontrager, M., Weitz, A.P., Hendricks, R. & Magarey, R. et 617 al. (2015). Phylogenetic structure and host abundance drive disease pressure in 618 communities. Nature , 520, 542 –544. 619 35. 620 Pauw, A. (2013). Can pollination niches facilitate plant coexistence? Trends Ecol. 621 Evol. , 28, 30 –37. 622 36. 623 Pavoine, S., Vela, E., Gachet, S., De Bélair, G. & Bonsall, M.B. (2011). Linking patterns 624 in phylogeny, traits, abiotic variables and space: A novel approach to linking 625 environmental filtering and plant community assembly. J. Ecol. , 99, 165 –175. 626 37. 627 Polidori, C., Scanni, B., Scamoni, E., Giovanetti, M., Andrietti, F. & Paxton, R.J. (2005). 628 Satellite flies (leucophora personata, diptera: Anthomyiidae) and other dipteran 629 parasites of the communal bee andrena agilissima (hymenoptera: Andrenidae) on 630 the island of elba, italy. J. Nat. Hist. , 39, 2745 –2758. 631 38. 632 R Core Team. (2016). R: A language and environment for statistical computing . R 633 Foundation for Statistical Computing, Vienna, Austria. 634 39. 635 Rees, M. (2013). Competition on productivity gradients –what do we expect? Ecol. 636 Lett. , 16, 291 –298. 637 40. 638 Runquist, R.B. & Stanton, M.L. (2013). Asymmetric and frequency-dependent 639 pollinator-mediated interactions may influence competitive displacement in two 640 vernal pool plants. Ecol. Lett. , 16, 183 –190. 641 41. 642 Silvertown, J. (2004). Plant coexistence and the niche. Trends in Ecology & Evolution , 643 19, 605 –611. 644 42. 645 Stump, S.M. & Chesson, P. (2017). How optimally foraging predators promote prey 646 coexistence in a variable environment. Theor. Popul. Biol. , 114, 40 –58. 647 43. 648 Teste, F.P., Kardol, P., Turner, B.L., Wardle, D.A., Zemunik, G. & Renton, M. et al.
17 649 (2017). Plant-soil feedback and the maintenance of diversity in mediterranean- 650 climate shrublands. Science , 355, 173 –176. 651 44. 652 Tilman, D. (1982). Resource competition and community structure . Princeton 653 university press. 654 45. 655 Tilman, D. (1994). Competition and biodiversity in spatially structured habitats. 656 Ecology , 75, 2 –16. 657 46. 658 Tur, C., Castro-Urgal, R. & Traveset, A. (2013). Linking plant specialization to 659 dependence in interactions for seed set in pollination networks. PloS ONE , 8, 660 e78294. 661 47. 662 Vamosi, J.C., Moray, C.M., Garcha, N.K., Chamberlain, S.A. & Mooers, A.Ø. (2014). 663 Pollinators visit related plant species across 29 plant –pollinator networks. Ecol. 664 Evol. , 4, 2303 –2315. 665 48. 666 Waites, A.R. & Ågren, J. (2004). Pollinator visitation, stigmatic pollen loads and 667 among-population variation in seed set in Lythrum salicaria. J. Ecol. , 92, 512 –526. 668 49. 669 Wäckers, F.L., Romeis, J. & Rijn, P. van. (2007). Nectar and pollen feeding by insect 670 herbivores and implications for multitrophic interactions. Annu. Rev. Entomol. , 52, 671 301 –323. 672 50. 673 Weber, M.G. & Strauss, S.Y. (2016). Coexistence in close relatives: Beyond 674 competition and reproductive isolation in sister taxa. Annu. Rev. Ecol. Evol. Syst. , 47, 675 359 –381. 676 677 678 679 680 681 682 683
18 684 Table 1 . List of species observed in Caracoles Ranch. Code and taxonomic family of 685 each species is provided. Sample size represents the total number of individuals 686 sampled for each focal species, and it is correlated with their natural abundance at 687 the study site.
Species Family Code Flora l visitors Sample size Beta macrocarpa Amaranthaceae BEMA Yes 289
Chamaemelum fuscatum Asteraceae CHFU Yes 162
Chamaemelum mixtum Asteraceae CHMI Yes 5
Centaurium tenuiflorum Gentianaceae CETE No 23
Frankenia pulverulenta Frankeniaceae FRPU No 5
Hordeum marinum Poaceae HOMA No 289
Leontodon maroccanus Asteraceae LEMA Yes 273
Melilotus elengans Fabaceae MEEL Yes 77
Melilotus sulcatus Fabaceae MESU Yes 229
Plantago coronopus Plantaginaceae PLCO No 171
Polypogon monspeliensis Poaceae POMO No 20
Pulicaria paludosa Asteraceae PUPA Yes 124
Scorzonera laciniata Asteraceae SCLA Yes 101
Spergularia rubra Caryophyllaceae SPRU Yes 44
Sonchus asper Asteraceae SOAS Yes 87
Suaeda splendens Amaranthaceae SUSP No 29
688 689 690 691
19 692 FIGURE LEGENDS 693 FIGURE 1. For the six focal species this figure shows: total number of visits of the 694 four groups of floral visitors (bees, beetles, butterflies and flies) (left panel) and 695 species abundance along the salinity gradient (right panel). The amount of salinity 696 experience during the life span of each species was measured as the sum of the 697 electric conductivity in Ds/m measured bi-monthly. Note that the three Asteraceae 698 species (a) Chamaemelum fuscatum , (c) Leontodon maroccanus and (e) Pulicaria 699 paludosa had an order of magnitude more floral visits than the non-Asteraceae 700 species (b) Beta macrocarpa , (d) Melilotus elegans and (f) Melilotus sulcatus . 701 FIGURE 2. Relationship between per capita seed production as a function of the 702 number of competitor within the neighbourhood according to three different 703 conditions of floral visitors and soil salinity. Here is represented the three focal 704 Asteraceae species ((a and d) Chamaemelum fuscatum , (b and e) Leontodon 705 maroccanus , and (c and f) Pulicaria paludosa ). Upper panels c ontainsfloralvisitors’ 706 effects with black curves representing no floral visitation, green curves representing 707 one or two visits, and red curves representing percentile 95 of floral visits (which 708 ranges from 6 visits in C. fuscatum to 9 visits in L. maroccanus and P. paludosa ). 709 Lower panels contains soil salinity effects with black curves considering no salt in 710 the soil, and green and red curves representing percentiles 50 and 95 respectively of 711 the soil salinity sum over focal species life span. Here, we grouped both conspecific 712 and heterospecific competitive interactions for visual purposes, but such 713 interactions are represented separately in Appendix S6. 714 FIGURE 3. Average fitness and stabilising niche differences for each pair of species 715 (denoted by a single point). Dashed arrows connect the scenario from not 716 considering the effect of floral visitors (a) or soil salinity (b) on the determinants of 717 competition outcomes (black open points) to the scenario in which each driver is 718 considered separately (black solid points). Dashed lines are blue when the identity 719 of the superior competitor changes across both scenarios. We used the condition for 720 mutual invasibility (eqn. 5) to know the identity of the superior competitor. Such 721 identity is written close to both ends of the lines (see Table 1 for species codes). For 722 those cases where both species of the pair are predicted to coexist, the superior 723 competitor is listed first. In case such identity does not change across scenarios then 724 dashed lines are black. The red curve separates the exclusion region from the region 725 where the condition for coexistence is met ( , where species j is the fitness 726 superior). 727 FIGURE 4. As in Figure 3, it is represented average fitness and stabilising niche 728 differences for each pair of species (denoted by a single point). Dashed arrows 729 connect the scenario from not considering the effects of floral visitors and soil 730 salinity on the determinants of competition outcomes (black open points) to the 731 scenario in which both drivers are jointly considered (black solid points). Dashed 732 lines are blue when the identity of the superior competitor changes across both 733 scenarios. We used the condition for mutual invasibility (eqn. 5) to know the
20 734 identity of the superior competitor. Such identity is written close to both ends of the 735 lines (see Table 1 for species codes). For those cases where both species of the pair 736 are predicted to coexist, the superior competitor is listed first. In case such identity 737 does not change across scenarios then dashed lines are black. The red curve 738 separates the exclusion region from the region where the condition for coexistence 739 is met ( , where species j is the fitness superior).
740
(a) Chamaemelum fuscatum (b) Beta macrocar pa
500 50 y
y 80 80
c c n n 400 40
60 60
e e u
u 300 30 q
q 40 40 e
e 200 20
r r F F 20 20 100 10 0 0 0 0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
(c) Leontodon maroccan us (d) Melilotus elegans
500 50 y
y 80 80
c
c n n 400 40
60 60
e e u
u 300 30 q
q 40 40 e
200 e 20
r r F F 20 20 100 10 0 0 0 0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
(e) Pulicar ia paludosa (f) Melilotus sulcatus
500 50 y
y 80 80
c c n n 400 40
60 60
e e u
u 300 30 q
q 40 40 e
e 200 20
r r F F 20 20 100 10 0 0 0 0 s s s e s e s e li s 0.0 0.5 1.0 1.5 2.0 s e li s 0.0 0.5 1.0 1.5 2.0 e tl rf e e tl rf e e e li e e e lie B e tt F B e tt F B u B u 741 B Salinity B Salinity 742 Figure 1
743
744
745
746
747
748
749
21 750
751
752
753
Floral visitor s eff ect (a) Chamaemelum fuscatum (b) Leontodon maroccanus (c) Pulicaria paludosa 2500 500 5000 no vis . no vis . no vis . lo w vis . lo w vis . lo w vis . high vis . high vis . high vis .
2000 400 4000
n
o
i
t
c
u d
o 1500 300 3000
r
p
d
e e
s 1000 200 2000
e
l
b
a i
V 500 100 1000
0 0 0
0 5 10 15 20 25 30 35 0 10 20 30 40 50 0 10 20 30 40 50
Salinity eff ect (d) Chamaemelum fuscatum (e) Leontodon maroccanus (f) Pulicaria paludosa 2500 500 5000 no sal. no sal. no sal. medium sal. medium sal. medium sal. high sal. high sal. high sal.
2000 400 4000
n
o
i
t
c
u d
o 1500 300 3000
r
p
d
e e
s 1000 200 2000
e
l
b
a i
V 500 100 1000
0 0 0
0 5 10 15 20 25 30 35 0 10 20 30 40 50 0 10 20 30 40 50 754 No . Neighbours No . Neighbours No . Neighbours 755 Figure 2
756
757
758
759
760
761
762
763
764
22 765
766
767
768
769
a) Floral Visitor s b) Salinity
No flor al visitors No salinity 10 Flor al visitors 10 Salinity Same super ior competitor Exclusion Same super ior competitor Exclusion
Changes in super ior competitor Changes in super ior competitor
) )
d d e
e 8 8
m m
r r
o o f
f PUP A
s s n
n PUP A
a
a
r r
T T .
. 6 6
g g
o
o
L L
( (
s s e
e BEMA
c
c
n n e
e 4 4 r
r PUP A
e
e
f f
f f
i
i
d d
s s s
s MEEL BEMA−PUP A
e
e n
n 2 2 BEMA−PUP A t
t PUP A
i
i F F PUP A CHFU−PUP A MEEL CHFU−PUP A MEEL PUP A−BEMA CHFU LEMA−CHFU
CHFU−LEMA Coe xistence Coe xistence 0 MESU−MEEL 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Niche diff erences Niche diff erences 770 771 Figure 3
772
773
774
775
776
777
778
779
23 780
781
782
783
a) Floral visitors and salinity
No flor al visitors and no salinity 10 Flor al visitors and salinity Same super ior competitor Exclusion Changes in super ior competitor
8 PUP A
PUP A
6 PUP A
BEMA 4
2 BEMA−PUP A Fitness differences (Log. Transformed) (Log. Fitnessdifferences
CHFU−PUP A MEEL CHFU Coe xistence 0 CHFU MESU
0.0 0.2 0.4 0.6 0.8 1.0
Niche differences 784 785 Figure 4
24