Skip to main content
NCBI home page
Annals of Botany logoLink to Annals of Botany
. 2020 Jul 25;126(7):1181–1191. doi: 10.1093/aob/mcaa138

Edge effects alter the role of fungi and insects in mediating functional composition and diversity of seedling recruits in a fragmented tropical forest

Meghna Krishnadas 1,2,✉, Kavya Agarwal 1, Liza S Comita 1,3
PMCID: PMC7684699  PMID: 32710752

Abstract

Background and Aims

In fragmented forests, proximity to forest edges can favour the establishment of resource-acquisitive species over more resource-conservative species. During seedling recruitment, resource-acquisitive species may benefit from either higher light availability or weaker top-down effects of natural enemies. The relative importance of light and enemies for recruitment has seldom been examined with respect to edge effects.

Methods

In a human-modified wet tropical forest in India, we first examined how functional traits indicative of resource-acquisitive vs. resource-conservative strategies, i.e. specific leaf area (SLA), leaf dry matter content, wood density and seed size, explained interspecific differences in densities of seedling recruits with distance to the forest edge. Then, we checked whether fungicide and insecticide treatments and canopy openness (proxy for light availability) explained edge effects on trait-mediated changes in seedling density. Finally, we examined whether light availability and natural enemy activity explained edge effects on functional diversity of seedling recruits.

Key Results

Up to 60 m from edges, recruit densities increased with decreasing seed size, but not at 90–100 m, where recruit densities increased with higher SLA. Trait-mediated variation in recruit densities changed with pesticides only at 90–100 m: compared with control plots, fungicide increased recruit densities for low SLA species and insecticide increased smaller seeded species. For SLA, wood density and seed size, functional diversity of recruits was higher at 90–100 m than at 0–5 m. At 90–100 m, fungicide decreased functional diversity for SLA and insecticide reduced seed size diversity compared with control plots. Canopy openness explained neither variation in recruit density in relation to traits nor functional diversity.

Conclusions

Altered biotic interactions can mediate local changes to trait composition and functional diversity during seedling recruitment in forest fragments, hinting at downstream effects on the structure and function of human-modified forests.

Keywords: Edge effects, fragmentation, functional diversity, India, plant–enemy interactions, seedling recruitment, traits, tropical forest

Introduction

Forest fragmentation can alter the functional composition of tree communities, but the mechanisms underlying functional changes remain unclear (Ewers and Didham, 2006; Haddad et al., 2015). Following fragmentation, fast-growing resource-acquisitive species increase in relative abundance compared with intact forest, whereas slower growing resource-conservative species decrease, particularly near forest edges (Laurance et al., 2006a, 2011; Tabarelli et al., 2010; Benchimol and Peres, 2015). Recruitment of resource-acquisitive species – typically with smaller seeds, lighter wood and thinner, cheaper leaves than resource-conservative species – can increase due to higher light availability near edges compared with shaded interiors of closed-canopy forests (Laurance et al., 2006a; Tabarelli et al., 2010; Santo-Silva et al., 2013; Benchimol and Peres, 2015). In addition to light, biotic interactions, such as between plants and their natural enemies, also structure plant communities (Freckleton and Lewis, 2006; Bagchi et al., 2014). The role of natural enemies in promoting species diversity can weaken with edge effects (Krishnadas and Comita, 2018; Krishnadas et al., 2018), but whether this also changes composition and diversity of traits remains to be assessed.

Plant enemies, such as insect herbivores and fungal pathogens, play a particularly prominent role during seedling recruitment, a crucial bottleneck in plant population dynamics (Grubb, 1977; Freckleton and Lewis, 2006; Mangan et al., 2010; Bagchi et al., 2014; Bever et al., 2015). By mediating density-dependent mortality of seeds and seedlings (Bell et al., 2006; Bagchi et al., 2014; Krishnadas et al., 2018), natural enemies curtail abundant species and help maintain plant diversity (Janzen, 1970; Connell, 1971; Freckleton and Lewis, 2006). However, susceptibility to enemies can vary among species, with resource-acquisitive species usually being more vulnerable to insects and fungi compared with resource-conservative species that tend to have better defences (McCarthy-Neumann and Kobe, 2008; McCarthy-Neumann and Ibáñez, 2013). Resource-acquisitive species may be especially vulnerable to natural enemies in the shaded understorey of forest interiors where fungal and insect activity is high and low light conditions impede recovery after enemy attack (Krishnadas and Comita, 2018). Insects and fungi may therefore curtail recruitment of acquisitive species more than conservative species. By extension, weaker effects of insects and fungi could increase seedling density of acquisitive species and shift the recruit community away from conservative traits that characterize a typical understorey in shaded, closed-canopy forests.

By affecting local recruit densities in relation to species traits, altered enemy activity could also modify functional diversity of seedlings. Recent evidence shows that a weaker effect of natural enemies in promoting diversity corresponds to lower species diversity of seedling recruits near forest edges (Krishnadas et al., 2018). Decreased species diversity can directly reduce functional diversity, regardless of whether some traits are being favoured or not (Swenson and Enquist, 2009; Bernard-Verdier et al., 2012; Siefert et al., 2013). However, natural enemies could alter functional diversity independent of changes in species diversity because some trait values, such as large seeds or thicker leaves, correlate with better defence against enemies (Gripenberg et al., 2018). By limiting the recruitment of poorly defended species, plant enemies may decrease functional diversity. Alternatively, specialized enemies could increase functional diversity if density-dependent infection tends to curtail abundant species regardless of their defensive capacity. Moreover, functional diversity can increase if similar defences or shared enemies in functionally similar species promote local establishment of species that differ in their traits (Gilbert and Webb, 2007; Parker and Gilbert, 2018).

Faster growing, shade-intolerant species often have thinner leaves and lighter wood than slower growing, more shade-tolerant species that exhibit more conservative life histories of slower growth but higher survival (Wright et al., 2010; Sterck et al., 2011). Leaf and wood traits associated with conservative life histories also correlate with better defence against natural enemies (Coley and Barone, 1996; Gripenberg et al., 2018). During early recruitment, seed size represents a trade-off between fecundity and tolerance of stresses such as shade and natural enemy attack (Muller-Landau, 2010). Recruitment differences in relation to these leaf, wood and seed traits of species can indicate how changes in abiotic and biotic conditions could modify the functional composition of plant communities (Kraft and Ackerly, 2010; Adler et al., 2014; Umaña et al., 2016). Spatial variation in functional composition and diversity of recruits can be especially prominent in fragmented forests, where edge effects create marked abiotic gradients and alter biotic interactions (Didham et al., 2012). Edge effects are also known to alter patterns of herbivory and pathogen infestation (Benítez-Malvido et al., 1999, 2018). However, whether edge effects mediate trait-mediated changes among recruiting seedlings remains poorly understood. In a human-modified wet tropical forest, we examined the relative role of light and natural enemies in structuring functional composition and diversity of seedling recruits of native tree species (henceforth recruits).

Specifically, we asked the following questions. (1) Do local densities of species with acquisitive vs. conservative traits in recruit communities vary with distance to the forest edge? Compared with interior sites, we expected sites near forest edges to have higher recruit densities of species with traits corresponding to resource-acquisitive than resource-conservative strategies. While this pattern has been well documented for later life stages, few studies have assessed whether this pattern emerges from edge effects at the initial recruitment stage itself.

(2) Along an edge-to-interior gradient, does trait-mediated variation in recruit densities correlate with light availability or decreased impact of natural enemies? Since differences in species abilities to withstand shade strongly influence recruitment dynamics in closed-canopy forests (Kobe, 1999; Nicotra et al., 1999; Balderrama and Chazdon, 2005; Valladares and Niinemets, 2008), we expected higher light availability to correlate with increased recruitment of resource-acquisitive species relative to resource-conservative species. Additionally, we expected trait-mediated recruitment patterns to be shaped by differences in natural enemy activity at the forest edge vs. the interior (Krishnadas and Comita, 2018). Specifically, we hypothesized that farther from edges, suppressing natural enemies would increase recruit densities for resource-acquisitive species more than resource-conservative species, but that this effect would weaken closer to edges.

(3) How does functional diversity of recruits vary with proximity to the forest edge, light availability, and natural enemy activity? We hypothesized that if warmer, drier conditions near edges filter out species, then functional diversity would decrease closer to edges and at sites with more open canopy. However, higher light availability could allow a greater range of species to recruit, in which case functional diversity would increase near edges and with canopy openness. Alternatively, or additionally, natural enemies may increase functional diversity by preventing dominance of a few species. Based on our previous findings of weaker enemy effects on seedling diversity near edges (Krishnadas et al., 2018), we expected that suppressing enemies would reduce functional diversity farther from, but not near, edges. Furthermore, if the examined traits governed species responses to light availability and/or natural enemy activity, we expected that proximity to an edge may decrease functional diversity even when controlling for differences in species richness.

MATERIALS AND METHODS

Study site

We conducted this study within Kadamane village limits (henceforth Kadamane), a human-modified landscape in the Western Ghats Biodiversity Hotspot, Karnataka, south India (12°56ʹN, 75°39ʹE). The 3600 ha landscape consists of fragments of wet tropical forest embedded in a patchwork of tea plantations, abandoned coffee plantations, roads and grassland. Nearly 60 % of the area comprises forest fragments. Mean annual rainfall is approx. 5000 mm, most of which occurs from June through October. Selective logging for timber occurred until 1996, when the Supreme Court banned all logging in forest lands.

Experimental design

We first identified and demarcated all forest fragments within Kadamane and identified 15 sampling locations across the landscape with a random selection procedure using Geographical Information Systems (GIS) and field surveys (see Krishnadas et al., 2018 for details). Compositional changes in seedling communities mainly occur within 50 m of edges (Murcia, 1995). So, after a preliminary survey, we set up 45 sampling replicates (hereafter stations; three stations per site across 15 locations) each at three distance categories from the edge, i.e. 0–5 m (E0), 20–30 m (E1) and 50–60 m (E2), and 15 stations at 90–100 m from the edge (E3). We were limited to 15 stations for the 90–100 m distance category because sites >100 m from the edge were often situated in topographically steep or rocky areas, where the structure and composition of tree communities varied markedly. Also, the rugged terrain made it difficult to access some forest patches in the study landscape, especially for weekly visits to spray pesticides. To minimize the confounding effects of disturbances other than edge distance, we avoided setting up our plots in large tree fall gaps and in sites with signs of selective logging or other human use.

To assess patterns of seedling recruitment per species, we established five 1 × 1 m seedling plots, demarcated at two diagonal corners using PVC pipes. To evaluate the role of fungal pathogens and insect herbivores in mediating species recruitment, we randomly assigned two seedling plots at every station to be sprayed with either fungicide or insecticide. For fungicide treatment, we used a combination of Amistar® (active ingredient, azoxystrobin), which offers broad-spectrum systemic protection against multiple pathogenic fungi, and Ridomil Gold® (active ingredients, mancozeb and metalxyl) that acts against oomycetes and other fungi upon contact. Previous studies in temperate grasslands, crops and in seedlings of two Neotropical tree species found that both fungicides have low toxicity to non-target organisms and minimally inhibited arbuscular mycorrhizal fungi (Gripenberg et al., 2014, and references therein). We used Actara® (active ingredient, thiamethoxam) as insecticide treatment, which provides systemic and contact protection against a broad range of insects. We followed standard instructions and dissolved each pesticide in 100 mL of water per 1 m2 plot (Amistar®, 0.01 g; Ridomil®, 0.5 g; Actara®, 0.021 g) and applied them with a hand mister to the soil and seedlings. We sprayed one plot with 100 mL of water and left one plot untreated. From September 2015 until November 2016, we applied treatments every 10 d during the dry season and every 5–7 d during the monsoon rains. Preliminary analysis showed that recruitment rates, species richness and functional diversity did not vary between untreated and ‘water’ plots, so for the analyses described below we used the mean seedling density per species and mean functional diversity of these two plots as ‘Control’ for each station.

Admittedly, using pesticides can be a somewhat crude approach to study plant–enemy interactions. For instance, it has been argued that more controlled experiments with autoclaved soil may offer cleaner insights into the mechanisms underlying plant–soil feedback through rhizosphere-associated fungi (Bever et al., 2015). However, such controlled experiments are often possible only for a small sub-set of species and, while instructive, the results may be hard to extrapolate to aggregate impacts at the community level. Since our focus was to understand community-wide impacts on recruitment patterns, we opted for the use of pesticides to suppress different groups of plant enemies (Bagchi et al., 2014; Krishnadas et al., 2018).

Because seedling densities of species with different traits can be influenced by their seed densities at a site, we also monitored seed rain in two seed traps, made of 1 m2 pieces of 0.5 mm PVC mesh, suspended 0.5–0.75 m above the ground and tied to nearby stems. In total, we monitored 292 seed traps and 730 seedling plots in 146 stations across the landscape. See Supplementary data Fig. S1 for a schematic of our experimental design.

Seed rain and seedling census

From September 2015 until November 2016, we recorded seed fall into seed traps every 10–15 d. We recorded only viable seeds without any mechanical or enemy-inflicted damage and classified seeds into morphospecies. Before we began pesticide treatments, we censused and tagged all existing seedlings. Then, we conducted one census after the dry season (March 2016) and a final census after the post-monsoon recruitment peak (November 2016), in which we tagged all new seedlings (hereafter referred to as ‘recruits’) and identified them to species. We matched 73 seedling species to their seeds, which constituted 98 % of the total seedling abundance. For this analysis, we only used seedlings that were recruited at the end of the 1 year period of pesticide treatment. Moreover, for both seeds and seedlings, we excluded the three non-native species that occurred in the data set (Camellia sinensis, Syzygium jambos and Lantana camara) and only analysed the data for native tree species.

Canopy openness

We characterized canopy openness (which correlates with microclimate, particularly understorey light availability) using hemispherical photos taken 0.25 m above-ground at the centre of each 1 m2 seedling plot. We used a digital camera (Nikon Coolpix 950, Melville, NY, USA) fitted with a fish-eye lens (Nikon FC-E8, Melville, NY, USA) and took all photos during early mornings from mid-June through mid-July, when the rainy season begins and skies are uniformly overcast. We analysed the images using Gap Light Analyzer (Frazer et al., 1999). Median canopy openness was significantly higher at E0 (0–5 m) than all other distances from the edge (Kruskal–Wallis test, χ 2 = 33.04, d.f. = 3, P-value < 0.001; Supplementary data Fig. S2).

Trait data collection

We measured traits for 72 species that comprised 90 % of adult individuals in our study (Supplementary data Table S1). We sampled traits for adult individuals because our goal was to test whether adults with different traits varied in their recruitment patterns in relation to edge effects. We did not aim to test the relative performance of seedlings with different traits. However, species ranks for traits have been found to be highly correlated among seedlings and adults in other tropical forests (Umaña et al., 2016). Between May and October 2016, we collected data on six functional traits: specific leaf area (SLA), leaf nitrogen percentage by weight, leaf carbon percentage by weight, leaf C:N ratio, leaf dry matter content (LDMC), wood density (WD) and seed size (SS). These traits were chosen because they represent the key dimensions of variation in plant life history (Wright et al., 2010; Reich, 2014). For each trait, we sampled five trees per species.

To estimate SLA, we collected five mature, healthy, sun-exposed canopy leaves per tree and immediately measured leaf fresh weight. We then scanned leaves using a flatbed scanner and used the images to calculate fresh leaf area with the Black Spot leaf area calculator (Varma and Osuri, 2013). We air-dried leaves for 2 weeks before returning them to the laboratory where we oven-dried leaves at 60 °C for 72 h to obtain leaf dry weight. SLA was calculated by dividing fresh leaf area by leaf dry weight. We estimated leaf dry matter content (LDMC) by dividing leaf dry weight by leaf fresh weight. We ground the oven-dried leaves using a standard leaf grinder and estimated per gram leaf carbon content and leaf nitrogen content with a C-N analyser using TruSpec CN Determinator (LECO Corporation). To estimate WD, we collected wood cores using a Haglöf increment borer from the trees sampled for leaves. We estimated fresh wood volume using the water displacement method. We oven-dried samples at 60 °C for 1 week to obtain wood dry weight, which we divided by wood volume to obtain WD. We calculated SS as the length of the longest dimension using five seeds per five trees per species. For the few species for which we were unable to find adequate samples, we compiled SS information from published databases and the literature (Osuri and Sankaran, 2016, and references therein).

Quantifying functional diversity

We quantified functional diversity using mean pairwise distance (MPD), which uses a trait dendrogram to calculate the mean of distances between each pair of species found in a sampling unit, weighted by their relative abundance (Swenson, 2014). Lower MPD values suggest a restriction on the functional trait space and co-occurrence of functionally similar species. To assess whether changes in functional diversity occurred independently of species richness, we standardized observed MPD values using null models to control for species diversity. For each null assemblage, we pooled all recruits per location, shuffled species occurrences and abundances among sites and calculated the expected MPD value per treatment per site per edge distance. In all randomizations, we maintained species occurrence frequencies and their total abundances. From the mean of 999 such null assemblages, we subtracted the observed functional diversity and divided by the standard deviation of null assemblages to obtain the standardized effect sizes; henceforth MPD.SES (Swenson, 2014).

Statistical analysis

The SLA was highly correlated with %N and LDMC with the C:N ratio, and model outputs and trend lines were similar for the correlated traits. Hence, we only present the results for SLA, along with LDMC, WD and SS. For each species, we pooled data from the three stations that were used to sample each edge distance category at a location.

Trait-mediated response of recruit density to edge distance and light

. First, we used data from only control plots to test the hypothesis that recruit densities of thinner leaved, light-wooded, smaller seeded species increase near forest edges relative to sites farther from edges. For each trait separately, we used a generalized linear mixed-effects model (GLMM) to analyse species recruit density in each plot as a function of the interaction between species trait values and distance to the edge. We modelled the error structure of all GLMMs using hurdle models with a zero-inflated negative binomial distribution to account for both the large number of zeros per plot and the large range of recruit densities across plots. The zero inflation component was included in relation to baseline variation in recruit densities among species to account for the fact that most plots had zero abundances of many species, as commonly seen for seedling censuses in tropical forests. Using a similar model, we tested whether trait-mediated differences in recruit densities between edge distance categories were due to differences in canopy openness (see above). For this, we set up a GLMM to model recruit densities in relation to an interaction between species trait value and canopy openness. We included species and sites as random intercepts in both components of the hurdle model for each trait.

Natural enemies and recruit density with distance to edge

. Next, we used the full data (control and pesticide treatments) to examine whether trait-mediated differences in recruit density were explained by differences in natural enemy effects at each distance category. We used the same GLMM as described above to model recruit densities in relation to three-way interactions between trait, edge distance and pesticide treatment.

Edge effects and functional diversity

. Finally, we assessed changes in functional diversity of each trait with distance to the forest edge and whether these changes were explained by differences in canopy openness or natural enemy activity. We tested our hypotheses using both raw (MPD.OBS) and standardized values (MPD.SES) of functional diversity for each trait separately. Using linear mixed-effects models (LMMs), we first used data from control plots to examine how functional diversity of a trait varied with distance to the edge and the interaction between edge distance and pesticide treatment. We also used LMMs to test whether canopy openness explained variation in functional diversity of control plots across the landscape. Sites were included as random intercepts in all LMMs. We used Tukey’s post-hoc tests of honest significant difference (Tukey HSD) to control for multiple comparisons in assessing pairwise differences among edge distances and treatments within edge distances.

Changes in seed rain with distance to edge

. Due to statistical issues associated with the potential spatial mismatch in seed and seedling numbers at a site (Detto et al., 2019), we did not directly test the relationship between seedling recruitment and seed arrival as a function of traits. Instead, we analysed patterns in the seed data separately to assess whether differences in seed arrival may have contributed to observed trait-mediated variation in recruit densities. We used GLMM with Poisson errors [which had a lower Akaike information criterion (AIC) than negative binomial models] to model variation in seed density in relation to an interaction between traits and distance to the edge or canopy openness. We used an LMM (with Gaussian errors) to assess change in functional diversity of seeds (for each of the four traits separately) with distance to the edge. We included sites and species as random effects in the GLMMs of seed densities and sites as a random effect in the LMMs of functional diversity.

Models with PCA axes of traits

. Because traits can covary, we decomposed trait variation among species with a principal components analysis (PCA) followed by varimax rotation to maximize loadings on individual axes (Supplementary data Table S2). We used the first three axes (Supplementary data Fig. S3) as composite dimensions of life history to set up models exactly as described above for individual traits. For each PC axis, we modelled interspecific differences in densities and functional diversity of seeds and seedlings with edge distance. Then, we modelled trait-mediated variation in recruit densities and functional diversity in relation to canopy openness. Finally, we modelled the interaction between edge distance and pesticide treatment in explaining recruit density and functional diversity.

We conducted all analysis in the program R version 3.6.1 (R Development Core Team, 2019). Package ‘picante’ (Kembel et al., 2010) was used to calculate functional diversity. We used package ‘glmmTMB’ (Brooks et al., 2017) to implement GLMMs for recruit densities and ‘nlme’ (Pinheiro et al., 2015) to implement LMMs for functional diversity.

RESULTS

Across 146 sites, we recorded a total of 71 species in the seedling plots, of which we included 51 species in the final analysis based on availability of trait data. This final set of species comprised approx. 96 % of total seedling abundance sampled in our study. See Supplementary data Table S3 for summary statistics of the numbers of species, seeds and seedlings in all combinations of distance from the forest edge and pesticide treatments. Models with PCA axes (Supplementary data Tables S4–S8) showed similar trends to individual traits and appeared to be driven largely by SLA. Because PCA models had fewer significant correlations with predictors, we focus on the results for individual traits as they captured species performance along key axes of life history variation (Butterfield and Suding, 2013).

Distance to the edge and trait–recruitment relationships

Of the four traits, SLA and SS interacted with distance to the edge in explaining variation among species in recruit densities in control plots. Specifically, recruit density decreased for species with higher SLA and larger SS at 0–5 m from edges relative to species with lower SLA and smaller seeds (Fig. 1A, Supplementary data Table S4). Conversely, at 90–100 m from edges, recruit density increased with higher SLA and did not vary appreciably with SS (Fig. 1D; Supplementary data Table S4). No trait interacted with canopy openness in explaining variation in recruit densities across the landscape (Supplementary data Table S5).

Fig. 1.

Fig. 1.

Open in a new tab

Tree species recruit densities (per m2) as a function of species trait values and distance to the forest edge in a wet tropical forest in the Western Ghats, India. Panels show trait-mediated variation in densities of seedling recruits of species in relation to (A) SLA, (B) LDMC, (C) WD and (D) SS. Coloured lines show the recruit density–trait relationships for each of four edge distance categories: E0 (0–5 m), E1 (20–30 m), E2 (50–50 m) and E3 (90–100 m). Recruit density in ‘Control’ plots (see the Materials and Methods) was modelled in relation to an interaction between trait and edge distance category, using generalized linear mixed-effect models with truncated negative binomial distribution and zero inflation. SLA and SS were log transformed and all traits were standardized prior to analysis. Sites and species were included as random intercepts. Separate models were run per trait, with ‘Control’ at E0 as the base level in each model. Points show observed data, for each species per site, jittered vertically for viewing. Solid lines show predictions from models.

Seed density in relation to SLA varied only at 90–100 m from edges where low SLA species increased relative to species with higher SLA (Supplementary data Table S6). For LDMC and WD, significant trait-mediated variations in seed density only occurred at 50–60 m from edges, where seed densities increased with increasing LDMC and decreasing WD (Supplementary data Table S6). For SS, species with smaller seeds had greater densities than larger seeded species at 0–5 m from edges. This SS-based difference significantly decreased at 20–30 m and 50–60 m, but sites at 90–100 m from the edge did not differ from E0 in correlation between SS and seed density (Supplementary data Table S6).

Distance to the edge, traits and natural enemies

The SLA and SS showed significant interactions with fungicide and insecticide treatments, respectively, in explaining variation in recruit densities, but only at 90–100 m from edges (Fig. 2). At 90–100 m, fungicide application resulted in decreased densities of high SLA species and increased densities of low SLA species relative to control plots (Fig. 2D). For SS, consistent with expectation, insecticide increased recruit density of smaller seeded species at 90–100 m from edges relative to control plots (Fig. 2H; Supplementary data Table S7). Trends were similar with fungicide, but not statistically significant. Unexpectedly, at 0–5 m from the edge, insecticide decreased the recruitment of smaller seeded species relative to control plots (Fig. 2E; Supplementary data Table S7). Neither LDMC nor WD explained changes in recruit densities with fungicide or insecticide treatments.

Fig. 2.

Fig. 2.

Open in a new tab

The effect of fungicide and insecticide treatment on trait-mediated variation in recruit densities at four edge distance categories: E0 (0–5 m), E1 (20–30 m), E2 (50–60 m) and E3 (90–100 m). Relationships are only shown for SLA (A–D) and SS (E–H), the two traits that showed significant interactions with fungicide and insecticide treatments. Relationships were evaluated using generalized linear mixed-effect models with truncated negative binomial distribution and zero inflation. SLA and SS were log transformed, and all traits were standardized prior to analysis. Sites and species were included as random intercepts. Separate models were run per trait, with ‘Control’ at E0 as the base level in each model. Points show observed data, for each species per site, jittered vertically for viewing. Solid lines show predictions from models.

Change in functional diversity

The MPD.OBS of SLA, WD and SS was significantly higher at 90–100 m from edges compared with other edge distance categories (Fig. 3; Supplementary data Fig. S4). Canopy openness did not predict differences in MPD.OBS across the landscape for any trait (Supplementary data Table S8). At 90–100 m from edges, MPD.OBS for SLA decreased significantly with fungicide and MPD.OBS for SS decreased significantly with insecticide treatment (Supplementary data Fig. S4). Fungicide reduced MPD.OBS of SS at 20–30 m from edges. For LDMC and WD, MPD.OBS did not vary with pesticide treatments at any edge distance category (Supplementary data Fig. S4). All trends remained qualitatively similar with standardized functional diversity (MPD.SES, Supplementary data Fig. S5), but pairwise differences between control and pesticide-treated plots at 90–100 m from edges were not significant after accounting for multiple comparisons (Tukey HSD, P = 0.1). In addition, at 0–5 m from the edge, MPD.SES of SLA significantly increased in insecticide and fungicide plots compared with control plots, and SS diversity among recruits increased with fungicide (Supplementary data Fig. S5A, D). For seeds, MPD.OBS did not differ among edge distance categories for any trait (Supplementary data Fig. S6). However, MPD.SES for SLA of seeds was higher at 90–100 m from edges compared with values at 0–5 m (Supplementary data Fig. S7), although the difference was marginally insignificant (Tukey HSD, P = 0.08). Hereafter, ‘functional diversity’ refers to MPD.OBS unless stated otherwise.

Fig. 3.

Fig. 3.

Open in a new tab

The effect of fungicide and insecticide treatment on functional diversity of seedling recruits at four edge distance categories: E0 (0–5 m), E1 (20–30 m), E2 (50–60 m) and E3 (90–100 m). Results are presented for observed mean pairwise distances (MPD.OBS) for (A) SLA and (B) SS, the two traits for which pesticide treatment had significant effects on functional diversity. Points represent mean functional diversity of recruits per edge distance per treatment, and bars show 95 % confidence intervals. Letters indicate significant differences among treatments within each edge category. Patterns were assessed using linear mixed-effect models with location as a random intercept and solid lines show 95 % confidence intervals for estimates. See Supplementary data Fig. S4 for coefficient estimates from models for all traits.

Discussion

We experimentally assessed whether edge effects altered the role of insects and fungi in mediating trait composition and functional diversity of seedling communities in a fragmented wet tropical forest. Only SS and SLA explained interspecific differences in recruit densities with distance from the forest edge. For these two traits, edge effects also corresponded to a weaker role for insects and fungi in regulating recruit densities at 90–100 m from edges compared with sites within 60 m of forest edges. Moreover, functional diversity for SLA, WD and SS was significantly higher at 90–100 m from edges compared with 0–5 m, and other than for SLA this was mainly due to lower species richness near edges. At 90–100 m, compared with control plots, functional diversity decreased with fungicide for SLA and with insecticide for seed size; fungicide decreased SS diversity at 20–30 m from edges. Canopy openness correlated neither with trait-mediated variation in recruit densities nor with functional diversity. Thus, altered biotic interactions due to edge effects, rather than light, seemed to mediate functional changes during seedling recruitment in this human-modified forest.

Edge distance and trait-mediated differences in recruit density

Closer to edges, larger seeded species had lower recruit densities than smaller seeded species. These differences may have been partly due to seed arrival; smaller seeded species had higher seed densities than larger seeded species near edges, consistent with patterns in other forests (Melo et al., 2006, 2007). Although sites at 90–100 m also had higher seed density of smaller seeded species, the larger seeded species attained higher recruit density. These results suggest that trait-mediated variation in seedling recruitment was decoupled from seed number in the forest interior. Larger seed size typically correlates with higher shade tolerance (Augspurger, 1984a; McCarthy-Neumann and Kobe, 2008). Low light in interior sites can limit smaller seeded species that would recruit better near edges due to higher light availability than in the interior, a common ‘edge effect’ (Melo et al., 2007; Tabarelli et al., 2008; Benchimol and Peres, 2015). However, canopy openness (a proxy for light availability) did not explain SS-based differences in recruit density here, despite edges having more open canopies than more interior sites.

That no trait interacted with canopy openness in explaining variation in recruit densities might reflect the facts that these are old fragments with the canopy largely intact, that edges often had fairly dense vegetation and that we avoided tree fall gaps or other disturbances when setting up our plots. Therefore, the modest differences in light availability captured in our study might not have sufficed for increased recruitment of species with the smallest seeds or cheap leaves and wood – heliophiles that increase near edges in fragmented forests (Laurance et al., 2006b; Benchimol and Peres, 2015). However, the results for canopy openness and seed arrival together suggest that other factors were driving changes to seedling recruitment near to vs. far from forest edges. Our experimental results indicate that one such factor may be the altered role of insects and fungi in mediating mortality during early seedling establishment (Krishnadas and Comita, 2019).

Edge effects weaken the role of natural enemies in recruitment

Trait-mediated variation in seedling recruitment could also arise from species having different vulnerability to natural enemies; acquisitive species are usually more susceptible than conservative species to depredation by fungi and insects (Augspurger, 1984b; Kobe, 1999; Valladares et al., 2016). Because proximity to edges can diminish the role of enemies in regulating seedling recruitment (Krishnadas and Comita, 2018; Krishnadas et al., 2018), we had expected that, farther from forest edges, suppressing enemies would increase the recruitment of species with acquisitive traits relative to conservative traits (Coley and Barone, 1996; McCarthy-Neumann and Kobe, 2008). Correlations of recruit density with SS reflected these expected trends. At 90–100 m from forest edges, recruit densities for smaller seeded – probably acquisitive – species increased with insecticide treatment relative to control plots. Hence, increased recruitment of smaller seeded species near forest edges could persist if edge effects weaken post-dispersal biotic interactions that regulate seedling recruitment – a factor that may contribute to the ‘arrested succession’ of fragmented forests (Tabarelli et al., 2008).

Differences in the interaction of SS with fungicide and insecticide in regulating recruit densities may reflect the greater host specificity of fungi than of insects (Gilbert and Webb, 2007; Benítez et al., 2013; Benítez-Malvido and Dáttilo, 2015). Suppressing fungi did not lead to as stark an increase in smaller seeded species as expected in the forest interior, because low adult densities of these tree species may have resulted in lower loads of their associated soil pathogens (Schroeder et al., 2018). In contrast, insects may have generally targeted smaller seeded species more than larger seeded species, but not necessarily in a host-specific manner. As a result, suppressing insects increased the recruitment of small seeded acquisitive species in the forest interior. These results highlight the need for experimental studies to understand the causal links in plant–enemy interactions and changes thereto with disturbance or environmental variation.

However, suppressing fungi did not unequivocally favour acquisitive species. At 90–100 m from edges, low SLA species, indicative of a conservative life history, increased in fungicide treatment (relative to control). Species response to fungi could occur via other traits such as seed thickness or chemistry (Gripenberg et al., 2018). Seed chemical traits are not well studied at the community level, but in one Neotropical forest, Gripenberg et al. (2018) found that species with thicker leaves and denser wood also had greater concentrations of defensive compounds in their seeds. If the same correlations hold in this forest, differences in seed chemistry may not be driving the patterns observed for SLA and LDMC. Alternatively, our results suggest that fungicide increased densities of whichever traits were less abundant in control plots. Thus, fungi might not favour conservative species per se, but mediate a density-dependent reduction in numbers for the traits that are more abundant at a site (Umaña et al., 2016).

It is possible that fungicide may have disrupted the beneficial effects of some fungi during seedling recruitment. Previous studies suggest that the fungicides used in this study do not negate the activity of beneficial fungi or have direct toxic effects on plants (Maron et al., 2011; Gripenberg et al., 2014), but we did not test that in our specific study system. Fungicides could have impacted mycorrhizae, but loss of mycorrhizae would be unlikely to affect seed to seedling transition and/or survival of the very young seedlings which were the focus of our study. Any loss of mycorrhizae will probably be more important at later stages when nutrient limitation or interspecific competition for resources becomes important for seedling survival (Lambers et al., 2008). However, given the complexity of interactions in the plant microbiome, we cannot rule out that pesticides may have decreased some micro-organisms that help break down the seed coat and facilitate germination.

Edge effects and functional diversity of seedling recruits

Functional diversity of SLA, WD and SS was higher at 90–100 m than at all other edge distances. This increase in functional diversity with distance to the edge was not attributable to changes in seed rain – functional diversity of traits among seeds did not vary with edge distance. Higher light availability due to more open canopies near edges could support a wider suite of functional strategies among recruits, but functional diversity of seedlings did not vary with canopy openness either. In comparison, at 90–100 m from edges, compared with control plots, fungicide decreased functional diversity of SLA, and SS diversity decreased in insecticide plots. Although fungicide decreased diversity of SS at 20–30 m from edges, in general, pesticides did not alter functional diversity at sites within 60 m of edges, indicating that edge effects weakened the role of natural enemies in mediating trait diversity.

Changes in functional diversity could reflect the fact that species diversity increased from the edge to the interior, where species diversity declined with fungicide treatment (Krishnadas et al., 2018). We used null models to verify whether changes in species diversity underlie the changes in functional diversity. While trends remained largely similar with raw and standardized values, the decrease in standardized functional diversity with fungicide at 90–100 m became marginally insignificant. This suggests that change in functional diversity was not entirely independent on species diversity, although low sample sizes might have been an issue in detecting significant differences when standardized for species richness (Götzenberger et al., 2016). However, even if changes in functional diversity occurred simply due to loss of species diversity, some traits were clearly more affected than others (e.g. SLA and SS), which may portend downstream consequences for ecosystem functions associated with those traits.

At 0–5 m from edges, both pesticides increased standardized functional diversity of SLA and fungicide increased SS diversity. This contrast with forest interiors suggests that fungi appeared to favour the recruitment of functionally similar species near forest edges, independent of effects on species diversity. Thus, at forest edges, fungi appeared to filter out some traits regardless of species identities, which might occur with generalist fungi eliminating poorly defended species. Less specialized enemies could attack species with similar morphological traits, especially if these species also had similar defences (or lack thereof) against natural enemies (Gripenberg et al., 2018). Our findings open up the question of whether interactions between plant hosts and their enemies become less specialized with edge effects (Benítez-Malvido et al., 2005; Benítez-Malvido and Dáttilo, 2015; Bagchi et al., 2018).

Conclusion

Seedling establishment is a crucial bottleneck in plant population dynamics (Connell and Green, 2000; Green et al., 2014), and persistent functional changes among seedlings can influence future ecosystem function of forest fragments. Across a fragmented tropical forest, we found that edge effects on plant–enemy interactions contributed to differences in recruitment of species with different SLA and SS, and hence functional diversity of seedlings. Since SS and SLA constituted separate axes of trait variation among species, fungi and insects appeared to affect seedling recruitment along different trait dimensions. Assessing a wider array of traits for seedlings could help discern the mechanistic links between traits and species’ susceptibility to fungi and insects at the younger life stages.

Notably, much of the variation in recruitment could be attributed to random effects of species, i.e. a lot of baseline variation among species that the four traits here did not explain. This calls attention to continued studies on species-level responses to biotic and abiotic changes with edge effects. Additionally, edge and interior habitats vary in the composition of host species (Krishnadas et al., 2019), which can change the composition of associated pests and pathogens. Since not all pathogens inflict the same extent of damage, change in pathogen composition can also alter the extent to which host densities are regulated, creating feedback loops. Only long-term studies can reveal whether changes in trait-mediated seedling recruitment have lasting effects on community composition and ecosystem functions, such as biomass accumulation or nutrient cycling. Maintaining ecosystem health in human-modified forests might require that restoration efforts include the recovery of key ecological interactions that regulate community structure (Laughlin, 2014).

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following. Figure S1: study landscape and experimental design. Figure S2: variation in canopy openness with distance to the forest edge. Figure S3: principal components analysis of all traits. Figure S4: edge effects and natural enemy impacts on raw functional diversity of seedlings. Figure S5: edge effects and natural enemy impacts on standardized functional diversity of seedlings. Figure S6: raw functional diversity of seeds at different distances from the forest edge. Figure S7: standardized functional diversity of seeds at different distances from the forest edge. Table S1: functional trait data for 71 species. Table S2: loadings for principal components analysis of traits. Table S3: summary statistics for seed and seedling densities. Table S4: model coefficients for trait-mediated variation in recruit density with edge distance. Table S5: model coefficients for trait-mediated variation in recruit density with canopy openness. Table S6: model coefficients for trait-mediated variation in seed density with edge distance. Table S7: model coefficients for trait-mediated variation in recruit density with pesticide treatment at different edge distances. Table S8: seedling functional diversity in relation to canopy openness.

mcaa138_suppl_Supplementary_Material
Click here for additional data file. (682.6KB, docx)

ACKNOWLEDGEMENTS

We thank Kadamane Estate Company and Mr Cariappa for logistical support during the conduct of this research. Netraprasad Sharma, Suresh Roy and Arun Kumar were indispensable during fieldwork. Sachin Sridhara helped with data collection and made the map of the study area using GIS data provided by Raghunath from Nature Conservation Foundation. Natalia Umaña provided helpful feedback on analysis and draft versions. All authors declare having no conflicts of interest. M.K. conceived and designed the study with inputs from L.S.C. M.K. and K.A. collected the data. M.K. analysed the data and wrote the manuscript with substantial contributions from L.S.C.

Funding

This work was supported by the Yale Tropical Resources Institute (Research Fellowship 2015), Garden Club of America (Award in Tropical Botany 2016) and Harvard Arnold Arboretum (Ashton Award for Student Research 2016).

LITERATURE CITED

  1. Adler PB, Salguero-Gómez R, Compagnoni A, et al.  2014. Functional traits explain variation in plant life history strategies. Proceedings of the National Academy of Sciences, USA 111: 740–745. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Augspurger CK. 1984a Seedling survival of tropical tree species: interactions of dispersal distance, light-gaps, and pathogens. Ecology 65: 1705–1712. [Google Scholar]
  3. Augspurger CK. 1984b Light requirements of neotropical tree seedlings : a comparative study of growth and survival. Journal of Ecology 72: 777–795. [Google Scholar]
  4. Bagchi R, Brown LM, Elphick CS, Wagner DL, Singer MS. 2018. Anthropogenic fragmentation of landscapes: mechanisms for eroding the specificity of plant–herbivore interactions. Oecologia 187: 521–533. [DOI] [PubMed] [Google Scholar]
  5. Bagchi R, Gallery RE, Gripenberg S, et al.  2014. Pathogens and insect herbivores drive rainforest plant diversity and composition. Nature 506: 85–88. [DOI] [PubMed] [Google Scholar]
  6. Balderrama SIV, Chazdon RL. 2005. Light-dependent seedling survival and growth of four tree species in Costa Rican second-growth rain forests. Journal of Tropical Ecology 21: 383–395. [Google Scholar]
  7. Bell T, Freckleton RP, Lewis OT. 2006. Plant pathogens drive density-dependent seedling mortality in a tropical tree. Ecology Letters 9: 569–574. [DOI] [PubMed] [Google Scholar]
  8. Benchimol M, Peres CA. 2015. Edge-mediated compositional and functional decay of tree assemblages in Amazonian forest islands after 26 years of isolation. Journal of Ecology 103: 408–420. [Google Scholar]
  9. Benítez-Malvido J, Dáttilo W. 2015. Interaction intimacy of pathogens and herbivores with their host plants influences the topological structure of ecological networks in different ways. American Journal of Botany 102: 512–519. [DOI] [PubMed] [Google Scholar]
  10. Benítez-Malvido J, García-Guzmán G, Kossmann-Ferraz ID. 1999. Leaf-fungal incidence and herbivory on tree seedlings in tropical rainforest fragments: an experimental study. Biological Conservation 91: 143–150. [Google Scholar]
  11. Benítez-Malvido J, Martínez-Ramos M, Camargo JLC, et al.  2005. Responses of seedling transplants to environmental variations in contrasting habitats of Central Amazonia. Journal of Tropical Ecology 21: 397–406. [Google Scholar]
  12. Benítez MS, Hersh MH, Vilgalys R, Clark JS. 2013. Pathogen regulation of plant diversity via effective specialization. Trends in Ecology & Evolution 28: 705–711. [DOI] [PubMed] [Google Scholar]
  13. Benítez-Malvido J, Lázaro A, Ferraz IDK. 2018. Effect of distance to edge and edge interaction on seedling regeneration and biotic damage in tropical rainforest fragments: a long-term experiment. Journal of Ecology 106: 2204–2217. [Google Scholar]
  14. Bernard-Verdier M, Navas ML, Vellend M, Violle C, Fayolle A, Garnier E. 2012. Community assembly along a soil depth gradient: contrasting patterns of plant trait convergence and divergence in a Mediterranean rangeland. Journal of Ecology 100: 1422–1433. [Google Scholar]
  15. Bever JD, Mangan SA, Alexander HM. 2015. Maintenance of plant species diversity by pathogens. Annual Review of Ecology, Evolution, and Systematics 46: 305–325. [Google Scholar]
  16. Brooks ME, Kristensen K, van Benthem KJ, et al.  2017. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal 9: 378–400. [Google Scholar]
  17. Butterfield BJ, Suding KN. 2013. Single-trait functional indices outperform multi-trait indices in linking environmental gradients and ecosystem services in a complex landscape. Journal of Ecology 101: 9–17. [Google Scholar]
  18. Coley P, Barone J. 1996. Herbivory and plant defenses in tropical forests. Annual Review of Ecology and Systematics 27: 305–335. [Google Scholar]
  19. Connell JH. 1971. On the role of natural enemies in preventing competitive exclusion in some marine animals and in rain forest trees. In: Den Boer PJ, Gradwell GR, eds. Dynamics of populations. Wageningen, The Netherlands: Centre for Agricultural Publishing and Documentation, 298–312. [Google Scholar]
  20. Connell JH, Green PT. 2000. Seedling dynamics over thirty-two years in a tropical rainforest tree. Ecology 81: 568–584. [Google Scholar]
  21. Detto M, Visser MD, Wright SJ, Pacala SW. 2019. Bias in the detection of negative density dependence in plant communities. Ecology Letters 22: 1923–1939. [DOI] [PubMed] [Google Scholar]
  22. Didham RK, Kapos V, Ewers RM. 2012. Rethinking the conceptual foundations of habitat fragmentation research. Oikos 121: 161–170. [Google Scholar]
  23. Ewers RM, Didham RK. 2006. Confounding factors in the detection of species responses to habitat fragmentation. Biological Reviews of the Cambridge Philosophical Society 81: 117–142. [DOI] [PubMed] [Google Scholar]
  24. Frazer GW, Canham CD, Lertzman KP.  1999. Gap Light Analyzer (GLA), Version 2.0: Imaging software to extract canopy structure and gap light transmission indices from true-colour fisheye photographs, users manual and program documentation. Simon Fraser University, Burnaby, British Columbia, and the Institute of Ecosystem Studies, Millbrook, New York. [Google Scholar]
  25. Freckleton RP, Lewis OT. 2006. Pathogens, density dependence and the coexistence of tropical trees. Proceedings of the Royal Society B: Biological Sciences 273: 2909–2916. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Gilbert GS, Webb CO. 2007. Phylogenetic signal in plant pathogen–host range. Proceedings of the National Academy of Sciences, USA 104: 4979–4983. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Götzenberger L, Botta-Dukát Z, Lepš J, Pärtel M, Zobel M, de Bello F. 2016. Which randomizations detect convergence and divergence in trait-based community assembly? A test of commonly used null models. Journal of Vegetation Science 27: 1275–1287. [Google Scholar]
  28. Green PT, Harms KE, Connell JH. 2014. Nonrandom, diversifying processes are disproportionately strong in the smallest size classes of a tropical forest. Proceedings of the National Academy of Sciences, USA 111: 18649–18654. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Gripenberg S, Bagchi R, Gallery RE, Freckleton RP, Narayan L, Lewis OT. 2014. Testing for enemy-mediated density-dependence in the mortality of seedlings: field experiments with five Neotropical tree species. Oikos 123: 185–193. [Google Scholar]
  30. Gripenberg S, Rota J, Kim J, et al.  2018. Seed polyphenols in a diverse tropical plant community. Journal of Ecology 106: 87–100. [Google Scholar]
  31. Grubb PJ. 1977. Maintenance of species-richness in plant communities: the importance of the regeneration niche. Biological Reviews 52: 107–145. [Google Scholar]
  32. Haddad NM, Brudvig LA, Clobert J, et al.  2015. Habitat fragmentation and its lasting impact on Earth’s ecosystems. Science Advances 1: e1500052. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Janzen DH. 1970. Herbivores and the number of tree species in tropical forests. The American Naturalist 104: 501–528. [Google Scholar]
  34. Kembel SW, Cowan PD, Helmus MR, et al.  2010. Picante: R tools for integrating phylogenies and ecology. Bioinformatics (Oxford, England) 26: 1463–1464. [DOI] [PubMed] [Google Scholar]
  35. Kobe RK. 1999. Light gradient partitioning among tropical tree species through differntial seedling mortality and growth. Ecology 80: 187–201. [Google Scholar]
  36. Kraft NJB, Ackerly DD. 2010. Functional trait and phylogenetic tests of community assembly across spatial scales in an Amazonian forest. Ecological Monographs 80: 401–422. [Google Scholar]
  37. Krishnadas M, Comita LS. 2018. Influence of soil pathogens on early regeneration success of tropical trees varies between forest edge and interior. Oecologia 186: 259–268. [DOI] [PubMed] [Google Scholar]
  38. Krishnadas M, Comita LS. 2019. Edge effects on seedling diversity are mediated by impacts of fungi and insects on seedling recruitment but not survival. Frontiers in Forests and Global Change 2: 1–11. [Google Scholar]
  39. Krishnadas M, Bagchi R, Sridhara S, Comita LS. 2018. Weaker plant–enemy interactions decrease tree seedling diversity with edge-effects in a fragmented tropical forest. Nature Communications 9: 4523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Krishnadas M, Kumar AN, Comita LS. 2019. Edge effects reduce α-diversity but not β-diversity during community assembly in a human-modified tropical forest. Ecological Applications 29: e01996. [DOI] [PubMed] [Google Scholar]
  41. Lambers H, Raven JA, Shaver GR, Smith SE. 2008. Plant nutrient-acquisition strategies change with soil age. Trends in Ecology & Evolution 23: 95–103. [DOI] [PubMed] [Google Scholar]
  42. Laughlin DC. 2014. Applying trait-based models to achieve functional targets for theory-driven ecological restoration. Ecology Letters 17: 771–784. [DOI] [PubMed] [Google Scholar]
  43. Laurance WF, Camargo JLC, Luizão RCC, et al.  2011. The fate of Amazonian forest fragments: a 32-year investigation. Biological Conservation 144: 56–67. [Google Scholar]
  44. Laurance WF, Nascimento HE, Laurance SG, et al.  2006.  a Rapid decay of tree-community composition in Amazonian forest fragments. Proceedings of the National Academy of Sciences, USA 103: 19010–19014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Laurance WF, Nascimento HE, Laurance SG, et al.  2006.  b Rain forest fragmentation and the proliferation of successional trees. Ecology 87: 469–482. [DOI] [PubMed] [Google Scholar]
  46. Mangan SA, Schnitzer SA, Herre EA, et al.  2010. Negative plant–soil feedback predicts tree-species relative abundance in a tropical forest. Nature 466: 752–755. [DOI] [PubMed] [Google Scholar]
  47. Maron JL, Marler M, Klironomos JN, Cleveland CC. 2011. Soil fungal pathogens and the relationship between plant diversity and productivity. Ecology Letters 14: 36–41. [DOI] [PubMed] [Google Scholar]
  48. McCarthy-Neumann S, Ibáñez I. 2013. Plant–soil feedback links negative distance dependence and light gradient partitioning during seedling establishment. Ecology 94: 780–786. [Google Scholar]
  49. McCarthy-Neumann S, Kobe RK. 2008. Tolerance of soil pathogens co-varies with shade tolerance across species of tropical tree seedlings. Ecology 89: 1883–1892. [DOI] [PubMed] [Google Scholar]
  50. Melo FPL, Dirzo R, Tabarelli M. 2006. Biased seed rain in forest edges: evidence from the Brazilian Atlantic forest. Biological Conservation 132: 50–60. [Google Scholar]
  51. Melo FPL, Lemire D, Tabarelli M. 2007. Extirpation of large-seeded seedlings from the edge of a large Brazilian Atlantic forest fragment. Ecoscience 14: 124–129. [Google Scholar]
  52. Muller-Landau HC. 2010. The tolerance–fecundity trade-off and the maintenance of diversity in seed size. Proceedings of the National Academy of Sciences, USA 107: 4242–4247. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Murcia C. 1995. Edge effects in fragmented forests: implications for conservation. Trends in Ecology & Evolution 10: 58–62. [DOI] [PubMed] [Google Scholar]
  54. Nicotra AB, Chazdon RL, Iriarte SVB. 1999. Spatial heterogeneity of light and woody seedling regeneration in tropical wet forests. Ecology 80: 1908–1926. [Google Scholar]
  55. Osuri AM, Sankaran M. 2016. Seed size predicts community composition and carbon storage potential of tree communities in rain forest fragments in India’s Western Ghats. Journal of Applied Ecology 53: 837–845. [Google Scholar]
  56. Parker IM, Gilbert GS. 2018. Density-dependent disease, life-history trade-offs, and the effect of leaf pathogens on a suite of co-occurring close relatives. Journal of Ecology 106: 1829–1838. [Google Scholar]
  57. Pinheiro J, Bates D, DebRoy S, Sarkar D, R Development Core Team R . 2015. nlme: linear and nonlinear mixed effects models. R package version 3; 1 https://cran.r-project.org/package=nlme. [Google Scholar]
  58. R Development Core Team 2019. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
  59. Reich PB. 2014. The world-wide ‘fast–slow’ plant economics spectrum: a traits manifesto. Journal of Ecology 102: 275–301. [Google Scholar]
  60. Santo-Silva EE, Almeida WR, Melo FPL, Zickel CS, Tabarelli M. 2013. The nature of seedling assemblages in a fragmented tropical landscape: implications for forest regeneration. Biotropica 45: 386–394. [Google Scholar]
  61. Schroeder JW, Martin JT, Angulo DF, et al.  2018. Community composition and diversity of Neotropical root-associated fungi in common and rare trees. Biotropica 50: 694–703. [Google Scholar]
  62. Siefert A, Ravenscroft C, Weiser MD, Swenson NG. 2013. Functional beta-diversity patterns reveal deterministic community assembly processes in eastern North American trees. Global Ecology and Biogeography 22: 682–691. [Google Scholar]
  63. Sterck F, Markesteijn L, Schieving F, Poorter L. 2011. Functional traits determine trade-offs and niches in a tropical forest community. Proceedings of the National Academy of Sciences, USA 8: 20627–20632. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Swenson NG. 2014. Functional and phylogenetic ecology in R. Cham: Springer. [Google Scholar]
  65. Swenson NG, Enquist BJ. 2009. Opposing assembly mechanisms in a neotropical dry forest: implications for phylogenetic and functional community ecology. Ecology 90: 2161–2170. [DOI] [PubMed] [Google Scholar]
  66. Tabarelli M, Lopes AV, Peres CA. 2008. Edge-effects drive tropical forest fragments towards an early-successional system. Biotropica 40: 657–661. [Google Scholar]
  67. Tabarelli M, Aguiar AV, Girão LC, Peres CA, Lopes AV. 2010. Effects of pioneer tree species hyperabundance on forest fragments in northeastern Brazil. Conservation Biology 24: 1654–1663. [DOI] [PubMed] [Google Scholar]
  68. Umaña MN, Forero-Montaña J, Muscarella R, et al.  2016. Interspecific functional convergence and divergence and intraspecific negative density dependence underlie the seed-to-seedling transition in tropical trees. The American Naturalist 187: 99–109. [DOI] [PubMed] [Google Scholar]
  69. Valladares F, Niinemets Ü. 2008. Shade tolerance, a key plant feature of complex nature and consequences. Annual Review of Ecology, Evolution, and Systematics 39: 237–257. [Google Scholar]
  70. Valladares F, Laanisto L, Niinemets Ü, Zavala MA. 2016. Shedding light on shade: ecological perspectives of understorey plant life. Plant Ecology and Diversity 9: 237–251. [Google Scholar]
  71. Varma V, Osuri AM. 2013. Black Spot: a platform for automated and rapid estimation of leaf area from scanned images. Plant Ecology 214: 1529–1534. [Google Scholar]
  72. Wright SJ, Kitajima K, Kraft NJ, et al.  2010. Functional traits and the growth–mortality trade-off in tropical trees. Ecology 91: 3664–3674. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mcaa138_suppl_Supplementary_Material
Click here for additional data file. (682.6KB, docx)

Articles from Annals of Botany are provided here courtesy of Oxford University Press

RESOURCES