When it comes to Pipers, more is better!

Piper by Ruud de Block

Editor's Introduction

Chemical similarity and local community assembly in the species rich tropical genus Piper

Diversity can be described as the differences between species in their environment. Tropical environments are some of the most diverse in terms of plant species. In particular, Pipers are one of the most species rich genera that inhabit tropical ecosystems. Studying the mechanism behind their diversity is important for their understanding. There are many ways biodiversity can be measured. In this study, the authors analyzed the chemical similarity and species coexistence of Pipers to understand how they are likely to be arranged. They investigated the phylogeny and chemical composition of various species within the genus and provided valuable insight to the scientific community. 

NOTE: This article is part of a Collection of student-annotated papers that are the product of the SitC team’s research into best practices for using primary literature to support STEM education. For this reason, these papers have undergone an alternate review process and may lack educator guides. To learn more, visit the main Collection page: SitC Lab.

Paper Details

Original title
Chemical similarity and local community assembly in the species rich tropical genus Piper
Original publication date
Volume 97, Issue 11 November 2016
Issue name


Community ecologists have strived to find mechanisms that mediate the assembly of natural communities. Recent evidence suggests that natural enemies could play an important role in the assembly of hyper-diverse tropical plant systems. Classic ecological theory predicts that in order for coexistence to occur, species differences must be maximized across biologically important niche dimensions. For plant–herbivore interactions, it has been recently suggested that, within a particular community, plant species that maximize the difference in chemical defense profiles compared to neighboring taxa will have a relative competitive advantage. Here we tested the hypothesis that plant chemical diversity can affect local community composition in the hyper-diverse genus Piper at a lowland wet forest location in Costa Rica. We first characterized the chemical composition of 27 of the most locally abundant species of Piper. We then tested whether species with different chemical compositions were more likely to coexist. Finally, we assessed the degree to which Piper phylogenetic relationships are related to differences in secondary chemical composition and community assembly. We found that, on average, co-occurring species were more likely to differ in chemical composition than expected by chance. Contrary to expectations, there was no phylogenetic signal for overall secondary chemical composition. In addition we found that species in local communities were, on average, more phylogenetically closely related than expected by chance, suggesting that functional traits other than those measured here also influence local assembly. We propose that selection by herbivores for divergent chemistries between closely related species facilitates the coexistence of a high diversity of congeneric taxa via apparent competition.



A classic goal of ecology has been to understand the processes that determine species community assembly (Weiher and Keddy 1999). Current theory predicts that the assembly of species at the local scale is determined by two major ecological processes: the interaction between a species and its abiotic environment, and interactions among the species themselves (Götzenberger et al. 2012). Within the same trophic level, theory also predicts that species that are more similar to each other will be less likely to coexist than more dissimilar species, due to competition for a limited set of resources (Wright 2002, Götzenberger et al. 2012). This mechanism, generally known as “species-limiting similarity” (Darwin 1859, MacArthur and Levins 1967), has been considered to be one of the major processes responsible for structuring biological communities at the local scale.

Numerous studies have shown that local species composition is constrained to those from the regional species pool that are most phenotypically or ecologically divergent (e.g., Cavender-Bares et al. 2004, Maherali and Klironomos 2007, Cooper et al. 2008, Kraft et al. 2008, Cornwell and Ackerly 2009, Graham et al. 2009, Ingram and Shurin 2009, Kursar et al. 2009, Maherali and Klironomos 2012, Sedio et al. 2012, Coley and Kursar 2014). In the Neotropics, however, one may find numerous closely related species co-occurring. For example, as many as 64 species of Piper are recorded to co-occur at a Peruvian lowland location (Marquis 2004). These very diverse local species assemblages from hyper-diverse plant groups (also known as “species swarms,” sensu Gentry 1982) would seem to challenge the limiting similarity hypothesis. In many of these genera (e.g., Piper and Peperomia [Piperaceae], Miconia and Clidemia[Melastomataceae], 
Elaphoglassum [Dryopteridaceae], Psychotria [Rubiaceae], 
Bursera and Protium [Burseraceae], Shorea [Dipterocarpaceae], and Inga[Mimosaceae]), there appears to be insufficient morphological and functional differentiation to explain local coexistence (Coley and Kursar 2014).

Currently, one of the most common strategies to assess species similarity is to describe their functional traits (McGill 2006). This approach measures species characteristics that determine the quantitative and qualitative use of a specific set of available resources, including enemy free space (Swenson 2013). Species with similar values of functional traits are expected to undergo stronger competitive interactions than species that differ more in said traits.

For plants, one of the most important functional traits is secondary chemistry (Hartmann 2007). While other functional traits are related to the use of a single resource, secondary metabolites can be associated with numerous critical functions for plants: protection against abiotic factors (Wahid et al. 2007), reproduction (pollinator and seed disperser attraction; Kessler and Baldwin 2007), competition via allelopathy (Ridenour and Callaway 2001), and plant defense (e.g., direct and indirect defense against herbivores and pathogens; Pusztahelyi et al. 2015). Given that plant secondary chemistry can play a critical role in plant–natural-enemy interactions, it is reasonable to expect that chemical similarity between sympatric taxa can also play an important role in determining species coexistence. The reason for this is that plant species with similar chemistries will share herbivores. As a result, they will suffer higher herbivore attack when co-occurring, i.e., they will experience apparent competition (indirect competition via a shared enemy; Holt and Lawton 1993).

From an evolutionary perspective, trait-based studies of species coexistence also provide a unique opportunity to assess the role of species evolutionary histories on patterns of community assembly, as well as to explore phylogenetic patterns of niche and trait evolution (Webb et al. 2002, Kembel and Hubbell 2006, Losos 2008, Cavender-Bares et al. 2009). An excellent example of this interdisciplinary approach for plant species coexistence is that by Kursar et al. (2009), who focused on secondary compounds as they confer defense against herbivores. They demonstrated that coexisting species of Inga (a hyper-diverse tropical taxon) were more different in their secondary chemical defenses than expected by chance. Moreover, they showed a lack of phylogenetic signal in these chemical defenses. Their results suggest the presence of divergent selection on anti-herbivore chemical defenses, and that such divergence is likely to play a pivotal role in structuring community assembly. Here, we address whether or not the results for Inga are generalizable to another species-rich genus, Piper.

Specifically, we assess the role that (1) plant secondary chemistry and (2) plant phylogeny play for species coexistence of Piper in a low land tropical forest. We apply both a species-pair and a community-based approach. We use gas chromatography mass spectroscopy (GC-MS) to assess chemical similarity across a range of secondary metabolite groups. We predicted that (1) local assemblages of Piper would consist of species more different in secondary metabolites than expected by chance and (2) chemistry profiles of individual species would be influenced by evolutionary history.


Site and system

This study was conducted at the La Selva Biological Station in Costa Rica (operated and owned by the Organization for Tropical Studies [OTS]) between May and August 2007. Located in the Atlantic lowlands of Puerto Viejo de Sarapiquí (Heredia), the station possesses more than 1,600 ha of tropical wet forest and receives approximately 4,000 mm of rainfall annually. Currently, 1,850 species of vascular plants have been cataloged in La Selva, 50 of which are in the genus Piper.

Piper is a pantropical genus with approximately 1,000 species in the Neotropics (Jaramillo 2006). The natural geographic range of the genus in the New World is from northern Mexico to northern Argentina. Piper species are very abundant in low- and mid-elevation forests (but rarely reaching 2,500 m) and are often among the most speciose plant genera in Neotropical forests understories (Gentry 1982, Marquis 2004). Most Piper species at La Selva occur in discrete, multi-species patches that can contain up to 21 different species (Salazar et al. 2012). Piper secondary metabolite diversity has been extensively studied and there is an important body of published methods for compound isolation, chemical profiling, and artificial synthesis (Parmar et al. 19971998, Kato and Furlan 2007). Furthermore, the effects of their secondary chemistry on herbivores, pathogens, and seed dispersal are well documented (Dyer et al. 20012004, Mikich et al. 2003, Fincher et al. 2008, Marques et al. 2010, Richards et al. 2015).

Species coexistence

To assess Piper species coexistence, we sampled 81 natural multi-species patches of Piper throughout the primary forest of La Selva. We established transects parallel to the station trails to locate naturally occurring Piper patches. All transects were between 50 and 100 m from the trail and a minimum 250 m from each other. To standardize the size of each plot sampled, a 10 m radius plot was set up in every patch. In each plot, we counted the number of individuals of Piper 1 cm or greater in diameter at ground level; all Piper plants were identified to species.

To assess species coexistence, we used two complementary indices. First, we calculated species co-occurrence c score index for all species pairs (based on presence/absence data; Stone and Roberts 1990). Second, to take into account the effect of species abundance, we assessed the degree of species coexistence for all species pairs using a niche overlap approach. Here, plot was considered as the studied niche and we used plot occupancy as a measure of niche use. Occupancy was calculated for all Piper species as the relative abundance of each species in every sampled plot. Finally, niche overlap was calculated using Pianka's Index (Pianka 1973) based on 1,000 iterations. Both measures of species coexistence were calculated using EcoSim 7.1 (Gotelli and Entsminger 2012) and yielded two pair-wise species matrices, one for the co-occurrence c score and another for Pianka's niche overlap index.

Chemical similarity

We collected leaf material from young, fully expanded leaves for all Piper species sampled in the plots; samples were obtained from at least four individuals of each species. Additionally, all samples selected had similar herbivore damage (between 5% and 10% leaf area missing, with damage assessed visually). Samples were immediately dried with silica gel after collection and transported to the University of Missouri, St. Louis for chemical analysis. Although it is possible that some high volatility compounds were affected during field collections, all samples were treated identically and therefore, all possible compound loss was likely to be systematic across all samples and species. From each sample, 0.4 g of material was ground under liquid nitrogen. To extract a broad range of secondary metabolites (polar and non-polar), samples were extracted using 1.5 mL of 1:1 methanol-chloroform solution. As an internal standard, 0.1 mg of piperine was added to all samples. Samples were finally filtered (0.2 μ) and stored in volatile organic compound (VOC) vials at −80°C until analysis.

Qualitative chemical analysis of the extracts was performed using GC-MS (HP 5890 coupled with a quadrapole Model 5988A mass detector, Hewlett Packard, USA) with helium as a carrier gas and a HP-5 capillary column (Hewlett Packard, USA) (30 m, 0.32 mm ID, 0.25 μm; see details for chromatographic conditions in the Supporting Information). It is important to note that because this study focuses only on compounds that are detectable with GC-MS, our chromatographic technique will not detect all secondary compounds that might affect plant herbivore interactions. Nevertheless, a review of more than 3500 records of secondary compounds found in Piper from NAPRALERT (Natural Products Alert Database; Loub et al. 1985; data available online)5 showed that GC-MS-detectable compounds account for more than 86% of all Piper secondary metabolites (Appendix S1: Fig. S1). Therefore, our approach is likely to capture a significant percentage of the total secondary chemical diversity of Piper species. Because the abundance of secondary compounds can vary between individuals due to factors such as induction, genetic variability, and resource availability, we only use presence and absence data of chromatographic features. We assessed chemical similarity between all sampled species by building a mass spectra library containing all chromatographic features for each species (one library per species). The libraries of each species were then cross-referenced across all species using AMDIS (Automated Mass Spectral Deconvolution and Identification System) to identify common as well as unique features based on mass spectra, molecular weight, and expected retention time (Stein 1999, Stein et al. 2005). It is important to underline that this particular approach can assess chemical similarity among species independently of chemical compound identification (for a proof of the concept of this approach, see Salazar 2013). This methodology yielded a species-pair matrix of chemical similarity between all sampled Piper species.

Using the chemical similarity data from the above methodology, we performed a hierarchical clustering analysis (Ward's algorithm, R package pvclust; Suzuki and Shimodaira 2006, R Core Team 2012) to construct a species chemical dendrogram. Subsequently, we extracted a species-pair matrix of chemical distances from the dendrogram. Additionally, we applied a principal component analysis (JMP 10.0; SAS Institute 2007) to the data on chromatographic features to generate continuous values of chemical diversity for the plant species in order to calculate the phylogenetic signal of overall chemical diversity across our Piper species pool (see Phylogenetic analysis in Materials and methods).

Finally, the mass spectra of the different compounds in the samples were compared with NIST/EPA/NIH and MassBank Databases (Horai et al. 2010) as well as primary literature. Metabolites that did not have a match from the available mass spectra databases were classified as unknown.

Phylogenetic analysis

We constructed a phylogenetic tree of all study species to determine the influence of phylogeny on chemical similarity. Samples of leaves were collected to perform sequencing of the ITS and the chloroplast intron psbJ-petA for phylogenetic analysis (following Jaramillo et al. 2008). Finally, to (1) control for phylogenetic non-independence of the chemical similarity between the sampled species and (2) to assess the effect of phylogenetic relatedness on species coexistence, we used the Picante R package (Kembel et al. 2010, R Core Team 2012) to generate a species-pair matrix of phylogenetic distances based on branch lengths.

Statistical analysis

To determine the effect that chemical similarity had on Piper species coexistence, we used two complementary approaches. First we used a species-pair approach to ascertain the effect that individual Piper species (thus specific species chemical compositions) had on species coexistence. Second, we used a community-based approach to quantify the combinatory or cumulative effect of multiple Piper species on local species coexistence.

Species-pair approach

To assess the correlation between species coexistence and chemical similarity, we performed (1) a Mantel test and (2) a partial Mantel test controlling for phylogenetic non-independence (10 ,000 permutations each). Each analysis was conducted with both measures (c score index and Pianka's Index) of species coexistence. Additionally, a simple Mantel test between species coexistence and phylogenetic distances was performed to quantify the relationship between phylogenetic similarity and coexistence (10, 000 permutations; PASSaGE 2.0; Rosemberg and Anderson 2011).

Community-based approach

We quantified the community phylogenetic and chemical over/underdispersion using (1) the Inverse Nearest Relative Index (-NRI), which measures tree-wide patterns of clustering, as well as (2) the Inverse Nearest Taxon Index (-NTI), which assesses clustering independently of deeper node clustering patterns (Webb et al. 2002, Webb and Donoghue 2005, Kembel and Hubbell 2006). To determine community chemical over/underdispersion we used the chemical dendrogram data as input for the analysis. Negative values of -NTI and -NRI indicate that similar species (phylogenetically and also, in this case, chemically) co-occur more than expected by chance; positive values indicate that similar species are not likely to co-occur. The randomization to generate null communities was done by shuffling phylogeny and chemical dendrogram tip labels in order to calculate the standardized effect sizes for -NRI and -NTI (abundance weighted model, n = 1,000 per community; Picante package). Additionally, we used the first principal component of the PCA derived from the presence/absence chemical data to calculate the Bloomberg's K for phylogenetic signal of secondary chemistry over our focal species (Picante package).


We sampled a total of 2,035 individuals from 27 species of Piper across the 81 sampled plots (Appendix S1: Table S1). The number of individuals present per plot was 25.2 ± 1.1 (mean ± SE; max–min = 4–51), and the number of Piper species per plot was 5.2 ± 1.4 (max–min = 3–11).

The GC-MS analysis yielded more than 1,100 chromatographic features. Approximately 40% of all features were present in all Piper species (e.g., phytol, stigmasterol, sitorterol, and tocopherol). Because these shared features were non-informative and most likely related to plant primary chemistry, they were not used for the clustering analysis. Among the remaining features, we found a great diversity of terpenes, phenylpropanoids, some lignans, flavonoids, and alkaloids (Appendix S1: Table S2). The hierarchical clustering showed five discrete chemical clusters (Fig. 1 and Appendix S1: Fig. S2).


Figure 1

Fig. 1. Comparison of the phylogenetic tree (left) and the chemical dendrogram (right) of the 27 Piper species sampled across the 81 natural patches.


The purpose of this figure was to determine the relationship between the evolution of certain species (visualized through the phylogentic tree) of Piper and their chemical similarities (visualized through the chemical dendrogram).

Phylogentic tree vs. chemical dendrogram

Compares a phylogenetic tree (which is constructed according to the species common ancestors or evolution tract) with a chemical dendrogram (which is constructed by grouping the presence or absence of certain chemicals).


The analysis suggests that closely related species are less similar in their secondary chemistry than expected from the Brownian motion model of evolution

A description of the chemical characteristics of the five Piper chemical clusters is included in the Appendix S1: Fig. S1. [Color figure can be viewed at wileyonlinelibrary.com]

Phylogenetic analysis yielded a local species phylogeny that concurs with the current phylogenetic Piper data (Appendix S1: Fig. S3). We did not find a strong phylogenetic signal for secondary chemical composition in our focal species (K = 0.03). This can be clearly seen in Fig. 1. The small value of K (K < 1) suggests that closely related species are less similar in their secondary chemistry than expected under a Brownian motion model of evolution. Nonetheless, a randomization test showed that K was not significantly different from 1 (ZPIC = −0.51, PPIC = 0.40).

For species coexistence, the species-pair approach showed a significant positive relationship between species chemical distance and the likelihood of species co-occurrence for both species presence/absence data (Gotelli's c score; Mantel test, r = 0.20, = 0.0001) and abundance-weighted data (Pianka's index; Mantel test, r = 0.17, = 0.0014). Similar results were obtained when controlling for phylogenetic non-independence: presence/absence co-occurrence (partial Mantel test, r = 0.20, = 0.002) and abundance-weighted co-occurrence (partial Mantel test, r = 0.18, = 0.0015). Both results suggest that chemically distinct species are more likely to co-occur. In contrast, we found a significant negative relationship between species phylogenetic distance and presence/absence co-occurrence (r = −0.16, = 0.01). Thus, more closely related species are more likely to co-occur. However, there was no significant relationship between phylogenetic distance and abundance-weighted species co-occurrence (r = 0.03, = 0.54).

In our community-based approach, we found that Piper species were, on average, more overdispersed with respect to their secondary chemical composition than expected by chance. However, only -NRI was significantly different from zero (-NRI, t = 1.83, df = 80, = 0.03; -NTI, t = 0.77, df = 80, = 0.22). In contrast, species composition within the plots was phylogenetically underdispersed. Both -NRI and -NTI were significantly different from zero (t = −5.24, df = 80, = 0.0001 and t = −2.26, df = 80, = 0.01, respectively; Fig. 2).


Fig. 2. Standardized chemical and phylogenetic community dispersion measured as Inverse Nearest Relative Index (‐NRI) and Inverse Nearest Taxon Index (‐NTI). Values above 0 indicate overdispersion and values below 0 indicate underdispersion. Daggers indicate values significantly different than 0 under a null model (communities assembled at random; †<0.05, ‡<0.005). Box shows 95% confidence intervals, vertical bars show full data range, middle box line shows group average, bar across plot represents grand mean.

Chemical data

Based on the plot, the chemical NRI and NTI means are positive values which indicate that similar species (phylogenetically and also, in this case, chemically) are not likely to co-occur. A caveat to this is that the boxes and vertical bars stretch into the negative region indicating levels of co-occurrences for a small percentage of the samples in the data set.  

Phylogenetic data

Based on the plot, the chemical NRI and NTI means are negative values which indicate that similar species (phylogenetically and also, in this case, chemically) co-occur more than expected by chance A caveat to this is that the boxes and vertical bars stretch into the positive region indicating levels of non-co-occurrences for a small percentage of the samples in the data set.  

In this study, we found that both chemical and phylogenetic similarity can have significant yet contrasting effects on species coexistence. Results from both of our approaches (species-pair and community-based) showed that Piper species with higher secondary chemical similarity were less likely to coexist in the same Piperpatch than communities assembled at random. In contrast, closely related species were more likely to coexist in the same patch. Furthermore, counter to one of our original predictions, we found that the overall composition in secondary metabolites was not phylogenetically conserved for the 27 studied Piper species.

In our species-pair approach, we found that, independent of the measure of coexistence used, chemical similarity had a significant negative effect on species coexistence. This result is consistent with patterns found for two other available studies of Neotropical species-rich genera, Bursera (Burseraceae; Becerra 2007) and Inga (Mimosaceae; Kursar et al. 2009). As in our case, Inga and Bursera species were less likely to coexist with conspecifics that had similar secondary metabolite composition. By controlling for phylogenetic non-independence, our results also suggest that the effect that chemical composition has on community assembly is not the result of chemical similarity due to common ancestry.

Phylogenetic distance was also found to be important for Piper community assembly. Contrary to the effect of chemical similarity, Piper species that were closely related were more likely to coexist in a particular patch. This pattern is likely the result of other unmeasured traits that, unlike chemical similarity, are strongly conserved across the phylogenetic history of our target Piper species. One possibility could be a strong environmental niche conservatism in which closely related species are more likely to have similar habitat preferences (Daws et al. 2002, Sedio et al. 2013). Furthermore, given that phylogenetic similarity was only significant for our presence/absence coexistence measures but not for the abundance weighted data, it seems probable that these unknown traits are not necessarily associated with ecological interactions that are density-dependent or that provide Piper species a strong competitive advantage. Additional evidence supporting this hypothesis can be found by close examination of the phylogenetic patterns of species abundances. In our data, the most abundant Piper species found in our plots were all from different sub-clades of the Piper phylogeny (P. trigonum, 346 individuals, Peltobryon clade; P. multiplinervium, 294 individuals, Pothomorphe clade; P. urostachyum, 240 individuals, Radula clade; P. cenocladum, 227 individuals, Macrostachys clade). Therefore, given that the most locally abundant Piper species are not closely related, these phylogenetically conserved traits are not likely to have a strong impact on interspecific interactions.

Our community-level results concur with those of the species-pair approach. Patches showed a significant overdispersion in terms of Piper chemical composition (positive values of -NRI and –NTI, Fig. 2) suggesting that species within a patch are less chemically similar than expected by chance. Furthermore, these results suggest that secondary metabolite composition is also important for the community assembly of complex multi-species patches. Nevertheless, the fact that this pattern was only found to be significant for –NRI suggests that the effect of chemical composition on community assembly is stronger between Piper species from different sub-clusters of the chemical dendrogram. Given that the major sub-clusters in our dendrogram differ mainly in terms of the richness of compounds from specific secondary metabolite groups (e.g., flavonoids, amides, phenylpropanoids; see Appendix S1: Fig. S1), it is likely that the effect of chemical composition on community assembly could be largely driven by differences with respect to representation from these major metabolite groups rather than by the presence or absence of each. It is important to note that, while the average value of -NRI is significantly different from zero, some plots showed chemical underdispersion. Thus, caution should be exercised in their interpretation. Nevertheless, given that the species-pair approach confirmed a general trend, we consider these results to be informative.

Conversely, Piper patches showed a significant phylogenetic underdispersion (negative values for -NRI and -NTI, Fig. 2). This suggests that Piper species in a patch are more closely related than expected by chance, a pattern that also agrees with the species-pair approach. Nonetheless, phylogenetic underdispersion was significant for both -NRI and -NTI, a result that supports the idea that the effect of phylogeny on community assembly is associated with strongly conserved traits not measured in this study.

Although we did not test the efficacy of secondary metabolites as anti-herbivore and anti-pathogen defenses, we believe that plant–natural-enemy interactions are responsible for a great proportion of the effect that chemical composition has on community assembly (Coley and Kursar 2014). Most of the chemicals found by our analysis are well known to confer direct and indirect anti-herbivore protection to plants (Appendix S1: Fig. S1). Furthermore, a recent paper by Richards et al. (2015) showed that secondary chemical diversity of all species used for this study predicts Piper herbivore species richness, leaf damage, and parasitoid attack at the La Selva Biological Station. Although Richards et al. (2015) measured chemical diversity as the diversity of chemical functional groups per Piper species, their findings strongly support the link between chemical plant species complexity and plant herbivore interactions.

Our results suggest that Piper species that are chemically dissimilar from those already present in a patch would be more likely to colonize and persist within the patch. The colonizing species would have to be sufficiently chemically distinct so as to prevent a potential increase in the number of shared herbivores with other plant hosts already in the community. This successful colonization is likely to be mediated via four distinct mechanisms. First, the abundance of the colonizing species would initially be low resulting in a low initial abundance of herbivores (the resource concentration hypothesis; Root 1973). Second, the colonizing Piper species is less likely to be chemically compatible with the metabolic constraints of the local herbivores. Third, because of the increase in local semiochemical complexity, extant herbivores are likely to suffer chemical disorientation (Zhang and Schlyter 2003). And fourth, if the colonizing species does not share herbivores with plants already present, the novel species would be less likely to be the victim of apparent competition (Holt and Lawton 1993). All advantages acting in concert could facilitate the coexistence of multiple, closely related, yet chemically distinct species. For these mechanisms to function, host chemical similarity should be correlated with host herbivore similarity. In other words, chemically similar plant species must share herbivores species, while chemically distinct plants must have unique herbivore species. Richards et al. (2015) provide some evidence that differences in chemistry among Piper species determines the nature each plant species’ herbivore fauna.

We acknowledge that, although community chemical similarity is likely to have a significant effect on plant–herbivore interactions, the potential role that plant pathogens could play on mediating coexistence through plant chemical similarity is an interesting possibility that is likely to generate similar patterns and, thus, requires formal testing. Nevertheless, fungal attack of La Selva Piper, although not entirely absent, is rare (R. J. Marquis, personal observation).

It is important to note that, for our study system, there seem to be two distinct sets of traits that are influencing coexistence in opposite manners. While communities tend to be chemically overdispersed, the same communities are also phylogenetically underdispersed. This is not the only example of species partitioning available niche space, but it has potentially interesting evolutionary outcomes. It is reasonable to expect that taxa under these two intrinsically different evolutionary pressures are likely to engender highly chemically diverse species pools with relatively low phylogenetic diversity.

Although we did not find a strong phylogenetic signal of chemical composition in our local species pool (a pattern that agrees with that of Kursar et al. 2009), phylogeny is likely to have a significant effect on Piper secondary chemistry at some, perhaps higher, taxonomic level. For example, although the amide piperine has been found in more than 20 Piper species, all of these species belong to the “tropical Asian Piperclade” (sensu Jaramillo et al. 2008). A similar pattern can be found for other compounds like piperlonguminine, methysticin, and yangonin (NAPRALERT [see footnote 5, data obtained 2011]; Loub et al. 1985). Finally, compounds like 4-nerolidylcatechol can be found in multiple Neotropical Piper species (especially, but not exclusively, in the Pothomorphe clade; NAPRALERT [see footnote 5, data obtained 2011]) but not in species of the Asian Piper clades.

Piper, like other species-rich plant groups, can have very high local species richness, yet it appears to have very low ecological diversity. Most species in the Neotropics are understory shrubs and small treelets of wet, lowland forest. Pollinated by generalist pollinators (Semple 1974, De Figueiredo and Sazima 2000) and dispersed by similar organisms (mostly a handful of species of the genus Carollia, Phyllostomidae; Fleming 19811985, Thies and Kalko 2004), Piper belongs to a small but abundant group of taxa that, due to the lack of obvious morphological and functional differentiation, challenge classical notions of ecological interactions and speciation processes (Frodin 2004, Kursar et al. 2009, Sedio et al. 2012). Given our results, we propose that the interaction between Piper and its natural enemies (mediated by secondary chemistry) is likely to play a major role in the community assembly and local coexistence of species in this genus. Finally, we put forward that the strong concordance between the finding of Kursar et al. (2009) and the present work could be a glimpse of a more widespread pattern, a pattern in which natural enemies and even multitrophic interactions may perhaps play a key role in the assembly of natural plant communities, as well as the evolutionary processes that have driven tropical plant radiation.


Financial support came from the Whitney R. Harris World Ecology Center, the Organization for Tropical Studies, and the National Science Foundation (DEB-1210643). Funding for GC analysis from DEB-1210643 and BIR-9419994.


  1. J.X. Becerra, Proceedings of the National Academy of Sciences USA 104, 7483–7488 (2007).
  2. J. Cavender-Bares, D. D. Ackerly, D. A. Baum, F. A. Bazzaz, American Naturalist 163, 823–843 (2004).
  3. J. Cavender-Bares, K. H. Kozak, P. V. A. Fine, S. W. Kembel, Ecology Letters 12, 693–715 (2009).
  4. P.D. Coley, T. A. Kursar. Science 343, 35–36 (2014).
  5. N. Cooper, J. Rodríguez, A. Purvis, Proceedings of the Royal Society B 275, 2031–2037 (2008).
  6. W.K. Cornwell, D. D. Ackerly, Ecological Monographs 79, 109–126 (2009).
  7. C. Darwin, On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life. First edition. (Murray, London, UK, 1859)
  8. M.I. Daws, D. F. R. P. Burslem, L. M. Crabtree, P. Kirkman, C. E. Mullins, et al. Functional Ecology 16, 258–267 (2002).
  9. R.A. De Figueiredo, M. Sazima,  Annals of Botany 85, 455–460 (2000).
  10. L.A. Dyer, C. D. Dodson, J. Beihoffer, D. K. Letourneau, Journal of Chemical Ecology 27, 581–592 (2001).
  11. L.A. Dyer, D. K. Letourneau, C. D. Dodson, M. A. Tobler, J. O. Stireman III, et al. Ecology 85, 2795–2803 (2004).
  12. R.M. Fincher, L. A. Dyer, C. D. Dodson, J. L. Richards, M. A. Tobler, et al. Journal of Chemical Ecology 34, 558–574 (2008).
  13. T.H. Fleming, Oecologia 51, 42–46 (1981).
  14. T.H. Fleming, Ecology 66, 688–700 (1985).
  15. D.G. Frodin, Taxon 53, 753–776 (2004).
  16. A.H. Gentry, Annals of the Missouri Botanical Garden 69, 557–593 (1982).
  17. N.J. Gotelli, G. L. Entsminger. 2012. EcoSim 7.72. (Acquired Intelligence, Jericho, Vermont, USA, 2012).
  18. L. Götzenberger, et al. Biological Reviews 87, 111–127 (2012).
  19. C.H. Graham, J. L. Parra, C. Rahbek, J. A. McGuire. Proceedings of the National Academy of Sciences USA 106, 19673–19678 (2009).
  20. T. Hartmann, Phytochemistry 68, 2831–2846 (2007).
  21. R.D. Holt, J. H. Lawton. American Naturalist 142, 623–645 (1993).
  22. H. Horai, et al. Journal of Mass Spectrometry 45, 703–714 (2010).
  23. T. Ingram, J. B. Shurin. Ecology 90, 2444–2453 (2009).
  24. M.A. Jaramillo, Biodiversity and Conservation 15, 1695–1712 (2006).
  25. M.A. Jaramillo, R. Callejas, C. Davidson, J. F. Smith, A. C. Stevens, et al. Systematic Botany 33, 647–660 (2008).
  26. M.J. Kato, M. Furlan. Pure and Applied Chemistry 79, 529–538 (2007).
  27. S.W. Kembel, S. P. Hubbell,  Ecology 87, 86–99 (2006).
  28. S.W. Kembel, P. D. Cowan, M. R. Helmus, W. K. Cornwell, H. Morlon, et al. Bioinformatics 26, 1463–1464 (2010).
  29. D. Kessler, I. T. Baldwin, Plant Journal 49, 840–854 (2007).
  30. N.J.B.  Kraft, R. Valencia, D. D. Ackerly, Science 322, 580–582 (2008).
  31. T.A. Kursar, K. G. Dexter, J. Lokvam, R. T. Pennington, J. E. Richardson, et al. Proceedings of the National Academy of Sciences USA 106, 18073–18078 (2009).
  32. J.B. Losos, Ecology Letters 11, 995–1003 (2008).
  33. W.D. Loub, N. R. Farnsworth, D. D. Soejarto, M. L. Quinn, Journal of Chemical Information and Computer Science 25, 99–103 (1985).
  34. R. MacArthur, R. Levins, American Naturalist 101, 377 (1967).
  35. H. Maherali, J. N. Klironomos, Science 316, 1746–1748 (2007).
  36. H. Maherali, J. N. Klironomos, PLoS ONE 7, 36695 (2012).
  37. J.V. Marques, A. De Oliveira, L. Raggi, M. C. M. Young, M. J. Kato. Journal of the Brazilian Chemical Society 21, 1807–1813 (2010).
  38. R.J. Marquis, A model genus for studies of chemistry, ecology, and evolution. (Kluwer Academic Press, 78–96, 2004).
  39. B.J. McGill, B. J. Enquist, E. Weiher, M. Westoby. Trends in Ecology & Evolution 21, 178–185 (2006).
  40. S.B. Mikich, G. V. Bianconi, B. H. L. N. S. Maia, S. D. Teixeira, Journal of Chemical Ecology 29, 2379–2383 (2003).
  41. V.S. Parmar, et al. Phytochemistry 46, 597–673 (1997).
  42. V.S. Parmar, et al. Phytochemistry 49, 1069–1078 (1998).
  43. E.M. Pianka, Annual Review of Ecology and Systematics 4, 53–74 (1973).
  44. T. Pusztahelyi, I. J. Holb, I. Pócsi, Frontiers in Plant Science 6, 573 (2015).
  45. R Core Team. R: a language and environment for statistical computing. (R Foundation for Statistical Computing, Vienna, Austria, 2012)
  46. L.A. Richards, L. A. Dyer, M. L. Forister, A. M. Smilanich, C. D. Dodson, et al. Proceedings of the National Academy of Sciences USA 112, 10973–10978 (2015).
  47. W.M. Ridenour, R. M. Callaway, Oecologia 126, 444–450 (2001).
  48. R.B. Root, Ecological Monographs 43, 95–124 (1973).
  49. M. Rosemberg, C. D. Anderson, Methods in Ecology and Evolution 2, 229–232 (2011).
  50. D. Salazar, Exploring the local and regional effects of plant diversity on plant herbivore interactions (Department of Biology, Saint Louis, Missouri, USA, 2013)
  51. D. Salazar, H. K. Detlev, R. J. Marquis, Ecology 94, 2444–2453 (2012).
  52. SAS Institute. JMP 10.0 (SAS Institute, Cary, North Carolina, USA, 2007).
  53. B.E. Sedio, S. J. Wright, C. W. Dick, Journal of Ecology 100, 1183–1193 (2012).
  54. B.E. Sedio, J. R. Paul, C. M. Taylor, C. W. Dick, Nature Communications 4, 2317 (2013).
  55. K.S.  Semple, Annals of the Missouri Botanical Garden 61, 868–871 (1974).
  56. S.E. Stein, Journal of the American Society for Mass Spectrometry 10, 770–781 (1999).
  57. S. Stein, Y. Mirokhin, D. Tchekhovskoi, G. Mallard, The NIST mass spectral search program for the NIST/EPA/NIH mass spectral library. (Scientific Instrument Services, Ringoes, New Jersey, USA, 2005)
  58. L. Stone, A. Roberts, Oecologia 85, 74–79 (1990).
  59. R. Suzuki, H. Shimodaira, Bioinformatics 22, 1540–1542 (2006).
  60. N.G. Swenson, Ecography 36, 264–276 (2013).
  61. W. Thies, E. K. V. Kalko, Oikos 104, 362–376 (2004).
  62. A. Wahid, S. Gelani, M. Ashraf, M. R. Foolad, Environmental and Experimental Botany 61, 199–223 (2007).
  63. C.O. Webb, M. J. Donoghue, Molecular Ecology Notes 5, 181–183 (2005).
  64. C.O. Webb, D. D. Ackerly, M. A. McPeek, M. J. Donoghue, Annual Review of Ecology and Systematics 33, 475–505 (2002).
  65. E. Weiher, P. Keddy, Ecological assembly rules: perspectives, advances, retreats. (Cambridge University Press, Cambridge, UK, 1999)
  66. S.J. Wright, Oecologia 130, 1–14 (2002).
  67. Q.H. Zhang, F. Schlyter, Oikos 101, 299–310 (2003).