Skip to main content
Dryad logo

Geographical variation in the trait-based assembly patterns of multitrophic invertebrate communities

Citation

Srivastava, Diane S. et al. (2022), Geographical variation in the trait-based assembly patterns of multitrophic invertebrate communities, Dryad, Dataset, https://doi.org/10.5061/dryad.vt4b8gtv4

Abstract

It has been argued that the mechanisms structuring ecological communities may be more generalizable when based on traits than on species identities. If so, patterns in the assembly of community-level traits along environmental gradients should be similar in different places in the world. Alternatively, geographic change in the species pool and regional variation in climate might result in site-specific relationships between community traits and local environments. These competing hypotheses are particularly untested for animal communities. Here we test the geographic constancy of trait-based assembly patterns using a widespread multi-trophic community: aquatic macroinvertebrates within bromeliads. We used data on 615 invertebrate taxa from 1656 bromeliads in 26 field sites from Mexico to Argentina. We summarized invertebrate traits with four orthogonal axes, and used these trait axes to examine trait convergence and divergence assembly patterns along three environmental gradients: detrital biomass and water volume in bromeliads, and canopy cover over bromeliads. We found no overall signal of trait-based assembly patterns along any of the environmental gradients. However, individual sites did show trait convergence along detrital and water gradients, and we built predictive models to explore these site differences. Sites that showed trait convergence along detrital gradients were all north of the Northern Andes. This geographic pattern may be related to phylogeographic differences in bromeliad morphology. Bromeliads with low detritus were dominated by detritivorous collectors and filter feeders, where those with high detritus had more sclerotized and predatory invertebrates. Sites that showed the strongest trait convergence along gradients in bromeliad water were in regions with seasonal precipitation. In such sites, bromeliads with low water were dominated by soft-bodied, benthic invertebrates with simple life cycles. In less seasonal sites, traits associated with short-term desiccation resistance, such as hard exoskeletons, were more important. In summary, we show that there are strong geographic effects on the trait-based assembly patterns of this invertebrate community, driven by the biogeography of their foundational plant species as well as by regional climate. We suggest that inclusion of biogeography and climate in trait-based community ecology could help make it a truly general theory. (excerpted from Srivastava, DS et al. 2022. Geographical variation in the trait-based assembly patterns of multitrophic invertebrate communities. Functional Ecology)

Methods

Bromeliad and macroinvertebrate data

We compiled data on the aquatic macroinvertebrates in tank bromeliads previously sampled in 26 different sites throughout the natural distribution of the tank bromeliads (Bromeliaceae family). Field sites were distinct from each other in space, elevation and the species composition of invertebrate communities. For every bromeliad, all water and detritus contained in the plant were removed, either by dissecting the plant or by pipetting. The water and detritus were examined for aquatic macroinvertebrates in small size-fractioned aliquots in white trays. Macroinvertebrates were identified to morphospecies in the field, and subsequently to the lowest possible taxonomic level. The detritus was oven-dried and weighed to determine dry mass. Bromeliads were sampled across a range of habitats, from exposed restinga (coastal sand-based shrub habitat) in Brazil to cloud forests on Caribbean mountaintops to rainforests in Central America. As no bromeliad genus was found in all field sites, we sampled the most common genera in sites. Our analysis included three environmental characteristics of each sampled bromeliad: (1) the dry mass of all detritus in the bromeliad (hereafter “detritus”, measured in grams); (2) the volume of standing water in each bromeliad on the day of sampling (hereafter “water”, measured in mL); (3) the openness of the canopy above the bromeliad (hereafter “canopy”, a binary variable with 1 = open canopy and 0 = closed canopy). These variables were chosen because previous site-specific research had established that they were important environmental drivers of community structure and function.

Trait data

Our analysis considers interspecific, but not intraspecific, differences in traits. In Céréghino et al. (2018), each bromeliad invertebrate morphospecies was scored in terms of twelve traits: aquatic developmental stage, body form, maximum body size, cohort production interval, dispersal mode, food, feeding group, locomotion, morphological defence, reproduction mode, resistance forms and respiration mode.Each trait was represented by several modalities or categories (e.g. the modalities for the trait “dispersal mode” were passive and active), and the affinity of the taxa for each modality was fuzzy coded. In total, the 12 traits were represented by 64 modalities.  Céréghino et al. (2018) reduced these 64 trait modalities to four orthogonal axes using PCA (original data:https://knb.ecoinformatics.org/view/doi:10.5063/F1VD6WMF). Since then, there have been modest updates to the trait data: some trait scores were improved, the number of missing values reduced, a few microscopic or terrestrial species were removed, and the taxonomic resolution of some species identifications was improved. We therefore reran the PCA analysis on the updated trait matrix, using the same R script as in Céréghino et al. (2018), and used the first four axes in our current study.

Site information

We collated information on biogeographic, bioclimatic, and sampling characteristics of each field site in order to better contextualize differences among sites. Biogeographic characteristics included position north and north west versus south and southeast of the Northern Andes (simplified hereafter as north versus south of Andes). The Northern Andes are known to be a dispersal barrier for both bromeliads  and bromeliad invertebrates. Although the Southern and Central Andes could be a potential barrier between the west coast and center of South America, we have no bromeliad data from the former and so do not analyse these mountain ranges. We examined species pool richness, estimated with Chao’s method (implemented in the vegan R package), to test if richer sites had stronger trait-environment matching. From the WorldClim database, we extracted site-specific estimates of four bioclimatic variables that a previous study (Guzman et al., 2020) found to underlie spatial variation in bromeliad macroinvertebrate traits: mean diurnal range in temperature (BC2), temperature annual seasonality (BC4), precipitation annual seasonality (BC15) and precipitation of the driest quarter (BC17). WorldClim data was extracted at the 1km2 scale; when field sites exceeded 1 km2 in size we averaged data over the relevant 1 km2 pixels. Finally, we examined sampling characteristics of each site, including the number of bromeliads sampled (which affects the power of tests) and the site mean of the focal environmental gradient (in case trait-environment relationships differ with site position on gradient).

Data processing

We organized the abundance data using the R package fwdata (developed by A.A.M.M, https://github.com/SrivastavaLab/fwdata). We then converted abundances to biomass by multiplying abundance by the estimated per capita biomass of each morphospecies, based on taxon-specific allometric relationships, using the hellometry R package (provided courtesy of P. Rogy, https://github.com/pierrerogy/hellometry).

Summary

In sum, the dataset consists of information on environmental attributes of bromeliads (n = 1656), macroinvertebrate morphospecies (n = 615), macroinvertebrate traits (n = 64), local environmental gradients (n = 3) and field sites (n = 26).

(text excerpted from Srivastava, DS et al. 2022. Geographical variation in the trait-based assembly patterns of multitrophic invertebrate communities. Functional Ecology)

Usage Notes

These data, in combination with archived R scripts, allow reproducibility of the results reported in the 2022 Functional Ecology ms "Geographical variation in the trait-based assembly patterns of multitrophic invertebrate communities" by DS Srivastava and co-authors. A README.TXT file is included with full metadata for all data files. The data files include input files for the R Scripts (visits_owners.csv, bromeliads_syncsa_ready.csv, biomass_syncsa_ready.csv, traits_pca_ready.csv, site_covariates.csv) as well as three output files (RES_pca_individuals_0.7.7_ranks_cleaned.txt, allout_lastrun.csv, allaxes1234_lastrun.csv). We include the output files linked to the published results for full reproducibility as permutation-based results can change slightly between runs.

Although this data is being made publicly available, we do ask that those intending to reuse the data for publication purposes to please first contact the data collectors (see "visits_owners.csv" file for names and emails of data collectors). We can provide you with valuable context and system-specific knowledge that will help you make best use of the data. We also point out that many data collectors are from the Global South and that there is a history in ecology of neocolonial practices regarding the unacknowledged flow of data from poorer to wealthier countries.

Spatial structure of data

We defined a distinct site as one that was distinct in terms of: space, elevation and taxonomic identity of its species pool. All bromeliads surveyed within a site must occur in an area <3.5 km in linear dimension and within a 300 m range of elevation. The invertebrate species pool should also be >50% dissimilar (Jaccard dissimilarity, based on presence-absence) between adjacent sites. For example, bromeliads surveyed along a transect covering an elevation gradient (Sonadora transect, Puerto Rico) are assigned to a low
elevation (400-700m) and high elevation (750-1000m) site based on differences in elevation and dissimilarity in the invertebrate species.

A dataset is defined as a distinct combination of time and location.  For example, if one habitat in a site was surveyed in two different years, these would be considered distinct datasets with one visit each. As datasets are often defined slightly differently by different data collectors, this is a less consistent spatial unit than sites and so is not the basis of our analysis.

Visits are defined as a survey of bromeliad macroinvertebrates distinct in habitat, time and location. Multiple visits may be nested within a dataset. If two distinct habitats were surveyed at the same location at the same time of year, these would also be considered two visits within a dataset.

Missing values are indicated by NA in all files.

Funding

French Foundation for Research on Biodiversity - Centre for the Synthesis and Analysis of Biodiversity

Agence Nationale de la Recherche, Award: Labex CEBA

BPE-FAPESP, Award: 2018/12225-0

BPE-FAPESP, Award: 2019/08474-86/01209-9

Royal Society, Award: NAF/R2/180791

Conselho Nacional de Desenvolvimento Científico e Tecnológico, Award: 307689/2014-0

Conselho Nacional de Desenvolvimento Científico e Tecnológico, Award: 301514/2017-8

Conselho Nacional de Desenvolvimento Científico e Tecnológico, Award: 312770/2014-6

Universidad Nacional de Rosario, Award: AGR-210

Universidad Nacional de Rosario, Award: AGR-290

Royal Society of Edinburgh

Carnegie Trust for the Universities of Scotland

National Science Foundation, Award: DEB-0218039

National Science Foundation, Award: DEB-0620910

USDA IITF, Award: 01-1G11120101-001

Saba Conservation Foundation

Natural Sciences and Engineering Research Council of Canada, Award: Discovery grants 2002-2022

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Award: 2014/04603-4

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior, Award: 20130877

Conselho Nacional de Desenvolvimento Científico e Tecnológico, Award: Ciências sem Fronteiras 401345/2014-9

Fundação de Amparo à Pesquisa do Estado de São Paulo, Award: 2016/09699-5

Agence Nationale de la Recherche, Award: ANR-10-LABX-25-01