Portal de Periódicos da CAPES

So Many Variables: Joint Modeling in Community Ecology

2015; Elsevier BV; Volume: 30; Issue: 12 Linguagem: Inglês

10.1016/j.tree.2015.09.007

1872-8383

David I. Warton, F. Guillaume Blanchet, Robert B. O’Hara, Otso Ovaskainen, Sara Taskinen, Steven C. Walker, Francis K. C. Hui,

Soil and Water Nutrient Dynamics

Technological advances have enabled a new class of multivariate models for ecology, with the potential now to specify a statistical model for abundances jointly across many taxa, to simultaneously explore interactions across taxa and the response of abundance to environmental variables. Joint models can be used for several purposes of interest to ecologists, including estimating patterns of residual correlation across taxa, ordination, multivariate inference about environmental effects and environment-by-trait interactions, accounting for missing predictors, and improving predictions in situations where one can leverage knowledge of some species to predict others. We demonstrate this by example and discuss recent computation tools and future directions. Technological advances have enabled a new class of multivariate models for ecology, with the potential now to specify a statistical model for abundances jointly across many taxa, to simultaneously explore interactions across taxa and the response of abundance to environmental variables. Joint models can be used for several purposes of interest to ecologists, including estimating patterns of residual correlation across taxa, ordination, multivariate inference about environmental effects and environment-by-trait interactions, accounting for missing predictors, and improving predictions in situations where one can leverage knowledge of some species to predict others. We demonstrate this by example and discuss recent computation tools and future directions. Many ecological questions require the joint analysis of abundances collected simultaneously across many taxonomic groups, and, if organisms are identified using modern tools such as metabarcoding, their number can be in the thousands. While historically such data have been analyzed using ad hoc algorithms, it is now possible to fully specify joint statistical models for abundance using multivariate extensions of generalized linear mixed models. These modern ‘joint modeling’ approaches allow the study of correlation patterns across taxa, at the same time as studying environmental response, to tease the two apart. Latent variable models are an especially exciting tool that has recently been used for ordination as well as for studying the factors driving co-occurrence. Many ecological questions require the joint analysis of abundances collected simultaneously across many taxonomic groups, and, if organisms are identified using modern tools such as metabarcoding, their number can be in the thousands. While historically such data have been analyzed using ad hoc algorithms, it is now possible to fully specify joint statistical models for abundance using multivariate extensions of generalized linear mixed models. These modern ‘joint modeling’ approaches allow the study of correlation patterns across taxa, at the same time as studying environmental response, to tease the two apart. Latent variable models are an especially exciting tool that has recently been used for ordination as well as for studying the factors driving co-occurrence. the extent to which a type of organism is present in a sample unit, measured either as a count, biomass, % cover, a factor with ordered levels, or presence/absence. a variable that can take any value within some interval (cf. discrete variable). Abundance is rarely continuous, complicating the modeling process. a variable that can take one of a countable number of distinct values. Abundance is often discrete, for example counts could be 0, 1, 2, 3,... a range of values with 95% (posterior) probability for a parameter (a Bayesian version of a 95% confidence interval). a regression model to predict a response variable assuming it comes from a distribution in the exponential family (Poisson, binomial, etc.), and assuming that some known transformation of the mean response is a linear function of predictor variables. a GLM with random effects included: that is, some of the coefficients are assumed to come at random from a larger population of potential values [23Bolker B.M. et al.Generalized linear mixed models: a practical guide for ecology and evolution.Trends Ecol. Evol. 2009; 24: 127-135Abstract Full Text Full Text PDF PubMed Scopus (6043) Google Scholar]. Of particular interest here is where the random effect is multivariate to account for correlation (Box 1). a parametric statistical model for the abundance of multiple taxa (usually species), accounting for correlation between taxa as well as response to predictor variables. a regression model for multivariate data that includes some unobserved (‘latent’) predictors that are usually introduced to model correlation or to account for missing predictors (Box 1). the function in a GLM defining the transformation of the mean to a linear function of predictors (e.g., logit or probit for presence/absence data, log for counts). Its main purpose is to map from the scale of the linear predictor (which can take any real-numbered value) onto the scale on which the mean response is defined (e.g., all positive numbers for abundance). joint analysis of multiple response variables; in particular, a joint analysis for abundance of multiple taxonomic groups. a visualization tool attempting to represent the main structures in multivariate data together with a reduced set of usually two or three axes. a variable used to predict the response of interest. In this paper these are treatments, environmental variables, or functional traits. correlation between response variables that is not explained by predictors in the model. Joint models can estimate this. variable of primary interest in analysis (e.g., for which predictions are required). In this paper these are typically abundances in taxa. the sampling unit at which abundances are measured for all taxa, often a site or transect. Extending Joint Models in Community Ecology: A Response to Beissinger et al.Warton et al.Trends in Ecology & EvolutionAugust 8, 2016In BriefThe joint modelling of many variables in community ecology is a new and technically challenging area with many opportunities for future developments. The possibility of extending joint models to deal with imperfect detection has been highlighted by Beissinger et al. as an important problem worthy of further investigation [1]. We agree, and previously pointed to this potential extension as an outstanding question [2], alongside models that can estimate phylogenetic repulsion or attraction, nonlinearity in the response to latent variables, and spatial or temporal correlation, because further developments in all these directions are needed. Full-Text PDF Incorporating Imperfect Detection into Joint Models of Communities: A response to Warton et al.Beissinger et al.Trends in Ecology & EvolutionAugust 13, 2016In BriefWarton et al. [1] advance community ecology by describing a statistical framework that can jointly model abundances (or distributions) across many taxa to quantify how community properties respond to environmental variables. This framework specifies the effects of both measured and unmeasured (latent) variables on the abundance (or occurrence) of each species. Latent variables are random effects that capture the effects of both missing environmental predictors and correlations in parameter values among different species. Full-Text PDF