Joint Species Distribution Models Research Articles

Species interactions weakly modify climate‐induced tree co‐occurrence patterns

AbstractAimsSpecies distributions are hypothesized to be underlain by a complex association of processes that span multiple spatial scales including biotic interactions, dispersal limitation, fine‐scale resource gradients and climate. Species disequilibrium with climate may reflect the effects of non‐climatic processes on species distributions, yet distribution models have rarely directly considered non‐climatic processes. Here, we use a Joint Species Distribution Model (JSDM) to investigate the influence of non‐climatic factors on species co‐occurrence patterns and to directly quantify the relative influences of climate and alternative processes that may generate correlated responses in species distributions, such as species interactions, on tree co‐occurrence patterns.LocationUS Rocky Mountains.MethodsWe apply a Bayesian JSDM to simultaneously model the co‐occurrence patterns of ten dominant tree species across the Rocky Mountains, and evaluate climatic and residual correlations from the fitted model to determine the relative contribution of each component to observed co‐occurrence patterns. We also evaluate predictions generated from the fitted model relative to a single‐species modelling approach.ResultsFor most species, correlation due to climate covariates exceeded residual correlation, indicating an overriding influence of broad‐scale climate on co‐occurrence patterns. Accounting for covariance among species did not significantly improve predictions relative to a single‐species approach, providing limited evidence for a strong independent influence of species interactions on distribution patterns.ConclusionsOverall, our findings indicate that climate is an important driver of regional biodiversity patterns and that interactions between dominant tree species contribute little to explain species co‐occurrence patterns among Rocky Mountain trees.

Disentangling biotic interactions, environmental filters, and dispersal limitation as drivers of species co‐occurrence

A key focus in ecology is to search for community assembly rules. Here we compare two community modelling frameworks that integrate a combination of environmental and spatial data to identify positive and negative species associations from presence–absence matrices, and incorporate an additional comparison using joint species distribution models (JSDM).The frameworks use a dichotomous logic tree that distinguishes dispersal limitation, environmental requirements, and interspecific interactions as causes of segregated or aggregated species pairs. The first framework is based on a classical null model analysis complemented by tests of spatial arrangement and environmental characteristics of the sites occupied by the members of each species pair (Classic framework). The second framework, (SDM framework) implemented here for the first time, builds on the application of environmentally‐constrained null models (or JSDMs) to partial out the influence of the environment, and includes an analysis of the geographical configuration of species ranges to account for dispersal effects.We applied these approaches to examine plot‐level species co‐occurrence in plant communities sampled along a wide elevation gradient in the Swiss Alps. According to the frameworks, the majority of species pairs were randomly associated, and most of the non‐random positive and negative species associations could be attributed to environmental filtering and/or dispersal limitation. These patterns were partly detected also with JSDM. Biotic interactions were detected more frequently in the SDM framework, and by JSDM, than in the Classic framework. All approaches detected species aggregation more often than segregation, perhaps reflecting the important role of facilitation in stressful high‐elevation environments.Differences between the frameworks may reflect the explicit incorporation of elevational segregation in the SDM framework and the sensitivity of JSDM to the environmental data. Nevertheless, all methods have the potential to reveal general patterns of species co‐occurrence for different taxa, spatial scales, and environmental conditions.

Joint Species Distribution Modeling: Dimension Reduction Using Dirichlet Processes

Species distribution models are used to evaluate the variables that affect the distribution and abundance of species and to predict biodiversity. Historically, such models have been fitted to each species independently. While independent models can provide useful information regarding distribution and abundance, they ignore the fact that, after accounting for environmental covariates, residual interspecies dependence persists. With stacking of individual models, misleading behaviors, may arise. In particular, individual models often imply too many species per location. Recently developed joint species distribution models have application to presence–absence, continuous or discrete abundance, abundance with large numbers of zeros, and discrete, ordinal, and compositional data. Here, we deal with the challenge of joint modeling for a large number of species. To appreciate the challenge in the simplest way, with just presence/absence (binary) response and say, S species, we have an S-way contingency table with 2S cell probabilities. Even if S is as small as 100 this is an enormous table, infeasible to work with without some structure to reduce dimension. We develop a computationally feasible approach to accommodate a large number of species (say order 103) that allows us to: 1) assess the dependence structure across species; 2) identify clusters of species that have similar dependence patterns; and 3) jointly predict species distributions. To do so, we build hierarchical models capturing dependence between species at the first or “data” stage rather than at a second or “mean” stage. We employ the Dirichlet process for clustering in a novel way to reduce dimension in the joint covariance structure. This last step makes computation tractable. We use Forest Inventory Analysis (FIA) data in the eastern region of the United States to demonstrate our method. It consists of presence–absence measurements for 112 tree species, observed east of the Mississippi. As a proof of concept for our dimension reduction approach, we also include simulations using continuous and binary data.

Joint species distribution modelling for spatio‐temporal occurrence and ordinal abundance data

AbstractAimSpecies distribution models are important tools used to study the distribution and abundance of organisms relative to abiotic variables. Dynamic local interactions among species in a community can affect abundance. The abundance of a single species may not be at equilibrium with the environment for spreading invasive species and species that are range shifting because of climate change.Innovation: We develop methods for incorporating temporal processes into a spatial joint species distribution model for presence/absence and ordinal abundance data. We model non‐equilibrium conditions via a temporal random effect and temporal dynamics with a vector‐autoregressive process allowing for intra‐ and interspecific dependence between co‐occurring species. The autoregressive term captures how the abundance of each species can enhance or inhibit its own subsequent abundance or the subsequent abundance of other species in the community and is well suited for a ‘community modules’ approach of strongly interacting species within a food web. R code is provided for fitting multispecies models within a Bayesian framework for ordinal data with any number of locations, time points, covariates and ordinal categories.Main conclusionsWe model ordinal abundance data of two invasive insects (hemlock woolly adelgid and elongate hemlock scale) that share a host tree and were undergoing northwards range expansion in the eastern U.S.A. during the period 1997–2011. Accounting for range expansion and high inter‐annual variability in abundance led to improved estimation of the species–environment relationships. We would have erroneously concluded that winter temperatures did not affect scale abundance had we not accounted for the range expansion of scale. The autoregressive component revealed weak evidence for commensalism, in which adelgid may have predisposed hemlock stands for subsequent infestation by scale. Residual spatial dependence indicated that an unmeasured variable additionally affected scale abundance. Our robust modelling approach could provide similar insights for other community modules of co‐occurring species.

Trait‐dependent distributional shifts in fruiting of common British fungi

Despite the dramatic phenological responses of fungal fruiting to recent climate warming, it is unknown whether spatial distributions of fungi have changed and to what extent such changes are influenced by fungal traits, such as ectomycorrhizal (ECM) or saprotrophic lifestyles, spore characteristics, or fruit body size.Our overall aim was to understand how climate and fungal traits determine whether and how species‐specific fungal fruit body abundances have shifted across latitudes over time, using the UK national database of fruiting records. The data employed were recorded over 45 yr (1970–2014), and include 853 278 records of Agaricales, Boletales and Russulales, though we focus only on the most common species (with more than 3000 records each). The georeferenced observations were analysed by a Bayesian inference as a Gaussian additive model with a specification following a joint species distribution model. We used an offset, random contributions and fixed effects to isolate different potential biases from the trait‐specific interactions with latitude/climate and time. Our main aim was assessed by examination of the three‐way‐interaction of trait, predictor (latitude or climate) and time.The results show a strong trait‐specific shift in latitudinal abundance through time, as ECM species have become more abundant relative to saprotrophic species in the north. Along precipitation gradients, phenology was important, in that species with shorter fruiting seasons have declined markedly in abundance in oceanic regions, whereas species with longer seasons have become relatively more common overall. These changes in fruit body distributions are correlated with temperature and rainfall, which act directly on both saprotrophic and ECM fungi, and also indirectly on ECM fungi, through altered photosynthate allocation from their hosts. If these distributional changes reflect fungal activity, there will be important consequences for the responses of forest ecosystems to changing climate, through effects on primary production and nutrient cycling.

Characterizing biotic interactions within the Order Lagomorpha using Joint Species Distribution Models at 3 different spatial scales

Characterizing biotic interactions within the Order Lagomorpha using Joint Species Distribution Models at 3 different spatial scales

Enhanced effects of biotic interactions on predicting multispecies spatial distribution of submerged macrophytes after eutrophication.

Water eutrophication creates unfavorable environmental conditions for submerged macrophytes. In these situations, biotic interactions may be particularly important for explaining and predicting the submerged macrophytes occurrence. Here, we evaluate the roles of biotic interactions in predicting spatial occurrence of submerged macrophytes in 1959 and 2009 for Dianshan Lake in eastern China, which became eutrophic since the 1980s. For the four common species occurred in 1959 and 2009, null species distribution models based on abiotic variables and full models based on both abiotic and biotic variables were developed using generalized linear model (GLM) and boosted regression trees (BRT) to determine whether the biotic variables improved the model performance. Hierarchical Bayesian‐based joint species distribution models capable of detecting paired biotic interactions were established for each species in both periods to evaluate the changes in the biotic interactions. In most of the GLM and BRT models, the full models showed better performance than the null models in predicting the species presence/absence, and the relative importance of the biotic variables in the full models increased from less than 50% in 1959 to more than 50% in 2009 for each species. Moreover, co‐occurrence correlation of each paired species interaction was higher in 2009 than that in 1959. The findings suggest biotic interactions that tend to be positive play more important roles in the spatial distribution of multispecies assemblages of macrophytes and should be included in prediction models to improve prediction accuracy when forecasting macrophytes’ distribution under eutrophication stress.

Design matters: An evaluation of the impact of small man-made forest clearings on tropical bats using a before-after-control-impact design

In recent years, large clearings (>1000ha) accounted for gradually smaller amounts of total annual deforestation in the Brazilian Amazon, whereas the proportion of small clearings (<50ha) nowadays represents more than 80% of annual deforestation. Despite the ubiquity of small clearings in fragmented Amazonian landscapes, most fragmentation research has focused on the effects of large-scale deforestation, leading to a poor understanding of the impacts of smaller barriers on Amazonian vertebrates. We capitalized on the periodical re-isolation of experimental forest fragments at the Biological Dynamics of Forest Fragments Project in the Central Amazon as a before-after-control-impact experiment to investigate the short-term effects of small clearings on bat assemblages. Over the course of three years we sampled six control sites in continuous forest, the interiors and edges of eight forest fragments as well as eight sites in the surrounding matrix. Sampling took place both before and after the experimental manipulation (clearing of a 100m wide strip of regrowth around each fragment), resulting in ∼4000 bat captures. Species were classified as old-growth specialists and habitat generalists according to their habitat affinities and a joint species distribution modeling framework was used to investigate the effect of fragment re-isolation on species occupancy. Following fragment re-isolation, species richness declined in all habitats other than fragment edges and, although responses were idiosyncratic, this decline was more pronounced for forest specialist than for generalist species. Additionally, fragment re-isolation led to a reduction in the similarity between assemblages in modified habitats (fragment interiors, edges and matrix) and continuous forest. Sampling of controls in continuous forest both prior to and after re-isolation revealed that much of the variation in bat species occupancy between sampling periods did not arise from fragment re-isolation but rather reflected natural spatiotemporal variability. This emphasizes the need to sample experimental controls both before and after experimental manipulation and suggests caution in the interpretation of results from studies in which the effects of habitat transformations are assessed based solely on data collected using space-for-time substitution approaches.

Parasite-microbiota interactions potentially affect intestinal communities in wild mammals.

Detecting interaction between species is notoriously difficult, and disentangling species associations in host-related gut communities is especially challenging. Nevertheless, due to contemporary methods, including metabarcoding and 16S sequencing, collecting observational data on community composition has become easier and much more common. We studied the previously collected datasets of intestinal bacterial microbiota and parasite compositions within longitudinally followed mouse lemurs by analysing the potential interactions with diversity metrics and novel joint species distribution modelling. Both methods showed statistical association between certain parasite species and bacterial microbiota composition. Unicellular Eimeria sp. had an effect on diversity of gut microbiota. The cestode Hymenolepis diminuta had negative associations with several bacterial orders, whereas closely related species Hymenolepis nana had positive associations with several bacterial orders. Our results reveal potential interactions between some, but not all, intestinal parasites and gut bacterial microbiota. Host variables contributed over half of the total variation explained with the model, and sex was the most important single host variable; especially with microbiota, there were sex-related differences in the community composition. This study shows how joint species distribution modelling can incorporate both within-host dynamics of several taxa and host characteristics to model potential interactions in intestinal community. These results provide new hypothesis for interactions between and among parasites and bacterial microbiota to be tested further with experimental studies.

Using joint species distribution models for evaluating how species‐to‐species associations depend on the environmental context

Summary Joint species distribution models (JSDM) are increasingly used to analyse community ecology data. Recent progress with JSDMs has provided ecologists with new tools for estimating species associations (residual co‐occurrence patterns after accounting for environmental niches) from large data sets, as well as for increasing the predictive power of species distribution models (SDMs) by accounting for such associations. Yet, one critical limitation of JSDMs developed thus far is that they assume constant species associations. However, in real ecological communities, the direction and strength of interspecific interactions are likely to be different under different environmental conditions. In this paper, we overcome the shortcoming of present JSDMs by allowing species associations covary with measured environmental covariates. To estimate environmental‐dependent species associations, we utilize a latent variable structure, where the factor loadings are modelled as a linear regression to environmental covariates. We illustrate the performance of the statistical framework with both simulated and real data. Our results show that JSDMs perform substantially better in inferring environmental‐dependent species associations than single SDMs, especially with sparse data. Furthermore, JSDMs consistently overperform SDMs in terms of predictive power for generating predictions that account for environment‐dependent biotic associations. We implemented the statistical framework as a MATLAB package, which includes tools both for model parameterization as well as for post‐processing of results, particularly for addressing whether and how species associations depend on the environmental conditions. Our statistical framework provides a new tool for ecologists who wish to investigate from non‐manipulative observational community data the dependency of interspecific interactions on environmental context. Our method can be applied to answer the fundamental questions in community ecology about how species’ interactions shift in changing environmental conditions, as well as to predict future changes of species’ interactions in response to global change.

How to make more out of community data? A conceptual framework and its implementation as models and software.

Community ecology aims to understand what factors determine the assembly and dynamics of species assemblages at different spatiotemporal scales. To facilitate the integration between conceptual and statistical approaches in community ecology, we propose Hierarchical Modelling of Species Communities (HMSC) as a general, flexible framework for modern analysis of community data. While non-manipulative data allow for only correlative and not causal inference, this framework facilitates the formulation of data-driven hypotheses regarding the processes that structure communities. We model environmental filtering by variation and covariation in the responses of individual species to the characteristics of their environment, with potential contingencies on species traits and phylogenetic relationships. We capture biotic assembly rules by species-to-species association matrices, which may be estimated at multiple spatial or temporal scales. We operationalise the HMSC framework as a hierarchical Bayesian joint species distribution model, and implement it as R- and Matlab-packages which enable computationally efficient analyses of large data sets. Armed with this tool, community ecologists can make sense of many types of data, including spatially explicit data and time-series data. We illustrate the use of this framework through a series of diverse ecological examples.

Species partitioning in a temperate mountain chain: Segregation by habitat vs. interspecific competition

Disentangling the relative influence of the environment and biotic interactions in determining species coexistence patterns is a major challenge in ecology. The zonation occurring along elevation gradients, or at bioclimatic contact zones, offers a good opportunity to improve such understanding because the small scale at which the partitioning occurs facilitates inference based on experiments and ecological modelling. We studied the influence of abiotic gradients, habitat types, and interspecific competition in determining the spatial turnover between two pipit and two bunting species in NW Spain. We explored two independent lines of evidence to draw inference about the relative importance of environment and biotic interactions in driving range partitioning along elevation, latitude, and longitude. We combined occurrence data with environmental data to develop joint species distribution models (JSDM), in order to attribute co‐occurrence (or exclusion) to shared (or divergent) environmental responses and to interactions (attraction or exclusion). In the same region, we tested for interference competition by means of playback experiments in the contact zone. The JSDMs highlighted different responses for the two species pairs, although we did not find direct evidence of interspecific aggressiveness in our playback experiments. In pipits, partitioning was explained by divergent climate and habitat requirements and also by the negative correlations between species not explained by the environment. This significant residual correlation may reflect forms of competition others than direct interference, although we could not completely exclude the influence of unmeasured environmental predictors. When bunting species co‐occurred, it was because of shared habitat preferences, and a possible limitation to dispersal might cause their partitioning. Our results indicate that no single mechanism dominates in driving the distribution of our study species, but rather distributions are determined by the combination of many small forces including biotic and abiotic determinants of niche, whose relative strengths varied among species.

Modeling joint abundance of multiple species using Dirichlet process mixtures

We present a method for modeling the distributions of multiple species simultaneously using Dirichlet process random effects to cluster species into guilds. Guilds are ecological groups of species that behave or react similarly to some environmental conditions. By modeling latent guild structure, we capture the cross‐correlations in abundance or occurrence of species over surveys. In addition, ecological information about the community structure is obtained as a by‐product of the model. By clustering species into similar functional groups, prediction uncertainty of community structure at additional sites is reduced over treating each species separately. The proposed model also presents an improvement over previously proposed joint species distribution models by reducing the number of parameters necessary to capture interspecies correlations and eliminating the need to have a priori information on the number of groups or a distance metric over species traits. The method is illustrated with a small simulation demonstration, as well as an analysis of a mesopelagic fish survey from the eastern Bering Sea near Alaska. The simulation data analysis shows that guild membership can be extracted as the differences between groups become larger and if guild differences are small, the model naturally collapses all the species into a small number of guilds, which increases predictive efficiency by reducing the number of parameters to that which is supported by the data.

Measuring and predicting the influence of traits on the assembly processes of wood‐inhabiting fungi

Summary The identification of traits that influence the responses of the species to environmental variation provides a mechanistic perspective on the assembly processes of ecological communities. While much research linking functional ecology with assembly processes has been conducted with animals and plants, the development of predictive or even conceptual frameworks for fungal functional community ecology remains poorly explored. Particularly, little is known about the contribution of traits to the occurrences of fungal species under different environmental conditions. Wood‐inhabiting fungi are known to strongly respond to habitat disturbance, and thus provide an interesting case study for investigating to what extent variation in occurrence patterns of fungi can be related to traits. We apply a trait‐based joint species distribution model to a data set consisting of fruit‐body occurrence data on 321 wood‐inhabiting fungal species collected in 22 460 dead wood units from managed and natural forest sites. Our results show that environmental filtering plays a big role on shaping wood‐inhabiting fungal communities, as different environments held different communities in terms of species and trait compositions. Most importantly, forest management selected against species with large and long‐lived fruit‐bodies as well as late decayers, and promoted the occurrences of species with small fruit‐bodies and early decayers. A strong phylogenetic signal in the data suggested the existence of also some other functionally important traits than the ones we considered. We found that those species groups that were more prevalent in natural conditions had more associations to other species than species groups that were tolerant to or benefitted from forest management. Therefore, the changes that forest management causes on wood‐inhabiting fungal communities influence ecosystem functioning through simplification of interactive associations among the fungal species. Synthesis. Our results show that functional traits are linked to the responses of wood‐inhabiting fungi to variation in their environment, and thus environmental changes alter ecosystem functions via promoting or reducing species with different fruit‐body types. However, further research is needed to identify other functional traits and to provide conclusive evidence for the adaptive nature of the links from traits to occurrence patterns found here.

Non‐stationarity in the co‐occurrence patterns of species across environmental gradients

Summary Quantifying the role of biotic interactions in driving community assembly often relies on analysing species co‐occurrence patterns, where segregated patterns are taken to indicate antagonistic interactions such as competition. It is unknown, however, if co‐occurrence patterns are stationary across environmental gradients, as it is possible that the strength of biotic interactions that drive these patterns also depends on the environment. In this study, we aim to understand how patterns of co‐occurrence change when species move from their environmental range centre towards their range periphery while isolating the potential signal of biotic interactions from confounding factors. We used two separate statistical approaches (null models and joint species distribution models) to quantify pairwise co‐occurrence patterns for tree species sampled in 9382 plots distributed across the conterminous US. We also analysed co‐occurrence patterns that emerged from a simple meta‐community model. We then assessed how patterns of species segregation and aggregation change in relation to habitat suitability while accounting for multiple factors known to confound co‐occurrence analyses. We found strong non‐stationarity in co‐occurrences, with patterns shifting from segregated at the environmental range centre towards aggregated at range margins for the majority of tree species. Patterns were in full agreement between model simulations and both empirical analyses. Model simulations suggest that this pattern is at least partly driven by variation in the relative abundances of non‐focal species even when no direct biological interactions are present. Synthesis. Patterns of tree species co‐occurrence vary across environmental gradients, with increased segregation when environmental conditions are optimal and increased aggregation when the environment is less suitable. This pattern may originate from a trade‐off between the abundance of the focal species pair, which decreases towards the environmental range margin, and the increasing abundance of non‐focal species to which the environment is more suitable. The strong dependence of co‐occurrence patterns on environmental conditions might limit the predictive ability of joint species distribution models, which couple species co‐occurrences and their environmental responses, because co‐occurrence patterns and environmental responses are confounded.

Joint species models reveal the effects of environment on community assemblage of freshwater mussels and fishes in European rivers

AbstractAimGiven that riverine systems exhibit longitudinal environmental gradients from headwater to the mouth of a river, habitat heterogeneity appears to be a major driver of spatial variation in community composition among riverine localities. As freshwater ecosystems are amongst the most endangered ecosystems in the world, community‐based conservation and multiple‐species management are necessary to maintain ecosystem integrity. We used joint species distribution models (JSDMs) to investigate the relative importance of abiotic and biotic factors that are responsible for the distribution and co‐occurrence of species in riverine ecosystems.LocationCentral and northern Europe.MethodsWe examined the general patterns of species assemblage of two endangered freshwater mussel species (Margaritifera margaritifera and Unio crassus) and their associated fish communities. We examined the patterns of positive or negative co‐occurrence in mussel and fish species and identified shared abiotic responses between mussel–host pairs.ResultsWe found that the relative importance of abiotic and residual factors and patterns of significant species correlations varied among taxa: significant residual correlations were prevalent among fish species, whereas mussel occurrences were exclusively explained by abiotic factors. Mussels and their fish hosts generally had shared abiotic responses with some mismatched responses between mussel–host pairs.Main conclusionsGiven that the composition of communities were tightly linked with abiotic factors and residual correlations, the results have significant implications for the conservation and restoration of aquatic communities. This study highlights the necessity to simultaneously consider environmental factors and species co‐occurrences in the modelling of species distributions and assemblages of riverine communities. Such a holistic community conservation approach can reveal ecological similarities and differences among species, which can help us avoid conflicts among target‐species conservation plans.

Wood‐inhabiting fungi with tight associations with other species have declined as a response to forest management

Research on mutualistic and antagonistic networks, such as plant–pollinator and host–parasite networks, has shown that species interactions can influence and be influenced by the responses of species to environmental perturbations. Here we examine whether results obtained for directly observable networks generalize to more complex networks in which species interactions cannot be observed directly. As a case study, we consider data on the occurrences of 98 wood‐inhabiting fungal species in managed and natural forests. We specifically ask if and how much the positions of wood‐inhabiting fungal species within the interaction networks influence their responses to forest management. For this, we utilize a joint species distribution model that partitions variation in species occurrences among environmental (i.e. resource availability) and biotic (i.e. species‐to‐species associations) predictors. Our results indicate that in addition to the direct loss of resource‐specialised species, forest management has indirect effects mediated through interactive associations. In particular, species with strong associative links to other species are especially sensitive to forest management.

The influence of climate on species distribution over time and space during the late Quaternary

Understanding the effect of climate on the composition of communities and its change over time and space is one of the major aims in ecology and paleoecology. Herein, we tackled on this issue by studying late Quaternary large mammal paleocommunities of Eurasia. The late Quaternary was a period of strong environmental instability, especially characterized by the occurrence of the last glacial maximum (LGM). We used community phylogenetics and joint species distribution models in order to understand the factors determining paleocommunity composition in the late Quaternary. Our results support the existence of strong climatic selection operating on the LGM fauna, both through the disappearance of warm-adapted species such as Elephas antiquus, Hippopothamus amphibious, and Stephanorhinus hemitoechus, and by setting the stage for the existence of a community characterized by cold-adapted large mammals. Patterns of abundance in the fossil record, co-occurrence between species pairs, and the extent of climatic forcing on faunal composition, differ between paleocommunities, but not between extinct and extant species, which is consistent with the idea that climate change, rather than the presence of humans, exerted a major effect on the survival of the late Quaternary megafauna.

Joint dynamic species distribution models: a tool for community ordination and spatio‐temporal monitoring

AbstractAimSpatial analysis of the distribution and density of species is of continuing interest within theoretical and applied ecology. Species distribution models (SDMs) are being increasingly used to analyse count, presence–absence and presence‐only data sets. There is a growing literature on dynamic SDMs (which incorporate temporal variation in species distribution), joint SDMs (which simultaneously analyse the correlated distribution of multiple species) and geostatistical models (which account for similarity between nearby sites caused by unobserved covariates). However, no previous study has combined all three attributes within a single framework.InnovationWe develop spatial dynamic factor analysis for use as a ‘joint, dynamic SDM’ (JDSDM), which uses geostatistical methods to account for spatial similarity when estimating one or more ‘factors’. Each factor evolves over time following a density‐dependent (Gompertz) process, and the log‐density of each species is approximated as a linear combination of different factors. We demonstrate a JDSDM using two multispecies case studies (an annual survey of bottom‐associated species in the Bering Sea and a seasonal survey of butterfly density in the continental USA), and also provide our code publicly as an R package.Main conclusionsCase study applications show that that JDSDMs can be used for species ordination, i.e. showing that dynamics for butterfly species within the same genus are significantly more correlated than for species from different genera. We also demonstrate how JDSDMs can rapidly identify dominant patterns in community dynamics, including the decline and recovery of several Bering Sea fishes since 2008, and the ‘flight curves’ typical of early or late‐emerging butterflies. We conclude by suggesting future research that could incorporate phylogenetic relatedness or functional similarity, and propose that our approach could be used to monitor community dynamics at large spatial and temporal scales.

Uncovering hidden spatial structure in species communities with spatially explicit joint species distribution models

Summary Modern species distribution models account for spatial autocorrelation in order to obtain unbiased statistical inference on the effects of covariates, to improve the model's predictive ability through spatial interpolation and to gain insight in the spatial processes shaping the data. Somewhat analogously, hierarchical approaches to community‐level data have been developed to gain insights into community‐level processes and to improve species‐level inference by borrowing information from other species that are either ecologically or phylogenetically related to the focal species. We unify spatial and community‐level structures by developing spatially explicit joint species distribution models. The models utilize spatially structured latent factors to model missing covariates as well as species‐to‐species associations in a statistically and computationally effective manner. We illustrate that the inclusion of the spatial latent factors greatly increases the predictive performance of the modelling approach with a case study of 55 species of butterfly recorded on a 10 km × 10 km grid in Great Britain consisting of 2609 grid cells.