Measuring and explaining the global debt of biotic homogenization in a changing world
The process of biotic homogenization corresponds to the increase in similarity between ecological communities through time. It is one of the most significant aspects of the current biodiversity crisis, and limiting its detrimental effects is gaining momentum in conservation biology. Lagged biodiversity responses to past environmental changes might result in disequilibrium states between community composition and present-day environmental conditions in many biomes and taxa across the planet. My hypothesis is that if delayed biodiversity responses affect the community compositions, they should also affect the similarity between communities. If so, lagged biodiversity responses to global changes might build a homogenization debt, here defined as a future increase of similarity between communities resulting from past environmental perturbation.
The INDEBT project will test the existence, quantify, and map the prevalence of the homogenization debt at a global scale, across biomes and taxa (terrestrial plants, birds, fish, and marine invertebrates). Using graphical, analytical and statistical methods, I will quantify disequilibrium states and dynamics of species turnover between communities, and formally test their link with a set of anthropogenic threats to biodiversity. The project will provide a novel assessment of the biotic homogenization at global scale, and a comprehensive assessment of its drivers. The first quantitative estimation and map of the homogenization debt will reveal areas where biodiversity is doomed to homogenize in the near future if no actions are undertaken. Thus, the project outcome will help to prioritize context-dependent conservation actions to mitigate future homogenization. Through this MSCA fellowship, I will acquire a new set of scientific (statistical development, big-data, new concept in ecology) and transversal (project and financial management) skills that will foster my capacity to reach a academic researcher position in France.
WP 1: Biological data assembly
The lack of reliable data has previously limited investigations of the spatio-temporal dynamics of β-diversity over multi-taxa. This project will use the BioTIME database, an open access global database41 of biodiversity time-series composed of species abundance records sampled through time, with a consistent methodology. BioTIME is the outcome of an ERC-funded project led by Maria Dornelas. The database is still growing and currently includes >8,777,000 records across >547,000 unique locations and >44,400 taxa. The database records span 118 years, with an average duration of 13 years.
I will focus on the four phyla most represented in the dataset: terrestrial plants, birds, fish, and marine invertebrates. For each time-series, quality of data will be checked42,43, and filtered out when not clearly resolved or display inconsistencies, or when time-series are not long enough to encompass ten generations of the phylum (43.5% of time-series are less than 10 years long).
WP 2.1: Estimating β-diversity responses to environment changes
β-diversity is defined as the compositional difference (turnover) between two assemblages, also called pairwise dissimilarity. I will rely on a versatile measure of β-diversity across scales: the distance decay, which quantifies the decrease of similarity with increasing geographic distance between sites (spatial decay, Figure 2a), or with increasing environmental difference (environmental decay, Figure 2c). I will build on the Multifaceted Biodiversity Modelling (MBM) that extends the Generalized Dissimilarity Modelling approach to estimate temporal variations in spatial and environmental decays. MBM models pairwise site dissimilarity in non-linear function of geographic and/or environmental (climate) distance and position while accommodating variation in the rate of
compositional turnover along environmental/spatial scales. Climatic conditions will be extracted from CHELSA database (https://chelsa-climate.org/). I will extend this modelling approach to explicitly incorporate temporal variation of dissimilarity along spatial/environmental distance (Figure 2b and 2c) in order to estimate scale-explicit temporal changes in β-diversity48. The development of temporal MBMs will provide a visual and statistical framework allowing to quantify summary statistics for each taxa, with associated errors and significance. These summary statistics will describe (i) the geographically and taxonomically explicit rate of BH across spatial scales, and (ii) the independent and relative effects of each global change predictors on dissimilarity temporal change.
WP 2 .2: Testing the effect of global change on Biotic Homogenization
The framework developed in WP 2.1 will be separately applied to each dataset/taxa resulting from WP 1 to quantify the relative effect of human-induced global changes on BH. Human-induced global changes information will be extracted from providing the most up-to-date maps of global changes across terrestrial and marine ecosystems at global scale, with a set of 22 anthropogenic drivers of biodiversity change (e.g climate change, human use, human population, pollution, alien species potential) summarized by a set of 11 gridded datasets of “Anthropogenic Threat Complex”. These analyses will provide the first comprehensive test of the additive and synergistic effects of human impact on biotic homogenization across spatial scales.
Figure 2. Basic features for the study of temporal changes in β-diversity environmental decay. (a) The spatial decay is the rate of decrease in similarity (y-axis) with increasing spatial distance (x-axis) between sites. (b) Spatial dependency of Biotic Homogenization (BH) is assessed by scale-specific (e.g 400 km, 800 km) quantification of temporal changes in distance decay (red arrows). (c) Environmental decay is similar as spatial
decay but based on environmental distances (e.g. temperature). (d) Scale-of-effect of temperature change on BH is assessed by comparing scale-specific temporal changes of decays (BH at 6 °C > BH at 2 °C). Preliminary results based on the French Breeding Bird Survey (2000-2016).
WP 3.1: Quantifying disequilibrium states of β-diversity
Recent theoretical advances based on dynamical systems theory propose to study several aspects of community disequilibrium, and empirical application of this approach has successfully shown its ability to quantify disequilibrium responses of any community attribute. I will bridge the so-called “community response diagram” approach with the temporal MBM framework developed in WP 1.1 to study disequilibrium states of β-diversity. This novel framework will quantify site-specific disequilibrium dynamics (Figure 3a) and states (Figure 3b) from the deviation between a temporal, site-specific decay (colored lines Figure 3) and a reference-spatial decay (red lines Figure 3) corresponding to expectation from equilibrium (estimated from spatial values based on all sites).
Figure 3. Quantitative estimations of the homogenization debt of French Breeding Birds (2000-2016). Red lines: spatial-reference thermal decay (expectations from equilibrium, based on temperature difference between sites). Colored lines : site specific temporal thermal decay (based on temperature differences between years). Line color in (a) : disequilibrium dynamics based on correlation between temporal and spatial values (0= disequilibrium, 1=equilibrium). Line color in (b) : disequilibrium states based on difference between temporal and spatial averaged value (-1 = homogenization debt, 1= diversity credit).
WP 3.2: The first global assessment of the homogenization debt across taxa
First, I will apply the framework developed in WP 3.1 to the empirical datasets from WP 1 for each “Anthropogenic Threat Complex” separately. This will quantify driver-specific disequilibrium (Figure 3, for temperature). Second, I will summarize the environmental space in two dimensions using multivariate analyses (e.g Principal Component Analysis) on
multiple environmental variables of interest. I will then assess general-environment disequilibrium based on this environmental space using the same approach. These results will be used to produce the first global quantification and mapping of the homogenization debt across all biomes and taxa present in our data (Figure 1). I will finally use random forest (RF) modeling to estimate the predictive power of each environmental variable on the general disequilibrium value while taking into account their multicollinearity.