This research project is led by Miguel B. Araújo

Setting standards for studies in the interface between biodiversity and climate change


Over the past 100 years, Earth’s climate has become warmer and precipitation regimes have changed. Can we predict the effects of these changes on biodiversity?

My research seek to improve understanding of the key mechanisms governing the distribution of life on Earth, with focus on species as unit of analysis. Basic research in my group tends to feed into the development of models to forecast species distributional dynamics under climate- and land-use change scenarios. To address these questions, I integrate large climate and species distributions databases with descriptions of behavioural and physiological traits of species, molecular phylogenies, and the fossil record.

Most research in my group involves statistical analyses of ecological data, including data mining, bioclimatic modeling, and mathematical simulations, but large-scale experiments, including mesocosm experiments, are now being devised for testing models and theory on species distributions and species coexistence.

The challenge

Researchers typically use bioclimatic envelope models (BEMs) to estimate the relationship between the distributions of species and climate, thus enabling projections of altered species distributions under climate change scenarios. However, models are based on a number of weak ecological assumptions and studies have shown that projections by alternative models can be so variable as to compromise the simplest assessment of whether species distributions should be expected to contract or expand for any given scenario.

A solution for inter-model variability is to fit several models (termed ‘ensembles’) and use appropriate techniques to explore the resulting range of projections. Ensemble forecasting has proved to be successful for characterizing and reducing ‘algorithmic uncertainties’ in models, and the approach is also widely used other branches of science such as climatology. However, unlike climate models, BEMs are not based on well-known bio-geophysical processes but on correlations between species distributions and aspects of climate. Careful choice of variables can make correlative models quasi-mechanistic, but in most cases it is difficult to guarantee that projections are driven by the mechanisms governing the species distributions.

To address ‘ecological uncertainties’ in the models, researchers have proposed the development of process-based models. In contrast with BEMs, process-based models begin with an analysis of the organism rather than its distribution; they determine the mechanistic interaction between an organism’s environment and its growth or fitness, usually based on theoretical inferences, experimental knowledge, or a combination of both. Process-based models have generally focused on some aspect of the physiology, phenology, or population dynamics of the target species, but they have generally failed to account for complex interdependencies between species.

Can uncertainty from biodiversity models be reduced? A current trend involved the development of ever more complex process-based models. Yet, there are ca. 1,9 million known species and many more remain unknown to science. Therefore, methods that require detailed specific information on all species for forecasting overall biodiversity change are not practical. The critical question is ‘What is the minimum level of model complexity that is required for making useful predictions of climate change impacts on biodiversity’.


My research group has pioneered the implementation and testing of ensemble forecasting methodologies in biodiversity sciences. We were the first to independently validate the performance of species distributions models under historical climate change, demonstrating the value of ensemble forecasting for handling and minimizing algorithmic uncertainties of the models.

With colleagues in Australia, Europe and the United States we were also the first to develop coupled-models to examine the effects of climate change on species distributions at population level. We also discovered, or co-discovered, several eco-geographic rules namely that 1) native species richness of several plan and animal groups have a positive relationship with human population density in Europe—a pattern later to be found general across most continents; 2) current native species richness is highly related to historical climate stability since the last glacial maxima, especially for species with limited dispersal abilities; 3) thermal tolerances to heat are conserved (non-adaptive) among terrestrial animals and plants while thermal tolerances to cold are labile (adaptive), and 4) the degree to which the geographical signature of biotic interactions scales-up to coarse resolutions depends on the strength of positive interactions.


The work in my group has changed the species distributions modeling practices commonly used until early 2000, helping set the standards for studies in the interface between biodiversity and climate change. This work has been extensively cited, with ‘google scholar’ counting nearly 20,000 citations since 2002.

The Future

An interdisciplinary research agenda is required to move the field forward. We are now exploring different modelling approaches and different strategies for testing models, using large-scale experimental approaches.

The Team

My group collaborates with researchers from a wide range of international universities and other institutions. Some include:

  • Resit Akçakaya, Stony Brook University
  • Barry Brook, University of Adelaide
  • Neil Burgess, UNEP World Conservation Monitoring Centre
  • Alexandre Diniz-Filho, Universidade Federal de Goiás
  • Damien Fordham, University of Adelaide
  • Antoine Guisan, Université de Lausanne
  • Lee Hannah, Conservation International
  • Pablo Marquet, Pontificia Universidad Católica de Chile
  • Richard Pearson, University College London
  • Townsend Peterson, University of Kansas
  • Drew Purves, Microsoft Research at Cambridge
  • Carsten Rahbek, University of Copenhagen
  • Wilfried Thuiller, Centre National de la Recherche Scientifique at Grenoble
  • Rob Whittaker, University of Oxford
  • Paul Williams, The Natural History Museum in London

Funders and Sponsors

My work has been funded through:

  • Delta Cafés
  • Energias de Portugal (EDP)
  • The BBVA Foundation
  • The Danish Science Foundation
  • The European Commission
  • The Imperial College’s Grand Challenges in Ecosystems and Ecology initiative
  • The International School for Biodiversity Sciences
  • The Portuguese Foundation for Science and Technology (ISOBIS)
  • The Spanish Ministry for Science and Education
  • The Spanish National Research Council (CSIC)