Florian Hartig

← Publications

Correlation and process in species distribution models: bridging a dichotomy

Carsten F. Dormann, Stanislaus J. Schymanski, Juliano Cabral, Isabelle Chuine, Catherine Graham, Florian Hartig, Michael Kearney, Xavier Morin, Christine Römermann, Boris Schröder & Alexander Singer

Journal of Biogeography, 39(12), 2119–2131 (2012)
Cite this
@article{dormann2012correlation,
  author = {Carsten F. Dormann and Stanislaus J. Schymanski and Juliano Cabral and Isabelle Chuine and Catherine Graham and Florian Hartig and Michael Kearney and Xavier Morin and Christine Römermann and Boris Schröder and Alexander Singer},
  title = {Correlation and process in species distribution models: bridging a dichotomy},
  journal = {Journal of Biogeography},
  volume = {39},
  number = {12},
  pages = {2119--2131},
  year = {2012},
  doi = {10.1111/j.1365-2699.2011.02659.x},
}

DOI: 10.1111/j.1365-2699.2011.02659.x
Citation statistics: 916 (Google Scholar) · 749 (OpenAlex), as of 29 October 2026

Google Scholar citations per year, as of 19 October 2026 (yearly counts are fetched separately from the total)

Funding

This study was supported through funding by LOEWE-BiK-F Biodiversity and Climate Research Centre Frankfurt (funded the initiating workshop); Helmholtz Association (to C.F.D.) (VH-NG 247); German Research Foundation DFG (to C.F.D.) (DO 686/5-1); Max Planck Society (to S.J.S.); DFG (to C.R.) (RO 3842/1-1).

View article (DOI)

Abstract

Within the field of species distribution modelling an apparent dichotomy exists between process‐based and correlative approaches, where the processes are explicit in the former and implicit in the latter. However, these intuitive distinctions can become blurred when comparing species distribution modelling approaches in more detail. In this review article, we contrast the extremes of the correlative–process spectrum of species distribution models with respect to core assumptions, model building and selection strategies, validation, uncertainties, common errors and the questions they are most suited to answer. The extremes of such approaches differ clearly in many aspects, such as model building approaches, parameter estimation strategies and transferability. However, they also share strengths and weaknesses. We show that claims of one approach being intrinsically superior to the other are misguided and that they ignore the process–correlation continuum as well as the domains of questions that each approach is addressing. Nonetheless, the application of process‐based approaches to species distribution modelling lags far behind more correlative (process‐implicit) methods and more research is required to explore their potential benefits. Critical issues for the employment of species distribution modelling approaches are given, together with a guideline for appropriate usage. We close with challenges for future development of process‐explicit species distribution models and how they may complement current approaches to study species distributions.

What the paper shows and why it matters

(AI-generated)

Correlative species distribution models relate occurrence records statistically to environmental data, mainly climate, leaving the processes implicit. The paper notes that their predictions are usually limited in biological realism and transferability. Process-based models, built around explicitly stated mechanisms, can address these deficits, but they often need many parameters and data of limited availability, and so have been used for far fewer species. The two are easily seen as rival camps, one presumed better. The paper questions this assumption of a sharp dichotomy with an intrinsically superior side.

The paper argues that correlative and process-based species distribution models lie on a continuum defined by how explicitly processes are represented, so that claims of one being intrinsically superior are misguided. The review compares the two extremes on thirteen features, among them assumptions, validation and uncertainty. Two results carry the argument. Once process-based models are fitted to distribution data, they share the statistical assumptions and uncertainty problems of correlative models. The data used for fitting cannot also test the model’s structure, and a good fit does not show that a model is right for the right reasons. Only forward process-based models, whose parameters come from independent data or theory, can in principle use distribution data to test their structure. Correlative models, in turn, yield correlations whose causal reading remains a hypothesis: a link with temperature does not necessarily show that temperature limits a species.

The paper concludes that neither camp should claim the moral high ground: the success of a correlative model may rest on spurious correlations, and a process-based model can only reveal effects of the processes put into it. Process-based approaches, it adds, lag far behind correlative ones in use and need more research. It highlights three avenues: Bayesian fitting of process-based models, forward models, and a combined workflow in which correlative models generate hypotheses that process-based models take up and turn into predictions for a test on independent data. Yates et al. (2018) cite the paper for their statement that the correlative–mechanistic distinction is usually blurred in practice.

References

  • Yates et al. (2018). Outstanding Challenges in the Transferability of Ecological Models. Trends in Ecology & Evolution 33, 790–802. doi:10.1016/j.tree.2018.08.001