Incorporating spatial constraints in different periods of the annual cycle improves species distribution model performance for a highly mobile bird species

Authors

  • Laura Cardador,

    Corresponding author
    1. Departament de Biologia Animal, Institut de Recerca de la Biodiversitat (IRBio), Universitat de Barcelona, Facultat de Biologia, Barcelona, Spain
    2. Forest Sciences Center of Catalonia (CTFC), Solsona, Spain
    Search for more papers by this author
  • Francesc Sardà-Palomera,

    1. Forest Sciences Center of Catalonia (CTFC), Solsona, Spain
    2. Catalan Ornithological Institute (ICO), Museu de Zoologia, Barcelona, Spain
    Search for more papers by this author
  • Martina Carrete,

    1. Department of Physical, Chemical and Natural Systems, Universidad Pablo de Olavide, Sevilla, Spain
    2. Estación Biológica de Doñana (CSIC), Sevilla, Spain
    Search for more papers by this author
  • Santi Mañosa

    1. Departament de Biologia Animal, Institut de Recerca de la Biodiversitat (IRBio), Universitat de Barcelona, Facultat de Biologia, Barcelona, Spain
    Search for more papers by this author

Abstract

Aim

The knowledge of both potential distribution and habitat suitability is fundamental for conservation planning and management of a variety of taxa world-wide. Species distribution models (SDMs) are increasingly applied as predictive tools for these purposes. Such models are based on the concept of ecological niche and assume that species distribute themselves based on niche spaces defined by climate and habitat features. However, this assumption can be violated due to the existence of pure spatial range constraints, a factor rarely accounted for in SDMs, particularly for highly mobile species. We analyse whether pure distance effects, niche-based environmental responses or a combination of both factors can play an important role in limiting the large-scale distribution of highly mobile species.

Location

Spain, southern Europe.

Methods

We modelled the spatial distribution of an expanding raptor species, the marsh harrier Circus aeruginosus, in Spain. We implemented one conventional statistical method (generalized linear model) and one nonparametric technique (maximum entropy, Maxent) using a large dataset on marsh harrier occurrence (n = 1586) in two different periods of the annual cycle, that is breeding and wintering seasons. We developed models that included environmental variables and that either ignored or incorporated spatial constraints using spatial eigenvector mapping (spatial filters).

Results

By comparing model accuracy, we found evidence that the distribution of marsh harriers in Spain was spatially constrained beyond environmental variables and that the effect of spatial constraints varies depending on the period of their annual cycle.

Main conclusions

Contrary to the equilibrium postulate, our results support the prediction that environmental specializations do not necessarily result in complete habitat matching due to dispersal limitations. Thus, ignoring spatial constraints in SDMs can lead to misunderstandings of the ecological mechanisms that explain species range limits.

Ancillary