Predicting how seed-eating passerines respond to cattle grazing in a semi-arid grassland using seed preferences and diet

Unveiling the mechanisms that act as causal links between habitat alteration and bird abundance conciliates results often interpreted as unpredictable or irreproducible. [Display omitted] •We modeled the main effects of cattle grazing on vegetation variables.•And used seed preferences of birds to predict their numerical response to grazing.•Grass seed specialists responded to the reduction of large grass seeds only.•Some expanding specialists responded to the reduction of all kinds of seeds.•Natural history knowledge conciliates results often interpreted as unpredictable. Numerical responses of animals to habitat perturbation often seem inconsistent, spreading skepticism about the predictive capacity of applied ecology. Domestic grazing changes several habitat variables that can affect seed-eating birds. Birds, in turn, show adaptations (e.g. in their feeding behavior) that could allow them to overcome habitat perturbations. Here we modelled habitat variables (e.g. cover of different plants, panicles, soil seed bank) in grazed and ungrazed (or lightly grazed) habitats of the central Monte desert, Argentina, to detect those affected by grazing activity. There was no effect of grazing on shrub and tree cover, but grazing reduced the abundance mostly of large grass seeds but also of small grass and forb seeds. Then, we used model’s outputs and knowledge of feeding preferences of the five most common seed-eating passerines in the Monte to make species-specific predictions: changes in abundance of grass seed specialists (Saltatricula multicolor, Microspingus torquatus and Porphyrospiza carbonaria) due to grazing activity should be consistent and should depend on large grass seeds (i.e. preferred seeds), whereas changes in abundance of more generalist species (Zonotrichia capensis and Diuca diuca) should be less consistent and explained also by the abundance of other seeds. The abundance of large grass seeds was sufficient to predict the abundances of S. multicolor, M. torquatus and P. carbonaria. The best model for predicting the abundance of Z. capensis included large grass seeds as well as small grass or forb seeds. No model including the abundance of seeds predicted the abundance of D. diuca. Therefore, feeding behavior explained the abundance of four out of the five bird species. A review of the literature showed that feeding behavior is also a good predictor of habitat use in other desert grasslands. Conservative range management should consider, and even