Response to agriculture by a woodland species depends on cover type and behavioural state: insights from resident and dispersing Iberian lynx

Conservation biology faces the challenge of ensuring species persistence in increasingly modified landscapes. Agriculture covers a large proportion of the Earth's surface, but the degree to which crop production is compatible with species use of the landscape is still uncertain, particularly for woodland carnivores with large territories. Here, we focus on the Iberian lynx Lynx pardinus, an emblematic endangered species that has coexisted for centuries with human‐modified Mediterranean mosaics, as a case study to unravel habitat and dispersal preferences in heterogeneous landscapes. We estimated species resource selection from ≈40 000 telemetry locations for 48 GPS‐collared individuals covering all the current Iberian lynx range, including more fragmented areas where the species was reintroduced from 2009. We differentiated GPS locations within home ranges (to estimate habitat suitability) and those corresponding to dispersal or exploratory movements (to estimate landscape permeability). We built mixed conditional logistic regression models with 12 land cover classes, terrain slope and roads as predictors. We found that lynx response to agriculture largely depends on the crop type and on the presence of natural vegetation remnants. Lynx largely avoided intensive cultivation areas such as irrigated arable lands when establishing home ranges, but frequently selected permanent crops (olive groves) and/or heterogeneous agricultural lands, which were used with smaller differences to the most preferred shrubland or forest covers than reported in previous studies. Such differences further narrowed down when lynx moved outside home ranges, with some agricultural covers being as permeable as shrublands for lynx dispersal. The species dispersal plasticity was also evidenced by a much weaker avoidance of roads and steep terrain when dispersing than when selecting territories. Synthesis and applications. We conclude that (i) the widespread consideration of all agricultural lands within a single (and usually regarded as unsuitable) class for the study and management of woodland or forest species is not supported and that (ii) the ability of woodland species to use fragmented and heterogeneous agricultural landscapes may have been underestimated, which may mislead conservation measures due to a priori assumptions that do not relate to the actual species responses to heterogeneous land covers. We suggest that Iberian lynx conservation and reintroduction may be successful in a wider set of more heterogeneous areas than previously thought, including mainly well‐conserved Mediterranean woodlands but also some extensive agricultural lands with permanent crops and natural vegetation remnants.