Habitat quality, configuration and context effects on roe deer fecundity across a forested landscape mosaic.

Effective landscape-scale management of source-sink deer populations will be strengthened by understanding whether local variation in habitat quality drives heterogeneity in productivity. We related female roe deer Capreolus capreolus fecundity and body mass to habitat composition and landscape context, separately for adults and yearlings, using multi-model inference (MMI) applied to a large sample of individuals (yearlings: fecundity = 202, body mass = 395; adults: fecundity = 908, body mass = 1669) culled during 2002-2015 from an extensive (195 km2) heterogeneous forest landscape. Adults were heavier (inter-quartile, IQ, effect size = +0.5kg) when culled in buffers comprising more arable lands while contrary to our prediction no effects on body mass of grassland, young forest or access to vegetation on calcareous soil were found. Heavier adults were more fertile (IQ effect size, +12% probability of having two embryos instead of one or zero). Counter-intuitively, adults with greater access to arable lands were less fecund (IQ effect of arable: -7% probability of having two embryos, instead of one or zero), and even accounting for greater body mass of adults with access to arable, their modelled fecundity was similar to or lower than that of adults in the forest interior. In contrast, effects of grassland, young forest and calcareous soil did not receive support. Yearling body mass had an effect on fecundity twice that found in adults (+23% probability of having one additional embryo), but yearling body mass and fecundity were not affected by any candidate habitat or landscape variables. Effect of arable lands on body mass and fecundity were small, with little variance explained (Coefficient of Variation of predicted fecundity across forest sub-regions = 0.03 for adults). More variance in fecundity was attributed to other differences between forest management sub-regions (modelled as random effects), suggesting other factors might be important. When analysing source-sink population dynamics to support management, an average value of fecundity can be appropriate across a heterogeneous forest landscape.