Increasing landscape heterogeneity as a win-win solution to manage trade-offs in biological control of crop and woodland pests.

Agriculture and forestry cover more than 75% of Europe, and invertebrate pests are a costly challenge for these two economic sectors. Landscape management is increasingly promoted as a solution to enhance biological pest control, but little is known on its effects on adjacent crop fields and woodlands. This study aims to explore the effect of the proportion of woodlands and permanent grasslands as well as crop diversity on biological pest control simultaneously in cereals fields and woodland patches, in south-western France. We used different types of sentinel prey as well as bird and carabid community metrics to assess biological pest control potential in these two ecosystems. We first show that land cover variables influence biological pest control both in cereal fields and woodland patches, but have antagonistic effects in the two ecosystems. Although results vary according to the biological control indicator considered, we show that increasing landscape heterogeneity represents a valuable solution to manage trade-offs and promote higher average predation rates across forests and cereal fields. Our study therefore calls for more integrative studies to identify landscape management strategies that enable nature-based solutions across ecosystems.

The influence of spatial and temporal discontinuities of forest habitats on the current presence of flightless saproxylic beetles.

Flightless saproxylic beetles were selected in order to study the impact of temporal and spatial discontinuity of forests. They were chosen because: (1) they are unable to fly, making them dispersal-limited species, (2) they have a saproxylic diet, which means they are closely linked to the forest, and (3), they have rarely been studied. Forest temporal continuity was expected to be the main factor explaining the presence of these species, modulated by the past and present amount of forest in the surrounding landscape. Twenty-seven forests, distributed into three zones, were sampled in southwestern France. Flightless saproxylic beetles were surveyed using a Winkler extractor and a Berlese funnel. Their presence/absence were modelled using generalised linear mixed models, with zone variable as random effect. Two species showed significant zone effect and were only or more present in the zone with the highest present forest amount in a 0.5 km radius. In the model that converged, the only selected variable was the past amount of forest in the landscape. The size of the forest, the presence of dead wood and the forest temporal continuity were not included in this model. The importance of the amount of forest in the landscape supports the hypothesis that dispersal-limited species are affected by landscape characteristics. This study demonstrates an important link between the presence of Dienerella clathrata and the amount of forest in the past, which led to an indicator species analysis being performed.

Is there an optimum scale for predicting bird species' distribution in agricultural landscapes?

Changes in forest cover in agricultural landscapes affect biodiversity. Its management needs some indications about scale to predict occurrence of populations and communities. In this study we considered a forest cover index to predict bird species and community patterns in agricultural landscapes in south-western France. We used generalized linear models for that purpose with prediction driven by wooded areas' spatial distribution at nine different radii. Using 1064 point counts, we modelled the distribution of 10 bird species whose habitat preferences are spread along a landscape opening gradient. We also modelled the distribution of species richness for farmland species and for forest species. We used satellite images to construct a 'wood/non-wood' map and calculated a forest index, considering the surface area of wooded areas at nine radii from 110m to 910m. The models' predictive quality was determined by the AUC (for predicted presences) and ρ (for predicted species richness) criteria. We found that the forest cover was a good predictor of the distribution of seven bird species in agricultural landscapes (mean AUC for the seven species = 0.74 for the radius 110m). Species richness of farmland and forest birds was satisfactorily predicted by the models (ρ = 0.55 and 0.49, respectively, for the radius 110m). The presence of the studied species and species richness metrics were better predicted at smaller scales (i.e. radii between 110 m and 310 m) within the range tested. These results have implications for bird population management in agricultural landscapes since better pinpointing the scale to predict species distributions will enhance targeting efforts to be made in terms of landscape management.