Multi-scale and multi-site resampling of a study area in spatial genetics: implications for flying insect species

Abstract

The use of multiple sampling areas in landscape genetic analysis has been recognized as a useful way of generalizing the patterns of environmental effects on organism gene flow. It reduces the variability in inference which can be substantially affected by the scale of the study area and its geographic location. However, empirical landscape genetic studies rarely consider multiple sampling areas due to the sampling effort required. In this study, we explored the effects of environmental features on the gene flow of a flying long-horned beetle (Monochamus galloprovincialis) using a landscape genetics approach. To account for the unknown scale of gene flow and the multiple local confounding effects of evolutionary history and landscape changes on inference, we developed a way of resampling study areas on multiple scales and in multiple locations (sliding windows) in a single large-scale sampling design. Landscape analyses were conducted in 3*104 study areas ranging in scale from 220 to 1,000 km and spread over 132 locations on the Iberian Peninsula. The resampling approach made it possible to identify the features affecting the gene flow of this species but also showed high variability in inference among the scales and the locations tested, independent of the variation in environmental features. This method provides an opportunity to explore the effects of environmental features on organism gene flow on the whole and reach conclusions about general landscape effects on their dispersal, while limiting the sampling effort to a reasonable level.

Strong gene flow across an urbanized coastal landscape in a dune specialist digger wasp

“…The deduced resistance values are values per grid cell and relative compared to the other considered landscape types. Consequently, the deduced resistance values of a category will depend on the habitat composition and the spatial configuration of how populations are embedded within the landscape (Richardson et al 2016;Haran et al 2017). In our study, urbanisation induces large-scale fragmentation compared to more natural processes like shrub and woodland encroachment.…”

“…Independent of the exact mechanism behind this potential (relative) increase in movement through the matrix, the eventual functional connectivity will remain dependent on the distance between habitats and thus the scale of the fragmentation. The same matrix typology or landscape category can therefore potentially facilitate or constrain connectivity in different landscapes, depending on the eventual distance-related costs in terms of survival and energy-use (Segelbacher et al 2010;Spear et al 2010;Bonte et al 2012;Richardson et al 2016;Haran et al 2017).…”

“…To effectively direct the limited resources available for conservation measures towards implementations with the highest success rate or biodiversity gain, implementations should be founded in empirical evidence (Sutherland et al 2004;Ferraro and Pattanayak 2006;Samways et al 2020). Especially for insectswhose decline and biological value get widespread attention (Didham et al 2020;Wagner et al 2021)-scientific evidence on the effects of landscapes on functional connectivity and gene flow is still limited (Keyghobadi et al 1999;Wilcock et al 2007;Keller et al 2012;Pérez-Espona et al 2012;Watts et al 2016;Haran et al 2017;Trense et al 2021).…”

The effect of sampling density and study area size on landscape genetics inferences for the Mississippi slimy salamander (Plethodon mississippi)

“…In such cases, it may become difficult to identify an environmental variable's true impact on gene flow (Keller et al, 2014). This idea is supported by outcomes from a study by Haran et al (2017), who repeatedly subsampled their dataset of individual-based genetic distances among pine sawyer beetles (Monochamus galloprovincialis) from the Iberian Peninsula. Through a series of partial Mantel tests examining the association between gene flow and environmental variables, they found significant relationships were more likely to be detected when using larger study areas.…”