Landscape genetic analyses of Cervus elaphus and Sus scrofa: comparative study and analytical developments

Abstract

Red deer and wild boar are two major game species whose populations are managed and live in areas impacted by human activities. Measuring and understanding the impact of landscape features on individual movements and spatial patterns of genetic variability in these species is thus of importance for managers. A large number of individuals sampled across Wallonia (Belgium) for both species have been genotyped using microsatellite markers (respectively > 1700 and > 1200 genotyped individuals) and some individuals have also been followed using a capture-mark-recapture (CMR) protocol. The combined data set represents an unprecedented opportunity to study and compare the environmental factors impacting the interconnectivity of these large mammals. The present study describes and uses a landscape genetic workflow to compare spatial patterns of genetic variability and the impact of environmental factors on genetic differentiation. For the latter analyses, we investigate the correlation between genetic and environmental distances (pairwise approach) and also between local genetic dissimilarity and environmental conditions (point approach). Preliminary analyses of CMR data confirm that motorways act as significant barriers to dispersal. However, analyses performed with the pairwise approach do not highlight any evidence of an impact of motorways on genetic differentiation, which is presumably due to their recent establishment. Complementary analyses performed with the point approach reveal that low altitude tends to be associated with higher genetic dissimilarity. From a methodological point of view, the present workflow illustrates the complementary application of both pairwise and point approaches, as well as univariate and multivariate analyses.

Unravelling the dispersal dynamics and ecological drivers of the African swine fever outbreak in Belgium

“…With the absence of detected crossing motorway events, even for individuals having their home range immediately next to a motorway, the GPS data further confirmed the significant impact of motorways on the dispersal frequency of these individuals, which was previously demonstrated with capture-mark-recapture data (Dellicour et al, 2019). The analysis of GPS movement data also revealed that on average, individuals spent ~74% of their time in forest areas, with ~59% of them spending at least 75% of their time in forest areas (Figure 1).…”

“…Because the motorway segment is an apparent effective barrier to ASF and wild boar dispersal, which is also confirmed by the analysis of individual capture-mark-recapture (Dellicour et al, 2019) and GPS data, the forest raster was preliminary modified to assign a very low conductance value (1/[1,000 × k]) to raster cells crossed by the motorway segment.…”

Important Dispersal Capacity of One Main Species Vector of Bluetongue and Schmallenberg Viruses Culicoides Obsoletus (Diptera: Ceratopogonidae) Revealed by Landscape Genetic Analyses 

“…The null raster is a uniform raster with all cell values equal to 1. As in Dellicour et al (2019), three values of k (10, 100 and 1000) were used to modify the potential impact of the environment on the resistance/conductance value. We used multiple regressions on distance matrices (MRDM) coupled with commonality analyses (CA; [76]) to identify unique and common contributions of predictors to the variance in the environmental distance (response variable).…”

“…In addition, we also performed univariate analyses by comparing (i) the determination coe cient R² obtained from the linear regression of genetic distances on distances computed on the environmental raster and (ii) the determination coe cient R² obtained from the linear regression of genetic distances on distances computed on the null raster. Only the environmental factors associated with a R² higher than that the one obtained from the linear regression based on environmental distances computed on the null raster were selected for the multivariate analyses (MRDM-CA) [78]. The same criterions to identify a suppressors were used for the multivariate analysis as in the commonality analyses descriptions [79,80].…”

Using empirical datasets to quantify uncertainty in inferences of landscape genetic resistance due to variation of individual-based genetic distance metrics

“…The proportion of shared alleles, D PS , (Draheim, Moore, Fortin, & Scribner, 2018; E. L. Landguth, Cushman, Murphy, & Luikart, 2010; Trumbo, Spear, Baumsteiger, & Storfer, 2013) and other kinship- or relatedness-coefficients (Dellicour et al, 2019; Renner et al, 2016) are popular choices and continue to be implemented in new applications developed for landscape genetic analyses (Savary et al 2020). However, many of these estimators have a large sample variance, which has recently been shown to negatively impact landscape genetic inferences (Winiarski, Peterman, & McGarigal, 2020).…”