Using connectivity to identify climatic drivers of local adaptation - data

Despite being able to conclusively demonstrate local adaptation, we are still often unable to objectively determine the climatic drivers of local adaptation. Given the rapid rate of global change, understanding the climatic drivers of local adaptation is vital. Not only will this tell us which climate axes matter most to population fitness, but such knowledge is critical to inform management strategies such as translocation and targeted gene flow. Targeted gene flow, for example, requires knowledge of where in the landscape we can find populations with pre-adapted climate-relevant traits; we cannot find these populations if we do not know the dominant climatic drivers of local adaptation. While simple assessments of geographic trait variation are useful, geographic variation (and its associations with environment) may represent plastic, rather than evolved differences. Additionally, the vast number of environment-trait combinations makes it difficult to determine which aspects of the environment populations adapt to. We argue that by incorporating a measure of landscape connectivity as a proxy for gene flow, we can differentiate between environment-trait relationships that are under selection versus those that reflect phenotypic plasticity. By doing so, we can rapidly shorten the list of environment–trait combinations that may be of adaptive significance. We demonstrate this method using data on geographic trait variation in a lizard species from Australia's Wet Tropics rainforest.
This dataset contains morphological and physiological measurements from the Australian Rainforest Sunskink (Lampropholis coggeri). Morphological measurements were obtained from 532 skinks from 32 sample sites. Physiological measurements were obtained from a smaller subset of these lizards: 259 skinks from 12 sites. All procedures involving lizards were approved by the James Cook University animal ethics committee (projects A1976 and A1726).
The following physiological measurements were taken from each skink during laboratory trials: critical thermal minimum (CTmin), critical thermal maximum (CTmax), thermal-performance breadth for sprinting (breadth80), maximum sprint speed (Rmax), temperature at which sprint speed is optimized (Topt), active body temperature as measured in a thermal gradient (Tactive), and desiccation rate (des). All physiological measurements were scaled prior to analysis.
The following measurements were taken from each skink using digital calipers: head width (HeadW); head length (HeadL); interlimb length (ILimbL); hindlimb length (HindLL). Left and right measurements were averaged to obtain one measurement for that trait. We also recorded snout–vent length (SVL), total length, and mass. All morphological variables were log-transformed and scaled prior to analysis.
The following climatic variables were extracted for each collection site from available climatic data layers: annual mean precipitation (AMP); seasonality of precipitation (Pcov); precipitation of the driest quarter (Pdry); annual mean temperature (AMT); coefficient of variation of temperature (Tcov); average minimum daily temperature (Tmin); average maximum daily temperature (Tmax); average variance of daily maximum temperature (TmaxVar); and average variance of daily Tmin (TminVar).
Our connectivity index is a measure of habitat suitability for our focal skink species, averaged over space using a species-specific estimate of dispersal potential. As our species is an obligate rainforest-dweller, grid cells in the landscape that are rainforest and that are surrounded by rainforest have high connectivity indices, while grid cells of rainforest surrounded by non-rainforest matrix have low indices.