Fig 1.
Point locations from where the samples were collected are depicted on a map of India. The samples were collected from distinct biogeographic zones. The map of India shown here was prepared with QGIS 3.8 [19] (Table A in S1 Text).
Fig 2.
Enzymatic activities of Russell’s viper venoms.
Bar graphs depicting the (A) PLA2, (B) protease and (C) LAAO activities of D. russelii venoms from across India are shown, where the x-axis represents sample IDs in Table A in S1 Text, while the y-axis represents specific activities (nmol/mg/min). The assays were performed in triplicates, and the mean activities were plotted, with the error bars indicating the standard deviation. Samples from each region/state are uniquely colour-coded. See Table A in S1 Text for details. Here, PB: Punjab (north India); RJ: Rajasthan (west India); MP: Madhya Pradesh (central India); MH: Maharashtra (southwest India); WB: West Bengal (east India); AP: Andhra Pradesh (southeast India); TN: Tamil Nadu (south India); GA: Goa (southwest India); KA: Karnataka (south India) and KL: Kerala (south India).
Table 1.
Bioclimatic predictors and their mathematical definitions. This table depicts the downselected bioclimatic predictors with their mathematical representations. The table also provides the temporal and spatial scales and the predictors’ biological relevance.
Fig 3.
Various MLR models were built to explain the variation in the D. russelii venom PLA2, protease and LAAO activities using the prevailing bioclimatic factors at the sampling locations. A combination of independent variables (IV), as depicted by yellow squares, was found to significantly affect the venom activities (A[enzyme]). The purple squares indicate variables that did not contribute substantially to the model. Logarithmic [ln(activity)], square-root [(activity)½], and inverse [1/(activity)] transformations of the dependent variables (DV) were also performed. The blue squares highlight the models that were downselected. Various model tests were also performed to assess normality (N), homoscedasticity (H), Linearity (L) and Multicollinearity (M) of the models. The green and red squares highlight if a particular MLR model passed or failed the corresponding model tests, respectively. The details of the bioclimatic variables included in the models are provided in Table 1. AMT: Annual mean temperature, TAR: Temperature annual range, AMDTR: Annual mean diurnal temperature range, I: Isothermality, TS: Temperature seasonality, APN: Annual precipitation and PS: Precipitation seasonality. The AIC values of the models are provided in Fig A in S1 Text.
Fig 4.
Predictive mapping of functional venom variation in D. russelii across the Indian subcontinent.
The predictive maps of D. russelii venom (A) PLA2, (B) protease and (C) LAAO activities based on the bioclimatic factors of the region are depicted on a map of India. The downselected MLR models, which establish the correlation between the bioclimatic variables and venom activities, have been used to build these predictive maps. The predicted enzymatic activity values across India are depicted by a colour gradient of blue to red, representing low to high activities. The grey graphs along the X- and Y-axis represent the activity at the individual pixels of 1 km2 resolution. The maps were generated using the R packages terra and rasterVis. These maps directly visualize raster data sourced from WorldClim (http://www.worldclim.org, accessed on 14.04.2023), clipped to the geographic extent of India (approx. 68.1°E to 97.4°E longitude, 6.75°N to 35.7°N latitude). No external basemap or proprietary geographic layer was used.