Lizard host abundances and climatic factors explain phylogenetic diversity and prevalence of blood parasites on an oceanic island

Description

1. Host abundance might favour the maintenance of a high phylogenetic diversity of some parasites via rapid transmission rates. Blood parasites of insular lizards represent a good model to test this hypothesis because these parasites can be particularly prevalent in islands and host lizards highly abundant. 2. We applied deep amplicon sequencing and analysed environmental predictors of blood parasite prevalence and phylogenetic diversity in the endemic lizard Gallotia galloti across 24 localities on Tenerife, an island in the Canary archipelago that has experienced increasing warming and drought in recent years. 3. Parasite prevalence assessed by microscopy was over 94% and a higher proportion of infected lizards was found in warmer and drier locations. A total of 33 different 18s rRNA parasite haplotype were identified and the phylogenetic analyses indicated that they belong to two genera of Adeleorina (Apicomplexa: Coccidia), with Karyolysus as the dominant genus. The most important predictor of between-locality variation in parasite phylogenetic diversity was the abundance of lizard hosts. 4. A combination of climatic and host demographic factors associated with an insular syndrome may be favouring a rapid transmission of blood parasites among lizards on Tenerife, which may favour the maintenance of a high phylogenetic diversity of parasites.

Steps to reproduce

The interlocal variation of both parasite phylogenetic diversity α-indices (q=0 and q=1) were best fitted to generalized linear models with quasibinomial error distribution and linked to a logit function (Westby et al., 2019). The resulting model residuals did not conform to parametric expectations and thus, in a second approach, non-parametric univariate Spearman’s correlations were used to test whether lizard abundance estimates, PC1_climate, PC2_climate, and median intensities of both mite vectors and blood parasites (magnifying glass and microscope counts, respectively), correlated with the α-indices of parasite diversity. The threshold of significance for these correlations was Bonferroni-corrected to 0.01 in order to reduce type I error. Since this approach did not consider all the factors simultaneously but one by one, a proxy of the α-diversity for q=0 was calculated ––the median prevalence of parasite haplotypes found per individual lizard. This latter metric was best fitted to a general linear model with Gaussian distribution model that included year, lizard abundance estimates, PC1_climate and PC2_climate, and median intensities of both mites and blood parasites as predictors. The number of samples molecularly analysed per locality was included as weighing term in this model. A test based on variance inflation factors (VIF; Schroeder et al., 1990) indicated a low autocorrelation of the variables (all VIF’s < 2). We used the R-package ‘MuMIn’ to implement a multimodel inference approach in the analysis of the models of parasite prevalence assessed by microscopy and the median prevalence of parasite haplotypes found per individual lizard (Barton, 2018). For this, we considered sufficiently informative all the models with ΔAICc ≤ 4 (Burnham and Anderson, 2004). We used model averaging to get a final model and calculate the relative importance of each predictor. For this, we considered only the models that included the effect (i.e., conditional average) to calculate the significance (α < 0.05) of the predictors, their z-standardized ß coefficient, and their standard error. The resulting final models were cross-validated using a k-fold split of 3 in the R-package ‘DAAG’ (Maindonald et al., 2015). Finally, we calculated the percentage of the variance explained by each significant predictor by means of their sum of squares.

Categories

Funding

Licence