Climate predictors and climate change projections for avian haemosporidian prevalence in Mexico

Abstract

Long-term, inter-annual and seasonal variation in temperature and precipitation influence the distribution and prevalence of intraerythrocytic haemosporidian parasites. We characterized the climatic niche behind the prevalence of the three main haemosporidian genera (Haemoproteus, Plasmodium and Leucocytozoon) in central-eastern Mexico, to understand their main climate drivers. Then, we projected the influence of climate change over prevalence distribution in the region. Using the MaxEnt modelling algorithm, we assessed the relative contribution of bioclimatic predictor variables to identify those most influential to haemosporidian prevalence in different avian communities within the region. Two contrasting climate change scenarios for 2070 were used to create distribution models to explain spatial turnover in prevalence caused by climate change. We assigned our study sites into polygonal operational climatic units (OCUs) and used the general haemosporidian prevalence for each OCU to indirectly measure environmental suitability for these parasites. A high statistical association between global prevalence and the bioclimatic variables ‘mean diurnal temperature range’ and ‘annual temperature range’ was found. Climate change projections for 2070 showed a significant modification of the current distribution of suitable climate areas for haemosporidians in the study region.

Keywords: Avian haemosporidia; avian malaria; climate change models; disease ecology; landscape parasitology.

Graphical abstract

Fig. 1.

Extent of the study region. Study sites are scattered across the states of San Luis Potosí and Veracruz, Mexico, and occur under a variety of climate conditions. To homogenize our sample data and increase the number of sampling locations, we grouped our localities on operational climate units (OCU) which facilitated the analysis and extraction of associated environmental data for each area.

Fig. 2.

Haemosporidan prevalence per representative habitat type, from the most humid to the most arid vegetation communities, is estimated by an additive generalized linear model using binomial distribution and logistic link function. Habitat types are as follow: MCF (montane cloud forest), TDF (tropical deciduous forest), TSD (tropical sub-deciduous forest), PERm (mature perennial forest), PERd (degraded perennial forest), MPF (Mexican pinyon forest), PF (pine forest), POF (pine-oak forest), OPF (oak-pine forest), MSm (mature mesquite-dominated shrub), MSd (degraded mesquite-dominated shrub), RDS (Rosetophyllous desert shrubland), MDS (Microphyllous desert shrub), LSA (livestock area), PL (pastureland), YS (Yucca shrubland). Error bars represent ±1 s.e.m. The highest prevalence trends are indicated with an asterisk, while the lowest prevalence trends are indicated with a tilde.

Fig. 3.

Mapped prediction of suitable climatic conditions associated with haemosporidian prevalence for the 2070 decade under climate change scenarios RCP 4.5 and 8.5. The white contour corresponds to the 50 km buffer applied for the models, and the coloured areas represent the different predictions regarding suitable climatic habitat for the observed haemosporidian prevalences under current climate trends: red grid cells represent a potential deficit of suitable climatic habitat under climate change scenarios RCP 4.5 and 8.5, yellow grid cells represent the area of climatic habitat that will potentially continue despite the effects of climate change, and the blue grid cells represent the area of climatic habitat added. The black contours within the buffer from which the coloured areas stem representing the polygons of each OCU.