Fig. 2: Best ranked models predicting viral loads in wild pollinators. | Nature Ecology & Evolution

Fig. 2: Best ranked models predicting viral loads in wild pollinators.

From: Species traits, landscape quality and floral resource overlap with honeybees determine virus transmission in plant–pollinator networks

Fig. 2: Best ranked models predicting viral loads in wild pollinators.

a–d, Averaged model estimates (modelled slopes of the relationships) and 95% confidence intervals of explanatory variables in best models (LMMs, ΔAICc < 2) explaining viral load of BQCV (n = 304 virus-positive individuals) (a,c) and DWV-B (n = 288 virus-positive individuals) (b,d) in wild pollinators (the response variable was log-transformed number of genome copies per microgram RNA of virus-positive individuals). a,b, display the results from modelling the effects of plant species traits and pollinator network roles (centrality measured as weighted betweenness (Betweenness), corolla length, specialization d′, proportion of dish-bowl flowers among visited flowers (Dish flowers) and floral resource overlap with honeybees (Resource overlap)). c,d, display the results from modelling the effects of pollinator community properties (Shannon diversity of flowers (Flower diversity), honeybee density (HB density) and honeybee viral load (HB BQCV or DWV-B load), wild pollinator abundance (WP abundance) and percentage cover of pollinator habitat (Pollinator habitat)) in a landscape.

Back to article page