We aim to present a comprehensive study to explore more aspects of environmental and biotic factors determining the population dynamics of insects that can act as vectors of plant pathogens. The vast variation in pest population sizes between years causes unpredictable disease outbreaks, and this is why further research on the factors that determine insect population dynamics is needed. To achieve this, we developed and cross-validated the prediction ability of Species Distribution Models (SDMs) based on self-collected count data. We developed climatic, land-cover and biotic interaction models as a function of biologically meaningful predictors. SDMs have been used effectively along spatio-temporal gradients in fields such as biogeography or invasion biology to predict species’ range shifting or future invasions/local extinctions, particularly in the current context of Global Change. However, SDMs have been rarely applied to pest management and non-natural habitats. We used potato psyllid populations weekly monitored since 2017 in the US Pacific Northwest as our model study system. Our main research questions were (i) how stable are pest population dynamics across seasons in our system? (ii) which environmental and biotic factors affect population size and phenology? Here we used two SDM analytical frameworks. We developed Maximum Entropy (MAXENT) models (based on presence-only data) to calculate climatic and spatial niche of the insect pest, and Generalized Boosted Models (GBMs) (based on abundance data) to monitor the population status of these insects. Relative influence of predictors was calculated for both models
Results/Conclusions
Presence-only models obtained an excellent performance score (mean AUC of the ROC curve = 0.915). Although insect density varied dramatically between years, MAXENT models were capable of characterizing the spatial and climatic niche, and for some species, predicted suitable areas of invasion in regions where populations are still not established. Abundance (GBM) models showed that biotic models outperformed climatic and land-use models (in that order), but all were statistically significant predicting insect abundance. For the biotic models, the densities of natural enemies were inversely correlated, although some of the predators did not show the expected effect. In the climatic and land-use models, the water availability seems to have an important role, since variables like “wetlands” and “distance to water bodies” (land-use), “precipitation” and “dewpoint temperature” (climatic) were ranked as top predictors. SDMs could thus have important applications in integrated pest management, and the development of specific yet applicable pest population models should be a powerful tool.