PS 92-167 - Using boosted regression tree analysis to identify bioclimatic variables correlated with Mormon cricket outbreaks

Friday, August 12, 2011
Exhibit Hall 3, Austin Convention Center
Patrick D. Lorch, Biological Sciences Department, Kent State University, Kent, OH
Background/Question/Methods

Insect population densities can become so high that they dominate the community, and have important ecological and economic impacts.  These outbreaks are partially regulated by weather parameters such as temperature and precipitation.  I use boosted regression tree analysis, applied to a unique dataset including the population density of Mormon crickets (Orthoptera:  Tettigoniidae:  Anabrus simplex) and modeled weather data, to gain insight into how temperature and precipitation affect insect outbreak dynamics.  

One of the most important contributors to outbreaks in arid habitats is release of eggs from diapause.  Mormon cricket eggs can sit in dry soil for up to five years before breaking out of diapause and hatching.  Recent findings from lab experiments indicate that long periods (>100 days) of storage at 4ºC and soil moisture between 25-50% saturation at key periods increase diapause release and hatching success.  This project was set up to test whether these lab findings could be extended to the field using historical survey data.  I used Mormon cricket census data from 2004-2006 across a 17 state area (>10,000 presence absence and density estimates per year), and paired each survey point with the nearest modeled weather estimate.  I then estimated the contribution of 12 bioclimatic summary variables (e.g., minimum temperature of the coldest month and quarter, precipitation of the driest month and quarter) to explaining both presence/absence and density estimates.  I used boosted regression tree analysis and weather data generated using the PRISM climate model.

Results/Conclusions

Three interesting relationships emerge from these analyses. First, increased mean temperature of the coldest quarter was negatively related to early season density and was associated with the likelihood that Mormon crickets were absent.  Second, levels of precipitation during the driest quarter were positively related to early season density and precipitation during the driest month was associated with the likelihood that Mormon crickets were present.  Both of these results fit nicely with recent empirical findings.  Both sets of results suggest that eggs must experience a prolonged cold period before they can be released from diapause, and that moisture before, during or after this period may be crucial for embryo development within the egg.  The third result shows that the effect temperature has on density can be positive or negative depending on whether it is above or below freezing.  Interactions between variables are also considered.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.