Does damage-dependent long-distance dispersal explain mountain pine beetle spread?

PhD project. University of Toronto. Supervisor: Marie-Josée Fortin.
In preparation for publication.  Meantime, find the details in my dissertation.

Quick summary: I investigated how well a discretized integro-difference model of mountain pine beetle population dynamics predicts the occurrence of new infestations in British Columbia.

Abstract

Mountain pine beetles can disperse long distances, but the role of long-distance dispersal
in the spread of infestations remains poorly understood. We expect that beetles should
emigrate when local resources are depleted, and therefore we hypothesized that a positively
damage-dependent long-distance dispersal model would improve predictions of new
mountain pine beetle infestations in previously uninfested habitat. We investigated how
well a discretized integro-di erence model of mountain pine beetle population dynamics
predicts the occurrence of new infestations in British Columbia, using aerial survey
data from 1960 to 2007. In all regions, we found that larger kernels (average dispersal
distances between 5 km and 20 km) predict better than the smallest kernels (average
dispersal distances of 1 or 2 km). We also found that prediction accuracy is highest when
we assume that many beetles disperse from new, low severity infestations, contrary to our
expectation that long-distance dispersal is positively damage-dependent. However, we
suspect that di erences in habitat quality and susceptibility among locations could cause
new infestations to arise near to other new infestations, even if beetles disperse from
old, high severity infestations. Weather and habitat gradients could also cause apparent
spread patterns. Thus, we cannot conclude that the patterns we observe are caused by
long-distance dispersal, or that beetles emigrate more from new, low severity infestations.
Even if it is not possible to identify causal processes from analysis of infestation patterns, predicting infestation risk is useful, and we found that large, distance-weighted kernels
improve predictions.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s