Modeling the spatial and temporal dynamics of the amber-marked birch leaf miner infestation in Anchorage, Alaska

Abstract

Since 1998, the invasive insect amber-marked birch leaf miner (Profenusa thomsoni Konow.) has been an issue for the birch trees in Alaska's Anchorage Bowl. P. thomsoni is native to Europe and an invasive defoliator of birch trees; its impacts are considered aesthetically unpleasing to Anchorage residents. In this study, a spatial and temporal model was constructed using a cellular automata (CA) method. Employing the statistical program R (R Development Core Team 2008), coupled with a custom library called RandomFields (Schlather 2012) and linear regression techniques, a CA model was created. Using five years of field data gathered between 2006 and 2010 (Lundquist et al. 2012), the CA model mimics the observed change in severity of the infestation based upon the severity in the previous year and if the region was in an area that increased or decreased in severity. A voracity test was used to compare the CA model output for the time period of the observed field data; a sensitivity analysis on various input parameters was also implemented. The CA model simulated results were analyzed for the time period 1998 to 2018 and indicated that P. thomsoni may follow three primary phases: 1) expansion, 2) decline, and 3) equilibrium. The expansion phase demonstrated a six-year spatial spread cycle, which can be described as random, disjointed regions of high infestation that move about the landscape. The expansion phase may be the result of an abundance of host, lack of natural enemies, and no density-dependent factors. The declining phase is depicted as a decrease in severity at an increase rate. The declining phase is possibly due to the combination of density-dependent factors and natural enemies. The equilibrium phase is a possible product of long-term plant defenses. The development of this spatial and temporal predictive CA model will allow resource managers to be proactive in order to mitigate and manage the P. thomsoni infestation. In addition, this modeling method can be used to simulate other forest pests and pathogens at the landscape level.