United States 

 'Ml Department of 

 P Agriculture 



Forest Service 



Northeastern Forest 

 Experiment Station 



NE-RN-338 



1^ 



n Op 



Mortality Risks for Forest 

 Trees Tlireatened with Gypsy 

 IVIoth Infestation 



Owen W. Herrick 

 David A. Gansner 



Abstract 



Presents guidelines for estimating potential tree mortality 

 associated with gypsy moth defoliation. A tree's crown con- 

 dition, crown position, and species group can be used to 

 assign probabilities of death. Forest-land managers need 

 such information to develop marking guides and implement 

 silvicultural treatments for forest trees threatened with 

 gypsy moth infestation. 



There is growing interest in using silvicultural treatments to 

 create forest stands that are less vulnerable to mortality 

 after defoliation by gypsy moth.i To implement silvicultural 

 treatments, however, forest-land managers need guidelines 

 to help predict the probability of individual tree mortality 

 associated with defoliation. 



In this study, we have applied a multi-variable screening 

 procedure to derive mortality probabilities for individual 

 trees faced with defoliation by gypsy moth. The procedure 

 has identified key characteristics from among many forest 

 stand and tree conditions thought to be good predictors of 

 tree mortality. Previous work based on the gypsy moth out- 

 break in northeast Pennsylvania (Pocono region) pointed to 

 tree crown condition, species, and aspect as important pre- 

 dictor variables (Herrick 1982). Similar procedures applied 

 to data from more recent infestations in central Pennsylva- 

 nia have selected tree crown position in addition to crown 

 condition and species as the best tree mortality predictors. 



The Study 

 Data 



Data for the analysis are from the gypsy moth episode of 

 the early 1980's in central Pennsylvania. Cumulative tree 

 mortality was measured from 1979 through 1985 on 574 

 one-tenth acre plots established in 1978 before infestation. 

 Tree and timber stand characteristics thought to be poten- 

 tial predictors of tree mortality formed 43 variables for anal- 

 ysis. Rationale for their inclusion was based on findings of 

 Campbell and Sloan (1977), Houston and Valentine (1977), 

 Herrick et al, (1979), and Herrick (1982). All variables are 

 easy to measure and account for some dimension among 

 the following attributes: 



Stand size-composition 

 Tree diameter 

 Species 

 Stand age 

 Crown position 

 Crown condition 



Site index 

 Land capability 

 Elevation 

 Aspect 

 Slope 



Position on slope 



Analysis 



Automatic Interaction Detection (AID) analysis (Sonquist et 

 al. 1973) was used to generate a set of probabilities for indi- 

 vidual tree mortality in gypsy moth infested stands. AID is 

 an iterative technique that partitions the observations 

 (14,684 trees in this study) through a series of two-way 

 splits into subgroups. Each tree is a member of one of the 

 subgroups. The subgroups formed at each step of the pro- 



iQottschalk, Kurt W. [In preparation.] Silvicultural guide- 

 lines for forest stands threatened by gypsy moth. Gen. 

 Tech. Rep. Broomall, PA: U.S. Department of Agriculture, 

 Forest Service, Northeastern Forest Experiment Station. 



The authors are research foresters, USDA Forest Service; 

 Northeastern Forest Experiment Station; Forest Inventory, 

 Analysis, and Economics. 



1 



