Dendrochronologia 30 (2012) 287–294 Contents lists available at SciVerse ScienceDirect Dendrochronologia journal homepage: www.elsevier.de/dendro Original article Linking aerial survey data of forest insect defoliation and tree ring data to estimate forest level growth losses Steen Magnussen ∗ , René I. Alfaro 506 West Burnside Road, Natural Resources Canada, Canadian Forest Service, Victoria, BC, V8Z 1M5 Canada article info Article history: Received 7 July 2011 Accepted 8 March 2012 Keywords: Tree ring analysis Autoregressive conditional heteroscedastic model Insect surveys Tree response to defoliation and climate abstract Aerial surveys provide valuable information on the scale and severity of defoliation and mortality caused by forest insects. However, methods to estimate the losses in growth using aerial survey data are not avail- able. This study presents a method to link aerial survey data of location and severity of insect defoliation to tree ring series of radial growth in order to quantify potential growth losses due to defoliation. The method includes estimation of potentially confounding effects of climate. As an example we used aerial surveys of hemlock looper defoliation in the Mt. Revelstoke region of British Columbia, and tree ring series of radial growth of 351 western hemlock trees located in 22 plots and exposed to various levels of periodic defoliation. An autoregressive conditional heteroskedasticity time-series model (ARCH) of standardized growth indices identified significant direct and lagged effects of climate and defoliation on radial growth between 1930 and 2005. Defoliation effects were statistically significant in all plots (P < 0.05). The varia- tion in both timing and strength of estimated growth reductions varied considerably among plots. During years of observed defoliation the radial growth was reduced by an average of 10%. Growth reductions attributed to defoliation were most pronounced in the second and third year of a looper outbreak. Our modelling approach linked tree ring data to aerial survey data of defoliation through a logistic distance dependent threshold function and affords a quantitative estimation of regional growth losses. Crown Copyright © 2012 Published by Elsevier GmbH on behalf of Istituto Italiano di Dendrocronologia. All rights reserved. Introduction Large scale insect defoliation of forest trees is a common phe- nomenon in Canada’s forests (Cooke et al., 2007). Information about the location, extent, and severity of defoliation is typically pro- vided by annual aerial surveys of insect defoliation (Harris and Dawson, 1979; Iqbal and MacLean, 2010). For an effective manage- ment of forest resources an assessment of the impact of defoliation on growth of forest trees is also needed. This information typi- cally arrives from well-studied, on the ground forest plots with a known defoliation history of individual trees (Zhang and Alfaro, 2002; Mayfield et al., 2005). However, it has generally not been feasible to link ground-based tree-level responses to aerial obser- vations of defoliation severity (Erdle and MacLean, 1999; Pothier et al., 2005), mostly due to unresolved scaling and accuracy issues (Harris and Dawson, 1979; MacLean and MacKinnon, 1996). This study demonstrates how tree ring series of radial growth increments in trees growing in areas exposed to repeated defolia- tion events can be used, in combination with aerial survey-based sketch maps of location, extent, and severity of defoliation, to ∗ Corresponding author. Tel.: +1 250 298 2423; fax: +1 250 363 0775. E-mail address: steen.magnussen@nrcan.gc.ca (S. Magnussen). compute plot-level estimates of growth losses due to defoliation. To our knowledge this is the first attempt of this kind. We demonstrate our generally applicable approach with an example of defoliation in western hemlock (Hw, Tsuga heterophylla (Raf.) Sarg.) by the western hemlock looper (Lambdina fiscellaria (Hulst) Lepidoptera: Geometridae) or HL for short. The HL occurs throughout most of North America (Jardine, 1969). The larvae feed on the foliage of most conifers and some broad-leaved species, but prefer western hemlock. The west- ern subspecies, lugubrosa, periodically increases in numbers and causes defoliation and subsequent tree mortality over limited, well-defined areas in British Columbia, resulting in large losses of timber during periodic outbreaks (Kinghorn, 1954). A detailed description of the looper biology can be found in Jardine (1969). Looper outbreak populations can remove nearly all of the new and old needles on a tree in a single season. The severity and inten- sity of defoliation, however, is known to vary considerably among areas affected during an outbreak (Kinghorn, 1954; Jardine, 1969; Harris et al., 1985; Alfaro et al., 1999). According to Stock et al. (2005) mature, dense, and uneven aged (multilayered) stands with a high proportion of western hemlock in valley bottoms (below 1400 m a.s.l.) are the most susceptible. Although outbreaks gen- erally last no more than three to four years, western hemlock is considered intolerant of defoliation, and mortality can occur after 1125-7865/$ – see front matter. Crown Copyright © 2012 Published by Elsevier GmbH on behalf of Istituto Italiano di Dendrocronologia. All rights reserved. http://dx.doi.org/10.1016/j.dendro.2012.03.001