Determination of the spatial distribution of trees from digital aerial photographs

This study examined the possibilities of using computerized digital aerial photograph interpretation in determining the spatial distribution of trees. The material of the study included eight mapped stands in the municipality of Hyytiälä (61°50′N and 24°18′E), in southern Finland. The aerial photographs used were taken in June 1995 at a scale of 1:5000. Two approaches for determining the spatial pattern of trees were used. Firstly, in the point-process based approach used in this study, the individual trees in the digital aerial photograph were segmented by a robust segmentation method, based on recognition of the pattern of tree crowns with sub-pixel accuracy. Secondly, the crown coverage was determined by region growing segmentation combined with active surface representation. The significance of the differences in the means of image coverage pattern indices between the various spatial distribution categories was tested with one-way variance analysis. Because the process misclassified clustered spatial patterns as regular patterns, and regular patterns as random patterns, the usability of digital aerial photographs seems to be limited for the point-process based determination of the spatial pattern of trees if the scale is 1:5000 or less. When image coverage pattern indices were applied, the differences in the means of the spatial distribution categories proved not to be clearly statistically significant due to the great variation within classes. However, interpretation of crown coverage could have applications in practical forestry due to the low resolution requirements for the images used.     

Tree Species Classification using Semi-automatic Delineation of Trees on Aerial Images

The purpose of this study was to develop a method for classifying tree species from remote sensing images by combining a semi-automatic pattern recognition technique and spectral properties of trees. Five stands in southern Finland were studied. Individual trees in the digital colour infrared (CIR) aerial photographs were segmented by a method based on the recognition of tree crown patterns at subpixel accuracy. The images were filtered with the Gaussian N-by-N smoothing operator and local maxima above a threshold level were segmented. The segments were classified into three tree species classes. The kappa coefficients for stands varied from 0.43 to 0.86 when the training data and test data were from the same aerial photograph. When training data from other photographs were used as reference data, the kappa coefficients ranged from 0.40 to 0.75. The method described provides an interesting approach for detecting tree species semi-automatically in digital aerial data.     

Tree Mortality,Foliage Recovery and Top-kill in Stands of Scots Pine (Pinus sylvestris) Subsequent to Defoliation by the Pine Looper (Bupalus piniaria)

In 1996, the pine looper (Bupalus piniaria) (L.) defoliated 7000 ha of Scots pine (Pinus sylvestris L.) forest at Hökensås, in southern Sweden. To study tree mortality, foliage recovery and top-kill, plots were laid out in stands with varying levels of defoliation in autumn 1997. Tree mortality peaked 2 yrs after defoliation, and amounted to 25% in stands suffering from 90–100% defoliation. Suppressed trees suffered higher mortality than intermediate and dominant trees. In stands suffering <90% defoliation, tree mortality did not exceed 8%. Foliage recovery in moderately and severely defoliated stands was not complete at the end of the 4 yr study period, whereas slightly defoliated stands had regained full foliage in 1998. Top-kill was most frequent in severely defoliated stands, and 50% of all trees in these stands suffered from top-kill at the end of the study period in spring 2001.     

Tree Mortality,Needle Biomass Recovery and Growth Losses in Scots Pine Following Defoliation by Diprion pini (L.) and Subsequent Attack by Tomicus piniperda (L.)

Tree mortality and growth losses following insect defoliation are poorly documented in Scandinavia. In 1990-1991, Diprion pini (L.) caused extensive defoliation to Scots pine in Lauhanvuori national park and surrounding areas in south-western Finland. Most trees lost all their foliage in 1990. In 1991, the outbreak area was sprayed with diflubenzuron (Dimilin®), except in the national park, where trees were severely defoliated again. No further defoliation occurred in 1992. In spring 1993, sprayed trees had ca 30% foliage, whereas unsprayed trees on average carried less than 10% of full foliage. The latter trees were susceptible to attack by Tomicus piniperda (L.), whereas the former largely escaped beetle attack. Beetle attacks peaked in 1993, and depletion of suitable host trees probably terminated the beetle outbreak in the area. Two years of severe defoliation resulted in substantial tree mortality and growth losses. In spring 1997, these unsprayed stands had suffered a ca. 50% loss in basal area which was mainly because of mortality, and about half of the dead trees had been attacked by T. piniperda. Surviving trees had ca 50% of full foliage, and radial growth had still not recovered. Basal area growth was reduced by 40-70%, depending on the amount of foliage left after the second year of defoliation. In contrast, tree mortality and beetle attack in the sprayed stands were negligible, and these trees had recovered full foliage and radial growth by spring 1997. Thus, one year of total defoliation resulted in an estimated loss in basal area growth of approximately 30% during ca. 5 yrs. In conclusion, the spraying operation was economically justified, as it prevented substantial tree mortality and reduced growth losses.     

The impact of hemlock looper (Lambdina fiscellaria fiscellaria (Guen.)) on balsam fir and spruce in New Brunswick, Canada

In 1989, the first recorded outbreak of hemlock looper (Lambdina fiscellaria fiscellaria (Guen.)) occurred in New Brunswick, Canada. Data were collected from ten plots established in an area infested from 1992–1994, to assess impacts of hemlock looper. Ocular and branch sample assessments of current defoliation and ocular assessments of total defoliation (all age classes of foliage) were conducted for balsam fir (Abies balsamea [L.] Mill.), white spruce (Picea glauca [Moench] Voss), and black spruce (Picea mariana (Mill.) B.S.P.). Stand response was assessed and related to cumulative defoliation. Ocular assessments were found to accurately estimate defoliation, which was significantly related to tree mortality. Ninety-two percent of balsam fir trees that had cumulative defoliation >90% died. Mortality of balsam fir was significantly (p < 0.05) related to tree size, in both lightly and severely defoliated plots; trees with DBH <11 cm sustained 22–48% higher mortality than larger trees. Mortality of balsam fir, in terms of both percent stems/ha and m³/ha merchantable volume, increased exponentially in relation to three estimates of cumulative (summed) plot mean defoliation. The strongest relationships (r² = 0.75–0.79) were between mortality and the ocular defoliation assessment for 1990–1993 foliage. Tree mortality caused by the looper outbreak ranged from 4–14% stems/ha in lightly defoliated and from 32–100% in severely defoliated plots; merchantable volume killed was 3–14 m³/ha and 51–119 m³/ha, respectively. Relationships between mortality and defoliation were similar when defoliation was assessed for 1987–1993 and 1990–1993 foliage age classes.     

Zuwachs und Lebensfähigkeit von Eichenbeständen nach Frades Schwammspinners,Lymantria dispar L., in Rumänien

Summary Effects of the gypsy moth defoliation on growth and mortality of oak stands Investigations were carried out in Rumania in three oak stands (Quercus pedunculata) situated in two different phytoclimatic regions where the defoliation was observed one year or more consecutive years. In four other forests were used artificial defoliations in young (15–20 years) stands.The growth (i_g) was estimated in permanent trial plots (with minimum 150 trees each plot) in defoliated stands and in control stands where gypsy moth were controlled. p ]Radial growth (i_r) was measured by microauxometer and by increment cores sampled with Pressler borer.The results show that one year defoliation influenced the stands growth in that year and in the next two years. In this case the whole growth difference — as compared with control stands — was almost similar to the equivalent of one year growth. If the defoliation occurred two consecutive years, the whole growth difference was similar to 1.3–2.2 years growths, in old stands and in young stands respectively.Oak stands defoliation, unassociated with other limitative factors, does not lead to oak mortality but only to increase of natural elimination process.     

Carbohydrates in individual poplar fine roots: effects of root age and defoliation

Late-summer starch accumulation in fine roots of poplars (Populus x canadensis Moench.) defoliated by gypsy moth (Lymantria dispar L.) lagged behind that in fine roots of undefoliated trees. If starch concentration declines with age, defoliation-induced changes in root system age structure could be partly responsible for this difference. To test this hypothesis, we measured fine-root starch and soluble sugar concentrations in roots of known age from trees in defoliated and undefoliated plots. There was a significant interaction between the effects of defoliation and root type (white, brown, or dead) on fine root soluble sugar concentration because of the high concentration of soluble sugars in white roots from trees in undefoliated plots. Both root starch and soluble sugar concentrations were variable among individuals of each age class. The population frequency distributions for starch and soluble sugar concentrations were both right-skewed, and fit by exponential functions. These data are most consistent with direct defoliation effects on a labile and dynamic pool of carbohydrates in poplar fine roots, rather than indirect defoliation effects on root system age structure.     

Induced response of the Siebold's beech (Fagus crenata Blume) to manual defoliation

Siebold's beech (Fagus crenata) was manually defoliated for two successive years. The beech caterpillar (Quadricalcarifera punctatella) was used in a bioassay to determine insect performance. Survival and body size were low on foliage from defoliated trees. Reduced foliar nitrogen and increased tannin content were probably the main causes of the low insect performance. Leaves were less tough on defoliated trees than in controls. Two sucessive years of manual defoliation caused stronger induced resistance than one year defoliation. The quality, as well as the quality of the foliage, decreased the year following manual defoliation; total weight of leaves on a tree was less than one half of that before treatment. Severe defoliation may cause a decrease of leaves the following year and starvation may limit populations. Delayed induced resistance of beech trees is proposed as a possible cause of the cyclical population dynamics ofQ. punctatella. The delayed induced response also affected folivorous insects other thanQ. punctatella.     

The Ingrowth Bag Method in Measuring Root Production on Peatland Sites

The aim of this study was to develop a method for segment-based forest inventory and determine whether segment-level inventories can be used in forest management planning. The study area covered 76 ha located in two different aerial photographs in eastern Finland. The study area was segmented into 220 segments with the aid of aerial photographs and the segment-level forest characteristics were assessed in the field using relascope sample plots and a field computer which displayed the aerial photographs, segment borders and surveyor's location on the screen. The segment estimates were calculated as weighted averages of k nearest neighbours (kNN) for the segments and the sample plots. The estimates were tested with a cross-validation technique. The averages and the standard deviations of the spectral values of aerial images extracted for the segments and the sample plots were used in the kNN estimation. The relative root mean square error of the mean volume was 58.1% (bias –6.4%) at the segment level and 57.9% (bias –0.9%) at the sample plot level. The segment-based approach studied here needs further research and improvement before it can be applied to forest management planning.     

Tree defoliation by winter moth Operophtera brumata L. during an outbreak affected by structure of forest landscape

Tree defoliation by winter moth Operophtera brumata L. (Geometridae, Lepidoptera) caterpillars was quantified in the Białowieża Forest (E Poland) during the recent outbreak (2002). To assess effects of forestry management on defoliation, comparisons were made between mixed species (Tilio–Carpinetum) old-growth stands of natural origin in strictly protected primeval and managed fragments. This forest type was continuous in the primeval part, but highly fragmented in the managed part. The old-growth remnants in the latter part were usually surrounded by younger tree stands (often coniferous plantations). We predicted the winter moth densities would be reduced in fragmented stands due to dispersal mortality incurred by the early stage larvae.As predicted, all species of host trees suffered greater defoliation in the primeval forest. Hornbeam Carpinus betulus L. was the heaviest defoliated host. Other trees, which developed leaves synchronously with hornbeam – maple Acer platanoides L., lime Tilia cordata Mill., and pedunculate oak Quercus robur L. – were strongly affected as well. While we do not show a causal mechanism, greater dispersal mortality in fragmented stands is a most plausible explanation for the observed patterns of defoliation.     

Effects of Atlas cedar (<Emphasis Type="Italic">Cedrus atlantica</Emphasis>) defoliation on performance of the pine processionary moth (<Emphasis Type="Italic">Thaumetopoea pityocampa</Emphasis>)

The pine processionary moth (Thaumetopoea pityocampa) is an important defoliating insect of native cedar stands in Northern Africa. In 2002 and 2003, we studied the size of needles of Atlas cedar trees that had been defoliated or not in the previous year, and compared the oviposition preference and larval performance of processionary moth on these two types of trees in cedar stands of central Algeria. Needle length and their number per needle cluster were twice as high on non-defoliated trees than on defoliated trees. There were about ten times fewer egg batches on defoliated than on non-defoliated trees in both study years. Young larvae raised in the laboratory on the foliage of previously defoliated trees were smaller compared to those fed on the foliage from healthy trees. Larval colonies transferred to trees in the field showed two times higher mortality on defoliated trees. It was concluded that the decline in population density in defoliated stands often observed after outbreaks results both from the fact that the female moths avoid laying their eggs on defoliated trees and the lower performance of larvae which later feed on the new-grown needles.     

Effects of defoliation by Eucalyptus weevil, Gonipterus scutellatus, and chrysomelid beetles on growth of Eucalyptus globulus in southwestern Australia

Plantations of Eucalyptus globulus in southwestern Australia are defoliated by Eucalyptus weevil, Gonipterus scutellatus, and a complex of chrysomelid and scarab beetles, yet there is no information on the impact of beetle defoliation to tree growth in southwestern Australia. To address this shortcoming, we used insect exclusion trials, to compare growth of insecticide treated (and thus relatively undamaged) trees with untreated (and thus defoliated) trees to determine whether defoliation by G. scutellatus and other beetles reduced the growth and harvest volume of E. globulus trees. Our results showed some evidence of beetle defoliation reducing growth of E. globulus. Mean defoliation levels of the growing tip of untreated trees ranged from 18% to 33% across the duration of the study and were significantly greater than mean defoliation levels of 5–16% on insecticide treated trees. Seasonal peaks in defoliation of 30–80% to the growing tip of untreated trees were recorded between late spring and early autumn. The greatest impact of defoliation on tree growth was evident during the 2.5 year period of insect exclusion, when higher relative growth rates were recorded for insecticide treated trees, which were significantly different from relative growth rates of untreated trees at two of the four plantations. However, our results showed only a limited impact of beetle defoliation on the total volume at harvest. Initially small trees tended to suffer more severe defoliation than initially large trees. Effects of insect exclusion treatment on harvest volume were modified by the initial tree size and the relationship between the initial tree size and levels of defoliation.     

Stand structure interacts with previous defoliation to influence herbivore fitness

The intensity and duration of Neodiprion abietis outbreaks have recently increased in forests of North America that were precommercially thinned more than a decade earlier. We tested the hypotheses that changes in stand structure following thinning increase the fitness (i.e., survival rate × fecundity) of N. abietis by either (a) increasing foliar availability and/or quality (i.e., increased availability of primary metabolites and/or reduced foliar defenses) or (b) by reducing any negative effects on foliar quality and/or availability resulting from herbivory that occurred during the preceding season(s). Effects of thinning and previous herbivory on N. abietis and its host plant (Abies balsamea) were determined through (i) a manipulative field experiment that evaluated the effects of experimental defoliation on N. abietis in a thinned stand, (ii) a manipulative field experiment that examined the effects of thinning on N. abietis in undamaged and naturally defoliated stands, and (iii) a field survey to estimate survival of N. abietis in natural populations.Defoliation caused reductions in the availability of different-aged foliage available to larvae and in the fitness of a subsequent N. abietis generation feeding on defoliated branches, but decreases in fitness were smaller in thinned than unthinned stands. In thinned stands, defoliation was associated with increases in foliage production and foliar contents of monoterpenes and nitrogen, as well as with a decrease in foliar contents of water. Conversely, only small changes in plant growth and foliar contents of nutrients and secondary chemicals were observed in defoliated unthinned stands. This suggests that deleterious effects of defoliation on sawfly fitness were offset by an increase in the foliar content of nitrogen, a primary compound known to improve larval growth in sawflies, which supports the hypothesis that thinning moderates negative effects of previous defoliation on sawfly fitness. The present study demonstrates that forestry practices that alter stand structure by reducing tree density may increase herbivory by affecting the way trees respond to insect attack, even after crown closure, with consequences on the buildup of herbivore populations in attacked trees.     

Tree mortality risk of oak due to gypsy moth

We present prediction models for estimating tree mortality resulting from gypsy moth, Lymantria dispar, defoliation in mixed oak, Quercus sp., forests. These models differ from previous work by including defoliation as a factor in the analysis. Defoliation intensity, initial tree crown condition (crown vigour), crown position, and species grouping classes were highly significant in categorical analysis of variance for mortality. Heavy defoliation intensity was shown to have a strong, consistent influence in increasing the probability of tree mortality. Classification and Regression Tree (CART) analysis, a binomial decision tree procedure, was used to develop prediction models of mortality risk for use by forest managers. The best decision tree had 65 groups that correctly classified 75% of the live trees and 76% of the dead trees. Models were run separately by defoliation class and provided correct classifications between 63 and 78% of the trees. Forest land managers can use these models to assign probabilities of death for moderate and heavy defoliation intensity levels and compare predicted mortality to mortality of undefoliated trees to determine how gypsy moth defoliation will affect their stands. The probabilities can be used to develop marking guides Lased on projected defoliation levels for implementing silvicultural treatments to minimize gypsy moth effects in forest stands prior to infestation.     

Impact and recovery of western hemlock following disturbances by forestry and insect defoliation

Impact and recovery of juvenile western hemlock stands were measured 5 years after the end of an outbreak of western blackheaded budworm Acleris gloverana (Walsingham) (Lepidoptera: Tortricidae) and hemlock sawfly Neodiprion tsugae Middleton (Hymenoptera: Diprionidae) on Haida Gwaii, British Columbia, Canada. Overall, whole-tree mortality was infrequent (3.6–6.8%) but aggregated in a few, severely defoliated stands. Top-kill of the apical leader of surviving trees was more frequent (7.2–22.1%) and many of these trees had not recovered a growing dominant leader 5 years after the original damage. These severe impacts were most common in the largest trees in the youngest stands which had been spaced just before the outbreak. Radial growth was also reduced for 2 years following defoliation. However, mean annual ring width and basal area increment in surviving trees recovered quickly and was even enhanced in post-defoliation stands, particularly in spaced stands that had been moderately defoliated. Management recommendations include planning for likely outbreaks by avoiding creation of expanses of young regenerating stands of western hemlock during harvest and modifying spacing treatments when an insect outbreak is imminent to reflect the greater risk to trees <20 years of age.     

Forest edge quantification by line intersect sampling in aerial photographs

There is a need for accurate and efficient methods for quantification and characterisation of forest edges at the landscape level in order to understand and mitigate the effects of forest fragmentation on biodiversity. We present and evaluate a method for collecting detailed data on forest edges in aerial photographs by using line intersect sampling (LIS). A digital photogrammetric system was used to collect data from scanned colour infrared photographs in a managed boreal forest landscape. We focused on high-contrast edges between forest (height ≥ 10 m) and adjoining open habitat or young, regenerating forest (height ≤ 5 m). We evaluated the air photo interpretation with respect to accuracy in estimated edge length, edge detection, edge type classification and structural variables recorded in 20 m radius plots, using detailed field data as reference. The estimated length of forest edge in the air photo interpretation (52 ± 8.8 m ha⁻¹; mean ± standard error) was close to that in the field survey (58 ± 9.3 m ha⁻¹). The accuracy in edge type classification (type of open habitat) was high (88% correctly classified). Both tree height and canopy cover showed strong relationships with the field data in the forest, but tree height was underestimated by 2.3 m. Data collection was eight times faster and five times more cost-effective in aerial photographs than in field sampling. The study shows that line intersect sampling in aerial photographs has large potential application as a general tool for collecting detailed information on the quantity and characteristics of high-contrast edges in managed forest ecosystems.     

Responses of condensed tannins in poplar roots to fertilization and gypsy moth defoliation

We examined the effects of fertilization and gypsy moth defoliation on condensed tannin concentration (%CT) of hybrid poplar (Populus x canadensis cv 'Eugeneii') fine roots in the summers of 1997 and 1998. This factorial experiment included two defoliation treatments (defoliated and a foliated control) and fertilization treatments (100 kg nitrogen (N) ha(-1) and an unfertilized control). Gypsy moth (Lymantria dispar L.) populations were experimentally increased to obtain defoliation in the summers of 1996, 1997 and 1998; fertilization subplots were supplemented with NH4NO3 (100 kg N ha(-1)) in the spring of each year. Despite the severity of defoliation, the effects were small, and significant on only two sampling dates: in May 1997, when fine root %CT was 23% lower in the defoliated trees, and in November 1997, when trees in the defoliated unfertilized plots had 35% higher root %CT than trees in all other plots. Defoliation effects on root %CT did not follow the same seasonal pattern as defoliation effects on root starch content, N uptake capacity or leaf %CT. Regulation of root condensed tannin concentration appeared to be partially uncoupled from these traits. The small transient effects on root defense reflect the resilience of this early successional tree to severe early season defoliation.     

Long-term realised and projected growth impacts caused by autumn gum moth defoliation of 2-year-old Eucalyptus nitens plantation trees in Tasmania,Australia

Insect damage to production forests has the potential to reduce financial returns by retarding tree growth and causing mortality, however, long-term realised quantification of these losses is rare. In order to help elucidate economic damage thresholds for making spray decisions we capitalised on a natural outbreak of autumn gum moth, Mnesampela privata, in a 2-year-old Eucalyptus nitens plantation. Following the partial chemical control of this insect outbreak we measured the tree growth variables diameter at breast height over bark and height of five differing tree defoliation classes for 75 months following tree damage. At the end of this period a threshold model was fitted to describe the relationship between tree defoliation and realised tree wood volumes. The model revealed that realised stand wood volume was not significantly affected up until defoliation exceeded 60% and then declined sharply after this defoliation level was reached. Further support for this defoliation threshold was evident from multiple comparisons among defoliation classes that showed 50% defoliated trees did not have significantly different wood volume compared to more lightly defoliated trees, but did have significantly greater wood volume compared to trees that were 72% or more defoliated. To determine if the realised differences in wood volume resulted in differences in yield over a plantation rotation the E. nitens growth model NITGRO was used to on-grow trees to age 15 years for a ‘best case’ (type 1 growth response, constant growth rates from last inventory until harvest) and ‘worst case’ (type 2 growth response, divergent growth rates from last inventory until harvest) scenario. The threshold model was then fitted to the outcomes of both scenarios and the economic consequences of defoliation were clearly dependent on the growth function assumed.     

Photosynthetic responses of field-grown Pinus radiata trees to artificial and aphid-induced defoliation

The phloem-feeding aphid Essigella californica represents a potential threat to the productivity of Pinus radiata plantations in south-eastern Australia. Five- and nine-year-old field trials were used to characterize the effects of artificial and natural aphid-induced (E. californica) defoliation, respectively, on shoot photosynthesis and growth. Photosynthetic capacity (A(max)) was significantly greater following a 25% (D25) (13.8 μmol m(-2) s(-1)) and a 50% (D50) (15.9 μmol m(-2) s(-1)) single-event upper-crown artificial defoliation, 3 weeks after defoliation than in undefoliated control trees (12.9 μmol m(-2) s(-1)). This response was consistently observed for up to 11 weeks after the defoliation event; by Week 16, there was no difference in A(max) between control and defoliated trees. In the D50 treatment, this increased A(max) was not sufficient to fully compensate for the foliage loss as evidenced by the reduced diameter increment (by 15%) in defoliated trees 36 weeks after defoliation. In contrast, diameter increment of trees in the D25 treatment was unaffected by defoliation. The A(max) of trees experiencing upper-crown defoliation by natural and repeated E. californica infestations varied, depending on host genotype. Despite clear differences in defoliation levels between resistant and susceptible genotypes (17 vs. 35% of tree crown defoliated, respectively), growth of susceptible genotypes was not significantly different from that of resistant genotypes. The observed increases in A(max) in the lower crown of the canopy following attack suggested that susceptible genotypes were able to partly compensate for the loss of foliage by compensatory photosynthesis. The capacity of P. radiata to regulate photosynthesis in response to natural aphid-induced defoliation provides evidence that the impact of E. californica attack on stem growth will be less than expected, at least for up to 35% defoliation.     

Susceptibility of northern British Columbia forests to spruce budworm defoliation

Stand susceptibility to defoliation by spruce budworm, Choristoneura fumiferana (Clem.), was examined in the Fort Nelson area of the Prince George Forest Region of British Columbia. In a retrospective study, defoliation maps of the study area were overlaid onto British Columbia Ministry of Forests cover type maps using a geographic information system. Analysis of the combined data identified forest characteristics associated with increased susceptibility to defoliation by spruce budworm. These were stands where the leading species was white spruce (Picea glauca (Moench) Voss), or where spruce was associated with aspen (Populus tremuloides Michx. and P. balsamifera L.) in mixed stands. Susceptibility to defoliation also was related to site quality, level of crown closure and stand age. Spruce stands on medium quality sites (site index 15 to 25 m, at reference breast height age 50 years) were more susceptible than stands on both poor- and high-quality sites. When spruce was mixed with aspen, stands on higher quality sites were more susceptible to budworm attack than poor sites. Open stands, where crown closure was <50%, were more susceptible to attack by spruce budworm than closed canopy stands. Older stands (120–199 years) were more susceptible to budworm attack than younger stands (40–110 years). In defoliated plots monitored for 6 years, tree mortality and top-kill reached a maximum of 30.4 and 47.2%, respectively. The losses varied with level of defoliation and were reduced by applications of the biological pesticide Bacillus thuringiensis.