Attack Dynamics, Impact and Biology of Pissodes Terminalis Hopping, in Regenerating Lodgepole Pine Stands

ATTACK DYNAMICS, IMPACT AND BIOLOGY OF PISSODES TERMINALIS HOPPING, IN REGENERATING LODGEPOLE PINE STANDS

L.E. Maclauchlan B. Sc., University of Victoria, 1980 M.P.M., Simon Fraser University, 1986

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in the Department of Biological Sciences

Lorraine E. Maclauchlan 1992 SIMON FRASER UNIVERSITY August, 1992

Name: LORPtAHNE E. MIACLAUCHLAN Degree: Doctor of Philosophy

Title of Thesis: ATTACK DYNAMICS, IMPACT AND BIOLOGY OF PIISSODES TERMIXALXS HOPPING IN REGENERATING STANDS OF LODGEPOLE PIPIE

Examining Committee: Chair: Dr. N.1-I. Haunerland, Assistant Professor

m~orden,Professor, Senior ~uperviso< Department of Biological Sciences, SFU

Dr. B. Roitberg, Associate Professor, Departmat nf @io$gical Sciences, SFU

Dr. R.k! Alfaro, Resea For ry Canada, Paci V'A t ia, B.C.

Public &+niner

br. E.A. ~arnero6~~rofes6r,. Department of Entomology, Penn State University, University Park, Pennsylvania External Examiner PARTIAL.COPYRIGHT LICENSE

I hereby grant to Slmon Fraser Unlverslty the right to lend my thesis, proJect or extended essay'(the ?ltle of whlch Is shown below) to users of the Slmon Fraser Unlverslty ~lbrir~,and to make partial or single coples on1 y for such usors or In response to a request f tom tho library of any othor unlverslty, or other educational Instltutlon, on

its own behalf or for, one of Its users. I further agree that permission for multlple copylng of thls work for scholarly purposes may be granted by me or the Doan of Graduate Studios. It is understood that,copylng or publlcatlon of thl s work for flnanclal galn shall not be allowed without my wrltten permlsslon.

Tltle of Thesls/Project/Extended Essay

Attack dynamics, impact and biology of Pissodes terminal i s Howina.

in regenerating lodgepole pine stands

Author: - ., (s lgnature)

~:Ix. 1992 (date) ABSTRACT

The impact of Pissodes terminalis Hopping on immature lodgepole pine, Rnus contorta var. latifolia Engelm. is greatest in the dry, low elevation Interior Douglas-fir zone (IDF), where up to 31 and 17% of the annual height growth is lost in the first and second years after attack, respectively. In the Montane spruce (MS) zone 25 and 14% of annual height growth is lost in the same two years. An increase in the area potentially available (APA) to a tree can cause impact of the weevil in the MS zone to approach that in the IDF. The cool, high-elevation Engelmann spruce-subalpine fir zone has the lowest hazard. The natural distribution of lodgepole pine approaches an aggregated pattern, as described by negative values of the Clark-Evans-Donnelly statistic (CED), as does the spatial pattern of attack by P. terminalis, with CED values ranging from -7.127 (P < 0.001) in the lowest density, IDF plots, to -3.98 1 (I?< 0.00 1) in the highest density, MS plots. As density decreases in the IDF and MS zones, the probability of individual stems being attacked by P. terminalis increases, with one stand in the MS having a density of < 800 stems per ha sustaining > 70% of stems attacked. Increases in the APA lead to increasingly severe defects after attack. For 161 defects caused by P. terminalis, followed for three growing seasons, the Tau-b statistic was 0.602, indicating little change in defect category, although 11.8 % increased in severity and 16.8 % decreased in severity. Various phenomena were disclosed that could affect the impact of P. terminalis, including variations in the life cycle between biogeoclirnatic zones, apparent resistance against the weevil by vigorous trees, and the possible existence of an oviposition deterrent pheromone. A hazard rating system was developed based on biogeoclimatic zone, stand age and density, APA, and tree girth. In high hazard stands that are thinned, 70% of stems could be weevilled. Silvicultural recommendations are to "feather" clumps of lodgepole pine during spacing, and in stands aged 5 to 20 years, to increase the APA gradually, removing weevilled stems at each entry. DEDICATION

To Connie Maclauchlan and W. M. Maclauchlan . ACKNOWLEDGEMENTS

I would like to thank Dr. J.H.Borden for his patience, guidance and encouragement during the course of this study. I appreciate and thank my committee members Drs. R.1 Alfaro and B.D. Roitberg for their advice and review of various drafts and manuscripts. I also thank: M.C.M. Matteau, A.L. Carroll, T.L. McMullan, S. Sirvio and especially I. Moe (and all three dogs), for field and laboratory assistance; F. Bellavance for statistical assistance; and L.J. Chong, Dr. G. Gries and others in B6220 for helpful discussions; and Dr. George Harvey, and Mrs. P.M. Roden, Forest Pest Management Institute, Forestry Canada, Sault Ste. Marie, Ontario, for the isozyme collaboration work. I would also like to thank D.W. Hutcheson for his tolerance at work over the duration of this study. In particular, I thank J.R. Thompson, my husband, for his help measuring all those "little" trees, and for his understanding and support. The research was supported in part by a G.R.E.A.T. Award from the Science Council of B.C. and by the FederalIProvincial Forest Resource Development Agreement (FRDA I) Grant No. F-52-41-107. Dr. M.A. Hulme served as Liason Officer for this grant, and his guidance and advice is appreciated. TABLE OF CONTENTS .. Approval ...... ii ... ABSTRACT ...... iii DEDICATION ...... v ACKNOWLEDGEMENTS ...... vi List of Tables ...... ix .. List of Figures ...... xi1 I . INTRODUCTION ...... 1 A . Life history ...... -3 B . Damage...... -6 C . Host selection in the genus Pissodes ...... 7 D . Pheromones in the genus Pissodes ...... 8 I1. OBJECTIVES ...... 10 I11 . DESCRIPTION OF BIOGEOCLIMATIC ZONES ...... 13 IV . DETERMINATION OF THE INCIDENCE, IMPACT AND SPATIAL DYNAMICS OF P . TERMINALIS ...... 21 A . Sampling methodology ...... 21 B . Okanagan Falls: a case study ...... 28 1. Attack dynamics and impact in relation to stand age and tree size ...... 31 2 . Spatial attack dynamics ...... 50 C . Lac le Jeune: a case study ...... 62 1. Spatial attack dynamics and impact of P. terminalis in different stand densities ...... -64 2 . Attack dynamics in relation to stand age ...... 86 V . SPATIAL ATTACK DYNAMICS AND IMPACT OF P. TERMINALIS IN DIFFERENT BIOGEOCLIMATIC ZONES ...... 94

vii VI . INFESTATION PHENOLOGY. EMERGENCE PATTERNS AND FECUNDITY ...... 115 A . Methods ...... 115 B . Results and Discussion ...... 116 VII . THE SEARCH FOR A PHEROMONE MARKER ...... 137 VIII . HAZARD RATING ...... -142 IX . CONCLUSIONS ...... -161 X . APPENDIX I ...... -162 XI . APPENDIX I1 ...... -173 XI1. APPENDIX 111 ...... -180 XI11. APPENDIX IV ...... -188 XIV . REFERENCES ...... -205

viii LIST OF TABLES

Table 1. Description of biogeoclimatic zones sampled in this study...... 19

Table 2. Location, year surveyed and summary of strip surveys done in fifty regenerating stands in the Kamloops Forest Region from 1987-1989. Surveys are grouped into general geographic areas, by biogeoclimatic zone and subzone, and the range of stand density, stand age and P. terminalis attack incidence is noted for each subzone...... -22

Table 4. Location, size and attributes of eleven stem-mapped plots established in 1987- 9 1 in the Kamloops Forest Region...... -24

Table 5. The distribution of defect categories in the Okanagan Falls spaced plot, expressed as total trees as assessed in 1987 and then again in 1990. The Tau-b statistic is 0.602, indicating little change in defect category over time...... 35

Table 6. Height and APA, expressed as percentiles of all tree heights and APA's in a given year, of attacked trees in the Okanagan Falls spaced plot from 1982 to \ 1988...... -43

Table 7. Comparison of spatial distribution of attacked and unattacked trees in the two Okanagan Falls plots by two statistical measures, the size of Voronoi polygons (area potentially available=APA) and the nearest neighbor distance (NN). The mean total height and dbh of attacked and unattacked trees in the two plots are also compared...... 47

Table 8. Data from a random subsample of trees cut near the Okanagan Falls spaced plot comparing mean incremental height growth in the year of weevil attack and in the years immediately before and after weevil attack. Height loss is expressed as a percent (in brackets) of the total height increment potential of unattacked growth years...... -49

Table 9. Distribution of P.teminalis attacks in the four Lac le Jeune plots, showing number of attacks per tree, attacks per hectare, and total percent stems attacked over the life of the stand...... 63 Table 10. Comparison of mean diameter at breast height (1.3 m), mean height of trees and mean height of weevil attacks (current and past) in the four Lac le Jeune plots...... 65

Table 11. Frequency of defect types and attack statistics from a sub-sample of trees from the four Lac le Jeune plots...... 66

Table 12. Comparison of annual increment of attacked and unattacked trees, expressed as percentiles, in the year immediately preceeding attack, the year of attack and the year after attack in the four Lac le Jeune plots. Data taken from a subsample of felled trees in each plot...... 71

Table 13. Incremental height growth of trees in the four Lac le Jeune plots combined (all spacings) in the year preceeding attack, the year of attack and the year after attack by P. terminalis (N=42). Data were gathered from a subsample of trees felled and measured from each spacing regime...... 72

Table 14. Comparison of mean annual height, diameter at 1.3 m (dbh), APA and NN distance of attacked and unattacked trees in all four plots at Lac le Jeune, measured in the summer of 1988...... -78

Table 15. Clark-Evan' s-Donnelly (CED) statistics for each year's P. terminalis attack in the four Lac le Jeune plots. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values 'r indicating randomness...... -8 1

Table 16. Summary of tree characteristics, attacked and unattacked by P. terminalis, in strip surveys from selected biogeoclimatic zones throughout the Kamloops Forest Region...... -87

Table 17. P. terminalis attack frequency at different stand densities in four biogeoclimatic subzones...... -95

Table 18. Comparison between P. terminalis attacked and unattacked trees in four plots from different biogeoclimatic subzones and of different stand densities...... 99

Table 19. Comparison between three biogeoclimatic zones of the number of P. terminalis attacks per tree, attacks per ha, average stem density, and the defect type resulting from the attack. The defects were coded 0,1,2,3 or 4 according to the defect exhibited, with 0 = no attack, 1=crease, 2 =crook, 3 =fork and 4=staghead...... 110

Table 20. Numbers and categories of accumulated defects in three plots, noting the change in defect over time expressed as the Tau-b statistic...... 113 Table 21. Fecundity of newly emerged, first year P. terminalis, and overwintered adults, collected from trees in early May, 1989, near Allendale Lake, B. C. Oviposition occurred from May through September, 1989...... I17

Table 22. Dimensions, and frequency of successful emergence by P. terminalis from 2,073 lodgepole pine leaders, collected from 1986 through 1989 in various locations throughout the Kamloops Forest Region...... I32

Table 23. Results of 2- and 3-choice bioassay testing oviposition and feeding preferences of ovipositing first-year female P. terminalis (1988 and 1991) and overwintered females (1989). The duration of choice bioassays was 24 h in 1988 and 1989 (48 h total) and 48 h in 1991 (96 h total)...... 139 LIST OF FIGURES

Figure 1. Map of the Kamloops Forest Region indicating the location of study sites, with the inset showing the location of the Kamloops Forest Region within British Columbia. Numbered study sites are as follows: 1) Laluwissin Creek; 2) Maka Creek; 3) Ketchan Creek; 4) Dillard Creek; 5) Peachland Main; 6) Okanagan Falls; 7) Ellis Creek and Allendale Lake; 8) Daves Creek; 9) Monte Lake; 10) Stump Lake; 11) Lac le Jeune, Chewell's Mountain and Cornwall Lake...... I5

Figure 2. Climatic regions of B.C. with the Southern Interior Dry Region, which encompasses most of the Kamloops Forest Region and all of the study sites, shown in black. The inset at lower right displays the elevational sequence of the biogeoclimatic zones in the Southern Interior Dry Region, with the three zones used in this study shaded in black. The zones illustrated in the inset are as follows: Alpine tundra (AT), Engelmann spruce-subalpine fir (ESSF), Montane spruce (MS), Interior Douglas-fir (IDF), Ponderosa pine (PP) and Bunchgrass (BG)...... 17

Figure 3. Yearly rate of attack by Pissodes terminalis, expressed as percent of total trees, in two plots at Okanagan Falls, 1980-1989. Spacing of trees was done in 1984...... ,32

Figure 4. Frequency of four defect categories caused by Pissodes terminalis attack on 125 trees from 1982-1989 in the Okanagan Falls spaced plot and on 27 trees from 1980-1991 in the Okanagan Falls unspaced plot...... 37

Figure 5. Total tree height (upper graph) and area potentially available (APA)(lower graph), expressed as percentiles, of trees attacked and not attacked, for 1987, in the Okanagan Falls spaced plot. The number of trees falling into each percentile group, expressed as a percent of total, is plotted with the mean heights and APA's indicated by the vertical arrows...... -41

Figure 6. Frequency distribution of P. terminalis attacked and unattacked trees in five height classes in the Okanagan Falls spaced plot...... 45

Figure 7. Spatial arrangement of all trees, both those attacked at least once between 1982 and 1988 by Pissodes temzinalis and those not attacked, in the Okanagan Falls spaced plot. Each tree is defined by a Voronoi polygon which represents its "area potentially available" (APA). The Clark-Evans-Donnelly statistic (CED) is given below the figure for "all the trees" in the plot and for "all attacked trees" (shaded areas). Values equal to or < 0 approach a clumped

xii pattern, and >2 approach a more regular pattern, with intermediate values indicating randomness...... 52

Figure 8. The number of trees in the spaced and unspaced Okanagan Falls plots having no attacks, 1 attack, 2 attacks or 3 attacks per tree. The percent of total trees in each plot in each attack category is shown above each bar...... 54

Figure 9. Spatial arrangement of all trees, both those attacked at least once between 1980 and 1988 by Pissodes terminalis and those not attacked, in the unspaced Okanagan Falls plot. Each tree is defined by a Voronoi polygon which represents its "area potentially available" (APA). The Clark-Evans-Donnelly statistic (CED) is given below the figure for "all the trees" in the plot and for "all attacked trees" (shaded areas). Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness...... 56

Figure 10. Spatial pattern of Pissodes terminalis-attacked trees over time in the Okanagan Falls spaced plot showing the value of the Clark-Evans-Donnelly (CED) statistic for each year. In all years the pattern of attack approaches a "clumped" distribution (e.g ., CED =- 1.655, P =0.097 indicates a 9.7 % level of significance for a two-tailed test). Voronoi polygons have been drawn around attacked trees to help illustrate the shifting pattern and density of attack. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness...... 58

Figure 11. Spatial plots of attacked trees in the Okanagan Falls spaced plot. Each successive graph incorporates an additional year of attacked trees. The Clark- Evans-Donnelly (CED) statistic is calculated for each graph with a description of attack pattern...... -60

Figure 12. Annual and cumulative percent of stems attacked by Pissodes terminalis in the four plots at Lac le Jeune, B.C...... 69

Figure 13. Percent height loss in the year of attack by P. terminalis and in the following growing season. Height growth of trees not attacked in a given year is compared to the length of the compensating lateral of trees attacked in that same year and in the year following attack...... 75

Figure 14. Stem map of all trees in each of the four Lac le Jeune plots with Voronoi polygons drawn around each tree designating APA. Attacked trees are designated by shaded polygons. The CED statistic for all the trees in each plot and for attacked trees in each plot is given. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness...... 79

xiii Figure 15. Distribution of defects, grouped by type, in the four Lac le Jeune plots. Bars within a type with the same letter above them are not significantly different (Tukey ' s test, P < 0.05)...... 84

Figure 16. Plot of annual Pissodes temnalis attack in 4 spacing regimes in a stand at Lac le Jeune. P > 0.05 for all lines. Regression equations are as follows: unspaced, y=0.30x-1.85; 2.4 m spacing, y=0.54x-3.24; 3.0 m spacing, y =0.49x-3.14; and, 3.7 m spacing, y =0.69x-4.19...... 88

Figure 17. A 3-dimensional linear regression plot of the percent stems attacked in relation to the average age of stands and stand density (r2=O.57, F = 16.62, df =25). Regression equation is, y =-0.001~~ + 1. 75x2 +O. 66. Each point represents a unique survey in a different location, from all three biogeoclimatic zones, in the Kamloops Forest Region...... 90

Figure 18. Frequency distribution of four defect types (forks and stagheads combined into one category) in six spaced stands and five unspaced stands located in the IDFdkl, MSdml, MSxk, and ESSFdcl subzones...... 97

Figure 19. Stem plots of all trees in the Dillard Creek spaced plot, upper diagram, and unspaced plot, lower diagram, (MSdml) with Voronoi polygons drawn around each tree to designate APA (area potentially available). Attacked trees are represented by the shaded polygons. The CED (Clark-Evans-Donnelly statistic) for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness...... I01

Figure 20. Stem plot of all trees in the Ketchan Creek unspaced plot (MSdml), with Voronoi polygons drawn around each tree to designate APA. Attacked trees are represented by the shaded polygons. The CED for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness...... -103

Figure 21. Stem plot of all trees in the Conkle Lake unspaced plot (MSdml), with Voronoi polygons drawn around each tree to designate APA. Attacked trees are represented by the shaded polygons. The CED for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness...... -105

Figure 22. Stem plot of all trees in the Ellis Creek, spaced plot (MSdml), with Voronoi polygons drawn around each tree to designate APA. Attacked trees are

xiv represented by the shaded polygons. The CED for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values

M indicating randomness...... I07

Figure 23. Mean fecundity per day of 17 female P. terminalis. Assessment every 3-5 days, beginning on 8 May 1989, and ending on 11 November 1989. Bars represent mean oviposition per female per day for a 3-5 day assessment period. 119

Figure 24. Mean numbers of days in 1989 from emergence to start of oviposition for P. terminalis females from three biogeoclimatic subzones. The number of weevils assessed (N) is indicated for each biogeoclimatic zone. Means followed by the same letter are not significantly different (Tukey 's test, P < 0.05). ... .I22

Figure 25. Frequency distribution of number of eggs per oviposition puncture, upper graph, and the relationship of number of feeding punctures to number of eggs deposited per 3-5 day period, lower graph. Data from 17 pairs (male and female) of P. terminalis, collected from Allendale Lake (ESSFdcl) on 2 May 1989, and allowed to feed and oviposit on lodgepole pine terminal sections over a five month period...... 125

Figure 26. Data from leaders collected from 4 biogeoclimatic subzones in 1989 and held in the laboratory at 20•‹C: a) mean length of infested leaders (+ S.E.); b) percent of infested leaders in which weevils were parasitized (by one or more species of parasite) and the percent of infested leaders which had successful P. terminalis emergence; c) mean number of weevils emerging per infested leader (+ S.E.); and d) percent leaders with secondary attack by Magdalis gentilis. N=250,57, 59 and 92 for MSdml, MSxk, IDFdml and ESSFdcl, respectively. Bars with tthe same letter are not significantly different (Tukey's test, P < 0.05)...... -127

Figure 27. Summary of leader characteristics, weevil emergence and parasitism over four years in collections from Okanagan Falls, B.C.: a) leader dimensions; b) number of weevils emerging per infested leader; c) percent of infested leaders with one or more weevils emerging; and d) percent of infested leaders with parasitized weevils. N=84, 59, 62 and 126 for 1986, 1987, 1988 and 1989, respectively...... -129

Figure 28. a) Pissodes terminalis emergence, b) incidence of parasitism, and c) secondary infestation by Magdalis gentilis, in 1986-attacked lodgepole pine leaders collected in 1987 from two biogeoclimatic zones and two subzones within each zone...... 134 Figure 29. Pooled data on dbh percentiles for unattacked trees and those attacked by P. terminalis in all plots in the IDF and MS biogeoclimatic zones. The mean dbh percentiles for attacked and unattacked trees are indicated by vertical arrows...... -147

P Figure 30. The relationship between APA and density of attacked and unattacked trees in the IDF and MS zones is illustrated. The lines are fitted using a log regression...... -150

Figure 31. Hazard rating guide for stands based on biogeoclimatic zone, APA, and age. Hazard is divided into low, moderate, or high in each zone dependent on stand age and mean APA...... 152

Figure 32. Area graphs showing the number of trees, both attacked by P. terminalis and unattacked, falling into each dbh percentile range in two locations in the IDF biogeoclimatic zone (Okanagan Falls spaced and unspaced, and Ketchan Creek unspaced). The mean dbh percentile for attacked and unattacked trees is indicated by a vertical arrow...... I54

Figure 33. Stand risk is represented by a 3-dimensional plot of attack probability, APA and biogeoclimatic zone. The biogeoclimatic zones are: 1 =IDF; 2 =MS ; and, 3=ESSF...... I55

xvi I. INTRODUCTION

The lodgepole terminal weevil, Pissodes terminalis Hopping, is one of the most commonly encountered insects in regenerating lodgepole, Pinus contorta var. latifolia Engelm., or jack pine, Pinus banksiana Lamb., stands in western Canada. Due to the relatively short history of harvesting in these pine forests and consequent lack of knowledge in managing the new stands, the impact of the weevil as well as that of other damaging agents is not fully understood. With the increasingly large areas of regenerating lodgepole pines, and accompanying silvicultural treatments, there is concern that P. terminalis will become a serious problem in certain situations. Increased utilization and management of lodgepole pine in western North America has drawn attention to high, although often localized, incidences of damage from insects, disease and small mammals (Bella 1985a). In order to maximize the volume production and reduce the rotation age, stand density is often controlled through early thinnings. It is in these low density stands that P. teminalis attack has been highest and appears to cause the most damage, despite conflicting reports (Stevens and Knopf 1974; Furniss and Carolin 1977; Maher 1982). Stevens and Knopf (1974) reported that dense stands are infested most frequently., In Bella's (1985a) report, damage from P. terininalis as well as other agents was greatest on good sites in the most open treatments of about 500 to 1000 stems per hectare. The comparative level of P. terminalis in different densities and a quantification of the damage caused in these various densities is not clear. For these reasons, a hazard rating system for regenerating lodgepole pine and viable pest management strategies for the weevil are needed. The lodgepole terminal weevil attacks three varieties of lodgepole pine, P. contorta var. latifolia, P. c. murrayana, P. c. bolanderi, and jack pine, P. banksiana, throughout its range in North America. In Canada, collections of P. terininalis have been made in British Columbia, Alberta, Manitoba, Saskatchewan (Drouin et al. 1 1963), and the Northwest Territories (Stevenson and Petty 1968). Others have recorded the weevil in immature lodgepole pines in the Sierra Nevada from Yosemite park north into Oregon (Salman 1935), in Wyoming and South Dakota (Furniss and Carolin 1977), Washington, Oregon and Idaho (Stark and Wood 1964), and Colorado (Stevens and Knopf 1974). P. terminalis has recently been especially damaging to young lodgepole pine stands in the Cariboo Forest Region located in the central interior of British Columbia (Maher 1982). In the Kamloops Forest Region, lodgepole pine is particularly important, as it accounts for almost 65 % of the forest cover, totalling 1,357,405 ha. In terms of mature volume, almost 79% of the Canadian inventory of lodgepole pine is found in B.C. The area of productive forest land in British Columbia, which includes all Crown land under mature forests, immature forests, NSR (not satisfactorily restocked) and NC (non-commercial) brush, totals 42.5 million ha1. The area of immature forests in B. C. totals 14.3 million ha, 34 % of the productive forest land base. The area of immature pine forest in the Kamloops Forest Region is approximately 430,562 ha, about 25.5% of the total immature forest area. British Columbia produces 46% of Canada's softwood volume, and of this 25 % is composed of lodgepole pine. This represents the largest percentage of any one species. B.C. 's interior forests, which include the Kamloops Forest Region, contribute 66% of the total harvest, or allowable annual cut (A.A.C.), of the province. Only in the past 20 years has lodgepole pine become a substantial portion of the provincial cut. Much of the increased harvesting of lodgepole pine is due to outbreaks of mountain pine beetle, Dendroctonus ponderosae Hoplcins, and the subsequent control and salvage harvesting. Currently, within the Kamloops Forest Region, in the southern portion of the Okanagan Timber Supply

British Columbia forest industry fact book. 1989. Council of Forest Industries, Vancouver, B. C.

2 Area, over 80% of the A.A.C. is in pine types to control the mountain pine beetle. The result of these intense harvesting efforts to reduce mountain pine beetle populations have resulted in large areas of young, pure stands of lodgepole pines. Many of these young stands are now becoming susceptible to P. temzinalis, and many more will follow.

A. Life history

Most commonly, P. temzinalis completes one generation per year and its life cycle is closely synchronized with leader phenology. It is the only member of its genus that consistently oviposits in the expanding terminal shoots of its hosts (Stark and Wood 1964). This habit differs from that of the white pine weevil, Pissodes strobi (Peck), which oviposits in the previous year's leader, causing its death as well as that of the current year's growth (McMullen 1976a). There are other marked differences in life cycles and behavioral patterns between these two species (Salman 1935), some of which could possibly play an important role in future management of P, terminalis. The lodgepole terminal weevil overwinters mainly as larvae in infested leaders (Stevens and Knopf 1974) during their first year, whereas the white pine weevil overwinters primarily as adults in the litter (Rose and Lindquist 1977) or on foliage (Kline and Mitchell 1979). However, Maher (1982) noted from observations made in the Cariboo that the lodgepole terminal weevil can also overwinter as pupae or adults. Kovacs and McLean (1990) observed 95 % of a study population of P. terminalis from one site in the Kamloops Forest Region to overwinter as fourth instar larvae; the remaining 5 % overwintered as pupae or adults. Adult weevils have low survival rates when they overwinter in the terminals. Adults are also thought to overwinter in the duff (Furnisss and Carolin 1977), as does P. strobi, where their survival would be much higher. In contrast, in jack pine forests, only the adults appear to overwinter (Drouin et al. 1963), and their overwintering site is thought to be in the duff. 3 The most recent work on the life history of P. terminalis (Cameron and Stark 1989) was done in California, and describes three types of life cycle. Types 1 and 2 are univoltine, with type 2 having three subtypes (Type 2A, B and C explained below), and type 3 is bivoltine. Cameron and Stark (1989) divide these life cycles into elevational ranges, with type 1 most common at altitudes < 2000 m, type 3 at altitudes averaging about 2500 m, and all types at 2000-2500 m. While these ranges must be adjusted to B.C.'s more northerly geographic location, an elevational cline is also evident in B.C. In the type 1 cycle, adults emerge from attacked leaders in the fall from eggs laid in elongating leaders in the summer of the same year, and presumably overwinter in the ground. Personal observations indicate that this cycle is rare in south central British Columbia. The type 2 life cycle is similar to type 1 except that overwintering takes place in the leaders, with fourth-instar larvae overwintering in type 2A, pupae in type 2B, and adults in type 2C. The type 2A life cycle was commonly found in southern B.C. @ersonal observations); occasionally, type 2B was also seen. In the type 3 life cycle, the first winter is passed as a third instar in the leader and the second as an adult, probably in the ground. This life cycle has been observed on some moist, high- elevation sites in the Kamloops Forest Region. Adult P. terminalis are active in late spring to early summer, when they can be found feeding on the tissues of the developing terminal shoot. Soon after this maturation feeding, mating occurs, and oviposition punctures are excavated in the bark of the new leader. Usually a single egg is deposited into each puncture, but up to three eggs have been found in a single puncture (Drouin et al. 1963; Stark and Wood 1964). Punctures can be found throughout the terminal and occasionally on the needles of lodgepole pine; however, on jack pine they are concentrated near the basal portion of the leader. Typically, oviposition punctures are located at the base of a needle fascicle, similar to P. strobi, suggesting a positive thigmotactic response in terms of close range host selection (Harris et al. 1990). Immediately after hatching, P. terminalis larvae feed in any direction in the phloem and cortex of the new growth for a brief period before becoming clearly negatively geotactic, after which they mine upwards toward the apical bud (Drouin et al. 1963). Usually the third instar moves into the pith where mining continues. Once in the pith, larvae mine upward toward the apical bud as well as downward toward the base of the leader. The larvae feed as individuals (Drouin et al. 1963), and occasionally cannibalize other larvae they may encounter during their feeding in the phloem. This behavior is opposite to that of P. strobi larvae, which feed gregariously downward in the cambium until pupation occurs (Belyea and Sullivan 1956; Wood and McMullen 1983). During the early stages of larval development, high mortality may occur from drowning in resin (Drouin et al. 1963). In many leaders all the larvae die before completing development (Drouin et al. 1963), but due to their spiral feeding pattern in the leader and mining of the pith the leader is killed despite unsuccessful weevil emergence. In the Kamloops Forest Region pupal chambers are constructed mainly in the pith and occasionally in the terminal bud. Drouin et al. (1963) also found pupal chambers in the xylem and phloem of jack pine. In central Saskatchewan, adult emergence occurs about mid-August to early autumn (Drouin et al. 1963). Feeding was the only activity observed after emergence, suggesting weevils overwintered once as adults and then commenced oviposition the following spring. Kovacs and McLean (1990) report first weevil emergence in south central B.C. to be mid- to late June with oviposition commencing as soon as two days after emergence. These varied observations md the work by Cameron and Stark (1989), which describes the variations in the life cycle of P. terminalis in California, reveal the need to elucidate the unique life history of the weevil in B.C. The longevity of lodgepole terminal weevil adults has not been studied in great detail, but the average longevity in the laboratory of 10 females (Kovacs and McLean 1990) was 112.8 days. Some white pine weevils may live and continue to reproduce for up to four years (McMullen and Condrashoff 1973). The possibility that P. terminalis could also live for more than one year should be considered when developing management strategies for the weevil because of the increased number of offspring that a weevil could produce if it lived and reproduced for 2 or more years.

B. Damage

Although there are slight behavioral and developmental differences in populations of P. terminalis breeding in jack and lodgepole pines, the injury caused as a result of P. terminalis attack is similar (Drouin et al. 1963). Numerous observers claim that the proportion of stems weevilled and the severity of the resultant stem deformity is greatest in low density stands (Keen 1952; Stark and Wood 1964; Stevenson and Petty 1968; Furniss and Carolin 1977; Maher 1982; Bella 1985a, 1985b). However, no study has quantified the defects caused by the weevil in stands of different densities. The weevil does preferentially attack long and thick terminal shoots, which would be most prevalent in a spaced stand exhibiting rapid growth. The most common result of weevil attack is a crook produced in the main stem, when a single lateral in the whorl directly below the attacked leader assumes apical dominance (Maher 1982), and a crease, which is a minor linear indentation with little or no stem curvature at the point of attack. However, if two laterals compete equally for dominance, a bifurcation of the main stem, or forked stem, results. A more severe stem deformity is seen when multiple attacks occur on the same tree, promoting the development of multiple leaders, or " stag-heads" (Stevens and Knopf 1974; Maher 1982). Occasionally a lateral subtending the one-year-old growth assumes apical dominance rather than a lateral directly below the new growth (personal observation). 6 The potential impact of the lodgepole terminal weevil includes minor volume loss due to height growth reduction and a lower grade of lumber as a result of grain aberrations at the site of a crook, to increased rotation age for the stand due to reduced height growth, and finally to greatly increased logging and manufacturing costs resulting from small-dimension, deformed logs from forked and stag-head trees (Maher 1982). Another Pissodes species, the yellow-spotted pine weevil, Pissodes nitidus Roelofs, is a serious pest of terminals on young Korean pine, Pinus koraiensis Sieb. and Zucc., in Northeast China (Liyuan 1989). This weevil can cause up to 20% stand volume loss after two to four repeated attacks with 0.25 to 0.33 m of height loss resulting for each attack (Liyuan 1989). P. nitidus attacks 1-year-old leaders of Korean pine, and has been recorded occasionally attacking several other Pinus species (Liyuan 1989). Similar to what has been observed with the North American Pissodes (Alfaro and Omule 1990), P. nitidus causes attacked trees to fork and crook at the point of attack. Liyuan (1989) suggests that management activities, such as release thinning, adversely affect infested stands as they enhance weevil survival. P. nitidus has highest survival rates (>50%) in stands with high light conditions and where relative humidity is low (Liyuan 1989). These observations on the bionomics and impact of P. nitidus parallel those of the North American leader-infesting Pissodes (Alfaro and Omule 1990).

C. Host selection in the genus Pissodes

Feeding behavior of phytophagous insects is characteristically mediated by host- produced chemical attractants (Jermy 1976). It has been concluded that there are various processes influencing the selection of a host by P. strobi. Visual orientation apparentiy plays an important role in initial selection of a host (VanderSar and Borden 1977a). In short-range host selection, vision is particularly significant (Harman 1975; VanderSar and Borden 1977a). In field and laboratory tests, both sexes of P. strobi oriented preferentially to the largest and most vertically oriented silhouettes (VanderSar and Borden 1977a; Wilkinson 1983). This choice has adaptive significance for reproduction because long, large-diameter leaders can accommodate larger weevil broods than small-diameter leaders. Similarly, P. terminalis seems to attack long, thick terminal shoots preferentially (Maher 1982). Successful weevilling of a vigorous host leader often results in a multiple-top crown, which would increase the availability of optimal oviposition sites (VanderSar and Borden 1977a) and thus favor rapid population growth. Once P. strobi has oriented visually to a host, the host is accepted or rejected as suitable for feeding andlor oviposition activity. Alfaro et al. (1980) found that certain monoterpenes act as synergists to non-volatile chemicals in the bark to enhance feeding, while other compounds completely deter feeding. In interpreting the results of feeding bioassays using host and nonhost plants, Alfaro and Borden (1982) hypothesized that acceptance of a host by P. strobi is probably mediated by specific levels or blends of feeding stimulants. It is possible that similar mechanisms of host acceptance and rejection are employed by P. terminalis. An understanding of host selection by P. terminalis could be of major importance in designing efficient traps for the weevils as well as in detecting and selecting resistant lodgepole pine genotypes.

D. Pheromones in the genus Pissodes

Two related volatile compounds, grandisol (cis-2-isopropenyl-l- methylcyclobutanethanol) and its corresponding aldehyde, grandisal, have been isolated from abdomens and hindguts of male P. strobi and P. nemorensis Germar (=P. approximatus, Phillips et al. 1987) (Fontaine and Foltz 1982; Phillips et al. 1984). Males of both P. strobi and P. nemorensis produced grandisol and grandisal only at times when cohort females were reproductively mature (Booth et al. 1983). The two volatiles have been implicated as aggregation pheromones for P. nernorensis, and although P. strobi produces grandisol and grandisal, P. strobi males apparently produce an allelochemic signal that interupts the response of P. nemorensis to its natural or synthetic aggregation pheromone (Phillips and Lanier 1986). As grandisol and grandisal are two of the four aggregation pheromones in the boll weevil, Anthonomus grandis Boheman (Tumlinson et al. 1969), they may occur in other curculionids as well. In support of this hypothesis, boll weevil pheromone compounds have been reported to be attractive to the pecan weevil, Curculio caryae Horn (Hedin et al. 1979) and the New Guinea sugarcane weevil, Rhabdoscelus obscurus (Chang and Curtis 1972). However, P. teminalis is a hybrid species, between P. strobi and P. schwarzi (=yosemite) ~opkins(Drouin et al. 1963), and it is uncertain which parental species might have the greatest impact. P. strobi apparently does not rely on pheromone communication (Phillips et al. 1986). Smith and Sugden (1969) were uncertain in separating P. nernorensis from P. schwarzi, which both have been found breeding in Pinus contorta root collars, and finally only distinguished them cytologically. P. nernorensis definitely communicates with pheromones (Booth et al. 1983) and P. schwarzi could have a similar habit. It is therefore possible that, like the pine-infesting P. nemorensis in the southern United States, P. teminalis produces and utilizes sex pheromones. If P. terminalis produces and uses a pheromone, the identification and synthesis of this pheromone might have considerable utility in the management and control of the species. 1. The main focus of this study was to determine the spatial distribution of hosts, both attacked and unattacked by P. terminalis, and the relationship of host density to attack dynamics. Attack dynamics were compared among ecosystems to elucidate any differences due to biogeoclimatic zone or subzone which could be used to develop a hazard rating system for lodgepole pine. Characteristics of the host, such as height, diameter, age and leader dimensions, were analyzed as to their role in host selection. To investigate aspects of this objective further, a long- term spacing trial was established in three biogeoclimatic zones (Appendix I). 2. The second objective was to investigate the various aspects of host selection by P. terminalis. The quality of the host, in terms of physical damage, or presence/absence of feeding or oviposition by P. terminalis, could influence host selection. The role of host quality, as described above, and the possibility of semiochemical communication were tested in a series of bioassays. The susceptibility, resistance or preference of various conifer species by P. terminalis could also lead to a better understanding of the resistance mechanisms of lodgepole pine and other potential hosts (Appendix 11). 3. The third objective of this study was to define the life history and habits of P. terminalis in southern British Columbia. A better understanding of the biology and host selection behavior of this insect would help predict attack patterns and aspects of host selection by P. terminalis. Knowledge of the population dynamics and variance in insect biology in the various ecosystems of interior B.C. could lead to the development of a hazard rating system for young lodgepole pine stands. 4. The final objective of this study was to use the knowledge gained in meeting the above objectives to develop a hazard rating system for immature lodgepole pine stands in the south central interior of British Columbia, incorporating into the system host selection parameters, stand density, population dynamics and the prediction of impact in terms af stem quality. Eventually, other insect pests, such as P. schwarzi, pathogens and other damaging agents could be inco~oratedinto such a hazard rating framework.

Due to the increasing pressure on forest managers to find alternatives to chemicals, any non-chemical form of insect or disease control would be of value. As with the Sitka spruce weevil, P. strobi, there are no acceptable methods of controlling the lodgepole terminal weevil, P. terminalis. Some experiments using chemical insecticides have been successful for P. strobi (Johnson 1965; Silver 1968) and would most likely work for P. terminalis, but such treatments would be costly. Possible alternative methods of control are the enhancement of natural enemies and use of various silvicultural methods. Hulme et al. (1986,1987) evaluated the potential for enhancing the natural enemies of the Sitka spruce weevil. Early work on P. strobi indicated that dense stands and shaded habitat are unfavorable for weevil development (Wallace and Sullivan 1985); a current area of research is the manipulation of stands to make them less favorable for weevil development or less susceptible to weevil damage (Steill 1979; Steill and Berry 1985; Mclean 1989; Alfaro and Omule 1990). Clipping and removal of attacked leaders has been recommended as a control strategy for P. strobi, and P. nitidus (Liyuan 1989) in China. Therefore, although not a primary objective of my thesis, the potential of leader clipping as a control for P. terminalis in lodgepole pine stands in B. C. was investigated (Appendix 111). Another weevil that has been observed in young managed stands of lodgepole pine in interior B.C. is P. schwani. Little is known of the biology and behavior of this weevil or of its pest status, and it is easily confused visually with P. terminalis. Because P. schwani was frequently encountered during my field studies, I took the opportunity to investigate aspects of its biology, pheromone communication, and genetic relationship with P. terminalis (Appendix IV) . The B.C. Ministry of Forests' biogeoclimatic ecosystem classification system is widely used in B.C. and gives foresters, biologists and other resource managers a common framework for developing, comparing and communicating management strategies. The biogeoclimatic classification system thus provides a basis for ecosystem management and other practical.decision making. The biogeoclimatic classification system developed by Dr. V.J. Krajina (1965) and his students was adopted by the B.C. Ministry of Forests in the mid-1970's because its hierarchical structure makes it well suited for provincial, regional and site-specific interpretations. It incorporates both biotic and environmental factors, and is therefore applicable to many resource uses (Lloyd et al. 1990). Zones are generally named after one or two dominant climatic climax tree species (Lloyd et al. 1990). A series of connotatively meaningful, climatically based subzone names and symbols are used to distinguish the subzones. In the interior of B. C., these names correspond to the precipitation and temperature regimes of the subzone, relative to other subzones in that zone (Lloyd et al. 1990; Meidinger and Pojar 1991). Two lower-case alphabetic characters are used to denote individual subzones. The first character refers to precipitation and the second to temerature (Lloyd et al. 1990). For interior B.C., five precipitation terms and six temperature terms are used in various combinations to portray the relative climatic regime of each subzone. The terms are:

Precipitation Regime Temperature Regime x - very dry (xeric) h - hot d-dry w - warm m - moist m - mild w - wet k - cool v -very wet c - cold v - very cold indicate a variant. Recognized variants are numbered geographically from south to north (Lloyd et al. 1990). Variants further reflect differences in regional climate and are generally recognized for areas that are slightly drier, wetter, snowier, warmer, or colder that other areas in the subzone (Meidinger and Pojar 1991). Within the Kamloops Forest Region (Fig. l), the biogeoclimatic zones (Krajina 1965) in which lodgepole pine is found includes the Interior Douglas-fir zone (IDF), Montane Spruce zone (MS), Interior Cedar-Hemlock zone (ICH) and Engelmann spruce-subalpine fir zone (ESSF) (Fig. 2). Due to a very low incidence of P. terminalis in the pine types within the ICH, study sites were not selected from this zone. Characteristics of sites in three subzones from each of the IDF (IDFdkl, IDFdk2 and IDFdml), and MS (MSdml, MSdm2 and MSxk) and one subzone in the ESSF (ESSFdc 1) are compared in Table 1. The IDF is characterized by a warm, dry climate, with a relatively long growing season in which moisture deficits are common (Lloyd et al. 1990; Meidinger and Pojar 1991). Typically, the IDF occurs at elevations below the Montane Spruce zone and, where the valleys are deep enough, above the Ponderosa Pine zone. The IDFdkl occurs from 1130-1460 m in the central Thompson Plateau, the Clear Range and the Similkameen and Ashnola Drainages. The IDFdkl occurs below the MSxk subzone, and below the IDFdkl there is no lodgepole pine. It is the coldest IDF subzone in the Region, being slightly cooler and drier than the IDFdk2. The climax tree species is Douglas-fir, Pseudotsuga menziesii (Mirbel) Franco, with lodgepole pine a sera1 species on zonal sites. The IDFdkl changes directly into the MSxk at elevations of 1450-1650 m. The MS zone has cold winters and moderately short, warm summers. The MSxk is the driest MS subzone in the Kamloops Forest Region and moisture deficits commonly occur throughout the growing season

, (Lloyd et GZ. 1990).

I Figure 1. Map of the Kamloops Forest Region indicating the location of study sites, with the inset showing the location of the Kamloops Forest Region within British Columbia. Numbered study sites are as follows: 1) Laluwissin Creek; 2) Maka Creek; 3) Ketchan Creek; 4) Dillard Creek; 5) Peachland Main; 6) Okanagan Falls; 7) Ellis Creek and Allendale Lake; 8) Daves Creek; 9) Monte Lake; 10) Stump Lake; 11) Lac le Jeune, Chewell's Mountain and Cornwall Lake.

Figure 2. Climatic regions of B.C. with the Southern Interior Dry Region, which encompasses most of the Kamloops Forest Region and all of the study sites, shown in black. The inset at lower right displays the elevational sequence of the biogeoclimatic zones in the Southern Interior Dry Region, with the three zones used in this study shaded in black. The zones illustrated in the inset are as follows: Alpine tundra (AT), Engelmann spruce-subalpine fir (ESSF) , Montane spruce (MS) , Interior Douglas-fir (IDF) , Ponderosa pine (PP) and Bunchgrass (BG).

Table 1. Description of biogeoclimatic zones sampled in this study.

Biogeoclimatic zones and subzones IDF MS ESSF Characteristics dkl dk2 dm1 dm2 dm1 xk dcl

% area of Kamloops Forest Region 7.6 5.5 0.6 4.8 1.O 6.0 1.O w \O Elevation (m) 1130- 1460 600-1300 560- 1300 1275-1530 1300-1600 1450-1650 1600-1950

Climax speciesa Fd Fd Fd Sxw, BI Sxw, B1 Sxw, B1 Se, B1

Annual precipitation (mm) 438 568 505 606 638 444 no record

Annual mean temperature (OC) 3.4 4.1 3.8 2.8 3.2 3.1 2.0

a Fd =Douglas-fir; Sxw =hybrid white spruce; B1= subalpine fir; Se=Engelmann spruce. The IDFdk2 is located higher in elevation than the IDFdkl having an elevational range of 600-1300 m and occurs west of Lillooet, to a small extent along the North Thompson River, and northeast of Princeton, along Hayes and Trout Creeks to Summerland. It is moister and warmer than the IDFdkl and lodgepole pine is present on its zonal and wet sites (Lloyd et al. 1990). Above the IDFdk2, the MSdm2 is characterized as moister than the MSxk and drier than the MSdml. Most of the lodgepole pine occurs on the zonal and dry sites, with scatterings on the wet sites. The IDFdml subzone lies below the MSdml at 560-1300 m on the east side of the Okanagan Valley from Kelowna to Osoyoos and in the Kettle River drainage (Lloyd et al. 1990). Lodgepole pine occurs mainly on the zonal sites. The MSdml occurs at 1300-1600 m (Table I), above the IDFdml and below the ESSFdcl . The MSdml is slightly warmer and moister than the MSxk. Seral lodgepole pine stands are prevalent on zonal, wet and dry sites in this subzone. The ESSFdcl occurs on the upper slopes and ridge tops of the Okanagan Highlands and is characterized by long, cold winters with a high snow cover and short, cool summers (Lloyd et al. 1990). Lodgepole pine is a successional species on all sites in this subzone. IV. DETERMINATION OF THE INCIDENCE, IMPACT AND SPATIAL DYNAMICS OF P. TERMINALIS

A. Sampling methodology The first phase of the study was to determine the incidence of weevilling throughout the Kamloops Forest Region in the three biogeoclimatic zones (IDF, MS and ESSF) which contain the majority of the lodgepole pine type. Density-dependent, variable-width strip surveys were done in randomly selected stands throughout the Kamloops Forest Region during the summers of 1987-89 (Table 2). Candidate stands were chosen on the basis of stand age (>6 years), treatment (spaced or unspaced) and accessibility. The strips varied from 2-5 m wide dependent on the stand density, and were 100-800 m long. At every 100 m interval along the survey strip, stem density was determined. This calculation of stem density gave the range of density, if any, within the stand. Fifty immature stands were surveyed. Stands within each of the three biogeoclimatic zones were then selected, based on P. terminalis incidence, stand age and density, for installation of permanent sample plots. The objective was to establish fixed area plots in each biogeoclimatic zone in both an unspaced and a spaced stand of comparable age. Plots were of varying dimensions, depending upon stand density (Table 3). The largest plot size was 50 m by 50 m (0.25 ha) and the smallest 10 m by 10 m (0.01 ha). The smaller plots were established to follow the development of defects over time and will not be discussed in detail at this stage of analysis. The larger plots (Table 4) were established to investigate the spatial dynamics of P. terminalis attack and are the primary components of the study. The target number of trees per plot was 250 to 500 trees. However, due to the differences in densities between ecosystems, and between spaced and unspaced stands, tree numbers varied in each plot. For example, in order to obtain a representative area

Table 3. Location, biogeoclimatic zone classification, and summary statistics for 21 permanent sample plots established in the Kamloops Forest Region for long term monitoring of P. teminalis attack dynamics and impact. Geographic Biogeoclimatic Plot Density Mean tree Year Attack intensitya location classification area (ha) (stemdha) age in 1989 evaluated (%stems attacked) . Okanagan Falls IDFdml 0.220 1,300 14 1988 48.1 0.010 1,300 14 1987 33.6 0.035 6.400 14 1991 17.9

Ketchan Creek IDFdkl 0.100 3,900 13 1987 21.4 0.010 4,200 13 1987 12.6

Willis Creek MSxk 0.010 10,400 11 1987 4.8 0.010 3,900 11 1987 10.3 0.010 4,400 11 1987 2.3

Dillard Creek MSxk 0.250 1,640 14 1989 22.3 % 0.023 23,500 14 1988 7.0 0.010 2,100 17 1987 14.3 0.010 1,900 17 1987 31.6

Lac le Jeune MSxk

Ellis Creek MSdml

Conkle Lake MSdml 0.044 11,840 11 1989 7.8

Monte Lake MSxk 0.010 18,300 11 1987 0.5

Beblow Road ESSFdcl 0.250 1,180 13 1990 4.4 a Attacks in the year of evaluation as well as old attacks qualified tree as "attacked". Table 4. Location, size and attributes of twelve stem-mapped plots established in 1987-91 in the Kamloops Forest Region.

Geographic Biogeoclimatic Plot size No. trees Density location zone (ha) in plot Stand treatment (stems per ha)

Okanagan IDFdml 0.220 285 spaced1984 .. 1,300 Falls 0.035 224 unspaced 6,400

Ketchan Creek IDFdkl 0.096 379 unspaced 3,900

Dillard Creek MSxk 0.023 528 unspaced 23,500 0.250 409 spaced 1986 1,600

Lac le Jeune MSxk 0.023 329 unspaced 14,600 0.160 386 spaced 1971 (2.4 m) 2,400 0.120 2 17 spaced 1971 (3.0 m) 1,800 0.160 113 spaced 1971 (3.7 m) 700

Ellis Creek MSdml 0.168 458 spaced 1984 2,700

Conkle Lake MSdm1 0.045 528 unspaced 11,800

Beblow Road ESSFdc1 0.250 295 spaced 1986 1,200 (ha) to map stem locations and follow the spatial distribution of attacks in the Dillard Creek unspaced stand, a minimum of 0.02 ha was needed. This gave over 500 trees in the plot. Therefore the corresponding spaced plot had to be 0.25 ha to sample a comparable number of trees (Table 4). P. terminalis attacks and kills the expanding terminal growth of young pines causing minimal diameter loss and variable height and quality loss to the tree. As discussed above, this insect has an extremely flexible life history (Cameron and Stark 1989) and causes a range of growth defects which apparently depend on geographic location, host parameters including genetics, age, size, and relative growing space available to the tree. Host density influences the severity of defect (Maher 1982), but it is debatable whether P. terminalis prefers stands of low or high density (Keen 1952; Stark and Wood 1964; Stevenson and Petty 1968; Stevens and Knopf 1974; Maher 1982; Desnoyers 1988). In a study of P. strobi in Sitka spruce, Picea sitchensis (Bong.) Carriere, plantations, denser plantations sustained a lower intensity of attack than the more open plantations, although stands of different densities had the same number of stem defects per tree (Alfaro and Omule 1990). The commonly accepted method of assessing stand density is to count trees in plots of known area (Brown 1965), and to calculate a mean density for these plots. This is also a common method of describing insect abundance, numbers of attacked plants, tree mortality, or defects due to insect attack over known areas. However, this method yields no indication of the degree of aggregation or repulsion of individuals. My objective was to describe all the morphological attributes of the host which may influence its probability of being attacked by P. terminalis, and to elucidate the spatial dynamics of attack. One method of investigating spatial dynamics is to compare the unique area potentially available (APA) to each tree (Brown 1965). The APA is also the space available to the insect seeking a host and may influence choice between hosts. Greater space around one host than another may indicate to the colonizing insect a more favorable microsite for the next generation of developing insects because of increased radiant energy, as well as a more succulent and nutritious host that has a larger growing space. On the other hand, densely growing hosts may suffer stress from competition, increasing their susceptibility to infestation. It must be emphasized that it is the spatial distribution of stems and insects infesting these stems that is in question, not the trees' occupation of an area in terms of lateral branch spread, root grafting or basal area. In any stand of trees, there will be variation in the density or clumping of trees within the stand even though the overall stand density may be expressed as a single value. One of the most common silvilcutural manipulations performed on a young stand is spacing. Spacing lowers the overall mean density of a stand by removing selected trees in a systematic fashion, usually leaving the remaining trees at a specific intertree distance, thus creating an area containing fewer, very regularly arranged trees. The assumption of APA as described by Brown (1965) is that in a scatter of trees in a stand, each tree has potentially available to it half the distance to the neighboring tree, everything else being equal. To determine the APA of each tree, a line is drawn at right angles to and bisecting the line joining the tree in question to each surrounding tree; the polygon thus created defines the available space boundary between trees. This set of polygons, each containing one distinct point, or tree, is called a Dirichlet tessellation or Voronoi polygon (Upton and Fingleton 1985), and is one of the most useful constructs associated with such a point configuration (Green and Sibson 1977). These Voronoi polygons and their associated Delaunay triangulation are being applied increasingly to the statistical analysis of spatial patterns (Green and Sibson 1977; Diggle 1983; Upton and Fingleton 1985). When competition or crowding is examined at the level of an individual, whether the individual be the host tree or insect attacking the tree, the "effective density" experienced by an individual is dependent on the location of its immediate or nearest neighbors (Kenkel 1988). Therefore, in addition to determination of the APA, a method of nearest neighbor (NN) analysis can be used to compare relationships between trees, attacked and unattacked by P. terminalis. There has been much interest recently in making maps of various organisms, such as trees, and analyzing spatial patterns with various statistical methods (Hall 1991). Some mapping methods use interpoint distances and least squares in mapping objects (Rohlf and Archie 1978) and suffer accumulation of error (Hall 1991). Hall (1991) analyzed the method of Rohlf and Archie (1978) which begins with three reference points and uses triangulation to locate each new point. From the distances measured from each new point to any three previously located points, three sets of estimated coordinates can then be generated. Thus, all points have coordinates generated and are adjusted to yield a least squares fit of measured distances to calculated distances (Rohlf and Archie 1978; Hall 1991). To increase the accuracy of tree location and inter-tree relationships, all tree coordinates were mapped using a field location technique described in the following section. Plot boundaries were delineated by compass, and plot comers located as reference points to which trees were referred. Distances were measured to the nearest 0.10 m using metric tapes, estimating the centre of tree boles as the point location (dbh was measured to the nearest 0.001 m). In Hall's (1991) re-examination of the use of interpoint distances and least squares analysis in mapping trees, he concluded there is no simple, rapid, accurate method for mapping locations using interpoint distances and the most accurate method, although laborious, is physical field mapping of tree coordinates. Trees with more available area tend to have longer and thicker terminal shoots, increasing weevil survival. The hypothesis that insects select host trees which have larger growing areas was tested within stands and between stands. The relationship of APA (m2) and NN (m) to host attributes such as height, dbh and growth defects as a result of weevil attack, were analyzed .using Chi-square analysis and t-tests (Zar 1984).

B. Okanagan Falls: a case study Between 1987 and 1991, 22 plots were established throughout the Kamloops Forest Region. The relationship between biogeoclimatic zone, stem density, attack dynamics and impact of P. temzinalis was analyzed using 12 of the 22 plots (Table 4), representing the following subzones: IDFdm 1; IDFdkl ; MSxk; MSdml, and ESSFdcl . The Okanagan Falls plots, located on Tree Farm Licence 15 (T.F.L.), 18 km east of Okanagan Falls, B.C., will be used initially as a case study to describe the methodology and analysis used for all plots in the study. The Okanagan Falls site was logged in 1974 and left to regenerate naturally. In 1984, when the average age of the stand was 8.4 years, portions of the stand were spaced and portions were left natural. In 1987, a 0.22 ha plot was established in the spaced portion of the stand and 285 trees were permanently tagged, and measured. All trees were stem mapped by dividing the plot into 2.5 m wide strips, laying a 50 meter tape along the length of the plot at 2.5 m intervals and then plotting the location of all trees within each strip. Measurements and observations recorded from 1987-1990 for each tree were: height; incremental height growth; diameter at 1.3 m (diameter at breast height or dbh); year(s) of weevil attack; height to attack; defect (crease, crook, fork or staghead); and any other pertinent observations. Only small, scattered portions of the stands were left in the natural state. - Therefore, the unspaced plot established in the Okanagan Falls site could be influenced by the spaced portions of the stand in terms of edge effects. To make comparisons to the unspaced situation, a 0.035 ha plot was established in 1991, containing a number of trees comparable to the spaced plot (Table 4). The (x,y) coordinates of each tree were determined from the stem map of the study plot. Using the (x,y) coordinates, Voronoi polygons were constructed using SYGRAPH (Wilkinson 1988). Each Voronoi polygon was measured with a digital planimeter, giving the APA for each tree. A program was written using LOTUS 1-2-3R (A.J. Stock and L.E. Maclauchlan, unpublished) which took one tree at a time, calculated its distance to every other tree in the plot (attacked or unattacked), determined the minimum distance, i.e., the nearest neighbor (NN), and then calculated the modified Clark-Evans-Donnelly statistic (CED) (Clark and Evans 1954; Donnelly 1978) on all the calculated NN distances (Sinclair 1985; Matlack and Harper 1986). The CED statistic uses the nearest neighbor distances to calculate the pattern of trees in the study plots. The same procedure was followed for trees which had been weevilled one or more times, excluding all other trees from the calculation, to examine the spatial pattern of attacked trees in the plot. The ratio of the observed mean nearest neighbor distance to the expected mean nearest neighbor distance serves as the measure of departure from randomness (Clark and Evans 1954). The Clark and Evans test (1954) is based on the mean distance ( y ) from each point to its nearest neighbor; small values of ( 7 ) indicate an aggregated distribution of points and large values indicate regularity. If there are n points in a region of area A, then the test statistic proposed by Clark and Evans (1954) is

(7)- EWSE(~) where the expected value and standard error of j are approximated by

~(j)= 0.5- and SE(~)= 40.0683Aln, respectively. However, this approximate distribution of the test statistic assumes independent nearest neighbor distances and an absence of edge effects. Because of the combined influences of the edge effects and interdependence, a study by Donnelly (1978) suggested that the Clark-Evans statistic could be considered to be an observation from a unit normal distribution, CED = (5,) - E(j)/SE(j) providing that the tree distribution was random and that the expectation and variance were calculated from

and

where there are n.points in a region of area A, and length L. The test statistic (CED) is evaluated and referred to the standard normal distribution, with spatial randomness being rejected in favor of aggregation or regularity for values in the lower or upper tails, respectively (Sinclair 1985). The following scenarios were analyzed: 1) all trees in plot; 2) all weevil-attacked trees over time; 3) weevil-attacked trees in any given year (n varies due to differing levels of attack each year); and 4) additive weevil attack over time (n increases yearly). The CED statistic has a standard normal distribution under the hypothesis of complete spatial randomness for n > 7 "in any region with a reasonably smooth boundary" (Sinclair 1985); therefore, the distribution of weevil-attacked trees in some years, due to low levels of weevil attack in that year, was unreliable due to a small n. A t-test was used to compare each of the parameters, APA, NN, height and dbh, of all attacked trees to unattacked trees. Tree heights in the spaced plot were divided into height classes and the distribution of attacked and unattacked trees falling into each of the height classes was analyzed by Chi-square analysis (Wilkinson 1989). - To compare between years, biogeoclimatic zones and plots, the yearly height increment of each tree was expressed as a percentile, and the mean growth of the stand, for a given year, was analyzed by Chi-square analysis (Zar 1984; Wilkinson 1989). Kendall's coefficient of concordance (Kendall and Stuart 1979), the tau-b statistic, was used to test for change in defect type over time. The notion underlying the use of tau is that of disarray; if two variables (x,y) are observed on each member of a sample, and the x's are arranged in increasing order, the extent to which their corresponding y's depart from increasing order indicates the weakness of the correlation between x and y. A subsample of 22 trees randomly selected from outside the spaced plot boundary, growing under the same conditions, was destructively sampled to quantify height loss. The sample included five unattacked trees and 17 attacked trees. The yearly height increment of those trees was measured and compared. The height increment in the year of attack (the length of the compensating lateral) and in the growing season following the year an attack occurred, were tabulated. The yearly increments for years other than the latter, for the 17 attacked trees, were used as unattacked comparisons of height growth for their respective year. The annual height increment of all 5 unattacked trees was used as comparison for each growing season. Only the same years' growth increment (e.g., 1989 to 1989) was compared on each tree to standardize differences in growth rates between years. The mean annual height growth of attacked trees and unattacked trees was compared by a t-test (Wilkinson 1989) for the year before each attack, the year of each attack, and the year following each attack. 1. Attack dynamics and impact in relation to stand age and tree size Based on Maher's (1982) report and personal observations, the Okanagan Falls study area was just reaching its susceptible stage in 1980 when the first attacks were recorded in the unspaced plot (Fig. 3). The first recorded weevil attacks in the spaced plot occurred in 1982 when the average stand age was 6.4 years and mean (+S.E.) tree height was 1.28 + 0.05 m. It is not known what the pre-spacing distribution of the trees was at that time and most likely some weevil-attacked trees were removed during Figure 3. Yearly rate of attack by Pissodes terminalis, expressed as percent of total trees, in two plots at Okanagan Falls, 1980-1989. Spacing of trees was done in 1984. 4 Unspaced

......

spaced

80 81 82 83 84 85 86 87 88 89 Year the spacing operation. By comparison, the first weevil attacks in the unspaced plot occurred in 1980, with < 2% of the total stems being attacked. Mean tree height in the stand in 1980 was < 1 m. The assumption was that the attack levels were not significantly different in the two plot locations prior to spacing in 1984; therefore, this would indicate the number of attacks removed during the spacing operation.

Almost half of the 285 trees in the spaced plot (125 trees, 43.9 %) were attacked by P. terminalis one or more times. There were in total 194 attacks on the 125 trees, or 1.6 attacks per attacked tree. In terms of all trees in the plot, there were 0.7 attacks per tree, compared to 0.2 attacks per tree in the unspaced plot. The unspaced plot was assessed in 1991 and there were 27 trees attacked (12.1 %), from 1980 to 1991, for a total of 40 attacks over a 0.035 ha area. In 1987, and in each subsequent year until 1990 (1990 assessment detects 1989 attack), every tree in the spaced plot was assessed and each attack was categorized into one of four stem defect categories modified from Maher (1982) and Alfaro (1989). In increasing severity, they are: 1) crease, minor defect comprised of a linear indentation, but little or no stem curvature at point of attack; 2) crook, a major defect, defined when a lateral assuming dominance is offset from the main stem by at least 112 the stem diameter; 3) fork, a major defect resulting when two laterals assume dominance; and 4) staghead, a major defect resulting from three or more laterals assuming dominance. The most common major defect resulting from weevil attack is a crook. At this early stage in the life of the stand it is difficult to assess the end result of attack definitively, but some trends are evident. For example, all of the four 1982 attacks in the spaced plot caused severe defects, namely, two crooks, one fork and one staghead. However, upon casual observation it appears that defects change over time as the tree grows. To test this observation, in 1987, each weevil attack in the Okanagan Falls plot was assessed and assigned a defect category. The plot was re-assessed annually, and at the last assessment in 1990, all weevil attacks Table 5. The distribution of defect categories in the Okanagan Falls spaced plot, expressed as total trees as assessed in 1987 and then again in 1990. The Tau-b statistic is 0.602 indicating little change in defect category over time.

1987 defect assessment Distribution of 1987 defect in 1990 Type Number Crease Crook Fork Staghead

Crease 61 46 9 4 2

Crook 30 6 22 0 2

Fork 59 11 8 38 2

Staghead 11 0 1 1 9 prior to 1987 were again assigned a defect category. There was no significant change in defect in 1990 from the original assessment done in 1987 (tau-b=0.602)(Table 5) although there were some minor fluctuations. From 1987 to 1990, 51 % of the defects did not change, but 24% of the trees attacked increased by 1 to 3 severity classes, and 25 % lessened in severity class. The weevil attacks originally classified as forks were the defects which changed most significantly as the trees grew. Of the 59 trees in 1987 which appeared to have forks forming (Table 5), 64 % of these continued to form forked tops, 4 % formed stagheads, and 32 % formed crooks or creases. From assessments in 1990, 64%, of the 125 attacked trees had a major defect (Fig. 4). Forks comprised 50% of the defects assessed in the unspaced plot (Fig. 4). The reason for this abundance of forks could be twofold: fxst, the plot has only been assessed once and, as noted for the spaced plot, attacked trees initially often appear to be forked but over time one lateral assumes dominance (32% of the time); second, in a more "crowded" environment, the competition for growing space and light is high and this competition may force both competing laterals to assume a more vertical orientation. This extremely dense scenario is common on MS sites which have naturally regenerated following fire or clearcut harvesting. IDF sites, due to their moisture limitations (20- 50% of the precipitation falls as snow) and hot summers (Meidinger and Pojar 1991), often have naturally lower stem densities than other more moist ecosystems such as the MS. Therefore, the competition for light due to overcrowding may not be as severe in unspaced IDF sites. The distribution of crooks and stagheads is similar between the spaced and unspaced plots at Okanagan Falls. To evaluate the influence of growing space on the type of defect formed after weevil attack, the APA of attacked trees was compared with the defect observed. When all defect types were compared by Chi-square analysis, there was no significant difference (P > 0.05) in APA among different defects in the spaced or unspaced plots. Figure 4. Frequency or four defect categories caused by Pissodes terminalis attack on 125 trees from 1982-1989 in the Okanagan Falls spaced plot and on 27 trees from 1980-1991 in the Okanagan Falls unspaced plot. Crease

Spaced plot

Crook

......

Fork Unspaced plot However, in the spaced plot, those trees developing stagheads did on average occupy

larger APA's, 7.5 f 0.5 m2 (mean $ S.E.), than trees developing other types of defect, 6.7 + 0.5 m2 (mean & S.E.), although there was no significant difference. If these trees had been the least crowded, the length and diameter of their laterals would have been greatest. The subtending laterals may not have been efficient at assuming a vertical position because there was abundant light, even without a directly vertical orientation. In addition, lignified wood in large, thick laterals would resist shifting to a vertical position, causing crooks, forks and stagheads, because of high lateral offsets from the axis of the main stem. These factors would combine to make a noticeable crook in the stem as apical dominance was achieved (Ballard and Long 1988). In a dense stand, however, a single lateral may assume dominance and grow with little noticeable defect. This hypothesis may explain why on average the distribution of post- spacing defects had two thirds major defects (crook, fork, and staghead) and one third minor defects (crease)(Fig. 4). In very dense young stands with close crowns, there is a strong vertical orientation, and creases, forks and stagheads are frequent due to the

strong competition by laterals for overhead light. However, within a short time ( < 5 years) one of the laterals assumes dominance and the defect changes, usually to a crease. In a spaced stand where the competition for overhead light is less, the defect is more pronounced due to the lignification of the wood at the point of the defect, and even if one lateral eventually assumes dominance there is still a noticeable defect, most commonly a crook. The first weevil attacks in the spaced plot occurred in 1982 with just over 2% of - the trees being weevilled, and increased to a peak of almost 17% in 1985 (Fig.3). Lodgepole pines enter their most susceptible stage to P. terminalis at about 5-6 years of age and 1.5-2.0 m in height (Stark and Wood 1964), and attacks are most common in 15-25 year-old stands (Bella and Stoszek 1988). The first recorded attack in both Okanagan Falls plots occurred when trees were 4-6 years old. From 1984-1988 '\ inclusive, attack by P. temzinalis in the spaced plot averaged 8 % per year (Fig.3). The percent stems attacked was significantly less in the unspaced plot but in terms of attacks on an area basis (attacks per ha), the attack level was not significantly different (Chi- square analysis, P>0.05). In four years, 1983 through 1986, the trees attacked by P. terminalis in the spaced plot were significantly older than trees not attacked (9.2, 9.0, 10.3 and 11.9 years compared to 7.3, 8.3, 9.2 and 10.3, respectively)(t-test, P < 0.05). However, due to the relatively small variation in tree age within an opening of lodgepole pine regeneration, age probably does not play a large part in the overall selection of hosts by P. temzinalis. Once trees begin to express dominance in a stand the influencing factors in host selection by the weevil would be height and growing space and not necessarily tree age. When a simple linear regression is fit to the data, the percentage of the total variation in attack that is explained by age ranges from 0.5 % to 2.3 % . The mean age of a stand in relation to the mean age of an adjacent stand could be of importance in terms of which stand would become susceptible to the weevil first. But even in this scenario, other parameters such as height and growing space would influence the weevils' choice between stands. Tree height and APA of trees in the spaced stand were converted to percentiles (Zar 1984) of the height and APA distribution for each year to allow comparisons between years, and between attacked and unattacked trees, independent of the mean value (Alfaro 1989). The distribution of heights and APA, sorted by year and attack status, were tested for normality using the Kolmogorov test (Zar 1984). The tests of normality indicated that tree heights and APA were distributed normally. Separate analysis of tree height arid APA distribution by attack status indicated some departure from normality, but the difference was only minor (Fig. 5). The distribution of tree height and APA in any given year followed the same pattern as Figure 5. Total tree height (upper graph) and area potentially available (APA)(lower graph), expressed as percentiles, of trees attacked and not attacked, for 1987, in the Okanagan Falls spaced plot. The number of trees falling into each percentile group, expressed as a percent of total, is plotted with the mean heights and APA's indicated by the vertical arrows. K not attacked (40.8)

- APA (percentiles) Table 6. Height and APA, expressed as percentiles of all tree heights and APA's in a given year, of attacked trees in the Okanagan Falls spaced plot from 1982 to 1988.

Percentiles (mean + S.E.)a Year Height APA

a Means in a column followed by the same letter are not significantly different, Tukey ' s test, P < 0.05. seen in 1987 (Fig. 5)., with the attacked trees falling in the higher range of both height and APA percentiles. The APA of trees attacked by P. terminalis did not vary significantly between years and trees attacked were on average from the 40 to 50 percentile range of APA (Table 6). As the stand aged, the trees attacked by P. terminalis were from significantly higher height percentiles (Table 6). For example, on average all attacked trees in 1984 were in the 40-50 height percentile, compared to the attacked trees in 1987 and 1988 being from the 50-60 height percentile (P < 0.05). The category of height percentiles of attacked trees in 1982 and 1983 were not significantly different from the height percentiles of trees attacked in 1986-1988. However, due to the removal of trees in the 1984 spacing, the distribution of height percentiles of attacked trees may be skewed to the upper end because taller trees are usually left during the spacing process. Dominance is expressed within a stand over time and as the expression of dominance takes place, P. terminalis also seems to orient more strongly to these larger, more vigorously growing hosts. Based on the range of tree heights in the spaced plot, five height classes were arbitrarily chosen (114 cm intervals). When height is broken down into five height classes, rather than ten as in percentiles, the distribution of attack is seen more clearly (Fig. 6). The highest proportion of attack in the spaced plot (29 %) falls in the second tallest height class (371-485 cm) (Fig. 6). Using this height class grouping, the distributions for attacked and unattacked trees differ significantly (Chi-square, P < 0.05). But as with the percentile breakdown, host selection appears to be mediated by an interaction of parameters rather than just host height alone. The trees being . selected by P. terminalis also have larger diameters than unattacked trees (Table 7). Incremental height growth, or leader length, seems to be an important parameter in the successful colonization of hosts by P. terminalis. Studies on Pissodes strobi have Figure 6. Frequency distribution of P. temzinalis attacked and unattacked, trees in five height classes in the Okanagan Falls spaced plot. - Attacked trees -+Unattacked trees

25-139 140-255 256-370 371 -485 >486 Height class (cm) Table 7. Comparison of spatial distribution of attacked and unattacked trees in the two Okanagan Falls plots by two statistical measures, the size of Voronoi polygons (area potentially available=APA) and the nearest neighbor distance (NN). The mean total height and dbh of attacked and unattacked trees in the two plots are also compared.

Mean + S.E.a Treatment Tree status N~ APA (m2) (m) Height (m)' DBH (cm)'

3 Spaced Attacked 116 7.20 + 0.21a 1.76 + 0.04a 4.22 + 0.07a 6.7 f 0.2a Unattacked 156 5.50 + 0.23b 1.47 + 0.05b 3.47 + 0.25b 4.9 + 0.2b

Unspaced Attacked 40 1.82+0.19a 0.71+0.06a 4.5640.27a 6.7+0.6a Unattacked 196 1.58 + 0.11a 0.60 + 0.03a 3.22 + 0.09b 3.3 + 0.2b a Means followed by the same letter in the spaced or unspaced plots are not significantly different, t-test, P < 0.05. b Some trees not included due to proximity to plot boundary. C Height and dbh measurements were made in 1989 for the spaced plot and 1991 for the unspaced plot. the tallest trees (VanderSar and Borden 1977a; Wood and McMullen 1983). However, a more recent study by Kiss and Yanchuk (1991) suggests that the genetic composition of individual trees may be more important than leader dimensions. Kiss and Yanchuk (1991) examined the pattern of P. strobi attack among families of interior spruce (the complex of white spruce, Picea glauca (Moench) Voss, Engelmann spruce, P. engelmannii Parry, and their hybrid swarms). Their data, comparing average percent weevil damage at ages 10 and 16 years, showed an inverse relationship (r=-0.51), indicating that faster growing families are less frequently damaged than the slower growing ones. A similar relationship was shown for dbh, indicating that families with larger average dbh are damaged less frequently than those with smaller dbh. Other studies in the genetic resistance of spruce to P. strobi involved Sitka spruce provenance trials having records of fifteen years of weevil attack (Ying 1991), revealing large differences among provenances in percent of trees attacked and numbers of attacks per tree. This provenance variation in weevil attack was repeated in a clonal test which revealed that provenances which showed especially high resistance to weevil attack were also fast growing (Ying 1991). There may be a similar mechanism, as seen in interior and Sitka spruce, at work in lodgepole pine. Such genetic variation could in part explain why trees in the very tallest percentile were not attacked frequently by P. terminalis. The annual height increment was analyzed from a subsample of 22 trees cut near the spaced plot. Of the 22 trees randomly selected, 17 had been weevilled 1 or more times and 5 had no weevil attacks. The mean number of attacks per subplot tree (+S.E.) was 1.4 + 0.2, with 70% of the attacks resulting in major defects (12 crooks, 8 forks, 1 staghead). There was no significant difference in mean height increment between attacked and unattacked trees in the growing season prior to attack occurring (Table 8). Both in the year of attack and the year following attack, the height increment Table 8. Data from a random subsample of trees cut near the Okanagan Falls spaced plot comparing mean incremental height growth in the year of weevil attack and in the years immediately before and after weevil attack. Height loss is expressed as a percent (in brackets) of the total height increment potential of unattacked growth years.

Increment \O Mean annual height growth (cm) f S.E.0 status N 1 year before attack Year of attack 1 year after attack Attacked 30 43.2 + 2.9a 30.7 f 1.6a (31.4%) 41.0 + 3.5a (17.2%) Unattacked 34 41.6 f 2.6a 45.2 f 2.5b 49.5 4 2.3b

a Means in each column followed by the same letter are not significantly different, t-test, P <0.05. of the compensating lateral of attacked trees was significantly less than leaders not attacked the previous year (Table 8). This difference in height growth can be quantified in terms of percent of potential height growth (Table 8). In the year of attack, the height loss is 3 1.4 % of the annual potential height increment. In the following growing season, the incremental growth is still only 83 % the potential of unattacked stems (Table 8). Cameron (1974) approximated 10% height reduction in trees less than 3 m tall and 25 % in trees 3-6 m tall when the longest lateral assumed dominance. When the second-longest lateral assumed dominance, 20 and 33% height reduction for the two tree heights, respectively, was recorded (Cameron 1974). The Okanagan Falls result'is slightly higher than the height loss observed in spruce when buds are infested with the spruce bud midge, Rhabdophaga swainei Felt (Diptera: Cecidomyiidae)(West 1990). West (1990) observed a 25 % loss in dominant shoot growth due to midge damage; in an experiment simulating midge damage, a 16% loss was observed (Cerezke 1972). Similarly, the effect of midge infestation was significant for only two years (Cerezke 1972; West 1990). The western pine shoot borer, Eucosma sonomana Kearfott (Lepidoptera: Olethreutidae), which mines the pith of elongating terminal shoots of ponderosa pines, Pinus ponderosa Laws., also causes losses of about 25 % of one year's vertical growth per attack (Sower and Shorb 1984; Sower et al. 1988). For trees which are only weevilled once or twice in their rotation, this height reduction is minor. However, if a tree sustains multiple weevilling, the cumulative height loss could increase the rotation age as well as decrease the quality of the stem due to defect formation. 2. Spatial attack dynamics The mean APA of all attacked trees was significantly different (t-test, P < 0.05) from the mean APA of unattacked trees (Table 7) in the Okanagan Falls spaced plot.

On an area basis, 59.9% of the spaced plot was occupied by weevilled trees and 40.1 % was occupied by unattacked trees by the time the stand reached an average age of 13.4 years (Fig. 7). Over half the trees in the spaced plot had one or more weevil attacks (Fig. 8), with 17.8% having two attacks and 2.4% having three attacks. The mean nearest neighbor distance of attacked trees was also significantly greater than that of unattacked trees (Table 7) in the spaced plot. The mean height and dbh of attacked trees were the parameters in the unspaced plot which were significantly different (P < 0.05). The mean APA and NN showed no statistical difference between attacked and unattacked trees in the unspaced plot but this could be explained by the spatial location of attacks in relation to the spatial distribution of the hosts (Fig. 9). The attacked trees seem to be distributed around the perimeter of the natural "clumps" of the host. If this is indeed the case, then the nearest neighbor of an attacked tree may be very near on the side bordering the clump of hosts, but may be a much greater distance from any neighbor on the edge facing away from the clump of hosts (Fig. 9). The APA and NN values for trees remain static over time, but the status of "attack" changes yearly; therefore, the mean values of APA and NN for attacked and unattacked trees change over time as does the spatial relationship of attacked trees to one another. The nearest neighbor distances were calculated and analyzed by the CED statistic, in terms of yearly attack and cumulative attack (Figs. 10,ll). Due to the effects of spacing, the distribution of all the trees in the plot (Fig. 7) approaches a very regular distribution (CED=3.14, P=0.001) with a mean spacing (+ S.E.) of 1.61 + 0.04 m. The pattern of attacked trees in any given year tended toward a clumped distribution (all CED values are negative) (Fig. 10). However, the cumulative pattern of attacked trees viewed over time, as each years' attack is added on to the previous years', begins to approach a more regular pattern (Fig. 1I), with a mean spacing (k S.E.) between attacked trees of 2.39 + 0.07 m. The patterns of attacked trees may have varied slightly prior to 1984, before spacing. Maher (1982) also noted that P. Figure 7. Spatial arrangement of all trees, both those attacked at least once between 1982 and 1988 by Pissodes terminalis and those not attacked, in the Okanagan Falls spaced plot. Each tree is defined by a Voronoi polygon which represents its "area potentially available" (APA). The Clark-Evans-Donnelly statistic (CED) is given below the figure for "all the trees" in the plot and for "all attacked trees" (shaded areas). Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness. 0 10 20 27.5 m ED,all trees3.14 ED,attacked trees=1.43 (P=0.001) (P4.076) Regular Random Figure 8. The number of trees in the spaced and unspaced Okanagan Falls plots having no attacks, 1 attack, 2 attacks or 3 attacks per tree. The percent of total trees in each plot in each attack category is shown above each bar. spaced unspaced

no attack 1 attack 2 attacks 3 attacks Attacks per tree Figure 9. Spatial arrangement of all trees, both those attacked at least once between 1980 and 1988 by Pissodes terminalis and those not attacked, in the unspaced Okanagan Falls plot. Each tree is defined by a Voronoi polygon which represents its "area potentially available" (APA). The Clark-Evans-Donnelly statistic (CED) is given below the figure for "all the trees" in the plot and for "all attacked trees" (shaded areas). Values equal to or < 0 approach a clumped pattern, and 22approach a more regular pattern, with intermediate values indicating randomness. m Okanagan Falls, unspaced plot

7 14 17.5 m

CED, all trees = -1 A34 (P=0.076) CED. attacked trees = -1.61 4 (M.054) Clumped Clumped Figure 10. Spatial pattern of Pissodes teminalis-attacked trees over time in the Okanagan Falls spaced plot showing the value of the Clark-Evans-Donnelly (CED) statistic for each year. In all years the pattern of attack approaches a "clumped" distribution (e. g ., CED =- 1.655, P =0.097 indicates a 9.7 % level of significance for a two-tailed test). Voronoi polygons have been drawn around attacked trees to help illustrate the shifting pattern and density of attack. Values equal to or < 0 approach a clumped pattern, and >2 approach a more regular pattern, with intermediate values indicating randomness. 1982 . CEDE-1 .I88 (P=0.2341 1983 CED =-0.770 (P -0.441) 1984 CED = -1.208 (P =O.226) Clumped Clumped Clumped Clumped

1986 CEO 1-0.223 (P=0.826) 1987 CED=-1 i676 (P=0.093) I988 CED =-0.206 (P=0.834) 1989 CEO=-1.655 (P=0.097) Clumped Clumped Clumped Clumped Figure 11. Spatial plots of attacked trees in the Okanagan Falls spaced plot. Each successive graph incorporates an additional year of attacked trees. The Clark-Evans- Donnelly (CED) statistic is calculated for each graph with a description of attack pattern. terminalis attack in the Cariboo appeared random throughout the stands he investigated. A clumped distribution of weevil attacks has also been reported for P. strobi (Graham 1951; Alfaro and Ying 1990).

C. Lac le Jeune: a case study Lac le Jeune is in the MSxk subzone, approximately 28 km south of Kamloops, B.C. The pure lodgepole pine stand on the site regenerated naturally after a wildfire in 1960. In 1971, when the average tree age was 8 years, small plots were spaced to three different spacing regimes, 2.4 x 2.4, 3.0 x 3.0 and 3.7 x 3.7 m. The spacing regimes represent target inter-tree distances, but as illustrated in Table 9, the actual stem density in each plot does not always reflect this target. In 1976 and 1980, much of the remaining stand was also spaced, but a small portion of the area was left in its natural state. In 1987, four plots were established in the Lac le Jeune site (Table 3), with one plot per density regime and one plot in the unspaced area. Plot size was dependent on density and ranged from 0.023 ha (329 trees) in the unspaced area to 0.16 ha (1 13 trees) in the 3.7 m spacing (Table 9). Plots had to fit within the three spacing regimes, 2.4, 3.0 and 3.7 m, established in 1971 and were therefore restricted in size. Plot edges were kept 10 m from the edge of the treatment area to minimize any edge effects. The plot in the unspaced area was the smallest in area, but a representative number of trees was assessed. The plot in the 3.7 m spacing area (0.16 ha) occupied most of the treatment area which was only 50 x 50 m (0.25 ha) in size. All trees were stem mapped and the APA and NN values determined as for the Okanagan Falls plot. Measurements and observations recorded in 1987-88 for each tree were height, diameter at 1.3 m, year(s) of weevil attack, height to attack, defect, and any other pertinent observations. The average age of the trees in the Lac le Jeune study area was 27 years with the mean tree height ranging from 5.8 1k0.11 m in the unspaced plot to 7.95 k0.16 m Table 9. Distribution of P. terminalis attacks in the four Lac le Jeune plots, showing number of attacks per tree, attacks per hectare, and total percent stems attacked over the life of the stand.

Plot Plot No. trees Total No. attacks Percent stems Attacks treatment size (ha) in plot no. attacks per treeb attacked per hectare Unspaced 0.023 329 87 0.17 + 0.02a 21.6% 3,867 a\ 2.4 m spacing 0.160 386 234 0.61 + 0.04b 42.0% 1,463 3.0 m spacing 0.120 2 17 176 0.81 + 0.07~ 48.4% 1,467

a Numbers in brackets on second line indicate the number of overstory, dominant trees, plus 35 understory (ingrowth) trees which were not included in the analysis or stem mapped. No attacks occurred on these 35 understory trees. b Means in column followed by the same letter are not significantly different, Tukey's test, P < 0.05. in the 3.7 m spacing (Table 10). To get a more accurate estimate of impact, a subsample of 20 to 25 trees per plot were destructively sampled in 1988 to quantify defects and height loss. Without creating excessive unnatural openings in the plots, four to five trees exhibiting one of each of the following defects, crease, crook , fork and staghead, plus four to five unattacked trees, were chosen for sampling. The yearly height increment of attacked and unattacked trees was compared: 1) in all plots in the growing season before an attack occurred; 2) in the growing season following the year an attack occurred, and 3) the length of the compensating lateral in the year of attack. Each plot was first analyzed separately, and then all plots were pooled. The same year's growth increment was compared on each tree to standardize differences in growth rates between years.

1. Spatial attack dynamics and impact of P. terminalis in different stand densities

In each of the four plots, the parameters APA, NN, height, dbh and age were compared for attacked and unattacked trees using a Chi-square analysis and Tukey's multiple range test (Wilkinson 1989). Defects were coded as 0 =no defect, 1=crease, 2=crook, 3 =fork and 4 =staghead and compared among plots, and attacks per tree were compared among plots using a Chi-square analysis and Tukey's multiple range test (Wilkinson 1989). A total of 90 trees (Table 11) was felled in the four plots for more detailed analysis. Due to relatively few trees developing stagheads, this category was not fully represented in the subsample of trees. Once the selected trees were felled and the limbs removed to measure the trees, it became apparent that most of the trees chosen for sampling, even the trees thought to be unattacked, had been weevilled one or more Table 10. Comparison of mean diameter at breast height (1.3 m), mean height of trees and mean height of weevil attacks (current and past) in the four Lac le Jeune plots.

Mean tree measurements + S.E.a Mean height of weevil Treatment DBH (cm) Height (m) attacks (m) Unspaced 4.7 f O.la 5.81 + 0.11a 4.21+ 0.18a 2.4 m spacing 9.5 + 0.2b 6.67 + 0.09b 5.51 + 0.09b 3.0 m spacing 13.2 + 0.7~ 7.30 + 0.11~ 5.49 + 0.11b 3.7 m spacing 14.1 + 0.3~ 7.95 + 0.16d 6.14 + 0.13~ a Means in columns followed by the same letter are not significantly different, Tukey's test (P < 0.01). Table 11. Frequency of defect types and attack statistics from a sub-sample of trees from the four Lac le Jeune plots.

No. of Defect frequency Total No. stems Attacks per Treatment trees Crease Crook Fork Staghead attacks attacked attacked tree Unspaced 2 1 15 3 3 0 2 1 15 1.40 2.4 m spacing 25 14 18 4 1 37 19 1.95 3.0 m spacing 22 17 20 6 0 43 19 2.26 3.7 m spacing 22 33 10 9 5 57 19 3.00 times. This observation indicates that the estimates of weevil attack in these plots, as assessed from the ground, are probably low. When assessing trees of this size (Table 10) it is difficult to see the entire bole clearly, especially in the more open grown stands in which the trees are excessively branchy. Many of the attacks which caused creases or crooks were not noticed until the tree was felled, hence the abundance of these categories (Table 11). What should have been an attack ratio of <1 attack/tree, according to the selection criteria, actually ranged from 1.0 attacMtree in the unspaced plot to > 2.5 attacksltree in the 3.7 m spaced plot (Table 11). Height and diameter of lodgepole pines increased as post-spacing density decreased (Table 10). In most cases, the mean height of P. terminalis attack increased as the mean height of trees increased (Table 10). At 3.7 m spacing, mean diameter at stump height was three times that of the trees in the unspaced plot, and the mean height of trees had increased by more than 35 %. Attacks by P. terminalis in the four regimes ranged from almost 4,000 attacks per ha in the unspaced plot to almost 900 attacks per ha in the widest-spaced plot (Table 9). The cumulative percent of stems attacked in the two intermediate density spaced plots did not vary greatly, 42% and 48.4% for the 2.4 m and 3.0 m spacing, respectively, nor did the number of attacks on an area basis, 1,463 and 1,467 attacks per ha, respectively. However, the lowest density spaced plot had a lower incidence of attack on a per hectare basis, 869 attacks per ha, but a higher proportion of trees being attacked by P. terminalis (Table 9). When the numbers of attacks per tree were compared between plots, the spaced plots differed significantly (P < 0.05) from the unspaced plot (Table 9). A density cline is observed with the lowest number of attacks per tree occurring in the unspaced plot, to the highest number of attacks per tree occumng in the 3.7 m spacing plot, with the number of attacks occurring per tree being significantly different between each of the densities (Table 9). Although host

67 trees in the 3.7 m spacing may be of superior quality for oviposition, having longer and thicker terminal shoots, the number of potential hosts is drastically reduced on a per hectare basis. The searching range of a female weevil is not known. If there are fewer trees present within this finite range, then the probability of a tree being attacked one or more times could increase. The mean number of attacks per tree (Tfi S.E.) in the unspaced area was only 0.17 + 0.02 compared to 1.20 + 0.09 attacks per tree in the lowest density plot (Table 9). As in the Okanagan Falls site, the yearly attack since 1980 averaged between 5- 10% (Fig. 12) with the lower density plots on average having higher annual attack rates. In the years 1970 through 1979 (Fig. 12), very few attacks were recorded due to the evidence being "erased" from the tree. There were numerous creases, crooks, forks, and stagheads in evidence at this early stage in the trees' development; however, when there was not an old terminal still on the tree at the point of defect, these defects could not definitely be attributed to weevilling. The sub-sample of trees which were felled for detailed analysis revealed that the length of the compensating lateral in all plots was significantly shorter than the leader of the unattacked trees in the year of attack (Table 12). In the lowest density plot, the 3.7 m spacing, there was a significant difference in height growth between attacked and unattacked trees that persisted in the year following attack (Table 12). A non-significant reduction in height growth in trees the year after attack was observed in the other three plots. In the year before trees were attacked the growth was not significantly different (P < 0.05) between trees which would be attacked the next year and trees which would remain unattacked. When the - incremental height growth in all plots was combined (Table 13), trees that were attacked showed no significant difference in height growth in the seasons prior to or after attack to trees which remained unattacked. However in the year of attack, the length of the compensating lateral was shorter than in the year prior to, or after attack,

68 Figure 12. Annual and cumulative percent of stems attacked by Pissodes terminalis in the four plots at Lac le Jeune, B.C. - 70 a) Natural spacing '01b) 2.4 x 2.4 m spacing

Cumulatlvo attack - Cumulative attack /'+ Annual attack - / 10 Annual attack

70-tC) 3.0 x 3.0 rn spacing d) 3.7 x 3.7 m spacing Cumulallvo attack .f

J Annual attack

Year Table 12. Comparison of annual increment of attacked and unattacked trees, expressed as percentiles, in the year immediately preceding attack, the year of attack and the year after attack in the four Lac le Jeune plots. Data taken from a subsample of felled trees in each plot.

Mean annual height growth (percentiles) + S.E.a Treatment Tree status N 1 year before attack Year of attack 1 year after attackb Unspaced Attacked 2 1 73.5 + 5.5a 59.4 + 4.4a 65.7 + 5.0a Unattacked 2 1 73.6 f 2.9a 77.6 + 3.2b 77.8 + 3.la

-1 r 2.4 m spacing Attacked 25 66.8 f 5.3a 67.8 f 3.9a 72.1 + 3.3a Unattacked 25 63.0 + 1.4a 86.7 4 3.3b 80.9 + 2.6a

3.0 m spacing Attacked 22 73.2 + 3.3a 60.1 + 4.4a 57.7 + 1.7a Unattacked 22 67.4 + 2. la 82.1 + 206b 67.1 + 4.4a

3.7 m spacing Attacked 22 75.4 + 4.7a 66.6 + 5.6a 69.2 + 6.la Unattacked 22 75.3 + 2.3a 91.3 + 2.6b 83.9 f 3.0b a Paired means (attacked vs unattacked) followed by the same letter are not significantly different, t-test, P < 0.05. b Trees which were attacked in the year following the attack year were excluded from the 'unattacked trees" category. Table 13. Incremental height growth of trees in the four Lac le Jeune plots combined (all spacings) in the year preceding attack, the year of attack and the year after attack by P. terminalis (N=42). Data were gathered from a subsample of trees felled and measured from each spacing regime.

Mean height increment (percentile) f s.E.~ Time perioda N Attacked trees Unattacked trees

1 year before attack 42 72.1 _+ 2.3a 69.6 + 1.3a

Year of attack 42 63.5 f 2.0b 84.4 + 1.6~

1 year after attack 42 66.1 f 2.5ab 77.3 + 1.9b a The lateral which assumed dominance was measured for attacked trees in the year of attack, and 1 year after attack. b Means in a column followed by the same letter are not significantly different, Tukey's test, P <0.05 for attacked trees; P <0.01 for unattacked trees. During the year of attack the height increment for attacked trees is significantly less than the increment for unattacked trees, t-test, P < 0.05. by about 11.5 and 7.9 cm, respectively. Similarly, in the years before and after attack, height growth was not significantly different between attacked and unattacked trees, but the length of the compensating lateral was significantly less (15.7 cm), than the height growth of unattacked tree leaders in the year of attack. The height increment of attacked trees, expressed as percentiles (according to the method described in section IV. B. 1.)(Table 13), was significantly less the year of attack (P < 0.05). In comparison, because trees which had been attacked were still exhibiting reduced height increment (due to laterals assuming dominance), the unattacked trees height increment was significantly greater for these two years in terms of percentile height growth for the stand (Table 13). When a stand of trees is spaced, as in the Lac le Jeune study, the number of trees in the stand is reduced, therefore decreasing the number of potential oviposition sites (leaders) available to P. terminalis. The remaining trees in a spaced stand will generally produce longer, thicker leaders which will in turn support more weevils. Due to fewer hosts being available for attack by P. terminalis, the numbers of attacks per host, over time, are greater in spaced stands. In unspaced stands which have comparable weevil populations, the expected frequency of attack per host over time would be less than in spaced stands due to the greater availability of hosts, in any given year, to P. terminalis. Ideally, replication of this spacing trial would have highlighted the differences in attack intensity and defect development in different stand densities. Due to the absence of additional treated stands in the Lac le Jeune area, no replication was - possible. One of the most common types of "controlled" experiment in field ecology involves a single "replicate" per treatment (Hurlbert 1984). Hurlbert (1984) states that when gross effects of a treatment are anticipated, or when only a rough estimate of effect is required, or when the cost of replication is very great, experiments involving

73 unreplicated treatments may be the only or best option. The validity of using unreplicated treatments depends on the experimental units being identical at the time of manipulation and on their remaining comparable to each other after manipulation, except insofar as there is a treatment effect (Hurlbert 1984). The uniform effect of the wildfire at Lac le Jeune created identical units, and other than the spacing, no other stand manipulations were done in the area. The growth of the internode before, at, and after attack was measured for each observed attack on the sub-sample of 90 trees felled from the four plots and was expressed as a percentage of the total potential height growth for that year. Thus, the mean percent height loss could be quantified for each density regime. Averaging across all densities, there was just under 25 % height loss in the year of attack and about 14 % height loss the year following attack (Fig. 13). There was little difference in percent height loss among the density regimes, but when expressed as real numbers the loss would be greatest in the lowest density stand where height increment was the greatest. Although the height loss due to one weevil attack is only one quarter that year's potential height increment, if there were multiple weevil attacks on a tree, the cumulative height loss could be significant. These results show a trend similar to the annual height loss figures for the IDF (Table 8). The height loss in the year following attack is comparable, at 17% vs. 14 % ; however, the height loss the year of attack is slightly greater in the IDF, at > 30% vs. 25 % . In real values (cm), growth rates are generally less in the IDF than in the slightly moister MS zone, but when expressed in relative terms, the height loss is greater. In the IDF, the subtending laterals, which assume dominance after attack by P. terminalis, are not as long as the laterals in the MS. Even though P. terminalis caused a height loss in the year of attack, the mean height and dbh of attacked trees were significantly greater than for unattacked trees in Figure 13. Percent height loss in the year of attack by P. terminalis and in the following growing season. Height growth of trees not attacked in a given year is compared to the length of the compensating lateral of trees attacked in that same year and in the year following attack. Year of attack year after attack

/ / / Natural 2.4 m 3.0 m 3.7 m Spacing each of the four plots (Table 14). The height loss, expressed as a percent of the potential height growth of that year, ranged from 21.8% to 27.1 %, in the 2.4 m spacing and 3.7 m spacing, respectively, in the year of attack (Fig. l3), and 10.9 % to 17.5% in the year following attack. The mean APA's of attacked trees in the 3.0 m and 2.4 m spacings were significantly greater than for the unattacked trees (Table 14). The absence of a significant difference between the APA's of attacked and unattacked trees in the highest and lowest density plots can be explained in part by studying the mapped trees and Voronoi polygons of these two plots (Fig. 14). The very regular spacing pattern of the 3.7 m spacing plot masks any "choice" based on APA that may be occurring on the part of the weevil within the small plot area studied. Conversely, the distribution of stems in the very dense unspaced plot approaches a random spacing, with CED=0.546 (Fig. 14a), again masking any "choice" by the weevil. Often an attacked tree was at the edge of a clump of trees, and although it has very close neighbors, this tree has a better chance of maximizing growth. The greatest difference in APA between attacked and unattacked trees occurred in the 3.0 m plot (Fig. 14, Table 14). This plot also had a mean NN distance for attacked trees that was significantly greater than for the unattacked trees (Table 14). The spatial pattern for all trees in each spaced plot approached a regular distribution, as did the pattern of all attacked trees (Fig. 14). The spatial pattern for all trees in the unspaced plot approached a random to clumped distribution, as did the cumulative attacked stems in this plot (Fig. 14). Due to the high density of the unspaced stand, a fairly small plot (0.023 ha) was established; if additional or larger areas had been mapped, perhaps the pattern of attacked trees would also have approached an even more clumped distribution. When the annual attack pattern in the unspaced plot is studied (Table 15), the pattern is clumped to random in the years 1971 to 1987. Table 14. Comparison of mean annual height, diameter at 1.3 m (dbh), APA and NN distance of attacked and unattacked trees in all four plots at Lac le Jeune, measured in the summer of 1988.

Means (+S.E.)a

Treatment Attack status Height (m) DBH (cm) APA (m2) NN (m)

Unspaced Attacked 6.56 + 0.14a 6.0 + 0.2a 0.65 If: 0.06a 0.42 & 0.03a Unattacked 5.57 + 0.13b 4.4 + O.1b 0.63 + 0.02a 0.84 f 0.41a

4 00 2.4 m spacing Attacked 7.36 + 0.06a 11.0 + 0. la 4.7 + O.la 1.37 + 0.04a Unattacked 4.97 + 0.18b 6.3 + 0.3b 3.7 + 0. lb 1.37 + 0.04a

3.0 m spacing Attacked 7.72 + 0.10a 14.6 + 0.9a 7.4 + 1.2a 1.80 + 0.07a Unattacked 4.08 + 0.25b 5.3 + 0.5b 4.4 + 0.3b 1.14 + 0.0%

3.7 m spacing Attacked 8.57 + 0.09a 14.7 If: 0.2a 13.6 + 0.3a 2.80 + 0.1 la Unattacked 7.64 + 0.39b 11.6 + 0.9b 13.6 + 1.3a 2.65 + 0.24a

a Paired means for each treatment followed by the same letter are not significantly different, t-test, P <0.05. Figure 14. Stem map of all trees in each of the four Lac le Jeune plots with Voronoi polygons drawn around each tree designating APA. Attacked trees are designated by shaded polygons. The CED statistic for all the trees in each plot and for attacked trees in each plot is given. Values equal to or < 0 approach a clumped pattern, and 22 approach a more regular pattern, with intermediate values indicating randomness. Unspaced 2.4 m spacing

CED, all trees =0.546 (P-0.582) Random CED, all trees = 1 1.125 (P < 0.001 1 Regular CED, attacked trees =0.013 (P =0.992) Clumped CED, attacked trees =5.992 (P<0.001) Regular

3.7 m spacing ,

3.0 m spacing

e m LO a CEO, all trees=4.461 (P 2 approach a more regular pattern, with intermediate values indicating randomness. Treatment Year CED P-value Spatial pattem Unspaced 197 1-76 -0.49 clumped clumped clumped random random random random clumped clumped random random random

2.4 m spacing random random

clumped clumped random clumped random random random clumped clumped

3.0 m spacing 1971-76 clumped 1977 clumped 1978 random 1979 clumped 1980 random 1981 random 1982 random 1983 clumped 1984 clumped 1985 random 1986 clumped 1987 random

3.7 m spacing 1971-76 1977 random 1978 random 1979 random 1980 random 1981 clumped 1982 clumped 1983 clumped 1984 clumped 1985 random 1986 clumped 1987 -0.02 0.984 clumped 82 Data in Table 15 present the yearly spatial patterns of attacked stems in the unspaced plot and 2.4, 3.0 and 3.7 m spacings, respectively. The patterns and CED statistic among the three spacing regimes varied each year, with more years having a random pattern of attack than a clumped pattern of attack in the 2.4 m and 3.0 m spacing. The 3.7 m spacing had more years exibiting a clumped pattern of attack than a random pattern of attack. The attack pattern in any given year was clumped to random. The amount of growing space available to a tree could influence stem form to varying degrees (Alexander 1960) in terms of stem taper, branch size, and angle. Open-grown, lodgepole pines tend to have very thick branches (Alexander 1960) which come off the main stem at near perpendicular angles. Therefore, when compensating for a killed terminal, open-grown trees could potentially form more severe defects than trees in dense stands. The fact that the major defect types (crooks, forks and stagheads) were more prevalent in the spaced plots than in the unspaced plot (Fig. 15) supports this hypothesis. This result agrees with observations by Maher (1982) that density influences formation of major or minor defects. Alfaro and Omule (1990) found with P. strobi attacks on Sitka spruce, that only the frequency of minor crooks was significantly higher in a wide spacing compared to a medium or close spacing, yet the close spacing had an overall beneficial effect on tree form. A significantly higher number of spruce trees were rated as having good form in the close and medium spacings (Alfaro and Omule 1990). In the Lac le Jeune study, however, there was little difference among the three spacing regimes (2.4 m; 3.0 m; 3.7 m). Forks and stagheads were grouped into one category because as soon as a "multiple-top" occurs a tree becomes of minimal commercial value above the point of defect, whereas a crook will be cut out of the tree at the mill, causing a loss of only a section of the tree above and below the crook. In most cases a forkedlstaghead tree is left on site, offsetting the cost of transporting it the mill (Paul Tearoe, pers. comm., Ministry of Forests, Figure 15. Distribution of defects, grouped by type, in the four Lac le Jeune plots. Bars within a type with the same letter above them are not significantly different (Tukey's test, P < 0.05). Unspaced plot 2.4 m spacing

03.0.... m spacing ISI 3.7 m spacing

Crease Crook Forkistag head Defect type

85 Kamloops, B. C.). 2. Attack dynamics in relation to stand age Pissodes temzinalis attack can be noticed in a stand when it is as young as 5-8 years of age, and the trees are still between 1-2 m in height, and < 2.0 cm dbh (Table 16). Trees appear to be most susceptible to weevil attack in the early stages of growth, between age 5 to 20 years. Number of weevil attacks was plotted against tree age for all four plots in the Lac le Jeune study (Fig. 16). The proportion of the total variation in percent stems attacked (Y) that is accounted for by age, is greater in the spaced stands than in the unspaced stand, ranging from 46% to 76% (Fig. l6), respectively. When the percentage of weevil attack of all four plots is plotted against age, approximately 77% (r2=0.774) of attack is explained by age. Attacks increase over the life of the stand to about 20 to 25 years of age. There are no data on levels of attack from stands over 26 years of age; however, it is predicted that attack levels would remain static or decrease after this age (Fig. 17). These stands were spaced at age 8, so some prior attacks may have been removed in the spacing process. In addition some attacks may have been "outgrown", and be no longer visible or attributable directly to P. terminalis. This outcome would be particularly true in the case of "bud attacks" when the bud is easily removed from the stem and the resultant defect may not be visible. The term "bud attack" has been used for the killing of the bud or partially- expanded terminal by early instar mining. In the 2.4 m and 3.0 m spacing (Fig. 16) the relationship between attack and age is 0.502 and 0.763, respectively, and shows a slightly increasing trend in weevil attack as the stand ages. In the 3.7 m spacing (Fig. 16) the relationship between attack and age yields an r2 of 0.499, indicating a similar relationship in the low density scenario. The difference in attack levels in the different densities at this post-juvenile Table 16. Summary of tree characteristics, attacked and unattacked by P. terminalis, in strip surveys from selected biogeoclimatic zones throughout the Kamloops Forest Region.

Geographic BEC Density Mean Mean Mean tree height (m)b Mean dbh (cm)b locations Subzone (stemsf ha) attacksf ha age Attacked Unattacked Attacked Unattacked Willis Cr. MSxk Willis Cr. (#222) MSxk Willis Cr. (#3 14) MSxk Willis Cr. (#243) MSxk Chewel's (#47 1) MSxk Chewel's Mtn. MSxk 00 4 Peachland (#196) MSdm2 Peachland (#697) MSdm2 Peachland Main MSdm2 Laluwissin Cr. (1988) MSdml Ellis Cr. (1989) MSdml Dillard Cr. (1989) MSxk 2,100 300 12 no no no no

a Areas were surveyed in the summer of 1987 except where indicated otherwise. b Differences between paired means, t-test, indicated by* P < 0.05, and NS =not significant. Where no data appear, tree measurements were not taken. no=no data collected. Figure 16. Plot of annual Pissodes termnalis attack in 4 spacing regimes in a stand at Lac le Jeune. P > 0.05 for all lines. Regression equations are as follows: unspaced, y=0.30x-1.85; 2.4 m spacing, y=0.54x-3.24; 3.0 m spacing, y=0.49x-3.14; and, 3.7 m spacing, y=0.69x-4.19.

Figure 17. A 3-dimensional linear regression plot of the percent stems attacked in. relation to the average age of stands and stand density (r2=0.57, F=16.62, df=25). Regression equation is, y =-0.001~~+ 1.75x2+0.66. Each point represents a unique survey in a different location, from all three biogeoclimatic zones, in the Kamloops Forest Region.

(>20 years) stage could be due in part to the host (leader) quality. The more widely spaced trees would have longer, thicker and more "attractive" leaders than the trees in the more dense plots. The brood success in these leaders would also be higher. By the age of 20 years, free-growing lodgepole pine has usually attained enough height to yield a sawlog from the lower bole (P. Lishman, pers. comm., Ministry of Forests, Kamloops, B. C .) . Therefore, prevention of defects caused by weevilling is extremely important for the first 20 years in spaced stands. There is variation in attack level based on site and stand parameters, but once a stand is post-juvenile, the incidence of weevil attack will represent a relatively minor impact to the host, although the incidence of weevil attack may not necessarily decrease (Fig. 17). However, if in a spaced stand, weevilling is not prevented during the first 20 years, continued weevil attack beyond this age could represent an impact to the host in terms of both increment and stem quality. These results show that early spacing, at 8 to 10 years, can increase diameter growth by two to three times (Table lo), supporting other demonstations on how stand density affects the growth and yield of lodgepole pine (Alexander 1974; Johnstone 1985). However, the number of clear stems produced is reduced as post-spacing density decreases (Fig. 17). In all three spacing regimes, nearly 50% of the stems remaining after spacing have been attacked at least once by P. terminalis (Table 9). Attacks per ha were greatest in the unspaced control area, 3,867 attacks per ha, but the percent stems attacked was only 2 1.6% compared to 74.3 % in the 3.7 m spacing (Table 9). In summary, the incidence of weevilling in lodgepole pine stands does not vary greatly among densities, but proportionally it increases with decreasing density. Relatively, the height loss attributable to successful weevil attack is similar among different stand densities, averaging 25 % height loss the year of attack and 14 % the following year. However, in absolute terms this height loss is greatest in low density stands. Successful attacks by P. tewninalis are aggregated, dependent on stand density and host spatial arrangement, and generally are on trees with ample growing space (APA). Defects appear to be more severe if a stand is spaced to a near-final harvest density at an early age. If spacing were delayed 5 to 8 years, provided that the stand was not stagnating, defects may in general be less severe and subsequent attacks would also have less impact on the trees' final form. A later spacing would also afford an opportunity to remove most weevil-attacked trees. V. SPATIAL ATTACK DYNAMICS AND IMPACT OF P. TERMINALIS IN DIFFERENT BIOGEOCLIMATIC ZONES The other plots, which include Ellis Creek and Conkle Lake in the MSdml, Ketchan Creek in the IDFdkl, two Dillard Creek plots in the MSxk, and Beblow Road in the ESSFdcl, were analyzed using the same methodology as for Okanagan Falls and Lac le Jeune. All the defects were coded, 0 =no defect, 1=crease, 2 =crook, 3 =fork and 4=staghead, and then compared among the plots from different ecosystems using a Chi-square analysis and Tukey's multiple comparison (Zar 1984; Wilkinson 1989). The number of attacks per tree was also compared in this manner. Plots were then grouped according to zone (IDF, MS and ESSF) and the number of attacks per tree and defect severity were analyzed as above. The differences in attacks per ha and percent stems attacked in the various densities is an artifact of spacing. Due to trees being removed in the spacing process, the total number of trees in an area is greatly reduced, therefore equal attacks over a ha of forest is very different in terms of percent stems attacked if one area has 1,000 stems and the other area has 5,000 stems or more. If trees are attacked at an early age there is a possibility that the defect caused by the weevil attack could change over time. This hypothesis was tested, using data collected from four plots, two spaced and two unspaced, using Kendall's tau-b statistic (Kendall and Stuart 1979). The plots used in this analysis were the spaced plots in Okanagan Falls and Dillard Creek, and the unspaced plots in Ketchan Creek and Conkle Lake. Assessment of defects was made in 1987 through 1990, except at Conkle Lake, which was assessed in 1989 and 1990. The - frequency of P. terminalis attack ranged from < 400 to > 1,600 attacks per ha in the Table 17. P. terminalis attack frequency at different stand densities in five biogeoclimatic subzones, noting the number of P. derminalis attacks per ha, percent stems attacked, attacks per tree and the defect type resulting from the attack. The defects were coded 1,2,3 or 4 according to the defect exhibited, with 1 =crease, 2 =crook, 3 =fork and 4 =staghead. Biogeoclimatic Age in Density Attacks % stems No. attacks per tre& Defect typeb Geographic location subzone 1990 (sterndha) per ha attacked (mean f S.E.) (mean + S.E.)

Ellis Creek MSdinl (spaced) ConMe Lake MSdml (unspaced) Dillard Creek MSxk (spaced) Dillard Creek MSxk (unspaced) Okanagan Falls IDFdml (spaced) Okanagan Falls IDFdml (unspaced) Ketchan Creek IDFdkl (unspaced) Beblow Road ESSFdcl (spaced) a Means followed by the same letter are not significantly different, Chi-square analysis and Tukey's multiple range test, P

I Figure 18. Frequency distribution of four defect types (forks and stagheads combined into one category) in six spaced stands and five unspaced stands located in the IDFdkl, MSdml, MSxk, and ESSFdcl subzones. MSdml Defect type

crease u~rook

O~ork/sta~head

Unspaced Spaced Unspaced Spaced

MSxk ESSFdcl

Unspaced Unspaced Spaced Spacing

98 Table 18. Comparison between P. terminalis attacked and unattacked trees in four plots from different biogeoclimatic subzones and of different stand densities. Biogeoclimatic Attack Means (4S. E.)a Plot location subzone status Height (m) DBH (cm) APA (1n2)~ NN (mlb Ellis Creek MSdml Attacked (spaced) Unattacked Conkle Lake MSdml Attacked (unspaced) Unattacked Okanagan Falls IDFdml Attacked (spaced) Unattacked IDFdml Attacked 'O Okanagan Falls 'O (unspaced) Unattacked Ketchan Creek IDFdkl Attacked (unspaced) Unattacked Dillard Creek MSxk Attacked (spaced) Unattacked Dillard Creek MSxk Attacked (unspaced) Unattacked Beblow Road ESSFdcl Attacked (spaced) Unattacked 2.49 4 0.05b 3.3 + O.1b 8.19 + 0.94a 1.84 + 0.19b a Paired means within columns followed by the same letter are not significantly different, t-test, P < 0.05. b APA =area potentially available, and NN =nearest neighbor a significant difference seen in the spaced plots in the MSdml, Ellis Creek, the IDFdml , Okanagan Falls, and in the ESSFdcl , Beblow Road (Table 18). In each of these plots the NN distance was greater for attacked trees than unattacked trees. The attacked trees in the spaced stands in the MSxk, Dillard Creek, had on average a greater NN distance, 1.75k0.07 m, than unattacked trees, 1.57+O.O3 m (Table 18), although not significantly greater. The fact that differences were apparent only in the spaced plots could be an indication that P. terminalis selects more open grown trees, which are growing clear of neighbors on all aspects as opposed to trees in natural, dense stands which have the larger trees growing on the edges of clumps with close neighbors on one or more sides. Part of the selection process by P. terminalis may be associated with the weevil perceiving the leader silhouette against other trees (in a dense stand or clump of trees) vs. perceiving the leader silhouette against open sky (low density stand or tree in opening). Although oviposition in leaders by P. terminalis may not result in successful emergence, the leader is usually killed by the mining of the larvae (Drouin et al. 1963). Unsuccessful attacks, which do not kill the leader, are relatively uncommon in the areas sampled in this study. Therefore, it can be stated with relative confidence that the weevils are choosing the larger leaders on hosts with ample growing space. The APA's of attacked and unattacked trees did not differ significantly in the two unspaced plots in the IDF zone or in the spaced ESSF plot (Table 18). The power of the test was small (1-1 =0.30); a larger sample size would have given a better idea of differences, if any, between APA of attacked and unattacked trees. For all other . plots examined, trees attacked by P. terminalis had significantly larger APA' s than unattacked trees (Table 18). The pattern of attack can be seen fairly well on the stem plots of these areas (Fig. 19-22). The unspaced stands, Dillard, Ketchan and Conkle, have a clumped distribution of trees as indicated by the highly negative CED statistics, Figure 19. Stem plots of all trees in the Dillard Creek spaced plot, upper diagram, and unspaced plot, lower diagram, (MSdml) with Voronoi polygons drawn around each tree to designate APA (area potentially available). Attacked trees are represented by the shaded polygons. The CED (Clark-Evans-Donnelly statistic) for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and >2 approach a more regular pattern, with intermediate values indicating randomness. Spaced

0 10 20 30 40 50m CED, all trees = 9.542 (P<0.001), Regular CED, attacked trees =-I .609 (P =0.1 O7), Clumped

Unspaced

0 5 10 15m CED, all trees =-7.275 (P<0.001), Clumped CEDI attacked trees =-3.981 (Pc0.001), Clumped Figure 20. Stem plot of all trees in the Ketchan Creek unspaced plot (MSdml), with Voronoi polygons drawn around each tree to designate APA. Attacked trees are represented by the shaded polygons. The CED for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and 22 approach a more regular pattern, with intermediate values indicating randomness. 0 10 20 30 m CED, all trees=-I1 .I 14 CED. attacked trees= -7.1 27 Pt0.001 Clumped P<0.001 Clumped Ketchan Creek Figure 21. Stem plot of all trees in the Conkle Lake unspaced plot (MSdml), with Voronoi polygons drawn around each tree to designate APA. Attacked trees are represented by the shaded polygons. The CED for all trees and attacked trees is shown below the plot diagram. Values equal to or 2 approach a more regular pattern, with intermediate values indicating randomness. 0 5 10 15 20 m CED. ail trees= -15.540 CED, attacked trees= -2.454 Pt0.001 Clumped P=O. 01 4 Clumped Conkle Lake Figure 22. Stem plot of all trees in the Ellis Creek, spaced plot (MSdml), with Voronoi polygons drawn around each tree to designate APA. Attacked trees are represented by the shaded polygons. The CED for all trees and attacked trees is shown below the plot diagram. Values equal to or < 0 approach a clumped pattern, and 22 approach a more regular pattern, with intermediate values indicating randomness. 0 10 20 30 am CED, all trees=lI .833 CED, attacked trees-1 .I 67 P<0.001 Regular Pz0.242 Random Ellis Creek -7.275, -11.114 and -15 SO,respectively (Fig. 19,20,21). In each year, for all plots, the attack pattern approaches a clumped to random distribution (data not shown). Unlike the Lac le Jeune and Okanagan Falls stem plots, the cumulative attack pattern remains clumped, except in the Ellis Creek (Fig. 22) plot which approaches a random pattern. Perhaps this is due to the extremely regular pattern of the host, CED = 11.833 (P < 0.01). If the spatial patterns are viewed on two levels, the distribution of the host, and the distribution of the insect's attack, it appears that many of the attacks occur on the edges of clumps of the host and thus create clumps of attack (Fig. 19-22). This pattern is not as clear in the Ellis Creek plot because of the highly regular pattern of the host (Fig. 22) created by the spacing treatment. In both subzones studied in the IDF, the IDFdml and IDFdkl, the attack rate was high, as was the severity of defects (Table 17, Table 19) relative to that observed in the MS and ESSF zones. The mean number of attacks per tree was over twice as high in the IDF as in the MS and was four times higher in the MS than in the ESSF zone (Table 19). The difference in defect severity was not as pronounced between zones, with defects in the IDF being 1.6 times as severe as defects formed in the MS and over twice as severe as the majority of defects observed in the ESSF (Table 19). The mean number of attacks per ha was not significantly different among zones when all plots were pooled but this is a reflection of the different densities within each of the zones (Table 19). Plots were established in a range of densities in each of the zones; therefore attacks per tree give a better description of the intensity of attack among zones. The incidence of major defects was high in both spaced and unspaced stands in the IDF zone and for that reason the timing and severity of juvenile spacing would be critical. There was no significant difference in tree age among the plots compared in Table 19. The incidence of weevilling was greatest in the spaced, IDFdml plot, having Table 19. Comparison between three biogeoclimatic zones of the number of P. terminalis attacks per tree, attacks per ha, average stem density, and the defect type resulting from the attack. The defects were coded 0,1,2,3 or 4 according to the defect exhibited, with O= no attack, 1=crease, 2=crook, 3 =fork and 4 =staghead.

N No. attacks per trd Attacks per hab ~ensit~b Defect typeb Biogeoclimatic zone (mean + S.E.) (mean rf: S.E.) (mean + S.E.) (mean + S.E.) r C IDF 0

ESSF 303 0.04 f 0.01~ 44f 9b 1666f519a 0.27 + 0.03~

a Means followed by the same letter are not significantly different, Chi-square analysis and Tukey's multiple range test, P < 0.01. b Means followed by the same letter are not significantly different, Chi-square analysis and Tukey's multiple range test, P < 0.05. 0.67+0.05 (+ S.E.) attacks per tree, or 48.1% of the stems attacked (Table 17). The number of attacks per tree in this plot was significantly higher (P < 0.01) with respect to the number of attacks per tree in the IDFdml unspaced plot, as well as having a higher rate of attacks per tree than any of the other subzones discussed in Table 17. The IDF has a harsher, drier climate than the other two zones, the MS and ESSF, and as a result trees are already spaced at a lower natural density; therefore the impact of weevilling could be quite severe in this zone. On average, the unspaced plots in the MS zone had fewer attacks per tree than spaced plots in the MS zone (Table 17). The MSxk, spaced plot had a significantly higher level of attack per tree than was observed in the MSxk unspaced plot, 0.26 k0.01 and O.O7_+O.Ol,respectively. Similarly, the mean number of attacks per tree was greater in the MSdm 1 spaced plot than in the unspaced plot, 0.20+0.02 and 0.13 k0.02 attacks, respectively, but the difference was not significant. There was no difference in level of attack in the spaced plots in the two subzones, within the MS zone, nor between the unspaced plots in these subzones. A cline appears between zones, dependent upon density. As the density is reduced in MS stands, the incidence of weevil attack approaches levels seen in higher density IDF stands; similarly, when stands in the ESSF are spaced, the attack levels approach those seen in unspaced MS stands. The incidence of weevilling in the spaced, MS zone was not significantly different from that observed in unspaced, IDF stands. Similarly, the number of attacks per tree in the unspaced MS zone was not significantly different from that seen in the spaced, ESSF plot (Table 17). The defect severity between zones and density regimes showed the same general trends as attack levels (Table 17). The severity of defect encountered in the spaced MSdml (0.83 k0.07) and IDFdml (1.03 k0.07) plots was significantly greater from all other plots (P < 0.05). The unspaced IDF and spaced MSxk plots had no significant difference in defect severity. The least severe defect as a result of weevilling was in the MSxk unspaced and ESSFdcl spaced, plots (Table 17). Overall, the greatest potential for the weevil to cause a major defect is in IDF sites and in low density MS sites, with the likelihood of defect formation decreasing in higher density MS and ESSF sites, respectively (Table 19). As observed in the spaced Okanagan Falls plot (Table 5), some defects change over time, usually to a less severe defect category, but many attack deformities are determined within the first one or two years following weevil attack (Table 20). Of the three plots analyzed, the defects assessed in the Conkle Lake plot changed the least, but the time period was only one year between assessments for this plot. The Dillard Creek (spaced, MSxk) plot showed the most shifting of defect from the or.iginal assessment in 1987 (Tau-b=0.363); however the majority of defects remained in the same category as originally assigned or shifted one defect category. Twenty-one percent of the attacks which were assessed as a minor defect in 1987 (crease) had developed into a major defect by 1990, either a crook or a fork (Table 20). The attacks typed as forks in + 1987 changed the most over time, with 73% of the forks becoming less severe and growing into crooks (4 attacks) or creases (7 attacks). Two thirds of the stagheads had changed into crooks by 1990. The defects typed as crooks in 1987 also became less severe with 37% changing to creases by 1990. The Ketchan Creek plot (unspaced, IDFdkl) showed a similar trend, with creases changing to more severe defects over the three year period (Tau- b=0.403)(Table 20). Twenty-nine percent of the attacks first coded as creases were assessed in 1990 as either crooks (10 attacks) or forks (5 attacks). However, in this plot there was less change in the crook and fork category over time, with 83 % and 75 % of the crooks and forks, respectively, remaining as such (Table 20). This result suggests that about 50 to 75 % of defects resulting from P. terminalis attacks assume a growth

112 Table 20. Numbers and categories of accumulated defects in three plots, noting the change in defect over time expressed as the Tau-b statistic.

Geographic 1987 assessment Distribution (numbers) of 1987 defects in 1990 location Type No. Crease Crook Fork Staghead Tau-b Dillard creek Crease (spaced) Crook Fork Staghead

Ketchan Cr. Crease C Y L, (unspaced) Crook

Staghead

Conkle Lakea Crease (unspaced) Crook Fork Staghead

a Distribution (numbers) of 1989 defects in 1990. form within the first 2-4 years post-attack and do not alter greatly from this form over time. The defects that do change generally become less severe. This early trend in defect formation will be further clarified when all plots are reassessed over the life of the stand. In summary, there is a difference between ecosystems in the number of weevil attacks per tree and the defect caused by attack. The highest incidence of weevilling occurs in the IDF and MS biogeoclimatic zones and these zones also exhibit the most severe defects as a result of P. terminalis attack. The spatial arrangement of lodgepole pine influences the patterns of attack seen in the different stand densities and biogeoclimatics zones. Trees having larger available growing areas were attacked more frequently in most stands. P. terminalis yearly and cumulative attack approached a clumped pattern, except in those stands which were very regularly spaced. VI. INFESTATION PHENOLOGY, EMERGENCE PATTERNS AND FECUNDITY

The activity of P. terminalis in a stand of young lodgepole pine could present a silviculturalist with a difficult situation, in which intensive silviculture to increase volume and enhance stem form may result in pest damage that has opposite effects. The principal goal of a forest manager is to promote the growth of healthy trees, and this can usually be done by reducing inter-tree competition. However, thinnings throughout the early years of lodgepole pine development to increase tree height and diameter growth would result in trees with long, thick terminal shoots, which in turn increases their susceptibility to weevil attack. The loss would be particularly severe in stands spaced early in the rotation age to the desired density at harvest, for there would be no further chance to cull out deformed trees. Greater understanding of the bionomics of P. terminalis would allow forest managers to understand the consequences of silvicultural prescriptions, such as spacing, on form and quality of trees attacked by P. terminalis. Therefore, the infestation phenology, emergence patterns and fecundity in the three biogeoclimatic zones were

' investigated.

A. Methods

From 1986 through 1991, leader collections of P. terminalis-attacked lodgepole pine were made in selected areas within the IDF, MS and ESSF biogeoclimatic zones. Leaders were held in the laboratory at 20-25OC, or outside in screen cages, so that conditions would simulate those in the field. Infested leaders were contained individually in rearing tubes or separated in groups by collection site in screen cages. Individual rearings with daily collection of insects generally ensured no mating of r:1 l emerging weevils. Numbers of emerged weevils were counted daily. When emergence was complete, each leader was assessed for basal diameter, total length, number of P. terminalis emergence holes, parasitism, and presence of other insects, e.g., Magdalis gentilis LeConte (Coleoptera: Curculionidae). M. gentilis was commonly found in terminal shoots of lodgepole pine during collections for P. terminalis. Emergent weevils were separated by sex (Harman and Kulman 1966) and used in various experiments over the course of the summer and fall months. Males and females from representative biogeoclimatic zones and subzones, ESSFdcl, MSxk, MSdml, MSdm2 and IDF dml, were paired and placed in 10 cm diameter petri dishes with a 6 cm-long section of fresh lodgepole pine leader. The lodgepole pine leaders used in all experiments were collected along road edges near the study areas from trees between 6 to 12 years of age. The pine section was replaced after 3-5 days and assessed for feeding punctures and oviposition. The objective of this experiment was to determine if time from emergence to onset of oviposition was similar in each ecosystem, and to determine the fecundity of newly emerged weevils from each ecosystem. On May 4, 1989, adult P. terminalis were collected from leaders and branches of lodgepole pine from an area in the ESSFdcl near Allendale Lake, south of Penticton, B.C. These weevils were presumed to have overwintered as adults in the duff. Seventeen pairs were collected and set up in petri dishes as described above for evaluation of their fecundity.

B. Results and Discussion

There wzs a significant difference in the mean fecundity of newly-emerged females and those which were collected as adults in the early summer (Table 21). This Table 21. Fecundity of newly emerged, first year P. terminalis, and overwintered adults, collected from trees in early May, 1989, near Allendale Lake, B.C. Oviposition occurred from May through September, 1989.

No. eggs laid per femalea P. terminalis status No. females Mean + S.E. Range

First year females 28 6.6 _+ 1.6a 0 - 33 Overwintered females 17 154.7 + 14.7b 75 - 289 a Means followed by the same letter are not significantly different, t-test, P

The term "operational leader clipping" refers to hand clipping and removal from stands of P. terminalis infested leaders during spring and early summer months, prior to weevil emergence. Figure 24. Mean numbers of days in 1989 from emergence to start of oviposition for P. teminalis females from three biogeoclimatic subzones. The number of weevils assessed (N) is indicated for each biogeoclimatic zone. Means followed by the same letter are not significantly different (Tukey 's test, P < 0.05). I ESSF (29)

IDF (40)

0 10 20 30 40 Mean number of days from emergence to first oviposition (+S.E.) 1% contained three or four eggs (Fig. 25). There was a significant positive relationship between the feeding by a female and the number of eggs deposited in a 3-5 day period (y=1.29x+8.05; r2=0.55; P

I 4 I

0 5 10 15 20 No. feeding punctures per 3-5 day period Figure 26. Data from leaders collected from 4 biogeoclimatic subzones in 1989 and held in the laboratory at 200C: a) mean length of infested leaders (+ S.E.); b) percent of infested leaders in which weevils were parasitized (by one or more species of parasite) and the percent of infested leaders which had successful P. teminalis emergence; c) mean number of weevils emerging per infested leader (+ S.E.); and d) percent leaders with secondary attack by Magdalis gentilis. N=250, 57, 59 and 92 for MSdml, MSxk, IDFdml and ESSFdcl, respectively. Bars with tthe same letter are not significantly different (Tukey ' s test, P < 0.05). Parasitism Emergence

IDFdml MSxk MSdm1 ESSFdcl MSxk MSdml ESSFdcl

1DFdml MSxk MSdml ESSFdcl IDFdml MSxk MSdml ESSFdcl Biogeoclimatic zone and subzone 128 Figure 27. Summary of leader characteristics, weevil emergence and parasitism over four years in collections from Okanagan Falls, B.C.: a) leader dimensions; b) number of weevils emerging per infested leader; c) percent of infested leaders with one or more weevils emerging; and d) percent of infested leaders with parasitized weevils. N=84, 59, 62 and 126 for 1986, 1987, 1988 and 1989, respectively. a)

- Lengrh -+- D~ameter

Year

1986 1987 1988 1 989 1986 1987 1988 1989 Year Year leaders, late in the summer, or that leaders continued to grow while under attack. Fewer than one weevil emerged on average per leader (Fig. 26c, Table 22), confirming data obtained for P. terminalis in California (Stark and Wood 1964). Mortality was very high, with over 60% of attacked leaders having no successful adult emergence (Table 22). In general, longer and thicker leaders were attacked, and the longer and thicker the leader, the greater was the successful weevil emergence observed (Table 22; Fig. 26). There was a significant difference in length and diameter (Tukey's test, P < 0.05) between attacked leaders which produced no adults and those that produced more than one (Table 22). Evidently the weevils prefer to oviposit on the longest leaders, which also occur on the tallest trees (Table 16). Thus oviposition late in the summer when leaders are fully elongated may be most advantageous to the overall fitness of the population in terms of numbers of weevils produced per stem. However, on some sites it may be advantageous for the weevil to attack elongating buds or only partially expanded leaders, in order to maximize development in a short growing season. Since the longest leaders can support the most larvae, there could be a higher likelihood of parasites discovering these larvae. The data in Fig. 26b support this hypothesis; the ecosystems exibiting the longest leaders, the IDFdml, MSdml and ESSFdcl, support the highest parasitism rates, suggesting a density dependent relation between parasite and host. In four years of data from the IDF subzone, the relationship of mean leader - length and diameter to the numbers of weevils emerging per leader and percent leaders with emergence was apparently obscured by high overwintering mortality and high .,i i t- levels of parasitism from 1987 through 1989 (Fig. 27a-d). In the IDF zone, most of the P. terminalis population exhibits the Type 2a life cycle (Cameron and Stark 1989). There is an early, but extended period of emergence @ersonal observation), and the

131 Table 22. Dimensions, and frequency of successful emergence by P. terminalis from 2,073 lodgepole pine leaders, collected from 1986 through 1989 in various locations throughout the Kamloops Forest Region.

Number of P. terminalis Mean leader dimensions (cm)a emerging per leader Frequency (%) Length Basal diameter

>3 2.1 48.3 + 1.4~ 1.29 + 0.03~ a Means followed by the same letter are not significantly different, Tukey's multiple range test, P<0.05. emerging weevils oviposit in fully expanded leaders the same summer they emerge. These new weevils have a lower fecundity than overwintered adults (Table 21). In 1987, 876 leaders attacked in 1986 were collected from the MS and IDF and two subzones within each zone. The highest P. terminalis emergence was from the MSxk subzone, coinciding with the lowest secondary attack by M. gentilis (Fig. 28). The highest incidence of M. gentilis was in the IDFdkl, directly below (in elevation) the MSxk which had the lowest incidence. In general, M. gentilis was most prevalent at low elevations, such as in the IDFdml, IDFdkl, and MSdml (Figs. 27,28 and personal observation). Silvicultural treatments, such as spacing, can result in high populations of M. gentilis, because the weevils are apparently attracted to slash even though it is not utilized by them (Fellin 1973). The MSdml also had fairly high secondary attack by M. gentilis, as well as the highest rate of parasitism and lowest P. terminalis emergence (Fig. 28). In my observations, incidence of parasitism was not constant between years (Fig. 27d), but did not vary greatly among biogeoclimatic zones and subzones (Fig. 28b). There was no clear relationship between parasitism and leader length or among parasitism and the other parameters recorded. More detailed study is needed on the parasite complex of P. terminalis and host parameters that may influence parasitism. The major mortality factors for P. terminalis are drowning in resin during the early stages of development (Drouin et al. 1963) and parasitism. The parasite complex varies geographically in both parasite diversity and abundance (Stark and Wood 1964;

- McLean and Kovacs 1989; Kovacs and McLean 1990), with Rhopalicus pulchripennis Crawford and two species of Eurytorna being most abundant in samples of P. terminalis infested leaders (Kovacs and McLean 1990). Parasitism levels can reach as high as 70% for Eurytornapissodes Girault (Stevens and Knopf 1974), and ranges on average Figure 28. a) Pissodes terminalis emergence, b) incidence of parasitism, and c) secondary infestation by Magdalis gentilis, in 1986-attacked lodgepole pine leaders collected in 1987 from two biogeoclimatic zones and two subzones within each zone. IDFdkl IDFdml MSxk MSdml b, (N=86) (N=115) (N =430) (N=245) b

IDFdrnl MSxk MSdml (N-245)

IDFdkl IDFdml MSxk MSdml (N=lIS) (N -430) Biogeoclimatic subzo~ personal observation). In the Kamloops Forest Region, P. terminalis exhibits a variety of life cycles with the most common being the type 2A and type 3 (Cameron and Stark 1989).There was much overlap in life cycles in the three biogeoclimatic zones; however the shortest time from emergence to first oviposition for female P. terminalis was in the ESSF. Frequently in this zone the type 3 life cycle was observed. Weevils which have overwintered once as adults are more fecund than newly emerged females; in addition, weevils can live for longer than two years (pers. observation). Therefore, if control such as leader clipping is considered (Appendix 111), certain aspects in the weevils' biology must be considered such as emergence dates, duration of life cycle and fecundity. Leader mortality caused by M. gentilis was fairly common, especially in the IDF; therefore it should be considered in any treatment for P. teminalis. Although mortality of P. terminalis is high, in part due to parasitism, these long-lived insects can lay an average of 154 eggs per female in a season. VII. THE SEARCH FOR A PHEROMONE MARKER Classically, the phases of host plant selection by insects have been termed host habitat finding, host finding, host recognition and acceptance, and host suitability (Kogan 1977; Prokopy and Owens 1983). However, another way to analyze this process, up to the point of insect arrival, is to focus on host plant resources at different hierarchical patch levels (i.e., forest, tree, leader), with the behavior of the forager defining patch and boundary level, and with a progressive narrowing of foraging activities as the spatial scale of patch level decreases in size. Insects appear to rank hosts, and show individual variance in acceptance of hosts (Futuyma and Peterson 1985; Courtney et al. 1989), based on vision, chemical cues, presence of a conspecific, or occupation of the host by eggs of another female (Prokopy and Owens 1983; Stansly and Cate 1984; Butkewich et al. 1987). The importance of the latter parameter should be related to the holding capacity or "overcrowding" of the resource unit (Romstock-Volkl and Wissel 1989), in this case the limited capacity of the terminal shoot. Some insects make use of a marker pheromone (Prokopy and Owen 1972; Prokopy 1981; Butkewich et al. 1987) to discriminate between occupied and unoccupied hosts, and still others such as the European apple sawfly, Hoplocampa testudinea (Klug)(Hymenoptera: Tenthredinidae), responds to wound exudates of host tissue in descriminating against apples already infested with conspecifics (Roitberg and Prokopy 1984). To determine if P. temzinalis females employ a strategy of marking their oviposition sites, choice bioassay experiments were conducted. In 1988 and 1991, newly emerged and mated P. temzinalis from several sites were used once they had begun to oviposit in late summer. In 1989, overwintered females from only the Allendale site (Fig. 1) were used. In 1988, a fecund weevil was presented with a choice "quality". The test sections were of similar dimensions (length and diameter) taken from the mid-section of a leader. All three sections in 1988 and two sections in 1989 and 1990 were cut at the same time and stored under similar conditions as the pre-test section until they were used. Initially, a female was allowed to feed and oviposit on one of the three sections for 24 h. At the end of 24 h the numbers of feeding and oviposition punctures were recorded. The second section was punctured with a small probe 6 times to simulate feeding or mechanical damage, and the third section was left untouched. The treatments were called "used", "damaged" and "clean", respectively. After one test female subsequently had 24 h with the three host choices, the leader sections were removed and assessed for feeding and oviposition. The difference in numbers of feeding and oviposition punctures recorded from the used, damaged and clean sections following the second 24 h period were analyzed statistically. Similar bioassays were run in 1989 and 1991, but excluding the "damaged" treatment, thus making it a two-choice bioassay. The numbers of feeding and oviposition punctures in the pre-test exposure were not significantly different from the sum of the test treatments in each experiment except for feeding punctures in 1988, when feeding was stimulated on damaged host sections (Table 23). Tn 1988 the used hosts were chosen for oviposition more frequently than clean hosts (Table 23) although there was no significant difference in oviposition between the used and damaged sections or damaged and clean sections. These results contrast with those in 1989 and 1991, in which P. terminalis fed and oviposited more in the clean than in the used hosts. The lower egg laying frequency in 1988 than 1989 may have been due to the youth of the weevils used and the biogeoclimatic zone from which they were collected. In 1991, 108 choice tests were performed with 18 mated pairs over a period from 12 August to 17 September. Replicates in which there was no Table 23. Results of 2- and 3-choice bioassay testing oviposition and feeding preferences of ovipositing first-year female P. teminalis (1988 and 1991) and overwintered females (1989). The duration of choice bioassays was 24 h in 1988 and 1989 (48 h total) and 48 h in 1991 (96 h total). Year, no. of replicates and No. feeding punctures No. oviposition niches status of test weevils Treatment (mean + S.E.)a (mean f s.E.)~ 1988 Pre-test exposure, 24 h 4.30 + 0.63 0.78 + 1.27 N=46 (newly-emerged weevils) Used (pre-test sections) 1.33 + 0.38a 0.46 f 0.17a Damaged 6.59 + 0.28b 0.13 + 0.06ab Clean 0.63 + 0.23a . 0.07 + 0.05b

+ 1989 Pre-test exposure, 24 h 13.43 + 0.63 4.83 + 0.32 W V) N= 106 (field-collected weevils) Used @re-test sections) 3.10 + 0.37a 1.46 + 0.21a Clean 7.55 + 0.58b 2.91 f 0.29b

1991 Pre-test exposure, 48 h 18.13 + 2.01 3.87 + 0.64 N=31 (newly emerged weevils) Used @re-test sections) 4.48 + 0.77a 0.58 + 0.23a Clean 8.29 $ 1.42b 2.42 f 0.50b

a Means, by year, followed by the same letter are not significantly different (1988, Tukey's test, P<0.05; 1989 and 1991, t-test, P < 0.05). b Means, by year, followed by the same letter are not significantly different (1988, Tukey's test, P < 0.05; 1989, t-test, P<0.05; 1991, t-test, P<0.01). oviposition or in which weevils did not oviposit on used or clean hosts were deleted from the analysis. Therefore, N=46, 106, and 31, for the 1988, 1989 and 1991 bioassays, respectively. In 1988, 33, 28 and 3 1% of the used, damaged and clean hosts, respectively, had no oviposition, whereas in 1989, 53 % of the used and 21 % of the clean hosts had no oviposition, indicating a significant initial preference (Chi-square test, P < 0.001) for clean hosts, that had not occurred in 1988. Similarly, in 1991, 74 % of the used and 42 % of the clean hosts had no oviposition (Chi-square test, P < 0.025). A strategy of avoiding used hosts would be of adaptive advantage for P. terminalis, in that repeated oviposition by the same or other females would be deterred. The solitary larva usually hatching from the single egg per puncture (Fig. 25) would then have a maximal chance of survival in the limited host resource of a new leader. Should the suggestion of a deterrent compound (Table 23; 1989, 1991) be upheld in future experiments, there may be practical utility of such a compound as a "biorational" pesticide. These trends were reflected in the selection of hosts for feeding. In 1989 only 10% of the clean sections had no feeding, compared with 43% of the used sections. In 1988, 27, 23 and 28 % of the clean, used and damaged sections, respectively, had no feeding. These results support the data in Fig. 25 which disclose a strong correlation between feeding and oviposition. The conflicting results between the two experiments could have been due to the fact that the ends of the host sections were not sealed in 1988; in the 1989 and 1991 bioassays the ends of all pine sections were sealed with - paraffin wax. This could have influenced the choice of the responding weevils in that there were more volatiles present in the test arena in the 1988 experiments. Therefore, either stimulatory or deterrent volatiles may have permeated the air within the petri dish arena's, masking the true response of the weevils. For example, host stimuli exuding from the cut ends of the used sections in 1988 could have ovemdden any deterrent

140 effect of prior feeding or oviposition. The bioassays run in 1991 were doubled in time duration to allow for more oviposition to occur during the test period (48 h). Future experiments are recommended that isolate the damaged from the used treatments. VIII. HAZARD RATING

A problem of concern to forest managers is the timing of stand entry to perform silvicultural treatments such as spacing or pruning. Timing of juvenile spacing in lodgepole pine, particularly in naturally regenerated stands, is especially critical because when left in a very dense state these stands can become "stagnant" or repressed. Spacing early in the life of a stand can prevent stagnation and promote rapid growth. Manipulating the space available to each tree in a stand offers the best and earliest opportunity for shaping the harvestable end product of both tree and stand. Available growing space has a profound influence on the diameter growth of individual stems, and consequently of their total size (Table 10)(Vyse 1985). After harvesting Crown timber in British Columbia, the forest licensee, or the Forest Service, must ensure that the new stand attains what is known as "free-to-grow" status, i.e., that the stand is stocked according to Ministry standards with trees of an acceptable species for that ecosystem, and is free of pest incidence3. In order to meet "free-to-grow" status, forest managers must predict levels of pest incidence that may be encountered, and levels of stand treatments that should be prescribed to achieve a vigorous, healthy, free- growing stand of lodgepole pine. Without an adequate hazard or risk rating system, making these predictions and prescriptions is very difficult. D. Handley4, a MacMillan Bloedel forester and veteran of the coastal "spacing wars", said this on the spacing topic: " .. .stocking standards are unquestionably more art than science, or

- -- Forest Act. 1979 (consolidated Nov. 3, 1989). Queen's Printer, Prov. of B.C., Victoria, B.C.

D. Handley. From minutes of Coastal Silviculture Committee meeting, February 12, 1985, Nanaimo, B.C. perhaps better, more one of hypothesis or opinion than firm knowledge - whether mensurational or economic. Each individual manager must be free to set and observe a standard that fulfills corporate strategy. The only criterion for non-acceptance should be an inability to provide enough data to show that the strategy is reasonable considering both biology and economics". With this statement in mind I present below a tool, supported by data, to which managers of young lodgepole pine stands can refer when determining when to space, where to space, and how much to space. The hazard rating guide presented in this section is a subjective system based on the probability of attack (Tables 8, 9, 18; Figs. 3, 9, 17) and the defect severity resulting after attack by P. terminalis (Figs. 5, 16; Tables 9, 11, 12). This hazard guide can be used by forest managers to predict what impact P. terminalis may have in a stand if certain spacing densities are applied, or conversely, what the impact may be if the natural stand density is kept intact. In this section, I will use the terms "hazard" and "risk" in the context of the following definitions: Hazard is determined by physical parameters such as characteristics of trees, stands, sites, and climate that influence susceptibility to attack by an insect given the presence of that insect. For individual trees, this means characteristics such as age or size, vigor, and location (Waters 1985). For a stand or area, hazard refers to characteristics conducive to population build-up of a particular insect species, e.g ., species composition, age-size structure, density, soil type, precipitation, disturbance, ecosystem and elevation. && is a function of insect abundance and distribution. It is the likelihood, or probability, of a damaging event occurring within a specified time frame and under a given set of conditions (Berryman and Stark 1985). Regardless of hazard, a significant number of insects must be in the general proximity for attack to occur (Amman and Anhold 1989). McMullen (1976b) examined ecological factors associated with P. strobi damage and developed an ecologically-based hazard rating system for Vancouver Island. This system is based on the interaction between degree days and weevil damage. In areas where accumulated heat in any given year is below 888 degree-days above 7.2OC, the minimum accumulated heat required for brood development (from egg to adult), Sitka spruce can be planted with little risk of serious weevil damage (McMullen 1976a,b; Heppner and Wood 1984). More recently, a survey was conducted to study the incidence of attack by P. strobi on white spruce in the Prince George Region of B.C., in relation to biogeoclimatic subzone, site quality class, and plantation age (Taylor et al. 1991). A general trend of increasing attack with increasing biogeoclimatic subzone moisture was found. A hazard rating system is now being developed for the Prince George Region based on McMullen's (1976b) work and the biogeoclimatic subzone relationship5. Hazardhisk rating systems have been used historically, with much success, to predict the occurrence and severity of forest fires (Lawson 1977). Hazard and risk rating methods are also used to assess infestation potential of forest pests, such as the mountain pine beetle (McGregor et al. 1981; Amman and Anhold 1989; Shore and Safranyik 1990). The Shore and Safranyik (1990) hazard/risk rating system takes into account both the susceptibility of the stand (hazard) i I I - and beetle population dynamics (risk), combining the two to produce a Stand Risk Index.

5 B. Seibans, graduate student, University of British Columbia, Forest Sciences; and, S. Taylor, B. C. Forest Service, Prince George Region, Prince George, B.C.

144 With the above def nitions of hazard and risk as a reference, I have developed hazard rating guide for immature lodgepole pine stands. The parameters that are used for hazard rating are age, area potentially available (.4PA), and ecosystem. Density, diameter (dbh), and height can be used to estimate the tree hazard within a stand. An accompanying risk assessment is a combination of the probability of attack in given hazard scenarios and an estimate of expected damage given certain levels of risk. The coarsest measure, but one of the most important in my hazard rating system is ecosystem, defined by biogeoclimatic zone. The biogeoclimatic ecosystem classification (BEC) groups similar segments of landscape, called ecosystems, into biogeoclimatic units and subunits; each represents a complex interaction of vegetation, animals, microorganisms and their physical environment (Lloyd et al. 1990; Meidinger and Pojar 1991). Biogeoclimatic zones represent large geographic areas with a broadly homogeneous macroclimate. These zones are readily distinguished on the basis of climax tree species for which the zones are named (Lloyd et al. 1990). Ten zones are found in the Kamloops Forest Region, namely: the Alpine Tundra (AT), Engelmann Spruce-Subalpine Fir (ESSF), Montane Spruce (MS), Sub-boreal Spruce (SBS), Sub- Boreal Pine Spruce (SBPS), Interior Cedar-Hemlock (ICH) , Interior Douglas-fir (IDF), Coastal Western Hemlock (CWH), Ponderosa Pine (PP), and Bunchgrass (BG). Zones occur along two distinct elevational sequences in the Kamloops Forest Region, which encompasses the dry central and dry western part of the Region where this study was done, and the wetter northeastern part of the Region. The zones take into account elevational clines and the sequence for the dry central and western part of the Region, from low to high elevation is, BG, PP, IDF, MS, ESSF and AT (Fig. 2). Several subzones are amalgamated to form a zone. Subzones are climatically less variable and geographically more confined than zones and often may be distinguished on the basis of succession (Lloyd et al. 1990). The scope of my study was not broad enough to detect variation among subzones but obvious variation and trends were seen among biogeoclimatic zones (Table 19; Appendix I, Table 5). The first step in hazard assessment of a stand is to determine the biogeoclimatic zone in which it is located. Without any additional information about that stand a very gross division could be made as follows: Biogeoclimatic zone Hazard IDF high hazard MS moderate hazard ESSF low hazard Parameters such as biogeoclimatic zone, age, and APA could be used to hazard rate stands. To determine the relative hazard of individual trees, or groups of trees, dbh or height could be used. By sampling a stand, the range of diameters (dbh) and relative frequencies can be determined and plotted as percentiles of the stand (Fig. 29). Figure 29 shows dbh as percentiles, and the frequency of attacked and unattacked trees within each percentile range from plots in the IDF and MS zones. To determine the hazard of a tree in a stand, the percentile in which the trees' dbh falls is compared to Fig. 29. Trees fitting into the 50 to 80 percentile range of dbh within a stand, in both the IDF and MS zones, are considered high hazard. Trees in the > 80 percentile range are moderate hazard and trees below fitting in the < 50 percentile range are low hazard. Within a stand, age would not play a very important role in determining hazard because of the seral nature of lodgepole pine. However, between stands, age is an g - important factor in hazard assessment. The greatest hazard would be for stands from the age of 5 years through 20 years. Stands above or below this age range would have lesser but equal hazard ratings. The third, and most critical parameter for hazard rating a stand, is area potentially available, or density. The area potentially available (APA) of any tree is

146 Figure 29. Pooled data on dbh percentiles for unattacked trees and those attacked by P. terminalis in all plots in the IDF and MS biogeoclimatic zones. The mean dbh percentiles for attacked and unattacked trees are indicated by vertical arrows. IDF zone

. .. . -- - ,. .. . x attacked (56.2)

DBH percentiles MS zone

DBH percentiles inversely related to density. In the IDF and MS zones, the difference in APA between attacked and unattacked trees remains fairly constant as density increases to a mid-range density (3,500 stemslha in the IDF and 7,500 stems/ha in the MS), and then decreases again at very high stem densities (Fig. 30). Therefore, in a stand of any given density up to about 5,000 stemslha in the IDF and about 10,000 stemslha in the MS, the trees with the most growing space have the highest hazard. APA is used as a hazard index rather than density, as it better describes the spatial arrangement of trees, whether the stands are naturally-regenerated, planted or spaced. The MS zone tends to have higher natural stem densities than are normally found in IDF zones having lodgepole pine as the seral species. Although attacked trees have greater APA, than unattacked trees, those in the 80 to 100 percentile range, are not as frequently attacked, as those in the 50 to 80 percentile range (Fig. 5). Thus some other components to host selection are possibly involved, such as a genetic component, or an interaction of host parameters influencing choice of hosts. The mean APA of a stand will give an estimate of the relative hazard depending on the biogeoclimatic zone and age of the stand (Fig. 31). Individual tree hazard can be estimated by calculating percentile ranges for the stand (Fig. S), in a similar fashion as hazard was estimated using dbh. Height and diameter of trees is highly correlated so dbh would be the choice when hazard rating individual trees, because it is easy to measure. For each stand, dbh is converted to percentiles and the number of attacked and unattacked trees falling into each percentile range is graphed, as in the examples in Fig. 32 for the three plots in the IDF zone. In all three cases, the mean dbh percentile of attacked trees was higher than the mean of unattacked trees (Fig. 32). When all the plots located in the MS zone were similarly plotted, the same relationship was observed. The plots were then pooled to produce an overall view of the trends in the IDF and MS zones (Fig. 29). Figure 30. The relationship between APA and density of attacked and unattacked trees in the IDF and MS zones is illustrated. Tine lines are fitted using a log regression. IDF zone

MS zone Figure 31. Hazard rating guide for stands based on biogeoclimatic zone, APA, and age. Hazard is divided into low, moderate, or high in each zone dependent on stand age and mean APA. APA AGE IDF

<5 w >5.0 m2 5-20 yrs >20 yrs

(4.0 m2 &20 yrs ~20yrs

<5 yrs ESSF <5.0 m2 5-20 yrs ':: >20 ys

<5 yrs 5.0-7.0m2H s2oyr~ ....a. . =-20yrs

>7.0 mi2 5-20 yrs ~20yrs

Low Figure 32. Area graphs showing the number of trees, both attacked by P. terminalis and unattacked, falling into each dbh percentile range in two locations in the IDF biogeoclimatic zone (Okanagan Falls spaced and unspaced, and Ketchan Creek unspaced). The mean dbh percentile for attacked and unattacked trees is indicated by a vertical arrow. Okanagan Falls spaced plot n~naltackedM~nacked I

Ketchan Creek plot (natural)

DBH percentiles 155 The chart in Fig. 31 incorporate the two hazard rating parameters for each biogeoclimatic zone. Because of the low stem densities in the IDF, an APA greater than 5 m2 is given the highest hazard in stands ranging in age from 5 to > 20 years, as well as in stands 5-20 years of age where the APA is in the range of 2.0-5.0 m2. The hazard varies among biogeoclimatic zones in order to account for the differences in probability of attack by P. terminalis based on my observations in these ecosystems. The hazard in the ESSF is always low to moderate even given fairly large APA because the attack levels seen in this zone were lower than the other two zones given comparable stand parameters. In the MS zone, stands with large average APA (low density), hazard can be as high as in the IDF. Lodgepole pine is slower growing in the IDF, putting on less height growth in an equivalent time to trees in the MS, therefor stands loder than 20 years can still be relatively high hazard. Typically, most stands in the MS have attained enough height increment by 20 years that one sawlog can be obtained. Therefor hazard decreases in the MS once a stand reaches 20 years. The lowest hazard zone is the ESSF. Due to fairly recent logging history in this zone, only a few sites were evaluated, and lodgepole pines in many hectares of the ESSF zone are just presently reaching the susceptible age of five years. APA is given greater ranges due to less defect severity in this zone. The largest stems of the very open-grown trees are the highest hazard trees within lodgepole pine stands in the ESSF. If lodgepole pine stands are rated as moderate to high hazard the next question is: what is the potential for damage to that stand? Frequency of weevil attacks per tree and the severity of defect in low and high density stands in the IDF and MS zones and a low density stand in the ESSF are compared in Table 17. The conclusion is that low density stands in the IDF, both spaced and unspaced, and in the MS zone have the damage when spacing at an early age to very low densities ( < 800 stems per ha) is clearly illustrated by the Lac le Jeune data (Figs. 14-16). In the MS zone, if a stand is spaced to below 800 stems per ha at about 10 years, up to two-thirds of the trees in the stand are at risk to attack by P. terminalis (Table 9, Fig. 12). The attack rate per tree can be as high as 2.59 (Table 11). The height loss in these open grown stands as a result of weevilling is about 25-30% of total annual height growth the year of attack and 15 % height loss the year following attack. This height loss compounded with a defect such as a crook, fork or staghead would decrease both volume and quality of the tree and stand, as well as prevent the stand from achieving "free-to-grow" status, or have the stand pronounced "free-to-grow" erroneously. Stand risk is described by the 3-dimensional drawing in Fig. 33. As one moves from the high elevation, moist zones, such as the ESSF, into the lower, drier elevation zones, the MS and IDF, and as the APA increases, the probability of attack, or risk, also increases. Stand age plays an important part in assessing hazard. Data plotted in Figure 18 indicate that, soon after stands reach five years or 1.5 m height, attack can occur. If between the ages of 5 and 10 years weevil attack is occurring, even at a rate as low as 2-3%of all stems attacked annually, managers should realize that the risk of weevilling will increase with age, especially if stem density is decreased. Density could be reduced to target levels in stages, thus lessening the impact of weevilling and removing deformed trees in the process. If stands are left with densities greater than 3,000 stems per ha, most weevil attacks will result in a crease being formed which is not a major impact to the tree or stand. The defect created because of weevilling can be fairly well assessed by the third growing season after attack. By this time the tree has determined the orientation of the laterals and only minor changes will occur over time (Tables 5, 19). Therefore, if stands are initially left at high densities, and are brought down to target density in two Figure 33. Stand risk is represented by a 3-dimensional plot of attack probability, APA and biogeoclimatic zone. The biogeoclimatic zones are: 1=IDF; 2 =MS ; and, 3 =ESSF.

to three entries, trees which have been weevilled more than three growing seasons prior to the spacing and which are only exhibiting a crease could be left. Conversely, trees weevilled three or more seasons prior to spacing and which bear major defects, such as crooks, forks and/or stagheads, should be removed during the spacing operation as they most likely will not compensate for their defects. The date when attack occurs, and consequently when the leader is girdled by larval feeding, influences formation of a defect. When a leader which has not fully expanded is attacked and girdled the resultant defect is usually less severe, than if attack occurs on a fully expanded leader late in the growing season. When only the bud is killed by the weevil, the tree compensates rapidly, usually with little noticeable defect at the point of attack. The selection of hosts by P. terminalis appears to be mediated by a number of factors. The spatial pattern of the host pines influences the attack pattern. When hosts are highly aggregated the attack pattern of the weevil is also very aggregated. When the pattern of the host becomes more regular, because of juvenile spacing or planting, the attack pattern of P. terminalis also tends to be more regular or random in pattern. During spacing, attention should be given to the spatial arrangement of stems, as well as attack levels; "feathering" the edges of the "clumps" of trees, removing attacked stems with defects in the process, may greatly improve the quality of the residual stand. "Feathering" a stand is a process which removes trees at varying intertree distances, leaving a range of large and small APA to trees in a stand rather than spacing to a rigid intertree distance (e.g., 3.0 m intertree distance), which is the standard practice at this . time in B.C. This style of initial spacing treatment will leave some clumps of lodgepole pine within the stand. If further attack by P. terminalis occurs after this initial spacing, these host aggregations could be spaced a second time, creating a more regular pattern of trees and remove attacked trees in the process. IX. CONCLUSIONS

My results indicate that P. terminalis may cause significant quality loss in the form of stem defects to lodgepole pines at levels far greater than previously realized. My interpretation of the results have demonstrated that several factors can influence both the likelihood of attack as well as the severity of the defect that can occur. Using Voronoi polygons to describe the area potentially available to a tree, statistical descriptions of the distribution of trees, and measurements of individual trees, I have shown that the most likely trees to be attacked as well as those most likely to sustain severe damage, are trees between 5 and 20 years of age and larger than normal size for their age, with ample growing areas. The early results of an experiment in which growing space was increased by experimentally thinning stands appear to support this conclusion. Surveys concentrated in three biogeoclimatic zones have indicated that there are variations in the weevils' life history, incidence and impact among zones. Biological studies have disclosed differences in the susceptibility of different potential host species to attack, the possibility of resistance to attack in the most vigorous trees, and the potential existence of an oviposition marker pheromone. All of these factors could influence host selection, spatial distribution and impact of P. terminalis. My research leads to two major points of practical importance. The first is that silvicultural treatments that result in early spacing of a stand can increase the intensity and severity of attack by P. terminalis. The second practical implication is that stands and trees can be rated for weevil hazard using an heirarchial evaluation of several factors, including biogeoclimatic zone, age, growing space, stand density and tree diameter (dbh). Hazard indexes compiled for individual stands could be used by forest managers to develop appropriate, stand-specific, management strategies. X. APPENDIX I

A. LONGTERMSPACING TRIALS

Five spacing trials were established in 1987-88 to evaluate the effect of density on both attack dynamics and defect severity. Four were located in the Kamloops and one in the Cariboo Forest Regions. The goal was to bring the density of natural stands of lodgepole pine down to near and below the operationally targeted goal of 1,600 stems per ha6 which is equivalent to 2.7 m target spacing. Therefore, the spacing regimes chosen for the trial were 2.0 x 2.0 m (2,600 stems per ha), 2.5 x 2.5 m (1,680 stems per ha) and 3.0 x 3.0 m (1,156 stems per ha). In each trial, areas were left at the natural stem density to compare the activity of P. terminalis in the natural density and artificially manipulated densities. The present targeted goal of approximately 1,600 stems per hectare is being re-evaluated and lower targets, closer to desired final stocking levels, are being considered. The target levels would be 600 to 1,000 stems per hectare or less.

1. Methods Each trial was located in a stand of lodgepole pine approximately 10 years of age at the time of establishment. No spacing had occurred in any of the chosen stands prior to establishment of the trials. Each spacing trial was a total of 4 hectares in size consisting of sixteen 50 m x 50 m blocks of the following specifications, which were the same for all five trials Four 50 x 50 m blocks each of: 2.0 m x 2.0 m spacing 2.5 m x 2.5 m spacing

ti Ministry of Forests, Silviculture Branch. Juvenile spacing guidelines for natural lodgepole pine stands based on 20 year age classes. September, 1987. 3.0 m x 3.0 m spacing unspaced control The 16 blocks in each 4 ha trial were laid out to best fit within the opening chosen and treatments were randomly assigned to each of the 16 blocks. Using a hip-chain to measure inter-tree distances, leave trees were marked at 2.0 m, 2.5 m or 3.0 m intervals within each block. The trees marked were as near to the desired spacing as possible and were relatively free of pest damage and weevil attack. Due to the natural spacing of trees and the aggregated pattern of weevil attack, sometimes a weevilled tree would be marked to be left in order to achieve the desired spacing. All trees not marked within the spacing blocks were cut out using either a brush saw or chainsaw. All trials located within the Kamloops Region, Stump Lake, Maka Creek, Monte Lake and Daves Creek, were spaced in the summer of 1987. The spacing trial in the Cariboo Region, located at Riske Creek in the IDF biogeoclimatic zone, was spaced in 1988. The geographic location, biogeoclimatic zone and subzone, dates of spacing and evaluation are summarized in Table 1. Each of the spacing trials was assessed in the summer of 1990. Three to four of each of the spacing treatments in each trial was surveyed (Table 2), assessing each tree for past or current weevil attack. All four control blocks were surveyed in the Monte Lake and Daves Creek trial, and two control blocks each in the Maka Creek, Stump Lake and Riske Creek trials were surveyed (Table 2). Surveys were done by dividing the blocks into narrow strips, ranging from 1 m to 5 m in width, depending on the density of the block. All trees within a strip were tallied and assessed for weevil attack and damage. The following records were taken for each tree encountered which was attacked one or more times by P. terminalis: height to attack; year of attack; total height; dbh; age; and the resultant defect. All attacked trees were stem mapped during the survey. Approximately 50 unattacked trees per block were randomly chosen and the Table 1. Location of five spacing trials in the Kamloops and Cariboo Forest Regions noting biogeoclimatic zone, subzone and date of spacing and evaluation.

Biogeoclimatic zone Geographic location and subzone Date of spacing Date evaluated Stump Lake IDFdkl Aug-87 Jul-90 Maka Creek IDFdk2 Sep-87 Jul-90 Riske Creek IDFbSa Jul-88 Jul-90 Monte Lake MSxk Jul-87 Jun-90 Daves Creek ESSFdcl Sep-87 May-90

- - a This plot represents the Cariboo Forest Regions biogeoclimatic zone classification. Table 2. Summary of survey assessments of the five spacing trials in the Kamloops and Cariboo Forest Regions. The three treatment regimes, 2.0 m, 2.5 m and 3.0 m spacing, plus the unspaced blocks are listed for each trial summarizing the total number of trees, by species, in each block, the number of P. terminalis attacks and the number of each type of defect observed.

Trial Treatment Total no. trees per block No. of Defect categoryb Locationa regime Pine Balsam Spruce attacks Crease Crook Fork Stag

Stump Lk. 2.0 x 2.0 IDFdkl 2.0 x 2.0 2.0 x 2.0 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 2.5'~2.5 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 unspaced unspaced

Maka Cr. 2.0 x 2.0 IDFdk2 2.0 x 2.0 2.0 x 2.0 2.0 x 2.0 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 unspaced unspaced

Riske Cr. 2.0 x 2.0 IDFb5 2.0 x 2.0 2.0 x 2.0 2.0 x 2.0 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 unspaced unspaced Monte Lk. 2.0 x 2.0 MSxk 2.0 x 2.0 2.0 x 2.0 2.0 x 2.0 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 unspaced unspaced unspaced unspaced

Daves Cr. 2.0 x 2.0 ESSFdcl 2.0 x 2.0 2.0 x 2.0 2.5 x 2.5 2.5 x 2.5 2.5 x 2.5 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 3.0 x 3.0 unspaced unspaced unspaced unspaced 883 232 I 1 2 a The Riske Creek trial is located in the Cariboo Forest Region and all other trials are located in the Kamloops Forest Region. 1989 attacks were not assessed for defect therefore are not included in the defect categories. following was recorded: total height; dbh; and age. An analysis of categorical procedure (SAS Institue 1985) was performed on the data comparing the mean number of weevil attacks per tree in each of the 4 density regimes in the IDF, MS and ESSF to estimate response probabilities (I?. Bellavance, pers. comm.). The data were then grouped by biogeoclimatic zone and analyzed for difference in attack levels (attacks per ha and percent stems attacked) between zones. Due to unequal variance, a log transformation was performed to stabilize the variance, and then tested using a Chi- square analysis and Tukey's test. 2. Results and discussion The Monte Lake location is the only trial to have undergone three seasons of weevil attack from the time of spacing to the date of evaluation. However, the assessment was done in May 1990; thei-efore 1989 attack may only be partially recorded since it is likely that many of the attacked leaders had not yet faded, therefore attacks may have been missed in the evaluation. Due to the timing of spacing in the other four spacing trials, only two years of potential weevil-attack occurred since spacing. These trials were established as long-term studies and this first evaluation was an early assessment to record any trends in attack intensity between biogeoclimatic zone or density regime. Defects were not analyzed at this time although they were recorded when the assessment was done. When the trials are assessed again in the future, defects will be analyzed and also the attacks now present will give information on the development and change of defects over time. All five trials will be assessed at five year intervals to detect any changes in weevil incidence and defect type over time. Table 3 summarizes mean stem density, attacks per hectare and percent stems attacked, by plot and by treatment. The density shown in Table 3 is the calculation of stems per ha based on the stem count and actual area (k0.25 ha) of each plot. Due to natural variation in tree spacing, the target density and actual density differed in several Table 3. Summary of density, number of Pissodes terminalis attacks per hectare and percent stems attacked in five spacing trials. The means of two-four blocks per treatment for each trial are used.

Geographic Biogeoclimatic Density Attacks Percent location zone Treatment (stemslha) per ha stems attacked Stump Lake IDFdkl 2.0 x 2.0 m 2.5 x 2.5 m 3.0 x 3.0 m unspaced

Maka Creek IDFdk2 2.0 x 2.0 m 2.5 x 2.5 m 3.0 x 3.0 m unspaced

Riske Creek IDFb5 2.0 x 2.0 m 2.5 x 2.5 m 3.0 x 3.0 m unspaced

Monte Lake MSxk 2.0 x 2.0 m 2.5 x 2.5 m 3.0 x 3.0 m unspaced

Daves Creek ESSFdcl 2.0 x 2.0 m 2.5 x 2.5 m 3.0 x 3.0 m unspaced cases (Table 3). Many of the P. terminalis attacks which occurred prior to spacing were removed in the spacing process. Therefore, the attack levels are fairly low in all of the trials and the majority of the attacks tallied occurred in the past two to three years. Attack level was highest in the three IDF trials and some trends are beginning to emerge as were seen in the other parts of this study. In the Maka Creek spacing trial, the attack level increases with decreasing density. The percent stems attacked are generally higher in the lower density blocks (Table 3) although there is no clear trend yet in the IDFdkl or IDFb5 subzones. Analyzing by biogeoclimatic zone, the three IDF trials were pooled, and the effect of treatment was tested. The difference among treatments may not have been great enough to distinguish between each of the spacing regimes, but the interaction between treatment and biogeoclimatic zone was significant (P < 0.00 l)(Table 4). The treatment effect was not the same for each zone. In the IDF the number of attacks in the two most severe spacings, 2.5 and 3.0 m, were similar, but they were higher than in the 2.0 m spacing and unspaced blocks (Table 4). In the MS all spacing treatments had higher attack levels than the controls, with no difference seen among the three spacing regimes (Table 4). Even though the attack level was low in the ESSF, there was a more pronounced difference among the different densities. The treatment effect was different in the ESSF compared to the IDF, with the lightest treatment, 2.0 m spacing, showing the highest response (Table 4). Pooling the treatments and comparing attack among biogeoclimatic zones, there was a significant difference in mean number of attacks per ha among zones (Table 5). The percent of stems attacked was significantly greater in the IDF than in the MS or ESSF (P < O.OS)(Table 5). At this early stage of the trial, attack levels are too low to show any strong trends, and that is why tieatment and ecosystem were pooled in the analyses. After further years of weevil pressure, it is expected that stronger trends will Table 4. The probability of P. temzinalis attack in each of the four densities in the five spacing trials grouped according to biogeoclimatic zone.

- - Response probabilitya Treatment IDF MS ESSF no treatment 0.087 0.023 0.018 2.0 m x 2.0 m 0.083 0.038 0.040 2.5 m x 2.5 m 0.109 0.04 1 0.021 3.0 m x 3.0 m 0.106 0.038 0.023 a The interaction between treatment and biogeoclimatic zone was significant (P < 0.001), with treatment effect being different in each biogeoclimatic zone. Table 5. The incidence of attack in each of the ecosystems in which the spacing trials were located, expressed as attacks per hectare and percent stems attacked (with log transformations), combining all density regimes.

Biogeoclimatic No. of Mean no. attacks per hab Percent stems attackedc zone plotsa Number Log transformation (%) Log transformation

IDF 4 1 188.9 2.12 + 0.04a 11.0 0.96 + 0.04a

ESSF 14 41.7 1.51 1 0.09~ 2.8 0.38 + 0.07b a Number of 50 m x 50 m plots assessed in each biogeoclimatic zone. b Means in columns followed by the same letter are not significantly different, Tukey's test (P < 0.01). Means in columns followed by the same letter are not significantly different, Tukey's test (P < 0.05). become apparent. The early indications from these trials agree with the results from the strip surveys and stem-map plot data (Chapter IV). Density influences the probability of an individual tree being attacked by P. terminalis and the IDF has the highest probability of attack, with mid- to low-density MS sites also having a high hazard of weevil attack. The ESSF generally seems to have the lowest attack levels of the three biogeoclimatic zones studied, although at this early stage of analysis when the overall attack intensity is low, the real differences between the MS and ESSF zones needs further study. XI. APPENDIX 11

A. INTRA- AND INTERSPECIFIC SUSCEPTIBILITY AND RESISTANCE TO P.

TERMZNA LZS In 1968, the British Columbia Forest Service initiated a comprehensive tree improvement program for interior spruce. Large, open-pollinated progeny trials were established and at the 10-year measurement, light to moderate attack levels of P. strobi were detected (Kiss and Yanchuk 1991). Cursory field examination indicated a pattern of differential attack among families. Thus a study was undertaken to determine the level of genetic variation in resistance to P. strobi attack, the magnitude of interactions among families and plantations for resistance, and to determine a relationship, if any, between weevil resistance and growth attributes (Kiss and Yanchuk 1991). The results from this study suggest that there is a moderate genetic basis for resistance to weevil attack in interior spruce and that selection for height and diameter growth may improve resistance to P. strobi attack. Evidence of genetic resistance to P. strobi has also been recently reported in Sitka spruce hybrids (Mitchell et al. 1990), and a suggestion of a genetic basis for weevil resistance has been described for Sitka spruce (Ying 1991). Similar resistance might be present in lodgepole pine to the lodgepole terminal weevil and should be investigated. Considerable research has been done on weevil-feeding mechanisms for P. strobi, related to host morphological and chemical stimulants or deterrents (VanderSar and Borden 1977b; Alfaro and Borden 1982). Recently, the concept of using a multicomponent index to characterize resistant trees has been proposed (Brooks et al. 1987; Brooks and Borden 1990) utilizing monoterpene composition and feeding deterrency. Efforts to identify chemicals that determine feeding deterrency by P. strobi produced variable results (VanderSar and Borden 1977b: Alfaro and Borden 1982), yet highlighted the fact that feeding responses probably play a major role in host selection. For example, partial or complete rejection of some non-host plants may be due to 173 inappropriate levels or blends of feeding stimulants or to the presence of feeding deterrents (Alfaro and Borden 1982). Feeding bioassays could lead to an understanding of the weevils' ability to change to a changing "host environment". P. strobi has great flexibility in what it will feed on in a force-feeding situation, and this may have implications to the field situation in terms of its adaptability to other conifer species that are not normally considered to be its host species. P. terminalis may also show this flexibility of feeding preference and if changing silvicultural practices offer the weevil a mix or change of species in its environment, the weevil may well adapt to feeding and breeding in new conifer species. 1. Intraspecific resistance In many young pine stands, cumulative weevilling may exceed 40% even when trees are still < 3 m high. However, it is common to see > 1 attack per tree (Maher l982), even when there appear to be acceptable hosts which are not attacked in the very near proximity to trees having multiple attacks (personal observation). A number of these potentially resistant trees, as well as apparently extra-susceptible trees were identified in the Okanagan Falls area (IDFdml) and tested in both field and laboratory bioassays. Following Brookes' and Borden's (1990) methodology, large lateral branches from susceptible and resistant trees were cut into 6 cm sections, attached end- to-end with an insect pin to simulate a single branch, and the exposed ends sealed with paraffin. These susceptible-resistant "branches" were placed with a female weevil in a large petri dish for 24 h. There was no difference in mean numbers of feeding punctures between the apparently susceptible or resistant sections (N=61). Due to a lack of oviposition (only one weevil oviposited three eggs in a section of susceptible pine) this parameter could not be compared. Fifteen pairs of susceptible and resistant trees were selected for field bioassay. In this trial, laterals from each tree were tied together and enclosed in a mesh sleeve , in a method adapted from VanderSar (1978). Male and female weevils were enclosed in the sleeve, and left to feed and oviposit for 10 day periods. There was no difference between potentially susceptible and resistant trees in terms of feeding by the weevils, and as in the laboratory bioassay, there was no oviposition by the test weevils. Therefore, while resistance to P. terminalis may occur in lodgepole pine, I do not foresee that it will be readily assessed by feeding or oviposition tests as appears to feasible for P. strobi (VanderSar 1978). Provenance or progeny trials, similar to those done by Ying (1991) and Kiss and Yanchuk (1991), with P. strobi on Sitka and Interior spruce, respectively, will have to be done with lodgepole pine to identify clearly any resistance to P. terminalis.

2. Acceptability of Ponderosa pine In 1986, choice feeding bioassays were done with various species of conifers indigenous to B.C.. In five experiments, adult P. terminalis were given the choice of four species in different combinations, but always including lodgepole pine, of 6 cm- long sections of approximately equal diameter, with ends waxed, placed at 12, 3, 6 and 9 o'clock in a 10 cm diameter petri dish. Because weevils of both sexes feed at equal rates (personal observation), weevils of both sexes were used in all five experiments. After 24 h, feeding punctures were counted on all sections. On sections of most species there was significantly less feeding than on lodgepole pine in at least one experiment (Table 1). However, ponderosa pine was fed on equally or more so than lodgepole pine. This result prompted additional experimentation.

I In Experiment #3 (Table I), the number of feeding punctures on western larch, i Nutt., was not significantly different from the number of feeding i Larix occidentalis punctures on either ponderosa pine, or lodgepole pine. However, in Experiment #2, which also tested western larch with ponderosa and lodgepole pine, there were Table 1. Results of choice feeding bioassays in terms of mean number of feeding punctures made by P. terminalis during a 24 h period. During each 24 h period, one weevil was given the choice of cut twigs of 4 conifer species.

Mean no. of feeding punctures in 24 h period + S.E.a Exp. #1 Exp. #2 Exp. #3 Exp. #4 Exp. #5 Species N=40 N=40 N=40 N=20 N=60 Lodgepole pine 1.9 + 0.66a 2.48 + 0.66a 2.45 + 0.66a 5.70 + 1.48a 3.35 + 0.57a Ponderosa pine 3.9 + 0362b 4.45 + 0.76b 2.00 + 0.52a 2.40 + 0.43a

Engelmann spruce 0.38 + 0.21a 0.15 + 0.09b Western larch 0.73 + 0.35~ 1.73 + 0.50a Western white pine 0.Ob Western red cedar 0.55 + 0.23b Scots pine 0.48 f 0.24b Norway spruce 0.33 + 0.18b Jack pine 2.00 + 0.85b a Means within a column followed by the same letter are not significantly different, Tukey's multiple range test, P <0.05. significantly less feeding punctures on larch compared to the pines. Western larch commonly occurs in the same stand as lodgepole pine, particularly in IDF and MS sites (Lloyd et al. 1990), and could be occasionally fed upon by P. terminalis in the field. A field trial was set up in which 5-year-old, potted ponderosa pine was placed next to a rooted lodgepole pine of corresponding age, height and diameter. A pair of weevils (1 male with 1 female) was then enclosed over these two trees with a mesh bag and allowed to feed and oviposit for three weeks. No oviposition and minimal feeding was found on the stem or leader of either species. The needles were not assessed but it is presumed that the majority of feeding occurred on them, or that the weevils did not feed extensively. In forced trials in 1990, weevils were enclosed with mesh sleeves on live ponderosa pine and allowed to feed and oviposit. The ponderosa pine was about 8 years old and the weevils used were newly emerged males and females collected that same season. Twenty pairs were enclosed on 20 trees from 1 August to 25 August, 1990. The weevils fed on the phloem but did not oviposit. This result could have been due to two factors; one, the weevils being first year females did not have a full egg complement, or two, the live ponderosa pine was not suitable for oviposition. In July-August 1988, 191 two-choice, laboratory bioassays using lodgepole and ponderosa pine were done. Lodgepole and ponderosa pine laterals of approximately equal diameters cut into 6 cm sections, were joined end-to-end as described above, and placed in the centre of a 10 cm diameter petri dish. One female P. temzinalis, which

. had been starved for 24 h, was placed in each dish and left for 24 h. In this . experiment, P. terminalis fed slightly more on ponderosa than on lodgepole pine, and no oviposition preference was shown (Table 2). It is difficult to base absolute conclusions on the results of these experiments, since the choice bioassays with living ponderosa and lodgepole pine were inconclusive.

177 Table 2. Results of a laboratory feedingloviposition bioassay in July and August, 1988, in which one female P. terminalis was allowed to choose between lodgepole pine or ponderosa pine for 24 h (N = 19 1).

No. feeding punctures No. oviposition niches Host (mean f S.E.)a (mean f S.E.)a

Ponderosa pine

Lodgepole pine a Means within each column followed by the same letter are not significantly different, t-test, P <0.05. On cut sections there appear to be clear preference for lodgepole and ponderosa pine over other coniferous species. However, there is equal preference for both lodgepole and ponderosa pine, or even a slight preference for ponderosa pine. If P. terminalis will not (or cannot) oviposit and feed on ponderosa pine saplings, as it does on cut branches, the interaction between the host and the insect may be similar to that found between western larch and the Douglas-fir beetle, Dendroctonus pseudotsugae (Hopkins)(Reed et al. 1987). On occasion, live, standing western larch have been attacked by Dendroctonus pseudotsugae, but with no successful brood emergence. However, successful brood production has been observed in felled larch (Ross 1967) at rates similar to that in Douglas-fir (Furniss 1976), the Douglas-fir beetle's primary host. Standing larch has a high 3-carene content in its xylem oleoresin, whereas felled larch has less. D. pseudotsugae attacks on felled larch were negatively correlated with 3-carene content (Reed et al. 1987). Further tests, including a monoterpene analysis of intact and severed ponderosa pine branches and leaders, should be done to determine if a similar situation to that described by Reed et al. (1987) is occurring. Should a biochemical difference be disclosed, a similar difference could be sought as a basis for resistance by lodgepole pine to P. terminalis. XII. APPENDIX m

A. LEADERCLIPPING AS A CONTROL OF'TION FOR P. TERMINALIS The concept of leader clipping is based on the hypothesis that by removing attacked leaders in the spring or early summer before adult weevils emerge, the resident population of insects in that stand will decrease; therefore attack and subsequent impact to the trees will also decrease. However, leader clipping trials with P. strobi in Sitka spruce have not conclusively supported this hypothesis (Jeklin 1980; Carlson and Wood 1984). Leader clipping trials for P. terminalis near Prince George (R.B. Hodgkinson, B.C. Forest Service, pers. comm.) in the early 1980's failed to have any effect on the weevil population. This trial was done in July, and depending on the emergence times of the weevil population in the Prince George Region, this late clipping date could have been post-emergence. It is critical to know the emergence dates of the particular population before attempting such control efforts. Other leader clipping trials were performed in the Cariboo Forest Region (L.J. Rankin, B.C. Forest Service, pers. comm.) from 1982 to 1984, again with inconclusive results. A factor which could mask the results of a clipping operation is the elimination of parasites with the clipping of weevil infested leaders. Hulme et al. (1987) suggested caging clipped leaders using a mesh size which was selective in allowing the smaller sized parasites to escape the cage but not allowing the larger girth P. strobi to escape. If clipping for P. terminalis seems to be operationally feasible, parasite enhancement should be given some consideration. Two regenerating stands of lodgepole pine in the Ellis Creek drainage, near Penticton, B. C., were chosen for a leader clipping trial for P. terminalis. Both are approximately 10 ha in size. Logging occurred in 1974 and the new stands were juvenile spaced in 1984. The goal was to clip and remove in early summer as close to 100% of the past summer's weevil-attacked leaders as could be achieved, for 4 consecutive years (1986 through 1989) in one of the plots. The other block served as an untreated control. To evaluate the result of the clipping treatment, strip surveys were done in May 1990 in both blocks to compare levels of attack and defect severity. The numbers of attacks per ha were higher in the treated block than the control block in 1984 and 1985, prior to commencement of treatment (Table 1). Attack levels rose steadily in the control block through to 1989. Attack levels also rose in the clipped block, except for a 40.5 % decline between 1988 and 1987. Overall, there were 4.9- and 3 .O-fold increases in attacks per year from 1986 to 1989 in the control and treated blocks, respectively. There was no significant difference in cumulative attack levels between the two blocks by 1989, with the numbers of attacks per hectare in the control and clipped block totalling 115.3 and 179.3, respectively (Table 1). There was a higher occurrence of creases in the clipped block than the control block (Fig. 1). This could be caused by the removal of competing laterals at the time of clipping, thereby allowing the trees to compensate more rapidly and with less stem curvature at the point of attack. Except for the frequency of stagheads, the occurrence of major defects was less in the clipped block than in the control block. In the clipped block, 46.9% of the trees attacked by P. terminalis incurred major defects compared to 84.9% of the attacked trees in the control block (Fig. 1). The distribution of defect types is significantly different in the two blocks (Chi-square, P < 0.05). The mean length and diameter of infested leaders removed increased in a linear fashion in both blocks over a three year period (Fig. 2a). The mean number of weevils successfully emerging per attacked leader decreased from 1987 to 1989, but the percent of leaders with at least one weevil emerging increased in 1988 and 1989, over that observed in 1987 (Fig. 2b-d). The percent of leaders which contained parasitised weevils increased in a similar pattern. Increased rates of parasitism could have caused the decline in number of weevils emerging per leader (Fig. 2d). Table 1. Comparison of annual P. terminalis attack, expressed as number of attacks per ha, in the leader-clipped block and control block located in Ellis Creek, B.C.

Attacks Der haa % change from previous year Year of attack Control Clipped Control Clipped 1982 2.6 0.0 1983 7.9 8.2 +67.1 1984 7.9 38.0 0 +363.4 1985 13.1 27.2 +65.8 -28.4 1986 23.6 59.8 +80.2 + 119.9 1987 34.1 114.1 +44.5 +90. 8 1988 55.0 67.9 4-61.3 -40.5 1989 115.3 179.3 + 109.1 + 164.1 a The cumulative number of attacks per hectare in the control and clipped blocks are not significantly different, t-test (P < 0.00 1). Figure 1. Comparison of the occurrence of four defect types caused by P. terminalis attack, expressed as a percent, in the leader-clipped block and unclipped, control block located in Ellis Creek, B. C. Crease 53%

Staghead 8% Clipped

Staghead 1/ 2% Fork 36%

Control Figure 2. Summary of leader characteristics, weevil emergence and parasitism over three years in collections from the Ellis Creek 'leader-clipped' block: a) leader dimensions; b) mean number of weevils emerging per infested leader (+ S.E.); c) percent of infested leaders with one or more weevil emerging; and d) percent of infested leaders with parasitized weevils. N=59, 62 and 126 for 1987, 1988 and 1989, respectively. 35l 'Length f Diameter 0.6 1988 1989 Year

1988 1987 1988 Year Year There are three main reasons why I conclude that leader clipping will not be a viable control option for P. temzinalis. 1. Timing of entries into a block to optimize the number of attacks that will be detected is difficult and variable, depending on site parameters and yearly weather. Leader fade varies depending on these two factors. If clipping is done too early in the spring, many attacked leaders will be missed because they have not yet faded. Early to mid-summer clipping allows a higher percentage to fade, and those that have still not faded can be detected because they are not "candling" (elongating). However, even a trained eye can miss these attacks, as well as "bud attacks". Therefore, multiple entries into a stand would be needed to approach 100 % efficiency. 2. The process of leader clipping is labor-intensive, and could be very costly if multiple entries into a stand were necessary. When trees are relatively small, clipping is easy. Once trees are > 4 m high, clipping becomes increasingly difficult, and will lead to leaving attacks in the stand. 3. P. temzinalis can live for more than two years, possibly up to five years as does P. strobi (McMullen and Condrashoff 1973), and as the data in this study indicate, overwintered weevils are more fecund than first-year, newly emerged adults (Table 4). Given these two factors, plus the fact that there are many areas of young, regenerating pine that are potential sources of immigrating weevils, even if clipping were 100% effective in a stand each year, there would be continued attack. xm. APPENDIX rv

A. ROOTWEEVILS: A POTENTIAL CONFOUNDING FACTOR A weevil morphologically identical to P. terminalis was observed feeding and ovipositing on the roots and lower bole of young lodgepole pines in and near three study areas, at Ellis creek, Daves Creek and Okanagan Falls. This fact was of particular interest because it is presumed that the duff and areas around the boles of trees is the overwintering site for P. terminalis, and is an occasional habitat throughout their active season. This root-dwelling weevil has been described as Pissodes schwarzi Hopk., the Yosemite bark weevil (Wood 1964; Stevens 1966). Work done by Smith and Takenouchi (1962) on chromosomal polymorphism in P. terminalis implicates P. schwarzi in the hybrid origin of P. terminalis. Pissodes schwarzi attacks and breeds in the bole, root collar and larger roots of stressed or dying trees (Wood 1964; Stevens 1966). Hopkins (1911) and Smith and Sugden (1969) list its hosts as Larix occidentalis Nutt. (Western larch), Picea engelmannii (Engelmann spruce), P. glauca (White spruce), P. mariana (Mill.) B.S .P. (Black spruce), P. pungens Engelm. (Blue spruce), Pinusponderosa (Ponderosa pine), P. albicaulis Engelm. (Whitebark pine), P. contorta (Lodgepole pine), P. flexilis James (Limber pine) and P. monticola Dougl. (Western white pine). In British Columbia, P. schwarzi is commonly found in lodgepole pine infected with Comandra blister rust, Cronartium comandrae Pk. (Furniss and Carolin 1977) or other damaging agents. Host selection by P. nemorensis has been shown to be pheromone mediated (Fontaine and Foltz 1982). Males release grandisol and grandisal, originally found in

. the boll weevil, Anthonornus grandis Boheman (Tumlinson et al. 1969), which attract both males and females (Phillips et al. 1984). Both P. schwarzi and P. nemorensis exhibit similar habits, attacking boles and root collars of young trees. This portion of the study investigates the hypothesis that P. schwarzi produces an aggregation pheromone and describes some aspects of the weevil's life history, genetics, and habits. 188 The root weevils were found breeding in the cut stumps left from the spacing operation at the site which was leader clipped in Ellis Creek, and had begun to attack living, although highly-stressed trees.

\ 1. Genetic relationship of P. teminalis and P. schwani Boyce et al. (1989) found that three species of pine weevil (Pissodes strobi, P. terminalis and P. nemorensis) possess a mitochondrial genome of unusually large size (30 to 36 kb). Weevils sampled in all three species (Boyce et al. 1989) exhibit from two to five distinct size classes of mitochondrial DNA (mtDNA). Using mtDNA polymorphisms in these three species has made it possible to assess the degree and nature of genetic differentiation. In 1988, a sample of approximately 20 adult P. temzinalis and 20 P. schwani were sent to Boyce7 for analysis. P. strobi, P. terminalis and P. nemorensis are all members of the Pissodes strobi species group, and are considered to be closely related (Phillips 1984). Yet, the mtDNA lineages of the three are distinct, differing in sequence by as much as 3.66% in the coding portion of the molecule (Boyce et al. 1989). However, the mitochondrial genomes of the three species retain a similar structure. P. strobi typically has the largest genome, and P. terminalis the smallest. The estimated divergence between P. strobi and P. nemorensis is 2.87%; between P. strobi and P. terminalis, 3.66%; and between P. nemorensis and P. terminalis, 1.83 % . P. schwarzi was found to be distinctly different from P. temzinalis, having a large genome, similar to P. strobi (T. M. ~oyce', pers. comm.). Therefore, although cross-breeding is possible between P. teminalis and P. schwarzi, they are most likely two distinct species.

T.M. Boyce, Section of Genetics, Biotechnology Building, Cornell University, Ithaca, New York. a. Isozyrne study methodology Because P. terminalis and P. schwarzi cannot be distiguished morphologically, and because they are found in the same host trees in B.C., an isozyme study was initiated in collaboration with Dr. George Harvey, and Mrs. P.M. Roden, Forest Pest Management Institute, Forestry Canada, Sault Ste. Marie, Ontario. Collections of P. schwarzi and P. terminalis were made from study sites within the Kamloops Forest Region (Table 1). In the terminal collections, Magdalis gentilis was also found. Insects of the "terminal population" (PIL and MAG)(Codes described in Table 1) were all shipped in the terminals; larvae were dissected out, adults were allowed to emerge (Table l)(Harvey et al. 1991). P. schwarzi was sent as larvae in root collars and large roots of infested lodgepole pine (PIR-1 and PIR-2) and as adults which were collected from root collars in the field (PIR-3). Adult P. strobi were collected from the Sault Ste. Marie area in 1990 and analyzed as a comparison. Electrophoresis was carried out using horizontal starch gels according to procedures adapted for the spruce budworm (Harvey and Sohi 1985). Insect numbers in several collections were too low to constitute a valid sample; therefore they were pooled by stage and collection site: root collars (PIR) versus leaders (PIL), larvae (L) versus adults (A)(Table 1). An attempt was made to verify bands with those reported elsewhere for P. terminalis and P. strobi (Phillips 1984).

b. Results and discussion Results of the analysis indicated that P. schwarzi and P. terminalis larvae are very similar. Among the Pissodes spp. genetic distances (mean D-value=2.29) between groups were fairly similar, except that P. strobi was more widely separated from the rest, and Root PIR-A was slightly separated from the other lodgepole pine collections (Table 2)(Harvey et al. 1991). The genetic distances between Magdalis and all the Pissodes samples including P. strobi were the greatest, consistent with taxonomic

190 Table 1. Collections of terminal and root collar weevils for isozyme analysis (Selected data from isozyme study done by G.T. Harvey and P.M. Roden, Forestry Canada, Ontario Region, Great Lakes Forestry Centre, Sault Ste. Marie).

Code Type Insectsc Date Location Tested GrouDd PIR- 1 Collar 55L 31/1/89 Kamloops area 37L PIR-L PIR-2 Collar 2L 6/2/89 Kamloops area 2L PIR-L PIR-3 Collar - 21A 21/6/89 Ellis Creek 19A PIR-A PIL- 1 Terminal 3L 3011189 Kamloops area 3L PIL-L PIL-2 Terminal 3L 6/2/89 Kamloops area 3L PIL-L p1~-3a Terminal 4L, 1A 6/5/89 Kamloops area 4L, 1A PIL-L, A PIL-4 Terminal 8A 26/7/89 Ellis Creek 7A PIL-A PIL-5 Terminal 6A 25/8/89 Ellis Creek 4A PIL-A GAL- 1b Terminal 8/90 Garden L. ON 11A GAL Ma@ Terminal 2P, 9A 6/5/89 Kamloops area 2P, 9A MAG a Collection was found to contain insects tentatively identified as Magdalis gentilis, as well as Pissodes terminalis. b Pissodes strobi from the Sault Ste. Marie area, Ontario. Stages of insects: A=adult; L=larvae; P=pupae. d Groups used in analysis. Table 2. Matrix of distance coefficientsa averaged by groupb (Selected data from isozyme study done by G.T. Harvey and P.M. Roden, Forestry Canada, Ontario Region, Great Lakes Forestry Centre, Sault Ste. Marie).

Species No. of pops. 1 2 3 4 Root-collar 2 0.173 (0.173-0.173) Leader 2 0.140 0.069 (0.016-0.264) (0.069-0.069) Strobi 1 0.465 0.290 *~(c***c (0.276-0.655) (0.244-0.336) (****-****I Magdalis 1 1.765 1.826 3.283 *s**c (1.640- 1.890) (1.567-2.086) (3.283-3.283) (****-****I a Coefficient: Nei, M. 1978. Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics 89583-590; unbiased genetic distance. Pissodes.mss Feb. 20, 1991lMay 7, 1991. Only one population included separation at the generic level (Harvey et al. 1991). Among the Pissodes groups D- values were much smaller (mean D=0.298) and were consistent with the distances expected among species or subspecies (Harvey et al. 1991). The smallest value within a group was that within leaders (D=0.069), confirming by its low value that isozymes of these 2 developmental stages (larvae and adult) are very similar or identical in this species. Within the root collar group, the genetic distance was 2.5 times as great (mean D =O. 173)(Table 2), indicating a greater divergence between larvae and adults in this group (Harvey et al. 1991). Comparison with published data on P. terminalis and P. strobi (Phillips 1984; Phillips and Lanier 1985) show that the Harvey et al. (1991) results are in agreement at the 4 loci in common. Taxonomically, P. terminalis and P. strobi are close and have very similar isozyme patterns and frequencies (Phillips 1984; Harvey et al. 1991). Phillips (1984) showed a genetic distance of approximately 0.170 between P. terminalis and P. nemorensis based on 11 loci. The Harvey et al. (1991) data, based on 9 loci (4 in common) produced a somewhat higher 'Dl value and the Root PIR-A fell between P. strobi and the P. terminalis groups (Harvey et al. 1991). The results from the isozyme study (Harvey et al. 1991) indicate that larvae and adults collected from lodgepole pine terminals and larvae from root collars are all very similar, and could possibly be the same species, most probably P. terminalis. Isozyme frequencies at 9 loci in 39 larvae from root collars were essentially indistinguishable from those of insects found in the terminals. Genetic distances among these 3 groups

- (PIR-L, PIL-L and PIL-A) averaged 0.043 (Harvey et al. 199 1). Weevils collected as adults feeding on root collars in June (PIR-A) showed some isozyme differences from the presumed P. terminalis collections and root larvae (PIR-L)(Table 3). Isozyme frequencies in this group of insects are consistent, indicating Table 3. Allozyme frequencies in 5 collectionsa of Pissodes sp. and 1 collection of Magdalis gentilis (Selected data from isozyme study done by G.T. Harvey and P.M. Roden, Forestry Canada, Ontario Region, Great Lakes Forestry Centre, Sault Ste. Marie).

LOCUS PIR-L PIR- A PIL-L PIL-A GAL MAG

PGN-1

MDH-1

IDH- 1

IDH-2

ME-1

LDH-1 AT- 1 (N) 39 19 9 A 0.038 B 0.026 C 0.064 0.079 0.056 D 0.872 0.895 0.944 E 0.026 AAT-2 (N) 30 1 6 A B 0.817 1.000 0.833 C 0.183 0.167 a Collection identities in Table 1. Major alleles bolded. Dummy values in parenthesis. that the collection was not a mixed one. However, there are marked differences in some enzymes from the other 3 groups from lodgepole pine (Harvey et al. 1991). The differences between the adults collected from the roots and larvae from the roots (PIR-A and PIR-L) could be from differences between developmental stages. However, there were insignificant differences between the larvae and adults from the leaders (Harvey et al. 1991). In addition, the larvae from root collars appeared closer to the leader population. There is a possibility that the adults collected from around the root collars are not the same species as the larvae developing in the root collar. However, the differences seen could be due to variations within a population, accentuated by the inter-stage differences. Although not conclusive, in part due to small sample sizes, isozyme analysis seems to show no difference at the species level, but perhaps a difference at a race level. However, the data generated are insufficient to refute the conclusion based on mtDNA that the two species are valid. Additional studies are needed to define clearly the similarities and or differences between these groups. If the larval population in the root collars is P. terminalis, this could have important implications on the future management of young lodgepole pine stands.

2. Pheromone-based communication and bionomics of P. schwani Pitfall traps modified slightly from those used to catch Hylobius abietis (L.) (Tilles et aE. 1986a,b; Nordlander 1987) were constructed from 30 cm lengths of PVC plastic drainpipe with a 10 cm inside diameter. The pipes were inserted into the soil so that 8 equidistant holes (6 mm diam.) drilled around the circumference at mid-point of - the pipe were at ground level. A thin coating of TanglefootR was applied every 3 weeks to the above-ground portion of the trap to catch any responding weevils which might climb the trap. Experimental traps all had a 4-5 cm long section of fresh lodgepole pine wrapped in a fine mesh fabric and suspended at ground level inside the trap. Four treatments were: 1) one male on a pine section; 2) one female on a pine section; 3) pine alone; and 4) an unbaited control. The bait weevils were collected 20 May 1989 on lodgepole pines infested with comandra blister rust. The traps were placed in the naturally-regenerated stand of lodgepole pine (average age 12 years) at Ellis Creek, 15 km east of Penticton, B.C. The stand was juvenile-spaced in 1983 and had sustained about 35 % infection by C. comandrae. Approximately one third of the infected trees, or 10% of the trees in the stand, showed past or current evidence of P. schwarzi infestation. Between 21-23 May 1989, 60 traps were placed in 4 rows of 15 traps each, spaced approximately 12 m apart, with 15 m between rows. At approximately weekly intervals from 1 June to 1 September 1989, captured weevils were collected, and the pine, and weevil baits if necessary, were replaced. Captured weevils were separated by sex (Harman and Kulman 1966). Observations from the trapping study were related to weather recorded by the B.C. Forest Service about 2 km from the Ellis Creek site, at the same elevation. To study the bionomics of P. schwarzi, 5 to 8 infested pine roots were collected at approximately weekly intervals from 1 June to 23 August 1988 from 3 sites in the Penticton area (Okanagan Falls, Daves Creek, and the trapping site, Ellis Creek). The roots were subdivided into bole, root-ball, and lateral roots, and dissected; the numbers of weevils in each life stage, as well as empty chip cocoons were recorded. Infested roots were collected from Ellis Creek on 21 May 1989 and adults were allowed to emerge in the laboratory. Oviposition by emergent weevils was studied by placing a male and female on a 6 cm length of fresh pine in a 15 x 2.5 cm petri dish. The pine was replaced every 3-5 days and assessed for oviposition. a. Results and discussion

, Forty-eight P. schwarzi females were captured throughout the summer in traps baited with males on pine sections (two of these were captured in the Tanglefoot). No weevils were captured in unbaited traps, or those baited with females on pine or pine sections alone. Forty-six of the P. schwarzi captured were females, indicating a male- produced sex pheromone. The lack of response to all but the males-on-pine treatment suggests that P. schwarzi does not respond to an attractive tree trunk silhouette for visual orientation, with or without host volatiles, similar to results obtained with H. radicus Buchanan (Hunt and Raffa 1989)g. The seasonal response of P. schwani to the male-baited pitfall traps indicates peak periods of activity in early June and mid- to late July (Fig. 1). Trap catches were generally highest in warm weather. The first seasonal peak probably represents overwintered adults, and the second peak newly emerged adults. Adults were collected from around the boles of stressed pine on 10 May 1989 and all 6 of the females collected were ovipositing. These females when paired with males laid 4.1 f 0.3 eggs per day (mean t S .E.) from 10 May to 1 June 1989. Dissections of infested roots collected periodically from three locations throughout the summer revealed a fairly high frequency of larvae in the host from early June, to the end of August. The frequency of pupae increased from late June through late July, and decreased in August. The frequency of adults in the host varied only slightly between sites, and generally increased from late July to early August. Separation of larval instars visually into early versus late disclosed that late instars were most frequent in early June and August, and early instars from mid-June through late July, corresponding to the observed activity of overwintered adults in the field. As also noted by Stevens (1966), all developmental stages were represented during July and August. However, oviposition in B.C. begins in early May and continues through

Three Hylobius warreni and one Magdalis sp. were caught in response to the male- on-pine treatment, suggesting a possible cross attraction to P. schwarzi. Figure 1. Numbers of Pissodes schwani caught in pitfall traps, by collection date, from June 1 to September 1, 1989, and corresponding average daily temperature for each period. All pitfall traps were located in the Ellis Creek drainage. 15 2 3 Trapcatch Y F Mean daily temp. a, 10 pE-

June July August Date (monthlday) August, as opposed to July in California (Stevens 1966), and overwintered adults can be found mating, feeding and ovipositing on boles as early as May. Because of the protracted oviposition period, overwintering larvae of all stages may be encountered (Stevens 1966). Developmental time varied depending on the spatial location of oviposition on the tree. The preferred oviposition site was the lower bole (> 80%) with the remainder occurring equally in the root-ball and lateral roots. Developmental time in the bole can be as much as a year shorter than in the root ball or lateral roots due to higher above- ground temperatures (personal observation). About 50 % of infested trees that were dissected from the three sites had empty chip cocoons, from which adults had already emerged and the majority of the emergence was from the bole. The collections made from Daves Creek (ESSF), had > 90% of the cocoons located in the above ground portion of the bole. This could be due to the cooler temperature regime in this biogeoclimatic zone. On 21 June 1989, 39 pairs of P. schwarzi were placed on pine sections. From 12-18 July 1989, the females began ovipositing and continued until 2 Nov. 1989 (Fig. 2). There appeared to be a major peak in late July through August, and then a lesser

one in October. The mean number of eggs laid per female (_f S.E.) was 22.6 + 1.82, with a maximum of 92 eggs laid by one female. There were up to 5 eggs deposited per puncture; however, of the 904 oviposition punctures examined, 86% contained 1 egg and 12%,2 eggs. According to Stevens (1966) mating occurs on the foliage, with oviposition taking place - throughout the summer. My observations and data (Fig. 2) support Stevens observation regarding oviposition; however, in the sites used in this study, mating was only observed on the bole of lodgepole pine. Mating locations may differ between geographic areas or climatic regimes, or perhaps it was both P. terminalis and P. Figure 2. Seasonal trend in numbers of eggs laid by newly emerged Pissodes schwarzi females per 3-7 day periods, from July through November, 1989. Aug. 1 Sep. 1 Oct. 1 Date schwani which were observed mating on the foliage in Stevens (1966) study. Aggregation pheromones were reported for P. nemorensis by Booth and Lanier (1974). Males produced a pheromone that when deployed together with host odors attracted conspecific males and females (Booth et al. 1983). P. strobi and P. nemorensis both produce grandisol (cis-2-isopropenyl-1-methylcyclobutaneethanol) , and its corresponding aldehyde, grandisal, which act together as aggregation pheromones for P. nemorensis (Booth et al. 1983; Phillips and Lanier 1986). Phillips and Lanier (1986) found that male P. strobi produce an unknown allelochemic that interrupts the response of P. nemorensis to its natural or synthetic aggregation pheromone. Although Booth and Lanier (1974) postulated that P. strobi uses a male-produced aggregation pheromone, repeated field tests have indicated that grandisol and grandisal are not pheromones for P. strobi (Booth 1974; Phillips 1981; Booth et al. 1983). I hypothesize that a similar relationship to that between P. nemorensis and P. strobi could occur between the lodgepole terminal weevil and P. schwarzi which spatially occupy similar host sites. Commonly, P. schwarzi infests trees stressed by rusts, Cronartium cornandrae, root rots and other insects, such as C'ylindrocopturus spp. (Coleoptera: Curculionidae)l2 (Wood 1964; Stevens 1966; Coulson and Franklin 1970), which are in themselves damaging or fatal. Therefore, P. schwarzi is not economically important at present, but with increasingly intensive silvicultural practices, e.g. spacing, and the probable onset of climatic warming trends, P. schwani could emerge as a problem in some circums+mces,particularly because its ability to infest apparently drought-stressed trees (personal observation). XIV. REFERENCES

Alexander, R.R. 1960. Thinning lodgepole pine in the Central Rocky Mountains. J. For. 58: 99-104.

Alexander, R.R. 1974. Silviculture of subalpine forests in the central and southern Rocky Mountains: the status of our knowledge. U.S. For. Serv. Res. Pap. R.M.-121.

Alfaro, R.I. 1989. Stem defects in Sitka spruce induced by Sitka spruce weevil, Pissodes strobi (Peck.). pp. 177-185. In Proc. IUFRO Working Group on Insects Affecting Reforestation, Vancouver, B. C. R. I. Alfaro and S .G. Glover (editors).

Alfaro, R.I. and J.H. Borden. 1982. Host selection by the white pine weevil, Pissodes strobi Peck: feeding bioassays using host and nonhost plants. Can. J. For. Res. 12: 64-70.

Alfaro, R.I. and S.A.Y. Omule. 1990. The effect of spacing on Sitka spruce weevil damage to Sitka spruce. Can. J. For. Res. 20: 179-184.

Alfaro, R.I., H.D. Pierce, Jr., J.H. Borden and A.C. Oehlschlager. 1980. Role of volatile and nonvolatile components of Sitka spruce bark as feeding stimulants for Pissodes strobi (Coleoptera:Curculionidae). Can. J. 2001. 58: 626-632.

Alfaro, R.I. and C. C. Ying. 1990. Levels of sitka spruce weevil, Pissodes strobi (Peck), damage among sitka spruce provenances and families near Sayward, British Columbia. Can. Entomol. 122: 607-615.

Amman, G.D. and J. A Anhold. 1989. Preliminary evaluation of hazard and risk rating variables for mountain pine beetle infestations in lodgepole pine stands. pp. 141-175. In: W.E. Waters, R. W. Stark and D.L. Wood, eds. Integrated pest management in pine-bark beetle ecosystems. Wiley Interscience. New York: John Wiley and Sons.

Ballad, L.A. and J.N. Long. 1988. Influence of stand density on log quality of lodgepole pine. Can. J. For. Res. 18:9ll-916.

Bella, I.E. 1985a. Pest damage incidence in natural and thinned lodgepole pine in Alberta. Forestry Chronicle 6 1: 233-238.

Bella, I.E. 1985b. Western gall rust and insect leader damage in relation to tree size in young lodgepole pine in Alberta. Can. J. For. Res. 15: 1008-1010. Bella, I.E. and K.J. Stoszek. 1988. Strategies for reducing impacts of terminal shoot insects. pp. 286-297, in: Proceedings- Future Forests of the Mountain West: A Stand Culture Symposium. U.S. Dept. Agr., For. Sew., Gen. Techn. Rep. INT-243.

Belyea, R.M. and C.R. Sullivan. 1956. Occurrence of Pissodes terminalis Hopp. (Coleoptera: Curculionidae) in Canada: life history, behavior and cytogenetic identification. Can. Entomol. 95: 70-76.

Berryman, A.A. and R.W Stark. 1985. Assessing the risk of lodgepole pine stand destruction by pests. pp. 163-169. In Lodgepole pine the species and its management: Symposium proceedings. D.M. Baumgartner, R.G. Krebill, J.T. Arnott and G.F. Weetman, eds. Washington State University, Pullman, Washington.

Bliss, M. and W.H. Kearby. 1970. Notes on the life history of the pales weevil and Northern pine weevil in central Pennsylvania. Ann. Entomol. Soc. Am. 63: 73 1-734.

Booth, D.C. and G.N. Lanier. 1974. Evidence of an aggregation pheromone in Pissodes approximatus and P. strobi. Ann. Entomol. Soc. Amer. 67: 992-994.

Booth, D.C., T.W. Phillips, A. Claesson, R.M. Silverstein, G.N. Lanier and J.R. West. 1983. Aggregation pheromone components of two species of Pissodes weevils (Coleoptera: Curculionidae). J. Chem. Ecol. 9: 1- 12.

Boyce, T.M., M.E. Zwick and C.F. Aquadro. 1989. Mitochondrial DNA in the bark weevils: size, structure and heteroplasmy. Genetics 123:825-836.

Brooks, J.E. and J.H. Borden. 1990. Development of a resistance index for Sitka spruce against the white pine weevil, Pissodes strobi Peck. FRDA , Res. and Devel. Sub-program, Final Report, 37 pp.

Brooks, J.E., J.H. Borden and H.D. Pierce, Jr. 1987. Foliar and cortical monoterpenes in Sitka spruce: potential indicators of resistance to the white pine weevil, Pissodes strobi Peck (Coleoptera: Curculionidae) . Can. J. For. Res. 17:740-745.

. Brown, G.S. 1965. Point density in stems per acre. New Zealand Forestry Res. Notes 38: 1-12.

Butkewich, S .L., R. J. Prokopy and T. A. Green. 1987. Discrimination of occupied host fruit by plum curculio females (Coleoptera: Curculionidae). J. Chem. Ecol. 13: 1833- 1841. Cameron, E.A. 1974. Bionomics and impact of the lodgepole terminal weevil, Pissodes terminalis Hopping (Coleoptera: Curculionidae), in the Sierra Nevada of California. Ph. D. dissertation, University of California, Berkeley.

Cameron, E. A. and R. W. Stark. 1989. Variation in the life cycle of the lodgepole terminal weevil, Pissodes terminalis Hopping (Coleoptera: Curculionidae), in California. Can. Entomol. 121: 793-801.

Carlson, J.A., E.M.J. Jeklin and P.M. Wood. 1984. Interim guidelines for the management of Sitka spruce weevil in the Vancouver Forest Region. Forest Service Internal Report PM-V-4.

Chang, V.C.S. and G. A. Curtis. 1972. Pheromone production by the New Guinea sugarcane weevil. Envir. Entomol. 1: 348-354.

Clark, P.J. and F.C. Evans. 1954. Distance to nearest neighbor as a measure of spatial relationships in populations. Ecology 35:445-453.

Clark, P.J. and F.C. Evans. 1955. On some aspects of spatial pattern in biological populations. Science 121: 397-398.

Coulson, R.N. and R.T. Franklin. 1970. The occurrence of Dioryctria amatella and other insects in Cronartiumjkijorm cankers. Can. Entomol. 102: 353-357.

Courtney, S.P., G.K. Chen and A. Gardner. 1989. A general model for individual host selection. Oikos 55: 55-65.

Desnoyers, E.A. 1988. An operational survey of Pissodes terminalis in Penticton Creek South, 1980 juvenile spaced area. Professional paper for A. B. C. P. F. Vancouver, B. C.

Diggle, P.J. 1983. Statistical Ananlysis of Point Patterns. Academic Press, Toronto.

Donnelly, K. 1978. Simulations to determine the variance and edge-effect of total nearest neighbor distance. Pages 91-95 in I. Hodder, editor. Simulation methods in archaeology. Cambridge University Press, London, England.

Drouin, J.A., C.R. Sullivan and S.G. Smith. 1963. Occurrence of Pissodes terminalis Hopp. (Coleoptera: Curculionidae) in Canada: life history, behavior, and cytogenetic identification. Can. Entomol. 95: 70-76.

Fellin, D.G. 1973. Weevils attracted to thinned lodgepole pine stands in Montana. USDA Forest Service, Res. Paper INT-136, Ogden, Utah. 20 pp. Fontaine, M.S. and J.L. Foltz. 1982. Field studies of a male-released aggregation pheromone in Pissodes nemorensis. Envir. Entomol. 11 : 88 1-883.

Furniss, M.M. 1976. Controlled breeding, comparative anatomy and bionomics of Dendroctonus simplex LeConte and Dendroctonus pseudotsugae Hopkins (Coleoptera: Scolytidae). pp. 109-120. In Univ. Idaho, Dep. Entomol. Anniv. Publ. W.F. Barr (ed.) Univ. Idaho, Moscow.

Furniss, R.L. and V.M. Carolin. 1977. Western Forest Insects. USDA Forest Service Misc. Publ. No. 1339.

Futuyma, D.J. and S.C. Peterson. 1985. Genetic variation in the use of resources by insects. Ann. Rev. Entomol. 36: 217-238.

Graham, K. 1951. The sitka spruce weevil. Can. Dep. Agric. Bi-monthly Prog. Rep. 7:3-4.

Green, P.J. and R. Sibson. 1977. Computing Dirichlet tessellations in the plane. The Computer Journal 2 1: 168-173.

Hall, R.B.W. 1991. A re-examination of the use of interpoint distances and least squares in mapping forest trees. Ecology 72: 2286-2289.

Harman, D.M. 1975. Movement of individually marked white pine weevils, Pissodes strobi. Envir. Entomol. 4: 120-124.

Harman, D.M. and H.M. Kulman. 1966. A technique for sexing live white pine weevils, Pissodes strobi. Ann. Entomol. Soc. Amer. 59: 3 15-317.

Harris, L.J., R.I. Alfaro and J.H. Borden. 1990. Role of needles in close range host selection by the white pineweevil on sitka spruce.

Harvey, G.T. and S.S. Sohi. 1985. Isozyme characterization of 28 cell lines from five insect species. Can. J. Zool. 63:2270-2276.

Harvey, G.T., L. Maclauchlan and P.M. Roden. 1991. Studies of isozymes of weevils on lodgepole pine in the Kamloops area of British Columbia. Misc. Report WPS #250, Forestry Canada, Great Lakes Forestry Centre, Sault Ste. Marie, Ontario.

Hedin, P. A,, J.A. Payne, T.L. Carpenter and W. Neel. 1979. Sex pheromones of the male and female pecan weevil, Curczilio caryae: behavioral and chemical studies. Envir. Entomol. 8521-523.

Heppner, D.G. and P.M. Wood. 1984. Vancouver Region Sitka spruce weevil survey results (1982-1983), with recomendations for planting Sitka spruce. British Columbia For. Serv. Int. Rep. PM-V-5.

Hopkins, A.D. 1911. Technical papers on miscellaneous forest insects. I. Contribution toward a monograph of the bark-weevils of the genus Pissodes. USDA Bur. Entomol. Tech. Ser. 20: 1-68.

Hulme, M. A., A.F. Dawson and J. W.E. Harris. 1986. Exploiting cold-hardiness to separate Pissodes strobi (Peck)(Coleoptera: Curculionidae) from associated insects in leaders of Picea sitchensis (Bong.) Carr. Can. Entomol. 118: 1115- 1122.

Hulme, M.A., J.W.E. Harris and A.F. Dawson. 1987. Exploiting adult girth to separate Pissodes strobi (Peck)(Coleoptera: Curculionidae) from associated insects in leaders of Picea sitchensis (Bong.) Carr. Can. Entomol. 119: 75 1- 753.

Hunt, D. W.A. and K.F. Raffa. 1989. Attraction of Hylobius radicis and Pachylobius picivorus (Coleoptera: Curculionidae) to ethanol and turpentine in pitfall traps. Environ. Entomol. 18: 35 1-355.

Jeklin, E.M.J. 1980. Management procedures for control of spruce terminal weevil on Nootka Island. M.O.F. internal report. Vancouver Forest Region. 21 pp.

Jermy, T. (Ed.). 1976. The host-plant in relation to insect behavior and reproduction. Symp. Biol. Hung. 16: 1-322.

Johnson, N.E. 1965. A test of 12 insecticides for the control of the Sitka spruce weevil, Pissodes sitchensis Hopkins. J. Econ. Entomol. 58: 572-574.

Johnstone, W.D. 1985. Thinning lodgepole pine. In Lodgepole pine the species and its management: Symposium proceedings. D.M. Baumgartner, R.G. Krebill, J.T. Arnott and G.F. Weetman, eds. Washington State University, Pullman, Washington.

Keen, F.P. 1952. Insect enemies of western forests. U.S. Dep. Agric. Misc. Publ. 273.

Kendall, M and A. Stuart. 1979. The Advanced Theory of Statistics. Vol. 2, 4th ed. Charles Griffin, London.

Kenkel, N.C. 1988. Pattern of self-thinning in Jack pine: testing the random mortality hypothesis. Ecol. 69: 1019-1024. Kiss, G.K. and A.D. Yanchuk. 1991. Preliminary evaluation of genetic variation of weevil resistance in interior spruce in British Columbia. Can. J. For. Res. 2 1:230-234.

Kline, L.N. and R.G Mitchell. 1979. Insects affecting twigs, terminals and buds. pp. 215-226. In Forest Insect Survey and Control. J.A. Rudinsky (Ed.). Oregon State Univ., Corvallis, Oregon.

Kogan, M. 1977. The role of chemical factors in insect/plant relationships. Proc. Int. Congr. Entomol., Washington DC, 1976: 21 1-227.

Kovacs, E. and J.A. McLean. 1990. Notes on the longevity, fecundity and development of Pissodes terminalis Hopping (Coleoptera: Curculionidae) in the interior of British Columbia, Canada. J. Entomol. Soc. Brit. Columbia 87: 68- 73.

Kovacs, E. and J.A. McLean. 1990. Emergence patterns of terminal weevils (Coleoptera:Curculionidae) and their parasitoids from lodgepole pine in the Interior of British Columbia, Canada. J. Entomol. Soc. Brit. Columbia 87:75- 79.

Krajina, V.J. 1965. Biogeoclimatic zones and classification of British Columbia. Ecol. of Western N. Amer. 1: 1-17.

Lawson, B.D. 1977. Fire weather index. The basis for fire danger rating in British Columbia. Forestry Canada, Pac. For. Res. Cent. BC-P-17.

Liyuan, J. 1989. Pissodes nitidus Roelofs, the yellow-spotted pine weevil (Coleoptera: Curculionidae): a serious pest of Korean pine plantations in northeastern China. pp. 186-193. In Proc. IUFRO Working Group on Insects Affecting Reforestation, Vancouver, B.C. R.I. Alfaro and S. G. Glover (editors).

Lloyd, D., K. Angrove, G. Hope and C. Thompson. 1990. A guide to site identification and interpretation for the Kamloops Forest Region. Res. Branch, Min. Forests, Victoria, B. C.

Maher, T.F. 1982. The biology and impact of the lodgepole terminal weevil in the Cariboo Forest Region. M.F. Thesis, Faculty of Forestry, University of British Columbia.

Matlack, G.R. and J.L. Harper. 1986. Spatial distribution and the performance of individual plants in a natural population of Silene dioica. Oecologia 70: 121- 127. McGregor, M.D., G.D. Amman and W.E. Cole. 1981. Hazard-rating lodgepole pine for susceptibility to mountain pine beetle infestation, pp.99- 104. In R.L. Hedden, S.J. Barras and J.E. Coster (Coordinators), Hazard-rating systems in forest insect pest management. Symp. Proc., U.S. Dep. Agric. For. Serv. Gen. Tech. Rep. WO-27.

McLean, J. A. 1989. Effect of red alder overstory on the occurrence of Pissodes strobi (Peck) during the establishment of a Sitka spruce plot. pp. 167-176. In Proc. IUFRO Working Group on Insects Affecting Reforestation, Vancouver, B. C. R.I. Alfaro and S.G. Glover (editors).

McLean, J.A. and E. Kovacs. 1987. Emergence patterns of the parasitoid complex of the lodgepole terminal weevil, Pissodes terminalis Hopping (Coleoptera: Curculionidae) and recommendations for enhancing parasitoid survival in leader clipping operations in British Columbia. FRDA Report No. F-52-31-008.

McMullen, L.H. 1976a. Effect of temperature on oviposition and brood development of Pissodes strobi (Coleoptera: Curculionidae). Can. Entomol. 108: 1167- 1172.

McMullen, L.H. 1976b. Spruce weevil damage - ecological basis and hazard rating for Vancouver Island. Info. Rep. BC-X-141. Pac. For. Res. Centre. Forestry Canada, Victoria, B.C.

McMullen, L.H. and S.F. Condrashoff. 1973. Notes on dispersal, longevity, and overwintering of adult Pissodes strobi (Peck) (Coleoptera: Curculionidae) on Vancouver Island. J. Entomol. Soc. B.C. 70: 22-26.

Meidinger, D. and J. Pojar. 1991. Ecosystems of British Columbia. Research Branch, Min. of Forests, Victoria, B.C.

Mitchell, R.G., K.H. Wright and N.E. Johnson. 1990. Damage by Sitka spruce weevil (Pissodes strobi) and growth patterns for 10 spruce and hybrids over 26 years in the ~acific~orthwest.US. Dep. Agric. For. Serv. Res. Pap. PNW-RP-434.

Nordlander, G. 1987. A method for trapping Hylobius abietis (L.) with a standardized bait and its potential for forecasting seedling damage. Scand. J. For. Res. 2: 199-213.

Phillips, T.W. 1981. Aspects of host preference and chemically mediated aggregation in Pissodes strobi (Peck) and P, approximatus Hopkins (Coleoptera: Curculionidae). MS Thesis. State University of New York. College of Environmental Science and Forestry. Syracuse. 94 pp.

Phillips, T.W. 1984. Ecology and systematics of Pissodes sibling species (Coleoptera: Curculionidae). PhD. Dissertation. State University of New York, Syracuse.

Phillips, T.W. and G.N. Lanier. 1985. Genetic divergence among populations of the white pine weevil, Pissodes strobi (Coleoptera: Curculionidae). Ann. Ent. Soc. Am. 78:744-750.

Phillips, T.W. and G.N. Lanier. 1986. Interspecific activity of semiochemicals among sibling species of Pissodes (Coleoptera: Curculionidae). J. Chem. Ecol. 12: 1587-1601.

Phillips, T.W., S.A. Teale and G.N. Lanier. 1987. Biosystematics of Pissodes Germar (Coleoptera: Curculionidae): seasonality, morphoiogy, and synonomy of P. approximatus Hopkins and P. nemorensis Germar. Can. Entomol. 119: 465- 480.

Phillips, T. W., J.R. West, J.L. Foltz, R.M. Silverstein and G.N. Lanier. 1984. Aggregation pheromone of the deodar weevil, Pissodes nemorensis (Coleoptera: Curculionidae): Isolation and activity of grandisol and grandisal. J. Chem. Ecol. 10: 1417-1423.

Prokopy , R. J. I981. Epideictic pheromones influencing spacing patterns of phytophagous insects, in D.A. Nordlund, R.L. Jones, and W.L. Lewis, (eds.). Semiochemicals: Their role in Pest Control. Wiley, New York.

Prokopy, R.J. and E.D. Owens. 1983. Visual detection of plants by herbivorous insects. Ann Rev. Entomol. 28: 337-364.

Reed, A.N., J.W. Hanover and M.M. Furniss. 1987. Douglas-fir and Western Larch: Chemical and physical properties in relation to Douglas-fir bark beetle attack. Tree Physiol. 1: 277-287.

Rieske, L.K. and K.F. Raffa. 1990. Dispersal patterns and mark-recapture estimates of two pine root weevil species, Hylobius pales and Pachylobius picivorus (Coleoptera: Curculionidae), in Christmans tree plantations. Environ. Entomol. 19: 1829-1836.

Rohlf, F.J. and J.W Archie. 1978. Least-squares mapping using interpoint distances. Ecology 59: 126-132.

Roitberg, B.D. and R.J. Prokopy. 1984. Host discrimination by adult and larval European apple sawflies Hoplocampa testudinea (Klug)(Hymenoptera: Tenthredinidae). Environ. En tomol. 13: 1000- 1003.

Romstock-Volkl, M. and C. Wissel. 1989. Spatial and seasonal patterns in the egg distribution of Tephritis conura (Diptera: Tephritidae). Oikos 55: 165-174.

Rose, A.H. and O.H. Lindquist. 1977. Insects of eastern spruces, fir and hemlock. Can. For. Serv. Tech. Report 23.

Ross, D.A. 1967. Wood and bark feeding Coleoptera of felled western larch in British Columbia. J. Entomol. Soc. B.C. 64: 23-24.

Salman, K.A. 1935. The effects of attack by Pissodes terminalis Hopping on lodgepole pine in California. J. Econ. Entomol. 28: 496-497.

SAS Institute. 1985. SAS user's guide: statistics. SAS Institute, Cary N.C.

Shore, T.L. and L. Safranyik. 1990. A system for rating the susceptibility and risk of lodgepole pine stands to mountain pine beetle. File Report 5203, Forestry Canada, Pacific Forestry Research Centre, Victoria, B.C.

Silver, G.T. 1968. Studies on the Sitka spruce weevil, Pissodes sitchensis, in British Columbia. Can. Entomol. 100: 93-1 10.

Sinclair, D.F. 1985. On tests of spatial randomness using mean nearest neighbor distance. Ecol. 66: 1084-1085.

Smith, S.G. and S.A. Sugden. 1969. Host trees and breeding sites of native North American Pissodes bark weevils, with a note on synonomy. Ann. Entomol. Soc. Amer. 62: 146-148.

Smith, S.G. and Y. Takenouchi. 1962. Unique incompatibility system in a hybrid species. Science 138: 36-37.

Sower, L.L., G.E. Daterman and C. Sartwell. 1988. Biology and impact of western pine shoot borer, Eucosma sonomana, in the western United States. In Proc. IUFRO Working Group on Insects Affecting Reforestation, Vancouver, B.C. R.I. Alfaro and S.G. Glover (editors), pp. 186-193.

Sower, L.L. and M.D. Shorb. 1984. Effect of western pine shoot borer (Lepidoptera:Olethreutidae) on vertical growth of ponderosa pine. J. Econ. Entomol. 77:932-935.

Stansly, P.A. and J.R. Cate. 1984. Discrimination by ovipositing boll weevils (Coleoptera: Curculionidae) against previously infested Hampea (Malvaceae) flower buds. Environ. Entomol. 13: 1361-1365.

Stark, R. W. and D.L. Wood. 1964. The biology of Pissodes teminalis Hopping (Coleoptera: Curculionidae) in California. Can. Entomol. 96: 1208-12 1 8.

Steill, W.M. 1979. Releasing unweevilled white pine to ensure first-log quality of final crop. For. Chron. 55: 142-143.

Steill, W.M. and A.B. Berry. 1985. Limiting white pine weevil attacks by side shade. For. Chron. (Feb. 86):5-9.

Stevens, R.E. 1966. Observations on the Yosemite bark weevil in California. Pan- Pacific Entomol. 42: 184-189.

Stevens, R.E. and J.A.E. Knopf. 1974. Lodgepole terminal weevil in interior lodgepole forests. Environ. Entomol. 3: 998-1002.

Stevenson, R. E. and J. J. Petty. 1968. Lodgepole terminal weevil (Pissodes terminalis Hopping) in the AlbertaINorthwest Temtories Region. Can Dep. of For. Bi- monthly Res. Notes 24:6.

Taylor, S., R.I. Alfaro and K. Lewis. 1991. Factors affecting the incidence of white pine weevil damage to white spruce in the Prince George Region of British Columbia. J. Entomol. Soc. B.C. 88: 3-7.

Tilles, D.A., G. Nordlander, H. Nordenhem, H.H. Eidmann, A. Wassgren and G. Bergstrom. 1986a. Increased release of host volatiles from feeding scars: a major cause of field aggregation in the pine weevil Hylobius abietis (Coleoptera: Curculionidae). Environ. Entomol. 15: 1050-1054.

Tilles, D. A., K. Sjodin, .G. Nordlander and H.H. Eidman. 1986b. Synergism between ethanol and conifer host volatiles as attractants for the pine weevil, Hylobius abietis (L.) (Coleoptera: Curculionidae). J. Econ. Entomol. 79: 970-973.

Tumlinson, J.H., D.D. Hardee, R.C. Gueldner, A.C. Thompson, P.A. Hedin and J.P. Minyard. 1969. Sex pheromones produced by male boll weevil: isolation, identification, and synthesis. Science 166: 1010-1012.

Upton, G. J. G. and B. Fingleton. 1985. Spatial Data Analysis by Example. Vol. I, Point pattern and quantitative data. John Wiley adSons, Toronto.

VanderSar, T.J.D. 1978. Resistance of Western White pine to feeding and oviposition by Pissodes strobi Peck in western Canada. J. Chem. Ecol. 6: 641-647.

VanderSar, T.J.D. and J.H. Borden. 1977a. Visual orientation of Pissodes strobi Peck (Coleoptera: Curculionidae) in relation to host selection behavior. Can. J. 2001. 55: 2042-2049. Vyse, A. 1985. Spacing lodgepole pine - how much space is enough? Unpublished report. Ministry of Forests, Kamloops Forest Region, Kamloops, B.c.

Wallace, D.R. and C.R Sullivan. 1985. The white pine weevil, Pissodes strobi (Coleoptera: Curculionidae): a review emphasizing behavior and development in relation to physical factors. Proc. Entomol. SOC.Ont. 16 (suppl.): 39-62.

Waters, W.E. 1985. Monitoring bark beetle populations and beetle-caused damage. pp. 141-175. In: W.E. Waters, R.W. Stark and D.L.Wood, eds. Integrated pest management in pine-bark beetle ecosystems. Wiley Interscience. New York: John Wiley and Sons.

West, R.J. 1990. The biology, damage and within-tree distribution of the spruce bud midge on black spruce in Newfoundland. Phytoprotection 71:31-36.

Wilkinson, L. 1988. SYGRAPH. Evanston, IL: SYSTAT, Inc.

Wilkinson, L. 1989. SYSTAT: the system for statistics. Evanston, IL: SYSTAT, Inc.

Wilkinson, R. C. 1983. Leader and growth characteristics of eastern white pine associated with white pine weevil attack susceptibility. Can. J. For. Res. 13: 78-84.

Wood, R.O. 1964. Notes on distribution and hosts of the weevils Pissodes schwarzi Hopk. and Pissodes curriei Hopk. in British Columbia and Yukon Territory. Proc. Entomol. Soc. B.C. 61: 42-44.

Wood, R.0 and L.H. McMullen. 1983. Spruce weevil in British Columbia. Can. For. 3 Serv. Forest Insect and Disease Survey Pest Leaflet No. 2.

Ying, C. 1991. Genetic resistance to the white pine weevil in Sitka spruce. B.C. Min. For., Research Branch, Victoria, B.C. Res. Note, no. 106, ISSN 0226-9368.

Zar, J.H. 1984. Biostatistical Analysis. Second Ed. Prentice Hall, Inc. Englewood Cliffs, New Jersey.