1427 Differential impacts of the southern , frontalis,onPinus palustris and

Nicholas A. Friedenberg, Brenda M. Whited, Daniel H. Slone, Sharon J. Martinson, and Matthew P. Ayres

Abstract: Patterns of host use by herbivore pests can have serious consequences for natural and managed ecosystems but are often poorly understood. Here, we provide the first quantification of large differential impacts of the southern pine bee- tle, Dendroctonus frontalis Zimmermann, on loblolly pine, Pinus taeda L., and , Pinus palustris P. Mill., and evaluate putative mechanisms for the disparity. Spatially extensive survey data from recent epidemics indicate that, per square kilometre, stands of loblolly versus longleaf pine in four forests (380–1273 km2) sustained 3–18 times more local infestations and 3–116 times more tree mortality. Differences were not attributable to size or age structure of pine stands. Using pheromone-baited traps, we found no differences in the abundance of dispersing D. frontalis or its predator Thanasi- mus dubius Fabricius between loblolly and longleaf stands. Trapping triggered numerous attacks on trees, but the pine spe- cies did not differ in the probability of attack initiation or in the surface area of bark attacked by growing aggregations. We found no evidence for postaggregation mechanisms of discrimination or differential success on the two hosts, suggest- ing that early colonizers discriminate between host species before a pheromone plume is present. Re´sume´ : Le patron d’utilisation des hoˆtes par les ravageurs herbivores peut avoir de se´rieuses conse´quences pour les e´co- syste`mes naturels et ame´nage´s mais elles sont souvent mal connues. Dans cet article, nous quantifions pour la premie`re fois les impacts du dendroctone me´ridional du pin, Dendroctonus frontalis Zimmermann, qui sont tre`s diffe´rents sur le pin a` en- cens, Pinus taeda L., et le pin des marais, Pinus palustris P. Mill., et nous e´valuons les me´canismes qui pourraient expliquer cette disparite´. Les donne´es d’un inventaire couvrant un vaste territoire et portant sur des e´pide´mies re´centes indiquent que les peuplements de pin a` encens ont subi 3–18 fois plus d’infestations locales et 3–116 fois plus de mortalite´ par kilome`tre carre´ que les peuplements de pin des marais dans quatre foreˆts (380–1273 km2). Ces diffe´rences n’e´taient pas attribuables a` la structure d’aˆge ou de dimension des peuplements de pin. A` l’aide de trappes appaˆte´es avec des phe´romones, nous n’avons observe´ aucune diffe´rence dans l’abondance de D. frontalis alors qu’il se dispersait ni de son pre´dateur, Thanasimus dubius Fabricius, entre les peuplements de pin a` encens et de pin des marais. Le pie´geage a provoque´ plusieurs attaques sur les ar- bres mais il n’y avait pas de diffe´rence entre les espe`ces de pin quant a` la probabilite´ du de´clenchement d’une attaque ni quant a` la superficie d’e´corce attaque´e par le regroupement d’un nombre croissant d’insectes. Nous n’avons trouve´ aucun in- dice de discrimination ou de diffe´rence de succe`s dues a` des me´canismes de regroupement a posteriori sur les deux hoˆtes, ce qui indique que les insectes pionniers distinguent l’espe`ce hoˆte avant qu’une traıˆne´e de phe´romone soit pre´sente. [Traduit par la Re´daction]

Introduction that serve as resources may reflect either active resource preference on the part of the pest or differential success fol- Within ecological communities, consumers inevitably use lowing indiscriminant attacks; either of these alternatives a subset of potential resources. Even within the niche of a could result from variation among host species in availabil- generalist consumer, some resources are used more than ity or suitability. Realized host availability could be a sim- others. In the case of forest pests, identifying the mecha- ple reflection of biomass but could also be influenced by nisms that determine which host species are impacted and age structure and (or) spatial dispersion. Realized host suit- to what degree is fundamental to managing forests for resil- ability could be a function of primary nutritive quality, effi- ience against widespread damage (Graham 1939; Carnus et cacy of defenses, or even differences in species associations al. 2006). For instance, differential impacts on populations between hosts and predators, competitors, or mutualists of

Received 31 October 2006. Accepted 5 January 2007. Published on the NRC Research Press Web site at cjfr.nrc.ca on 15 September 2007. N.A. Friedenberg,1 B.M. Whited,2 S.J. Martinson, and M.P. Ayres. Department of Biological Sciences, Dartmouth College, Hanover, NH 03755, USA. D.H. Slone.3 USDA Forest Service, Southern Research Station, 2500 Shreveport Highway, Pineville, LA 71360, USA. 1Corresponding author (e-mail: [email protected]). 2Present address: Graduate Program in Organismic and Evolutionary Biology, University of Massachusetts, Amherst, MA 01003, USA. 3Present address: US Geological Survey Sirenia Project, 2201 NW 40th Ter., Gainesville, FL 32601, USA.

Can. J. For. Res. 37: 1427–1437 (2007) doi:10.1139/X07-008 # 2007 NRC Canada 1428 Can. J. For. Res. Vol. 37, 2007 the pest. The mechanisms governing host use and impact are D. frontalis in the Kisatchie NF that occurred during the of particular interest in aggressive pathogens and pests that late 1970s and an epidemic in the Oakmulgee RD that oc- kill their hosts, which not only have maximal impact on the curred from 1998 to 2001. Measurements during the epi- forest but also create the potential for rapid consumer–host demic in the Kisatchie NF were collected as part of an coevolution (Ehrlich and Raven 1964; Van Valen 1973). unusually thorough and careful survey of infestations (Lorio Variation in host impact is evident in bark of the and Sommers 1981). All data for the Oakmulgee RD were genus Dendroctonus (Coleoptera: : Scolyti- collected by the same experienced employee of the USDA nae). Dendroctonus includes 16 North and Central American Forest Service (USFS) as part of the Southern Pine Beetle species that attack conifers (mainly Pinus spp.). Most of the Infestation Survey. Data included the number of trees beetle species attack multiple host species, but variably so killed by D. frontalis within each local infestation. (13%–100% of the pine species within each beetle’s geo- We characterized the prevalence of host species in each graphic range; Wood 1982; Kelley and Farrell 1998). While ranger district using the ‘‘stand’’ layer of the USFS GIS da- some sympatric beetle–host associations are common, others tabase. The stand layer divides each ranger district into pol- are only rarely observed (Salinas-Moreno et al. 2004). ygons that represent management units and are classified The southern pine beetle Dendroctonus frontalis Zimmer- with respect to age, area, forest type (dominant species), mann has been observed to attack all of the pine species in and location. From the air or the ground, stand boundaries its range (Feldman et al. 1981). Attacks draw large numbers are usually clearly visible as abrupt changes in forest age or of beetles to individual trees via aggregation pheromones. species composition. Thus, stands have biological meaning Host trees are usually girdled and killed within weeks. Tree as units of habitat because they tend to be parcels of forest death is required for beetle reproduction, a rare example of that are relatively homogenous internally by virtue of com- obligatory virulence in –plant systems (Berryman et al. mon landform, soil type, and management history. We cal- 1989). Dendroctonus frontalis ranks among the most signifi- culated the total number and area of stands classified as cant forest pests in North America (Price et al. 1997; Ayres longleaf pine (USFS forest type = 21) and loblolly pine and Lombardero 2000). It has been widely observed that lo- (USFS forest type = 31) for each ranger district under study. blolly pine, Pinus taeda L., is a common and highly im- The infestation survey data for each epidemic allowed us pacted host of D. frontalis (Wahlenberg 1960, pp. 101–106; to associate each local infestation (detected by aerial survey Feldman et al. 1981; Blanche et al. 1983; Walker and Os- as a group of fading trees and referred to as ‘‘spots’’) with wald 2000, p. 168), while longleaf pine, Pinus palustris P. the stand in which the spot occurred. In the Oakmulgee RD, Mill., is rarely attacked or killed (Wahlenberg 1946, we also obtained longitude and latitude coordinates of every pp. 169–170; Blanche et al. 1983; Boyer 1990). The appa- spot by digitizing the field-checked locations marked on rently greater resistance of longleaf pine to bark beetles is 1 : 24 000 topographic maps. These coordinates allowed us often attributed to its greater production of oleoresin (a mix- to generate a map of the Oakmulgee RD that overlaid the ture of monoterpenes and resin acids; Hodges et al. 1977, location of spots relative to longleaf and loblolly stands. 1979), which also made longleaf pine a mainstay of the na- Data from the USFS D. frontalis survey data were merged val stores industry (Butler 1998). with the GIS stand layer data to obtain the total number of Owing to its economic and ecological relevance, the infestations and total number of trees infested in longleaf D. frontalis system has been the subject of intense biolog- and loblolly stands. Spots with less than five trees were ical study for many years (e.g., St. George and Beal 1929; omitted from analyses because such small spots are not con- Coulson 1979; Payne 1980; Wood 1982), yet the relative sistently reported. impact of D. frontalis on longleaf and loblolly pine has We analyzed the impact of D. frontalis on longleaf and never been quantified and mechanisms for the putative dis- loblolly in terms of the number of local infestations and the parity have not generally been tested (but see Hodges et al. number of trees killed. Both parameters were standardized 1977, 1979). Using historical records, we measured the de- to the area covered by each host species (e.g., loblolly trees gree to which loblolly pine has been a disproportionately killed per 10 km2 of loblolly forest). Each parameter was more common host of D. frontalis infestations than long- compared between species within forests using w2 tests leaf pine. We then tested whether this disparity in impact against the expectation of equal per-area impact, setting the could be explained by host age structure, the size of host acceptable Type 1 error rate to 0.05/4 = 0.0125 to correct stands, the distribution of dispersing adult beetles, or the for multiple comparisons. relative abundance of an important specialist predator. Age distribution of attacks Methods We tested whether different attack rates on the the two pine species could be explained by different age structures Historical analysis of the pine species. Dendroctonus frontalis infestation risk is strongly correlated with age in loblolly pine (Lorio and Impact on hosts Sommers 1981), and age distributions of loblolly and long- The Catahoula, Evangeline, and Kisatchie Ranger Dis- leaf pine typically differ owing to changes in management tricts (RD) are all part of the Kisatchie National Forest priorities and regeneration strategies during the last century (NF) in central Louisiana, USA. Each district is separated (Clarke et al. 2000; Walker and Oswald 2000) (Fig. 1). Thus, by ~75 km and has a unique history of management. The different attack rates could reflect differences in the relative Oakmulgee RD is located in the Talladega NF in Alabama, abundance of stands that are of an age where trees are suitable USA. We studied historical records from an epidemic of hosts for beetles rather than intrinsic differences between the

# 2007 NRC Canada Friedenberg et al. 1429

Fig. 1. Age distribution of stands of longleaf pine (Pinus palustris) Field survey and loblolly pine (Pinus taeda) during the studied outbreaks in four ranger districts in Louisiana and Alabama. Note that the scales Site selection of vertical axes vary. Years of epidemics were (a) 1975–1980, We compared stands of loblolly and longleaf pine with re- (b) 1975–1977, (c) 1975–1977, and (d) 1998–2001. spect to the abundance of dispersing adult beetles that could initiate attacks. Studies were done in the Chickasawhay RD of the DeSoto NF, Mississippi, and the Oakmulgee RD of the Talladega NF, Alabama. These ranger districts had re- cent beetle activity, had sufficient components of longleaf and loblolly pine, and were separated by a logistically rea- sonable distance. In 2004, the six counties comprising the Oakmulgee RD reported D. frontalis activity levels of 0.033–0.33 infestations per 1000 host acres with five coun- ties reporting more than 0.1. The five counties comprising the Chickasawhay RD reported 0.031–0.22 infestations per 1000 host acres in 2004 with two counties reporting more than 0.1 (www.srs.fs.usda.gov/econ/data/spb/). We sampled local beetle abundance within 10 stands of each pine species within each ranger district. Because stand age could matter (Ylioja et al. 2005), we stratified the sam- pling into ‘‘young’’ and ‘‘old’’ stands. To choose the sample stands, we first queried the GIS database for each ranger district for stands of loblolly and longleaf pine. We then stratified the stands into those planted before 1975 (old, >30 years) and those planted between 1975 and 1985 (young, 20–30 years old). We then identified all of the age- matched pairs of longleaf and loblolly stands that shared a common boundary (maximizing the similarity of abiotic fac- tors and local D. frontalis abundance for pairwise compari- son). From these, we selected five pairs of stands of each age group in each ranger district to maximize geographic distribution while still allowing for reasonable accessibility (<0.75 km from a drivable road). This final step in our site selection was not random, but it was performed ahead of time using only the GIS information, thereby eliminating subjective decisions in the field. pine species. We applied a two-sample Kolmogorov–Smirnov Across the 40 stands in which we sampled beetle abun- test to compare the pine species with respect to the age dance, tree size and stand structure differed between young distribution of infested stands relative to that of all stands. and old stands but were very similar between longleaf and We used an aggregate data set of all four ranger districts, loblolly pine. Average diameter of in young versus representing population structure and host use at a regional old stands (diameter at breast height ± SE) was 19.8 ± 1.4 scale (truncated at 105 years of age, which excluded <1% versus 33.0 ± 0.9 cm, respectively. Among young stands, of stands). For visualization, we placed stands into 5-year but not old stands, longleaf had a marginally smaller diame- bins and calculated the observed and expected (based on ter at breast height (difference = 5.5 cm, t[18] = 2.06, P = the null hypothesis of age-independent host impact) per- 0.054). Basal area of conifers in young versus old stands centage of stands hosting infestations in each bin based on was 23.2 ± 1.4 versus 16.7 ± 1.0 m2/ha, respectively, and the total number of spots in the four ranger districts. Data did not differ between species. Using measurements of tree were smoothed for presentation using a five-point weighted diameter, height, and height to lower live crown, we esti- moving average (Shiyomi et al. 1998). Stand ages were mated the amount of beetle habitat per tree and per hectare calculated for the median year of epidemics. using the method of Baldwin and Feduccia (1987) (defining Instead of using stand age for these analyses, we might beetle habitat as the area of outer bole from 1.5 m to the have used direct measures of stand structure (e.g., tree diam- lower live crown). There was significantly more beetle hab- eter, basal area, and stems per hectare, which are usually re- itat in young stands versus old stands: 1769 ± 186 versus 2 garded as the more proximate basis for changes in beetle 1153 ± 85 m /ha (t[38] = 3.017, P = 0.0045), but there was risk with changes in stand age), but the USFS database no difference between pine species. lacked reliable, spatially extensive measurements of stand structure, while the stand age data were very reliable. Our Trapping protocol and analysis interpretations allow for the fact that stands of longleaf pine Our measurements of the abundance of D. frontalis and typically mature somewhat more slowly than loblolly pine, its major predator Thanasimus dubius Fabricius (Coleoptera: partly because of having a juvenile ‘‘grass’’ stage where ) followed the logic, timing, and trapping protocol height growth is initially attenuated (Wahlenberg 1946). of a southwide beetle monitoring program that has had de-

# 2007 NRC Canada 1430 Can. J. For. Res. Vol. 37, 2007

Table 1. Comparison of southern pine beetle (Dendroctonus frontalis) impact on loblolly pine (Pinus taeda) versus longleaf pine (Pinus palustris) during epidemics in four National Forest ranger districts in Louisiana and Alabama.

Catahoula RD (Louisiana) Evangeline RD (Louisiana) Kisatchie RD (Louisiana) Oakmulgee RD (Alabama) Parameter Loblolly Longleaf Loblolly Longleaf Loblolly Longleaf Loblolly Longleaf Years of epidemic 1975–1984 1975–1977 1975–1977 1998–2001 Total district area (ha) 45 627 38 567 40 761 12 726 Total spotsa 451 122 185 705 Stands 633 150 409 234 762 419 963 1874 Area (ha) 23 124 6065 11 744 5954 15 078 9147 14 888 24 813 % of total area 51 13 31 15 37 22 12 20 Number of spots 275 14 74 9 109 4 353 176 % of all spots 61 3 61 7 59 2 50 25 Spots/10 km2 11.4 2.1 6.0 1.3 8.3 0.5 20.4 6.0 w2 for spotsb 43.3*** 19.6*** 63.7*** 169.4*** Trees infested/10 km2 371 122 689 105 648 6 1937 457 w2 for trees infestedb 929*** 2736*** 5647*** 20 122***

aIncludes 25%–39% of spots that occurred in stands classified as P. Mill. (USFS forest type 32), mixed pine–hardwood (USFS forest types 13, 46–49), or hardwood (USFS forest type 53). bw2 statistics test null hypotheses that the number of spots and number of infested trees were proportional to the total area of longleaf versus loblolly stands; in all cases, the null hypothesis was rejected with P < 0.001. monstrable success in predicting beetle outbreak levels (Bill- (Turchin and Odendaal 1996) and 1.4–15.9 m for other ag- ings 1988; Billings and Upton 2002). We sampled beetle gressive scolytid bark beetles (Byers 1999). abundance in early April 2005 during the spring dispersal Although it was not our intention, the pheromone lures period when new beetle spots are typically initated triggered D. frontalis attacks on pine trees near 36 of 80 (Thatcher and Pickard 1964; Coulson et al. 1999). Traps traps (in 30 of 40 stands). We recorded the total number of were standard 12-funnel hanging traps (Lindgren 1983) trees that had come under attack in the vicinity of each trap baited with D. frontalis aggregation pheromone (synthetic 4–6 days after removing our traps. This permitted us to frontalin, elution ~5 mg/day) and pure -pinene (170 g of compare loblolly and longleaf pine stands with respect to 75% (–) and 25% (+) in a UHR plastic sleeve), a behavioral the tendency of D. frontalis to initiate attacks. Using our synergist of frontalin (Billings 1985). Attractants were pur- measurements of beetle habitat per tree (m2), we also calcu- chased from Synergy Semiochemicals, Burnaby, British Co- lated the average surface area of tree bole that was attacked lumbia. Our trapping protocol matched the southwide per day per trap site. monitoring program except that we used pure -pinene instead of turpentine extracted from loblolly pine, and the elution rate We analyzed the trap captures of D. frontalis and T. dubius for -pinene was unintentionally higher. Baited funnel traps (log10 transformed) with a full-factorial ANOVA model are also attractive to T. dubius (Mizell et al. 1984). that included forest, stand age, tree species, and trap distance as fixed main effects and site nested within forest and stand At each pair of age-matched pine stands, we deployed one age as a random effect. The main effects of forest and stand pair of traps 50 m to either side of the boundary between age were tested against the error among sites, while all pine species and another pair 150 m to either side of the other effects and interactions were tested against the mean boundary (two traps within each stand, near and far from square error. Traps that captured more D. frontalis were as- the boundary between pine species; a total of 80 traps in sociated with somewhat higher attack rates on nearby trees the study). Our objective was to test whether D. frontalis pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 preferentially fly through loblolly stands at the time of year (log10 captures ¼ 2:16 þ 0:32 m attacked=day ; r = 0.19, when most infestations are initiated. If relatively high attack F[1,78] = 17.88, P < 0.001). This could have been partly be- rates in stands of loblolly pine are due to a host bias of dis- cause the pheromone and host volatiles from nearby at- persing D. frontalis, trap captures should have been higher tacks increased trap captures. So we also ran the ANOVA in the loblolly stands compared with their adjacent paired using trap captures corrected for the relationship with local longleaf stand, especially in the traps farther from the stand attack rates (grand mean plus residuals from the above lin- boundaries. Following Billings and Upton (2002), traps were ear regression). Thanasimus dubius captures were uncorre- hung from lines tied between hardwood trees >10 m from lated with attacks and were not adjusted. the nearest live conifer. Trap placement within stand pairs was mainly determined by the availability of hardwoods, Results but we also avoided roads and clearings to preclude edge ef- fects. Traps were set up and taken down in the same order Historical analysis to ensure 6 days of trapping (±2 h) at each site. Our inter- pretations depended on traps preferentially capturing beetles Host impact that were flying near each trap relative to the distance Historically, D. frontalis infestations were 3–18 times among traps (100–300 m), but this is justified by previous more common and affected 3–116 times more trees per unit studies indicating that pheromone-baited funnel traps have area of loblolly forest compared with longleaf forest an effective attractive distance of only ~18 m for D. frontalis (Table 1). The difference in impact on the two pine species

# 2007 NRC Canada Friedenberg et al. 1431

Fig. 2. Southern pine beetle (Dendroctonus frontalis) host use dividuals (ranging from 33 to 2988) and 107 T. dubius across age classes of longleaf pine and loblolly pine during out- (ranging from 13 to 312). Similar proportions of traps in break years in four ranger districts in Louisiana and Alabama. The each species triggered attacks on nearby trees (19 of 40 expected curve was based on the null hypothesis that probability of versus 17 of 40 in loblolly versus longleaf; w2 = 0.20, df = 1, D. frontalis infestation was independent of stand age. Deviations P = 0.65). More traps triggered attacks on nearby pine from the expected curve indicate age-biased infestation rates. trees in the Oakmulgee RD, where D. frontalis captures were higher, than in the Chickasawhay RD (28 of 40 versus 16 of 40; w2 = 7.27, df = 1, P = 0.007). Traps in young stands were more likely to trigger attacks than in old stands (28 of 40 versus 12 of 40; w2 = 7.39, df = 1, P = 0.007). Attacks were more frequent in young than in old loblolly stands but age independent in longleaf stands (16 young and 5 old versus 12 young and 11 old; w2 = 5.25, df = 1, P = 0.022). Based on analyses of trap captures, we reject the hypoth- esis that D. frontalis tend to be more abundant in stands of loblolly pine compared with longleaf pine during spring dis- persal (Fig. 3; Table 2). Dendroctonus frontalis captures were about twice as high in Alabama as in Mississippi (Fig. 3a; Table 2). Beetle abundance was significantly lower in young than in old stands (Fig. 3b; Table 2). Distance from the stand boundary between species did not affect trap captures (Fig. 3c; Table 2). Variance in D. frontalis among sites was not significant. Figures and tables show D. fronta- was highly significant in all ranger districts studied (Table 1), lis capture data corrected for local attack rates (see Meth- although total impact varied among districts. ods). Conclusions were the same from analyses of uncorrected captures except that the effect of stand age on Age distribution of attacks D. frontalis captures was not significant (F[1,16] = 1.67, P = The probability of D. frontalis infestation was partly re- 0.21). Conclusions were also the same from analysis of a re- lated to stand age, but the relative availability of different duced data set that excluded traps that were associated with age classes of trees (Fig. 1) did not explain differences in attacks. impact between pine species (Fig. 2). For both hosts, ages The clerid predator of D. frontalis, T. dubius, was abun- from 1 to 105 were well represented in the combined popu- dant in both forests and captured in all traps. Captures of T. lations of all four ranger districts, yet only 12% of the spots dubius did not vary with host species, stand age, or distance were in longleaf pine compared with 56% that were in lo- from stand boundaries (Table 2). Unlike D. frontalis, abun- blolly stands. (Remaining spots were in other stand classifi- dance of predators did not differ between forests but varied cations, mainly mixed loblolly pine – hardwood, USFS significantly among sites within forests (36% of random var- forest type 13, and shortleaf pine, USFS forest type 32.) iance in T. dubius captures was attributable to site within The age distribution of longleaf stands infested by D. fronta- forest and stand age) (Table 2). lis did not differ from the age distribution of all longleaf The above tests for differences between pine species in stands of ages 1–105 at the time of infestation (D[2543,170] = the captures of D. frontalis and clerids were quite robust. 0.08, P = 0.24) (Fig. 2), indicating that age did not influence For both beetle species, the probability of detecting a differ- impact. However, the age distribution of loblolly stands in- ence if it existed (statistical power) was >80% for a 10% fested by D. frontalis showed fewer spots than expected in difference in captures (logarithmic scale). stands <20 years old and more spots than expected in stands between 20 and 55 years old (D[2604,738] = 0.29, P < 0.0001) Discussion (Fig. 2). Our historical analysis of southern pine beetle outbreaks Size distribution of stands in four USFS ranger districts confirmed the conventional Differences in the number of spots and trees killed by wisdom that longleaf stands sustain fewer D. frontalis infes- D. frontalis in loblolly versus longleaf pine stands were tations than loblolly stands. Most importantly, it provides not a result of differences in the size of stands because the first quantification of biotic disturbance regimes in two the distributions of stand sizes were very similar for lo- native (and frequently alternative) forest types in the south- blolly versus longleaf in all four ranger districts: mean ± eastern United States. Across four ranger districts, 50%– SD = 34 ± 41 versus 40 ± 45, 23 ± 21 versus 26 ± 32, 61% of all D. frontalis spots were in loblolly stands (USFS 20 ± 17 versus 21 ± 21, and 15 ± 18 versus 13 ± 12 ha forest type 31) compared with only 3%–25% in longleaf for Catahoula, Evangeline, Kisatchie, and Oakmulgee RDs, stands (USFS forest type 21). Remaining spots were distrib- respectively (numbers of stands shown in Table 1). uted among shortleaf pine and mixed pine–hardwood stands. Based on the Southern Pine Beetle Infestation Surveys, Field survey D. frontalis activity killed 371–1937 trees/10 km2 of lo- The 6 day total trap captures averaged 346 D. frontalis in- blolly forest, compared with only 6–457/10 km2 of long-

# 2007 NRC Canada 1432 Can. J. For. Res. Vol. 37, 2007

Fig. 3. Southern pine beetle captures (+SE) in loblolly and longleaf stands over 6 days in the spring of 2005. (a) Captures in the Oakmulgee RD, Alabama, and in the Chickasawhay RD, Mississippi; (b) effect of stand age on trap captures; (c) effect of distance of a trap from the boundary between host species. Means and SEs were back-transformed from log10-transformed data detrended for the effect of attacks on nearby trees. See Table 2 for corresponding ANOVA.

Table 2. ANOVA for effect of forest (Oakmulgee versus Chickasawhay), stand age (20–30 versus >30 years), pine species (loblolly versus longleaf), distance from stand boundary (50 versus 150 m), and site within forest and stand age on trap captures of southern pine beetle and its chief predator Thanasimus dubius over 6 days.

Southern pine beetle T. dubius Source df MS FPMS FP Forest 1,16 1.61 9.52 0.007 0.13 0.70 0.41 Age 1,16 1.00 5.89 0.028 0.02 0.13 0.72 Species 1,48 0.01 0.11 0.74 0.00 0.01 0.91 Distance 1,48 0.09 0.87 0.36 0.00 0.01 0.94 Age species 1,48 0.03 0.26 0.60 0.20 3.70 0.06 Forest age species 1,48 0.00 0.00 0.95 0.15 2.67 0.11 Forest age distance 1,48 0.01 0.09 0.76 0.12 2.14 0.15 Age species distance 1,48 0.00 0.01 0.92 0.10 1.75 0.19 Forest age species distance 1,48 0.01 0.05 0.82 0.07 1.20 0.28 Site[forest, age](random) 16,48 0.17 1.55 0.12 0.18 3.30 0.0007 Error 48 0.11 ..0.05 .. Note: Significant effects in bold. Analysis was a full factorial, but the table excludes interaction terms with both F <1. leaf forest (Table 1). Even in the Oakmulgee RD, where As with any pest or pathogen population, the effective longleaf is the most common pine and loblolly stands ac- virulence of D. frontalis is mediated by a suite of mecha- count for only 12% of the district’s area, loblolly was still nisms operating at different points in the chronology of a lo- the more common host. Furthermore, our results are con- cal infestation (Fig. 4). Most spots begin in the spring servative because they overestimate D. frontalis infestations (Thatcher and Pickard 1964; Coulson et al. 1999). Those of longleaf pine in the Oakmulgee RD, where spots within spots where D. frontalis reproduction is successful can com- stands classified as longleaf pine were almost always plete three to six generations (Ungerer et al. 1999) before within parts of the stand infiltrated by loblolly pines (R. Lee, emigration during the following winter and spring ends the USFS, Oakmulgee RD, personal communication; note that life of the spot. Most spots are not detectable by aerial sur- the conditions for longleaf regeneration are very restrictive vey until after the emergence of progeny from initial attacks, compared with those for loblolly pine, so stands classified meaning that no data are available for local infestations that as loblolly pine in these forests very seldom contain fail to establish or that fail to persist after colonizing one or patches of longleaf pine). These results could aid decisions two trees. We discuss our findings in the context of the that affect the future species composition of southern pine D. frontalis infestation life cycle in an effort to focus fu- forests because disturbance rates are critical for projecting ture research on key elements of the chronology. the ecological attributes and socioeconomic value of forests (Kirkman et al. 2004; Zhou and Buongiorno 2006). Dispersal Such a consistent and marked difference in impact on two The first step in spot formation is the dispersal of beetles similarly abundant native host species should have an obvious into potential host stands (Fig. 4). During the spring disper- driver, but no mechanism has yet been confirmed. Identifying sal period, in two different forests, we found a similar num- the mechanism of differential host impact in the D. frontalis ber of D. frontalis at all sites within each forest. The lack of system could benefit forest management and add to our under- spatial variation in D. frontalis numbers at a scale of standing of niche limitation in other aggressive forest pests. ~1000 km2 suggests a lack of dispersal limitation and a con-

# 2007 NRC Canada Friedenberg et al. 1433

Fig. 4. Diagram of the life history of a local southern pine beetle infestation (spot). Terms inside the circle indicate steps in the life history of a local beetle population. An effect of host species at any step could theoretically generate the observed paucity of spots in longleaf pine compared with loblolly pine. Terms outside the circle indicate mechanisms that are likely to affect the initiation and subsequent success of a spot at each step in the cycle. The broken line indicates that spots are typically detected only after the beetle population is growing and attacking new trees. The arrow represents several generations of beetles before the population dissipates via emigration. Spot growth typi- cally begins in the spring and usually ends before the following spring.

comitant lack of spatiotemporal autocorrelation in infesta- function of age were generally congruent with these pre- tion risk. This was somewhat surprising given previous di- vious assessments (Fig. 2). Our analysis of longleaf risk as rect measurements of movements by marked beetles. For a function of age (the first that we know of) indicated a example, Turchin and Thoeny (1993) estimated a median lack of age dependence. Even stands younger than 15 years dispersal range for D. frontalis of only 690 m, and Cronin of age were attacked in proportion to their risk. This might et al. (2000) estimated that only 5% of D. frontalis reflect a lack of adaptation in D. frontalis for distinguishing move >2.5 km. Such movements would not easily account the age of host plants that are only colonized occasionally, for homogenization of abundance over sites separated by and (or) it could be a result of lower statistical power for 10–40 km. The simplest explanation for this disparity is detecting age-dependent patterns in a host for which we that D. frontalis disperse substantially farther during spring, have fewer records of occupancy by D. frontalis. when we conducted our trapping and when most new spots Our field study indicated that more beetles dispersed form (Hedden and Billings 1977; Payne 1980; Roberts et through old stands than through young stands. This may re- al. 1982; but see Coppedge et al. 1994), than during the flect a preference for dispersing through the more open summer and fall, when most measurements of D. frontalis understory habitat of older stands and (or) reduced trapping dispersal have been conducted and when reemerging bee- efficiency within the relatively dense understory of younger tles and their adult progeny generally attack new trees stands (Turchin and Odendaal 1996). As in our historical within 10–20 m of previously attacked trees (Gara and study, attacks in our field study were correlated with age in Coster 1968; Johnson and Coster 1978; Schowalter et al. loblolly but not in longleaf, producing the impression that 1981). Although D. frontalis captures were surprisingly D. frontalis interacts with loblolly pine in a more ordered uniform over 10–40 km within a ranger district, there way. Given that tree size and stand structure can greatly were marked differences between forests separated by affect the success of infestation growth, failure to discrim- ~200 km, which presumably reflects limits on the dispersal inate between young and old stands is maladaptive and of D. frontalis. may contribute to the lower rate of infestation success in longleaf pine. Effects of stand age and stand size Stand size can also affect risk of D. frontalis infestation The disproportionately low impact of D. frontalis on long- (Ylioja et al. 2005), but because longleaf and loblolly stands leaf was not a result of host age structure. Dendroctonus in our study forests were of a similar range of sizes, we re- frontalis impact is known to vary with stand age in loblolly. ject the hypothesis that reduced infestation risk in longleaf Previous assessments of D. frontalis impact on loblolly have stands is a result of differences in stand size. Basal area is variously indicated that stands <15 years old are generally frequently reported as a key factor in risk-rating of loblolly unsuitable for D. frontalis (Wahlenberg 1946), that the risk pine stands (Lorio and Sommers 1981; Ylioja et al. 2005) of infestation increases in stands of age 35 or older (Lorio and might be lower on average in longleaf vs. loblolly and Sommers 1981), and that risk increases with age to stands (B.M. Whited and M.P. Ayres, unpublished data), about 40 years and then decreases with further increases in but the frequency distributions of basal area are broadly age (Ylioja et al. 2005). Our analyses of loblolly risk as a overlapping (e.g., there was no detectable difference be-

# 2007 NRC Canada 1434 Can. J. For. Res. Vol. 37, 2007 tween species in our sample of 40 stands), and the relation- tial impacts of D. frontalis on longleaf versus loblolly pine ship between risk and basal area for loblolly pine cannot ex- because we used traps baited with pheromone. plain a 3- to 100-fold difference in risk between longleaf We note here that in correcting trap captures for attacks and loblolly pine. on neighboring trees, we found no effect of host species on the landing area attacked, suggesting that there is no differ- Habitat choice ence in secondary attraction of D. frontalis to longleaf pine Theoretically, discrimination by beetles between stands of relative to loblolly pine. More young stands (20–30 years one pine species versus another during dispersal could lead old) experienced attacks than old stands. This might mean to a bias in the probability of D. frontalis spots being initi- that secondary attraction is stronger in young stands, perhaps ated in loblolly versus longleaf pine. The use of particular because of changes in the attractiveness of the pheromone- habitat types as dispersal corridors, for instance, arises from volatile mixture with stand age. Alternatively, the greater discriminatory behaviors at habitat boundaries (e.g., Haddad spacing among trees in older stands may contribute to dissi- 1999; Morales 2002). However, our results reject this possi- pation of the pheromone signal through attenuation with dis- bility because we observed no effect of host species on trap tance (Gara and Coster 1968) and loss through the open captures during spring dispersal. Active host selection at the canopy (Thistle et al. 2004). level of stands would also have been reflected by decreasing Differences among potential hosts in primary host selec- average trap captures with increasing distance from loblolly tion and secondary attraction can yield different attack rates pine, but captures were spatially uniform within stands. The via behavioral mechanisms, but demographic processes of lack of host bias held even when traps that triggered attacks births, deaths, and migration can become at least as impor- were excluded from the analysis. tant at the scale of a landscape (Ylioja et al. 2005). In the our historical study, longleaf and loblolly pine were mainly Primary host selection, secondary attraction, and segregated into stands of 20 ha or more. Our experiment re- demographics jected the hypothesis of a host effect on dispersal through Given no differences in the tendency of beetles to dis- neighboring stands; differences in D. frontalis abundance perse into stands dominated by different potential host spe- may be evident between more spatially disparate samples cies, the next level at which discrimination could be owing to host effects on demographic rates. We know of no expressed is in the choice of individual host trees (Tunset et published data that compare per capita reproductive success al. 1993; Kogan 1994). Host selection by bark beetles can of D. frontalis in loblolly versus longleaf pine. All trees that involve visual, olfactory, and contact chemoreceptive signals were attacked as a result of our pheromone trapping had to (Strom et al. 1999). A distinction is made between the abil- be cut down promptly after our experiment to prevent forest ity to select hosts before landing (primary selection) and damage from infestation growth, so we do not know whether host selection exercised by tactile, gustatory, or chemorecep- there would have been successful reproduction by D. fronta- tive sampling after landing on potential hosts at random lis within the longleaf pines. (Moeck and Simmons 1991). Dendroctonus pseudotsugae Hopkins and Dendroctonus rufipennis Kirby recognize ap- propriate hosts via olfaction of plant volatiles in the absence Predation of an existing pheromone plume (primary attraction), Theoretically, differential impact of an herbivore on two whereas Dendroctonus ponderosae Hopkins fails to distin- potential host species could result from differential preda- guish between host and nonhost volatiles (Pureswaran and tion on the herbivore when it attempts to attack alternative Borden 2005) and appears instead to select hosts by gusta- host species. The specialist clerid predator T. dubius kills tory cues after landing on trees at random (Raffa and Berry- up to 60% of landing adult D. frontalis during the attack man 1982). There is no conclusive evidence for primary of a tree (Reeve 1997) and may play a key role in deter- attraction in D. frontalis (Payne 1980). Whether the first at- mining whether enough of the colonizing beetles survive to tack on a tree is influenced by prelanding cues, postlanding propagate the mass-attack necessary to overwhelm tree de- cues, or some combination, any tendency for discrimination fenses and permit beetle reproduction. However, the numer- between host tree species could lead to a bias in which po- ical response of T. dubius to D. frontalis pheromones did tential host species are most impacted by D. frontalis; not differ between longleaf and loblolly stands. This largely henceforth, we refer to the generic process of host discrimi- rejects the hypothesis that reduced use by D. frontalis of nation in the absence of aggregation pheromone as primary longleaf versus loblolly pine is a result of increased preda- host selection. Dendroctonus frontalis within mixed stands tion of D. frontalis when they attack longleaf versus lo- of loblolly and Virginia pine ( P. Mill.) blolly pine. were more likely to initiate attacks on Virginia pine and Although differential predation does not seem to explain had higher attack rates on Virginia pine once attacks were the low impacts of D. frontalis on longleaf pine, T. dubius initiated, indicating a role for both primary host selection may nonetheless play a role in the spatial patterning of in- and secondary attraction in the discrimination by D. fronta- festations. Unlike their prey, the abundance of T. dubius var- lis of these two pine species (Veysey et al. 2003; Ylioja et ied substantially among sites within forests. This was al. 2005). There is some evidence to suggest a role for gus- surprising because studies of T. dubius have indicated that tatory cues in host selection (Thomas et al. 1981). Although it disperses much more widely than D. frontalis; Cronin et we can conclude that D. frontalis does not avoid stands of al. (2000) estimated that 5% of T. dubius disperse >5 km. longleaf pine, the present study permits no inference about Thanasimus dubius can emerge from dead trees for up to 2 the role of primary host selection in generating the differen- years after oviposition by virtue of an extended diapause

# 2007 NRC Canada Friedenberg et al. 1435

(Reeve et al. 1995). These lasting point sources for migrants studies. Kier Klepzig, Maria Lombardero, Deepa Pure- may maintain spatial variation in T. dubius abundance. swaran, and Brian Sullivan provided valuable comments on the manuscript. Support was provided by a cooperative Causes of differential host impacts agreement with the USFS Southern Research Station and The life cycle of a D. frontalis spot (Fig. 4) begins with USDA NRI grant 2004-35302-14820. host selection and aggregation by dispersing beetles. We have shown that D. frontalis has a far greater impact on lo- References blolly pine than on longleaf, but we have been unable to at- Ayres, M.P., and Lombardero, M.J. 2000. Assessing the conse- D. frontalis tribute that disparity to the local abundance of quences of climate change for forest herbivores and pathogens. or its predator or to secondary attraction. It remains un- Sci. Total Environ. 262: 263–286. known whether D. frontalis discriminates between longleaf Ayres, M.P., Wilkens, R.T., Ruel, J.J., Lombardero, M.J., and Val- and loblolly pine during primary host selection, but this lery, E. 2000. Nitrogen budgets of phloem-feeding bark beetles could be evaluated with preference tests using host volatiles with and without symbiotic fungi. Ecology, 81: 2198–2210. to test for primary attraction (e.g., Pureswaran and Borden doi:10.2307/177108. 2005) and bolts to measure host acceptance after landing. Baldwin, V.C., and Feduccia, D.P. 1987. Loblolly pine growth and Studies that compare D. frontalis attack rates and reproduc- yield prediction for managed West Gulf plantations. USDA For. tive success in longleaf versus loblolly pine when the tree Serv. Res. Pap. SO-236. species are mixed could indicate whether the broad pattern Berryman, A.A., Raffa, K.F., Millstein, J.A., and Stenseth, N.C. of lower impact on longleaf stands is a result of behavioral 1989. Interaction dynamics of aggregation and coni- choices by individuals, demographics within D. frontalis fer defense rates. Oikos, 56: 256–263. doi:10.2307/3565345. spots, or both. Billings, R.F. 1985. Southern pine bark beetles and associated in- The most common explanation for differential impact is sects: effects of rapidly-released host volatiles on response to that longleaf pine is better defended than loblolly pine aggregation pheromones. Z. Ang. Entomol. 99: 483–491. (Hodges et al. 1977, 1979). However, recent related studies Billings, R.F. 1988. Forecasting southern pine beetle infestation have found no difference between loblolly and longleaf pines trends with pheromone traps. In Integrated control of scolytid in constitutive defenses (as measured by oleoresin flow) or bark beetles. Edited by T.L. Payne and H. Sarenmaa. Virginia in the probability of tree mortality after attack (S.J. Martin- State University, Blacksburg, Va. pp. 295–306. son and M.P. Ayres, unpublished data). A remaining hy- Billings, R.F., and Upton, W.W. 2002. How to predict southern pothesis is that larval survival is poor in longleaf trees, pine beetle infestation trends with pheromone traps. Tutorial. perhaps because qualitative differences in resin chemistry Texas Forest Service, Lufkin, Tex. influence the community of fungal symbionts that are cru- Blanche, C.A., Hodges, J.D., Nebeker, T.E., and Moehring, D.M. 1983. Southern pine beetle: the host dimension. Miss. Agric. cial to the survival of D. frontalis larvae (Ayres et al. For. Exp. Stn. Bull. 917. 2000; Lombardero et al. 2003; Hofstetter et al. 2005). Boyer, W.D. 1990. Pinus palustris Mill. longleaf pine. In Silvics of However, we doubt that larval survival is an adequate ex- North America. Vol. 1. Conifers. Edited by R.M. Burns and planation because trees are normally killed by oviposition, B.H. Honkala. USDA Agric. Handb. 654. pp. 405–412. whether the larvae survive or not, and groups of more than Brockway, D.G., Outcalt, K.W., Tomczak, D.J., and Johnson, E.E. five dead trees would be detected in aerial surveys (and 2005. Restoration of longleaf pine ecosystems. USDA For. Serv. represented in our analyses) even if there were no spot Gen. Tech. Rep. SRS-83. growth after the initial attacks. Butler, C.B. 1998. Treasures of the longleaf pines: naval stores. It would be surprising if the appearance of host selection Tarkel Publishing, Shalimar, Fl. in D. frontalis is purely an artifact of differential reproduc- Byers, J.A. 1999. Effects of attraction radius and flight paths on tive success. Although it might be that longleaf pine presents catch of scolytid beetles dispersing outward through rings of such a poor habitat (sensu Pulliam 1988) that D. frontalis is pheromone traps. J. Chem. Ecol. 25: 985–1005. doi:10.1023/ unable to adapt to its use (Holt and Gaines 1992), we A:1020869422943. know of no published data indicating that longleaf pine is Carnus, J.-M., Parrotta, J., Brockerhoff, E., Arbez, M., Jactel, H., actually a poor host as we and others have generally as- Kremer, A., Lamb, D., O’Hara, K., and Walters, B. 2006. sumed. It seems more likely that longleaf pine is avoided Planted forests and biodiversity. J. For. 104: 65–77. via some behavioral mechanism that we did not measure Clarke, S.R., Evans, R.E., and Billings, R.F. 2000. Influence of (e.g., differences in primary host selection). Identifying the pine bark beetles on the West Gulf Coastal Plain. Tex. J. Sci. causes and consequences of host selection by D. frontalis 52: 105–126. could inform some forest management decisions, especially Coppedge, B.R., Stephen, F.M., and Felton, G.W. 1994. Variation those related to longleaf restoration (Brockway et al. 2005), in size and lipid content of adult southern pine beetles, Dendroc- and would contribute to the general understanding of proc- tonus frontalis Zimmermann (Coleoptera: Scolytidae) in relation esses that drive niche evolution in heterogeneous land- to season. J. Entomol. Sci. 29: 570–579. scapes. Coulson, R.N. 1979. Population dynamics of bark beetles. Annu. Rev. Entomol. 24: 417–447. Acknowledgements Coulson, R.N., McFadden, B.A., Pulley, P.E., Lovelady, C.N., Fitz- gerald, J.W., and Jack, S.B. 1999. Heterogeneity of forest land- We thank personnel of USFS Forest Health in Region 8 scapes and the distribution and abudance of the southern pine and staff of the Chickasawhay and Oakmulgee RDs for their beetle. For. Ecol. Manage. 114: 471–485. doi:10.1016/S0378- cooperation in providing GIS information and felling at- 1127(98)00376-4. tacked trees. Khai Tran and Paul Marino assisted with field Cronin, J.T., Reeve, J.D., Wilkens, R., and Turchin, P. 2000. The

# 2007 NRC Canada 1436 Can. J. For. Res. Vol. 37, 2007

pattern and range of movement of a checkered beetle predator tera: Scolytidae), to bolts of lodgepole pine. Can. Entomol. 123: relative to its bark beetle prey. Oikos, 90: 127–138. doi:10. 299–304. 1034/j.1600-0706.2000.900113.x. Morales, J.M. 2002. Behavior at habitat boundaries can produce Ehrlich, P.R., and Raven, P.H. 1964. Butterflies and plants: a study leptokurtic movement distributions. Am. Nat. 160: 531–538. in coevolution. Evolution, 18: 586–608. doi:10.2307/2406212. doi:10.1086/342076. Feldman, R.M., Wagner, T.L., Sharpe, P.J.H., Gagne, J.A., and Payne, T.L. 1980. Life history and habits. In The southern pine Coulson, R.N. 1981. Within-tree life process models of the beetle. USDA Tech. Bull. 1631. Edited by R.C. Thatcher, southern pine beetle, Dendroctonus frontalis. Ecol. Model. 13: J.L. Searcy, J.E. Coster, and G.D. Hertel. USDA Forest Ser- 247–259. doi:10.1016/0304-3800(81)90031-4. vice, Combined Forest Pest Research and Development Program, Gara, R.I., and Coster, J.E. 1968. Studies on the attack behavior of Pineville, La. pp. 7–28. the southern pine beetle. III. Sequence of tree infestation within Price, T.S., Dogget, H.C., Pye, J.M., and Smith, B. 1997. A history stands. Contrib. Boyce Thompson Inst. 24: 77–85. of southern pine beetle outbreaks in the southeastern United Graham, S.A. 1939. Principles of forest entomology. McGraw-Hill, States. Georgia Forestry Commission, Macon, Ga. New York. Pulliam, H.R. 1988. Sources, sinks and population regulation. Am. Haddad, N.M. 1999. Corridor use predicted from behaviors at habi- Nat. 132: 652–661. doi:10.1086/284880. tat boundaries. Am. Nat. 153: 215–227. Pureswaran, D.S., and Borden, J.H. 2005. Primary attraction and Hedden, R.L., and Billings, R.F. 1977. Seasonal variations in fat kairamonal host discrimination in three species of Dendroctonus content and size of the southern pine beetle in East Texas USA. (Coleoptera: Scolytidae). Agric. For. Entomol. 7: 219–230. Ann. Entomol. Soc. Am. 70: 876–880. doi:10.1111/j.1461-9555.2005.00264.x. Hodges, J.D., Elam, W.W., and Watson, W.F. 1977. Physical prop- Raffa, K.F., and Berryman, A.A. 1982. Gustatory cues in the orien- erties of the oleoresin system of the four major southern pines. tation of Dendroctonus ponderosae (Coleoptera: Scolytidae) to Can. J. For. Res. 7: 520–525. host trees. Can. Entomol. 114: 97–104. Hodges, J.D., Elam, W.W., Watson, W.F., and Nebeker, T.E. 1979. Reeve, J.D. 1997. Predation and bark beetle dynamics. Oecologia, Oleoresin characteristics and susceptibility of four southern 112: 48–54. doi:10.1007/s004420050282. pines to southern pine beetle (Coleoptera: Scolytidae) attacks. Reeve, J.D., Ayres, M.P., and Lorio, P.L., Jr. 1995. Host suitability, Can. Entomol. 111: 889–896. predation, and bark beetle population dynamics. In Population Hofstetter, R.W., Mahfouz, J.B., Klepzig, K.D., and Ayres, M.P. dynamics: new approaches and synthesis. Edited by N. Cappuc- 2005. Effects of tree phytochemistry on the interactions among cino and P.W. Price. Academic Press, New York. pp. 339–357. endophloedic fungi associated with the southern pine beetle. J. Roberts, E.A., Billings, R.F., Payne, T.L., Richerson, J.V., Beris- Chem. Ecol. 31: 539–560. doi:10.1007/s10886-005-2035-4. ford, C.W., Edden, R.L., and Edson, L.J. 1982. Seasonal varia- PMID:15898500. tion in laboratory response to behavioral chemicals of the Holt, R.D., and Gaines, M.S. 1992. Analysis of adaptation in het- southern pine beetle Dendroctonus frontalis. J. Chem. Ecol. 8: erogeneous landscapes — implications for the evolution of 641–652. doi:10.1007/BF00989633. fundamental niches. Evol. Ecol. 6: 433–447. doi:10.1007/ Salinas-Moreno, Y., Mendoza, M.G., Barrios, M.A., Cisneros, R., BF02270702. Macias-Samano, J., and Zuniga, G. 2004. Areography of the Johnson, P.C., and Coster, J.E. 1978. Probability of attack by genus Dendroctonus (Coleoptera: Curculionidae: Scolytinae) in southern pine beetle in relation to distance from an attractive Mexico. J. Biogeogr. 31: 1163–1177. doi:10.1111/j.1365-2699. host tree. For. Sci. 24: 574–580. 2004.01110.x. Kelley, S.T., and Farrell, B.D. 1998. Is specialization a dead end? Schowalter, T.D., Pope, D.N., Coulson, R.N., and Fargo, W.S. The phylogeny of host use in Dendroctonus bark beetles (Scoly- 1981. Patterns of southern pine beetle (Dendroctonus frontalis tidae). Evolution, 52: 1731–1743. doi:10.2307/2411346. Zimm.) infestation enlargement. For. Sci. 27: 837–849. Kirkman, L.K., Goebel, P.C., Palik, B.J., and West, L.T. 2004. Pre- Shiyomi, M., Okada, M., Takahashi, S., and Tang, Y.H. 1998. Spa- dicting plant species diversity in a longleaf pine landscape. tial pattern changes in aboveground plant biomass in a grazing Ecoscience, 11: 80–93. pasture. Ecol. Res. 13: 313–322. doi:10.1046/j.1440-1703.1998. Kogan, M. 1994. Plant resistance in pest management. In Introduc- 00266.x. tion to insect pest management. Edited by R.L. Metcalf and St. George, R.A., and Beal, J.A. 1929. The southern pine beetle: a W.H. Lickmann. John Wiley & Sons, New York. pp. 73–128. serious enemy of pines in the south. USDA Farm. Bull. 1586. Lindgren, B.S. 1983. A multiple funnel trap for scolytid beetles Strom, B.L., Roton, L.M., Goyer, R.A., and Meeker, J.R. 1999. Vi- (Coleoptera). Can. Entomol. 115: 299–302. sual and semiochemical disruption of host finding in the south- Lombardero, M.J., Ayres, M.P., Hofstetter, R.W., Moser, J.C., and ern pine beetle. Ecol. Appl. 9: 1028–1038. doi:10.2307/2641348. Klepzig, K.D. 2003. Strong indirect interactions of Tarsonemus Thatcher, R.C., and Pickard, L.S. 1964. Seasonal variations in ac- mites (Acarina: Tarsonemidae) and Dendroctonus frontalis (Co- tivity of the southern pine beetle in east Texas. J. Econ. Ento- leoptera: Scolytidae). Oikos, 102: 243–252. doi:10.1034/j.1600- mol. 57: 840–842. 0706.2003.12599.x. Thistle, H.W., Peterson, H., Allwine, G., Lamb, B., Strand, T., Hol- Lorio, P.L., Jr., and Sommers, R.A. 1981. Central Louisiana. In sten, E.H., and Shea, P.J. 2004. Surrogate pheromone plumes in Site, stand, and host characteristics of southern pine beetle infes- three forest trunk spaces: composite statistics and case studies. tations. Edited by J.E. Coster and J.L. Searcy. USDA Tech. Bull. For. Sci. 50: 610–625. 1612. pp. 23–29. Thomas, H.A., Richmond, J.A., and Bradley, E.L. 1981. Bioassay Mizell, R.F., III, Frazier, J.L., and Nebeker, T.E. 1984. Response of of pine bark extracts as biting stimulants for the southern pine the clerid predator Thanasimus dubius (F.) to bark beetle phero- beetle. USDA For. Serv. Res. Note, SE-302. mones and tree volatiles in a wind tunnel. J. Chem. Ecol. 10: Tunset, K., Nilssen, A.C., and Andersen, J. 1993. Primary attrac- 177–187. doi:10.1007/BF00987655. tion in host recognition of coniferous bark beetles and bark wee- Moeck, H.A., and Simmons, C.S. 1991. Primary attraction of vils (Col., Scolytidae and Curculionidae). J. Appl. Entomol. 115: mountain pine beetle, Dendroctonus ponderosae Hopk. (Coleop- 155–169.

# 2007 NRC Canada Friedenberg et al. 1437

Turchin, P., and Odendaal, F.J. 1996. Measuring the effective sam- tion, protection, growth, and management. Charles Lathrop Pack pling area of a pheromone trap for monitoring population den- Forestry Foundation, Washington, D.C. sity of southern pine beetle (Coleoptera: Scolytidae). Environ. Wahlenberg, W.G. 1960. Loblolly pine. Its use, ecology, regenera- Entomol. 25: 582–588. tion, protections, growth and management. School of Forestry, Turchin, P., and Thoeny, W.T. 1993. Quantifying dispersal of Duke University, Durham, N.C. southern pine beetles with mark–recapture experiments and a Walker, L.C., and Oswald, B.P. 2000. The southern forest. Geogra- diffusion model. Ecol. Appl. 3: 187–198. doi:10.2307/ phy, ecology, and silviculture. CRC Press, New York. 1941801. Wood, S.L. 1982. The bark and ambrosia beetles (Coleoptera: Sco- Ungerer, M.J., Ayres, M.P., and Lombardero, M.J. 1999. Climate lytidae) of North and Central America, a taxonomic monograph. and the northern distribution limits of Dendroctonus frontalis Great Basin Nat. Mem. 6: 1–1359. Zimmermann (Coleoptera: Scolytidae). J. Biogeogr. 26: 1133– Ylioja, T., Slone, D.H., and Ayres, M.P. 2005. Mismatch between 1145. doi:10.1046/j.1365-2699.1999.00363.x. herbivore behavior and demographics contributes to scale- Van Valen, L. 1973. A new evolutionary law. Evol. Theory, 1: 1–30. dependence of host susceptibility in two pine species. For. Veysey, J.S., Ayres, M.P., Lombardero, M.J., Hofstetter, R.W., and Sci. 51: 522–531. Klepzig, K.D. 2003. Relative suitability of Virginia pine and lo- Zhou, M., and Buongiorno, J. 2006. Forest landscape management blolly pine as host species for Dendroctonus frontalis (Coleop- in a stochastic environment, with an application to mixed lo- tera: Scolytidae). Environ. Entomol. 32: 668–679. blolly pine – hardwood forests. For. Ecol. Manage. 223: 170– Wahlenberg, W.G. 1946. Longleaf pine. Its use, ecology, regenera- 182. doi:10.1016/j.foreco.2005.10.068.

# 2007 NRC Canada