The enemy of my enemy is still my enemy: competitors add to predator load of a tree-killing bark beetle

Abstract

1 The mountain pine beetle Dendroctonus ponderosae is a major tree-killing bark beetle in North America. We evaluated how the subsequent arrival of a competing bark beetle Ips pini influences the arrival of predators and their impact on both species.

2 The predators Temnochila chlorodia and Enoclerus sphegeus were strongly attracted to pheromones of D. ponderosae. By contrast, Enoclerus lecontei was mostly attracted to I. pini pheromones. The host compound myrcene synergized attraction of both D. ponderosae and E. sphegeus to the pheromone of D. ponderosae. However, it inhibited attraction of both I. pini and E. lecontei to I. pini’s pheromone.

3 Dendroctonus ponderosae were more attracted to trees than logs treated with its pheromones, whereas I. pini were more attracted to logs than trees treated with its pheromones. Some 78% of T. chlorodia were captured at hosts baited with D. ponderosae pheromones, whereas 83% of E. lecontei were captured at hosts baited with I. pini pheromones. We characterized the sequence of arrival to live trees baited with pheromones of D. ponderosae as: D. ponderosae, T. chlorodia, E. sphegeus, I. pini, E. lecontei.

4 Various combinations of I. pini and predators were added to logs colonized by D. ponderosae in the above sequence of arrival observed in live trees baited with D. ponderosae aggregation pheromones. Ips pini reduced D. ponderosae adult brood production. However, the combination of I. pini and E. lecontei did not raise D. ponderosae brood production above that observed with only I. pini present. Similarly, the combination of I. pini and T. chlorodia did not reduce D. ponderosae brood production below that observed with I. pini alone. By contrast, the combination of I. pini, T. chlorodia and E. lecontei caused more brood loss to D. ponderosae than I. pini alone.

5 Enoclerus lecontei did not reduce brood production by T. chlorodia, whereas T. chlorodia substantially reduced brood production by E. lecontei.

6 Secondary bark beetles that exploit the resource created by primary tree-killing species exert negative effects through both competition and increased predator load. Implications to the population dynamics, ecology and evolution of tree-killing bark beetles are discussed.

Introduction

Populations of some bark beetle (Coleoptera: Curculionidae: Scolytinae) species undergo dramatic outbreaks that exert important economic and ecological effects. However, a number of biotic and abiotic factors often constrain bark beetles from developing outbreaks and most generations are spent within an endemic, low-density state (Berryman, 1982; Raffa et al., 2005). Among biotic constraints, the most important factors are tree resistance (Raffa & Berryman, 1983; Franceschi et al., 2005), predation (Reeve, 1997; Turchin et al., 1999) and interspecific competition (Berryman, 1974; Coulson, 1979; Rankin & Borden, 1991).

Predators of bark beetles consist primarily of beetles in the Cleridae, Trogositidae and Histeridae, and flies in the genus Medetera (Dolichopodidae) (Berryman, 1966; Bickel, 1991; Coulibaly, 1993). Predaceous beetles appear to be particularly important because they attack both the colonizing adults and the developing brood (Reeve, 1997). These predators are highly attracted to the pheromones that bark beetles produce to coordinate mass attacks and overcome tree defences (Dahlsten & Stephen, 1974; Stephen & Dahlsten, 1976; Wood, 1982; Billings & Cameron, 1984; Ross & Daterman, 1995; Zhou et al., 2001; Dahlsten et al., 2004; Fettig & Dabney, 2006). Attraction of predators to bark beetle pheromones has been observed in many systems, and typically lasts several days (Bowers & Borden, 1992; Poland & Borden, 1997; Gaylord et al., 2006; Hulcr et al., 2006). These predaceous beetles are considered habitat specialists but feeding generalists, in that they develop almost exclusively in trees killed by bark beetles, but feed on the diverse guild of insects within this resource (Person, 1940; Berryman, 1966; Erbilgin & Raffa, 2001).

Other species of bark beetles that are more typically associated with dead or severely stressed trees rapidly exploit the resource created by tree-killing species (Safranyik et al., 2000). Most of these beetles also produce aggregation pheromones when colonizing host tissue (Wood, 1982). When colonization by a tree-killing species is followed by a species more commonly associated with dead or severely stressed trees, the terms ‘primary’ and ‘secondary’ are commonly applied to designate this sequence (Amman & Safranyik, 1985; Rankin & Borden, 1991). This nomenclature is adopted in the present study, recognizing that these terms refer to the sequence of cohabitation, and not strict categories. The so-called ‘secondary beetles’ are generally considered more efficient competitors than species that kill vigorous trees (Rankin & Borden, 1991; Safranyik et al., 1996, 2000).

We currently have a limited understanding of how interactions among competitors and predators affect the population dynamics of eruptive bark beetles. In many insect systems, multispecies interactions yield outcomes that would not be predicted based on paired interactions alone (Price, 1997; Losey & Denno, 1998; Venzon et al., 2001; Denno et al., 2002). For example, if a predator preys more heavily on one of two competing species, it can reduce the superior competitors’ impacts on the less consumed species, a relationship termed ‘asymmetric apparent competition’ (Settle & Wilson, 1990; Holt & Lawton, 1994; Holt & Barfield, 2003). It is not known whether such interactions occur among bark beetles and their associates, from either a behavioural or population dynamics perspective.

The mountain pine beetle Dendroctonus ponderosae Hopkins is the major tree-killing bark beetle affecting ponderosa pine Pinus ponderosa and lodgepole pine Pinus contorta in the Rocky Mountains (Logan & Bentz, 1999; Carroll et al., 2004). As with other tree-killing bark beetles, most generations are restricted to stressed trees, a resource that is sparsely distributed and of relatively poor nutritional quality, and in which interspecific competition can be high (Carroll & Safranyik, 2004). During eruptions, however, mountain pine beetle escapes these constraints (Raffa et al., 2008), and can generate positive feedback through its mass attack behaviour. During initial phases of host tree colonization, mountain pine beetle primarily produces the pheromone components exo-brevicomin (Rudinsky et al., 1974) and trans-verbenol (Pitman et al., 1968), which attract both sexes. Attraction to these pheromones is synergized by a host compound, myrcene (Miller & Borden, 2000). Aggregation typically lasts 3–4 days (Wood, 1982; Amman & Cole, 1983).

Among the mountain pine beetle’s competitors is another bark beetle, the pine engraver Ips pini (Say) (Rankin & Borden, 1991; Safranyik et al., 1999, 2000). Extensive colonization by the pine engraver has often been observed in trees attacked by mountain pine beetle (Andrews, 1987; Humphreys & Ferris, 1987; Unger & Stewart, 1987). They partially partition the resource, with the pine engraver mostly colonizing the upper bole, and the mountain pine colonizing the lower (Furniss & Carolin, 1977). However, colonization by the pine engraver has been shown to substantially reduce mountain pine beetle brood survival and replacement rates (Furniss & Carolin, 1977; Amman & Safranyik, 1985; Rankin & Borden, 1991; Safranyik et al., 2000). The pine engraver produces the pheromone ipsdienol that attracts both sexes. In some regions, including Montana, it also produces lanierone, which is not attractive by itself but synergizes its attraction to ipsdienol (Miller et al., 1997; Steed, 2003).

The present study aimed to evaluate how: (i) attraction of a secondary beetle (I. pini) might affect the arrival of predators that also attack the primary tree-killing beetle (D. ponderosae) and (ii) interactions between secondary bark beetles and predators affect the reproductive success of the tree-killing beetle. Specifically, we evaluated the attraction of the three major coleopteran predators of bark beetles in Montana to pheromones of the mountain pine beetle and the pine engraver. We also compared how the overall assemblage varies under dynamic conditions (i.e. when responders to an initial source of volatiles can enter and likewise emit volatiles). Finally, we compared the effects of various combinations of competitors and predators on reproduction by the mountain pine beetle.

Materials and methods

Attraction of predators and bark beetles to synthetic lures

Experiments were conducted in a predominantly ponderosa pine Pinus ponderosa Douglas ex Lawson & C. Lawson forest at the University of Montana Lubrecht Experimental Forest in Greenough, Montana (46°53.30′N, 113°26.00′W). Preliminary assays conducted in 2003 demonstrated that D. ponderosae was present but at relatively low densities, identified the major predators and characterized seasonal flight patterns. Detailed information on the sites can be found at: http://nmviewogc.cr.usgs.gov/viewer.htm?bbox=-113.555,46.826,-113.302,46.965.

A behavioural choice assay was conducted in 2004, consisting of six treatments: (i) Control: no lure; (ii) myrcene; (iii) 97% -(–)-ipsdienol and lanierone; (iv) 97% -(–)-ipsdienol, lanierone plus myrcene; (v) exo-brevicomin plus trans-verbenol; and (vi) exo-brevicomin, trans-verbenol plus myrcene. The treatments were arranged in a randomized complete block design (RCBD) consisting of four sites with three blocks per site. Each block contained six twelve-unit multiple funnel traps (Lindgren, 1983) arranged in a line, with the collection cup approximately 75 cm from the ground. Each trap contained one randomly assigned treatment. Spacing was at least 10 m between traps within a block, 100 m between blocks and 500 m between sites. Overall, this yielded 12 replicates per treatment across all sites.

Treatments were placed in the field 9 June and samples were collected at 4-day intervals for 11 weeks ending 23 August. The lures were re-randomized at each sampling period. The lures were removed and replaced with new ones on 18 July to ensure continuous delivery.

All lures were obtained from Pherotech International Inc. (Delta, British Columbia). Trans-verbenol (85–95% W/W) was dispensed from polyvinyl bubble-cap lures with a release rate of 2.0 mg/day at 24°C. Exo-brevicomin (96–98%, w/w) was dispensed from a 2.5-cm polyurethane flexlure in a PVC sleeve with a release rate of 0.3 mg/day at 20 °C. Myrcene (90–92%, w/w) was dispensed from 20-mL low density polyethylene bottles with a release rate of 100 mg/day at 24 °C. Ipsdienol and lanierone were formulated in 1,3-butanediol. Ipsdienol (94–97%, w/w) was dispensed from polyvinyl bubble-cap lures with a release rate of 0.2 mg/day at 24 °C. The (–) stereoisomer was used because western populations of I. pini use this form (Birch et al., 1980). Lanierone (> 95%, w/w) was dispensed from polyvinyl bubble-cap lures with a release rate of 0.02 mg/day at 24 °C.

The numbers of I. pini, D. ponderosae and coleopteran predators captured in traps were recorded. Data were pooled across sampling dates and analyzed according to a RCBD using proc GLM in SAS (SAS Institute, 2003). Data were log(x + 1) transformed prior to analyses to reduce heteroscedasticity, and suitability of the resultant models was judged by visual inspection of their residual plots and Levene’s test. Means separation using t-tests were performed where significant treatment effects occurred. Untransformed means are presented.

Arrival sequence of bark beetles and predators under dynamic conditions to hosts baited with various pheromones

This experiment was conducted in 2004, in the same general location as the semiochemical attraction tests. It was designed to characterize the arrival sequence and time intervals of bark beetles and coleopteran predators as a dynamic process beginning from various initial conditions.

This experiment included six treatments: (i) P. ponderosa log plus three male/female pairs of I. pini; (ii) P. ponderosa log plus three male/female pairs of D. ponderosae; (iii) P. ponderosa log plus I. pini two-component lure: 97% -(–)-ipsdienol plus lanierone; (iv) P. ponderosa log plus D. ponderosae three-component lure: trans-verbenol, exo-brevicomin, plus myrcene; (v) living P. ponderosa tree plus I. pini two-component lure; and (vi) living P. ponderosa tree plus D. ponderosae three-component lure.

The treatments were arranged in a RCBD of five sites and three blocks per site. Each block contained six treatments, yielding 15 replicates per treatment. Spacing was at least 30 m between treatments, 100 m between blocks and 500 m between sites. Treatments were deployed in the field between 16 and 19 July to coincide with the beginning of the mountain pine beetle’s emergence flight. Pinus ponderosa trees approximately 20 cm in diameter at 1.4 m in height were used. For treatments requiring logs, one 16-unit funnel trap (Lindgren, 1983) and one panel trap (IPM Tech, Portland, Oregon) were suspended from ropes stretched over the log at 3.0 m height so that the collection cups were just above the log. The rope was attached to either neighbouring trees or aluminum poles placed in the ground. Treatments were administered the day after trees were felled. Treatments containing beetles received three pairs that were introduced by drilling small holes in the bark and phloem at 2.0, 2.5 and 3.0 m. At each entrance site, we placed one member of the host selecting gender (female: D. ponderosae; male: I. pini), followed by a mate. A 2 ´ 2 cm piece of window screening was placed over the hole and sealed with duct tape. For treatments involving standing trees, traps were suspended from ropes stretched between trees so that the collection cups were approximately 75 cm from the ground. Synthetic lures were in place for the first 5 days.

Arrival sequence was monitored by sampling insects collected in traps daily for 10  days. The species and numbers of predators and bark beetles were recorded, and live insects were retained for use in the controlled manipulation experiment. Insects caught in panel and funnel traps were pooled for analyses. Data were also pooled across sampling dates and analyzed according to a RCBD using proc GLM in SAS (SAS Institute, 2003). Data were log(x + 1) transformed for analyses. Mean separation using t-tests were performed where significant treatment effects occurred. Mean arrival date for each species within treatments was calculated by the equation:

• 

Where t is the day after treatment was applied (from 1–10 days), N_t is the number of insects captured on day t, and N_total is the total number of insects captured for the treatment.

Reproduction of mountain pine beetle in logs containing varying mixtures of pine engravers and predators

The controlled manipulation experiment consisted of five treatments in P. ponderosa logs: (i) D. ponderosae alone; (ii) D. ponderosae plus I. pini; (iii) D. ponderosae, I. pini plus Temnochila chlorodia (Mannerheim); (iv) D. ponderosae, I. pini plus Enoclerus lecontei (Wolcott); and (v) D. ponderosae, I. pini, T. chlorodia plus E. lecontei. Densities of beetles and sex ratios were selected to emulate field conditions: D. ponderosae: 68 females/m², 1 : 1 female : male (Raffa & Berryman, 1983); I. pini: 100 males/m², 1 : 2 male : female (Rankin & Borden, 1991). For each predator species, eight individuals were used. At least three females were used in each treatment. The gender of the predators was determined by observing mating. The predator treatments were selected based on the results of preliminary studies conducted in 2003 indicated that three species were the predominant predators, and the two field experiments in 2004 (see Results).

Live uninfested ponderosa pine trees approximately 25 cm in diameter at 1.4 m in height were felled on 28 July and cut into 50-cm sections. Within the next 2 days, the ends were sprayed with 10% sodium hypochlorite solution to prevent microbial contamination. The solution was allowed to dry, and the ends were sealed by dipping them in paraffin wax. Logs were stored indoors under a tarp for approximately 1 week at ambient temperatures (unheated room) to protect them from colonization by wild beetles until they were placed in a rearing chamber. The rearing chambers consisted of 60 ´ 40.6 cm diameter Fastform (Boyce Lumber & Design Center, Missoula, Montana) cardboard concrete form tubes. The ends were covered with aluminum window screening, which was then covered with opaque black cloth to block light from entering the chamber. A collection cup was fastened to one end of the chamber by cutting the base off of a 473-mL clear plastic cup leaving approximately 5 cm of the top of the cup to be taped with duct tape to a hole cut in the end of the chamber. The cup was then covered with a second intact cup to capture emerging beetles. The cup contained paper tissue that provided a surface on which the beetles could walk. The rearing room was maintained at 18–20°C under 24 h. light to promote emergence.

Insects for each treatment were introduced into the rearing chamber over a 9-day period to simulate the arrival sequence that we observed in the treatment emulating D. ponderosae attacking live trees in the arrival sequence experiment. The sequence and number of beetles introduced was: day 1: D. ponderosae females (26); day 2: D. ponderosae males (26); day 4: T. chlorodia (8); day 6: I. pini males (40); day 7: I. pini females (80), and day 9: E. lecontei (8). The beetles were introduced by placing them on a piece of tissue paper and placing the paper on the log within the chamber. The treatments were administered as a completely randomized design with four replications of treatments two to five and five replications of treatment one. Because there were only four replicates for most treatments, no correction for experiment-wide error was used.

The treatments began 5 August and one replicate of each treatment was initiated each day for 4 days. During the insect introductions, the holes for the collection cups were sealed with dark cloth to prevent beetles from escaping by flying towards light. After 1 week, the cloth was removed and the collection cups were monitored every other day for approximately 11 weeks. The logs were then dissected beginning 26 October. The number of adults, immatures, and gallery structures were recorded for each beetle species. Some data were log(x + 1) transformed prior to analyses to reduce heteroscedasticity, and suitability of the resultant models was judged by visual inspection of their residual plots. The numbers of dark and teneral adults were pooled, and data were analyzed using proc GLM in SAS (SAS Institute, 2003) and mean separation using t-tests were performed where significant differences occurred.

Results

Attraction of predators and bark beetles to synthetic lures

The most common predators were T. chlorodia (819), Enoclerus sphegeus (448) and E. lecontei (280). Temnochila chlorodia had two major peaks, 23–27 June and 14–18 July. Enoclerus sphegeus peaked from 18–27 June. Enoclerus lecontei peaked from 4–7 August.

Overall, E. lecontei was more attracted to the I. pini lure (Fig. 1A), whereas T. chlorodia and E. sphegeus were more attracted to the D. ponderosae lure (Fig. 1B, C). Attraction by E. lecontei to 97%-(–)-ipsdienol plus lanierone was reduced by 42.2% when myrcene was added to this combination. By contrast, its attraction to trans-verbenol plus exo-brevicomin was not affected by myrcene. Enoclerus sphegeus showed the opposite pattern of E. lecontei. Attraction by E. sphegeus to trans-verbenol plus exo-brevicomin increased 24.1% with the addition of myrcene to trans-verbenol plus exo-brevicomin. However, its attraction to ipsdienol plus lanierone was not affected by myrcene. Temnochila chlorodia was not affected by the presence of myrcene. No predators were attracted to myrcene alone.

Figure 1. Response (mean + SE) of the predominant bark beetle predators responding to pine engraver and mountain pine beetle pheromones and the host compound myrcene in western Montana in 2004 (A) Enoclerus lecontei (F_5,55 = 17.13, P < 0.0001), (B) Temnochila chlorodia (F_5,55 = 44.27, P < 0.0001) and (C) Enoclerus sphegeus (F_5,55 = 23.09, P < 0.0001). Means with the same letter are not significantly different (µ = 0.05).

A total of 2320 D. ponderosae and 2849 I. pini were captured. As expected, D. ponderosae was attracted to trans-verbenol plus exo-brevicomin plus myrcene and I. pini was attracted to ipsdienol plus lanierone. Their responses to the host compound myrcene were markedly different. Addition of myrcene increased the attraction of D. ponderosae to its pheromone by 78.1%, whereas it reduced the attraction of I. pini to its pheromone by 96.3% (Fig. 2).

Figure 2. Response (mean + SE) of Dendroctonus ponderosae (F_5,55 = 18.87, P < 0.0001) and Ips pini (F_5,55 = 7.53, P < 0.0001) responding to bark beetle pheromones and the host compound myrcene in western Montana in 2004. Means with the same letter are not significantly different (µ = 0.05).

Arrival sequence of bark beetles and predators under dynamic conditions to hosts baited with various pheromones

Results from the field experiment in which various sources of beetle pheromones were applied to logs or trees, and arriving insects could generate additional sources of attraction, are shown in Tables 1 and 2. A total of 1555 D. ponderosae, 11957 I. pini, 328 E. lecontei, 274 E. sphegeus and 211 T. chlorodia were captured. The number of arriving I. pini was 18.2-fold higher when its lures were attached to logs rather than to trees (Table 1). By contrast, the number of D. ponderosae was 1.9-fold higher when its lures were attached to trees rather than to logs. Enoclerus lecontei was most attracted to logs baited with I. pini lures, and in particular was 22.1-fold more attracted to logs baited with I. pini lures than to trees baited with I. pini lures. Enoclerus sphegeus likewise was most attracted to logs baited with I. pini lures. By contrast, the highest number of T. chlorodia arrived at trees and logs that had been baited with D. ponderosae lures.

Table 1.  Response (mean ± SE) of bark beetles and predators to host plants treated with various sources of beetle pheromones in western Montana, 2004

	Ips pini	Dendroctonus ponderosae	Enoclerus lecontei	Enoclerus sphegeus	Temnochila chlorodia	Total predators
Means within a column followed by different supescript letters are significantly different (µ = 0.05).
Log + Ips pini beetles	66.6 ± 21.65^b	1.33 ± 0.48^d	2.2 ± 0.98^b	0.9 ± 0.27^bc	1.5 ± 0.58^cd	4.5 ± 1.15^c
Log + Dendroctonus  ponderosae beetles	1.8 ± 0.63^c	6.33 ± 1.56^c	1.0 ± 0.39^b	0.5 ± 0.22^c	1.3 ± 0.61^cd	2.9 ± 0.99^cd
Log + Ips pini lures	676.7 ± 75.17^a	2.2 ± 0.90^d	14.7 ± 1.67^a	9.9 ± 5.20^a	2.1 ± 0.68^bc	26.7 ± 6.77^a
Log + Dendroctonus  ponderosae lures	7.5 ± 3.25^c	30.0 ± 5.45^b	1.2 ± 0.58^b	3.7 ± 1.61^b	4.1 ± 1.07^ab	9.0 ± 2.21^b
Tree + Ips pini lures	37.2 ± 9.67^b	5.67 ± 2.52^d	0.7 ± 0.33^b	0.3 ± 0.16^c	0.4 ± 0.19^d	1.4 ± 0.43^d
Tree + Dendroctonus  ponderosae lures	7.4 ± 4.46^c	58.1 ± 7.04^a	2.1 ± 0.73^b	2.9 ± 8.56^b	4.7 ± 1.16^a	9.7 ± 1.73^b
F5,78	72.09	39.25	40.03	7.62	6.67	24.93
P	< 0.0001	< 0.0001	< 0.0001	< 0.0001	< 0.0001	< 0.0001

Table 2.  Arrival sequence of bark beetles and predators, expressed as days after treatment applied

	Ips pini		Dendroctonus ponderosae		Enoclerus lecontei		Enoclerus sphegeus		Temnochila chlorodia
	Peak	Mean	Peak	Mean	Peak	Mean	Peak	Mean	Peak	Mean
Log + Ips pini beetles	2	4.13	5	4.21	2	4.58	2	4.08	9	6.35
Log + Dendroctonus  ponderosae beetles	1	2.89	2	2.32	2	3.73	7	6.38	8	5.55
Log + Ips pini lures	2	4.75	2	2.72	8	5.23	9	6.30	9	7.81
Log + Dendroctonus  ponderosae lures	2	4.29	2	3.00	5	4.22	2	5.50	2	3.95
Tree + Ips pini lures	3	4.63	3	2.26	2	3.00	1	4.40	1	4.00
Tree + Dendroctonus  ponderosae lures	8	7.02	2	3.76	8	7.45	5	5.39	2	4.29

The various bark beetle and predator species showed distinct patterns of arrival with each treatment (Table 2). Ips pini arrived at logs with D. ponderosae lures at a mean of 4.3 days and a peak at 2 days, whereas it did not arrive at trees with the same treatment until after a mean 7 days and a peak of 8 days. Dendroctonus ponderosae generally arrived more quickly than I. pini, but showed little variation among treatments. Enoclerus lecontei arrived more rapidly to trees with I. pini than D. ponderosae lures, but equally to logs with these sources, apparently responding to the early arrival of I. pini to logs. Temnochila chlorodia generally arrived more quickly to sources that included either D. ponderosae or trees than I. pini or logs. All logs were eventually colonized by I. pini. All trees were attacked by D. ponderosae but not in numbers required to kill the tree.

The sequence of arrival to D. ponderosae pheromones on a live tree was: (i) D. ponderosae, which peaked on day 2 post-placement of pheromones with a mean arrival date of 3.8 days; (ii) T. chlorodia, which peaked on day 2 with a mean arrival date of 4.3 days; (iii) E. sphegeus, which peaked on day 5 with a mean arrival date of 5.4 days; (iv) I. pini, which peaked on day 8 with a mean arrival date of 7.0 days; and (v) E. lecontei, which peaked on day 8 with a mean arrival date of 7.5 days. This sequence was used to guide the introduction sequence in the controlled manipulation experiment.

Reproduction of mountain pine beetle in logs containing varying mixtures of pine engravers and predators

In the controlled manipulation assay, the total number of brood adult D. ponderosae was reduced by the addition of I. pini, either predator alone with I. pini, and by both predators with I. pini (Fig. 3). The interspecific competition effect due to the presence of I. pini reduced the number of D. ponderosae adults by approximately 22.2%. The addition of both I. pini and the predator attracted to D. ponderosa, T. chlorodia, reduced the adult brood production of D. ponderosae by 25.8%, relative to when D. ponderosae developed alone. The combination of I. pini and the predator responding to pheromones of I. pini, E. lecontei, did not raise reproduction of D. ponderosae, but rather resulted in the same reduction as I. pini alone or I. pini plus T. chlorodia. By contrast, the addition of E. lecontei to the combination of I. pini and T. chlorodia was the only treatment that reduced D. ponderosae reproduction below that resulting from the competitive effect of I. pini. Logs with T. chlorodia, E. lecontei and I. pini yielded 32.5% fewer D. ponderosae adults than did those with D. ponderosae alone, and 15.7 % fewer than with D. ponderosae plus I. pini. This reduction in the replacement rate of D. ponderosae was largely cumulative across stages, but appeared to result mostly from mortality during the transition from pupal to adult stages (Table 3).

Figure 3. Emergence of adult Dendroctonus ponderosae from Pinus ponderosa logs containing various treatments of Ips pini, Temnochila chlorodia and Enoclerus lecontei (F_4,12 = 13.17, P = 0.0002). Means with the same letter are not significantly different (µ = 0.05).

Table 3.  Mean ± SE of various life stages of bark beetles and predators in the presence of competitors and/or predators in western Montana, 2004

Variable	Dendroctonus ponderosae alone	+Ips pini	+Ips pini	+Ips pini	+Ips pini	F 4,12	P
					+Temnochila chlorodia
			+Temnochila chlorodia	+Enoclerus lecontei	+Enoclerus lecontei
Means followed by a different superscript letter within a row are significantly different (α = 0.05).
Dendroctonus ponderosae
Ovipositional galleries	20.40 ± 1.29^a	15.00 ± 1.47^b	16.75 ± 1.65^ab	16.75 ± 1.38^ab	14.50 ± 2.02^b	1.74	0.205
Galleriy length (cm)	532.8 ± 25.4^a	524.6 ± 40.6^a	549.5 ± 41.2^a	509.8 ± 27.9^a	522.6 ± 67.6^a	0.12	0.972
Larval galleries	707.2 ± 41.8^a	572.0 ± 24.4^c	658.8 ± 27.7^a.b	652.3 ± 26.8^ab	613.8 ± 19.9^bc	2.56	0.093
Pupal chambers	334.6 ± 14.6^ab	274.0 ± 12.4^bc	257.8 ± 25.4^c	347.3 ± 25.3^a	317.3 ± 11.4^abc	4.17	0.024
Larvae	79.4 ± 12.9^a	53.0 ± 4.26^ab	46.25 ± 6.38^b	64.5 ± 6.55^ab	59.3 ± 10.3^ab	1.72	0.210
Pupae	82.4 ± 7.43^a	70.0 ± 1.58^a	68.75 ± 7.79^a	77.5 ± 7.08^a	65.3 ± 19.9^a	0.38	0.819
Teneral adults	121.6 ± 5.42^a	103.0 ± 6.52^b	91.5 ± 4.11^bc	88.25 ± 2.84^bc	79.8 ± 10.7^c	4.27	0.022
Hardened adults	109.4 ± 9.88^a	82.0 ± 6.87^b	80.0 ± 2.48^b	85.25 ± 5.2^b	76.25 ± 3.47^b	6.96	0.004
Ips pini
Ovipositional galleries	0^d	4.5 ± 1.32^a	2.75 ± 0.25^b	3.0 ± 0.41^ab	1.25 ± 0.48^c	15.34	< 0.001
Gallery length (cm)	0^d	24.73 ± 4.89^a	14.03 ± 2.08^b	16.85 ± 1.67^b	12.18 ± 5.02^c	15.70	< 0.001
Larval galleries	0^c	37.0 ± 6.10^a	26.25 ± 4.39^a	32.0 ± 5.58^a	20.75 ± 9.53^b	8.21	0.002
Pupal chambers	0^d	31.75 ± 5.57^a	16.0 ± 1.58^b	20.25 ± 2.53^b	12.25 ± 5.44^c	11.32	< 0.001
Larvae	0^b	5.75 ± 2.29^a	4.5 ± 1.32^a	6.0 ± 2.35^a	3.75 ± 1.65^a	8.65	0.002
Pupae	0^c	8.00 ± 2.12^a	5.0 ± 1.08^ab	5.25 ± 2.29^ab	4.75 ± 2.21^b	7.87	0.002
Teneral adults	0^d	13.25 ± 3.20^a	11.0 ± 4.62^ab	7.25 ± 3.57^bc	5.00 ± 2.65^c	12.42	< 0.001
Hardened adults	0^c	12.50 ± 3.48^a	9.75 ± 1.11^a	9.50 ± 2.60^a	7.25 ± 3.09^b	14.49	< 0.001
Total adults	0^e	25.75 ± 6.07^a	20.75 ± 5.65^b	16.75 ± 4.07^c	12.25 ± 5.69^d	21.64	< 0.001
Temnochila chlorodia
Larvae	0^c	0^c	2.75 ± 1.25^b	0^c	5.00 ± 0.82^a	18.92	< 0.001
Adults	0^b	0^b	2.50 ± 1.44^a	0^b	2.25 ± 1.25^a	4.28	< 0.001
Total	0^c	0^c	5.25 ± 2.56^b	0^c	7.25 ± 1.97^a	14.28	< 0.001
Enoclerus lecontei
Larvae	0^c	0^c	0^c	3.75 ± 0.75^a	1.50 ± 0.65^b	14.51	< 0.001
Adults	0^b	0^b	0^b	1.75 ± 0.85^a	0.25 ± 0.25^b	5.50	0.009
Total	0^c	0^c	0^c	5.50 ± 0.65^a	1.75 ± 0.75^b	27.45	< 0.001

Ips pini also showed significant responses to predator treatments. Production of brood adults was reduced more by E. lecontei than T. chlorodia, and most strongly by the combination (Table 3). The greatest impact occurred during establishment of ovipositional galleries, which were reduced by 72% when both predators were present compared with when both were absent. Additional reduction due to predators occurred in larval galleries, pupal chambers, pupae, teneral adults and hardened adults.

Predator reproduction also varied among treatments (Table 3). Enoclerus lecontei and T. chlorodia performed equally well when introduced alone. Addition of E. lecontei did not reduce brood production by T. chlorodia. By contrast, T. chlorodia reduced brood production of E. lecontei by 68%.

Discussion

Exploitation of trees killed by D. ponderosae by I. pini appears to add to the predator load experienced by D. ponderosae. The chemical signals produced by I. pini are highly attractive to E. lecontei, which otherwise might not locate and predate on D. ponderosae to the extent observed. Overall, 65% of the E. lecontei that arrived were in traps containing ipsdienol (Fig. 1). Moreover, the later arrival of pine engravers to trees (Table 2), and its longer flight season (Rankin & Borden, 1991; Safranyik et al., 1996, 2000), can extend the period during which trees colonized by mountain pine beetle are attractive to predators. Mountain pine beetle brood require a full season or more to develop under natural conditions, and hence are susceptible to predators that track the pheromones of I. pini, which often arrives substantially after D. ponderosae has ceased producing pheromones (Rankin & Borden, 1991; Safranyik et al., 1999).

The additional predators recruited by I. pini reduce reproduction by D. ponderosae. Overall, I. pini reduced D. ponderosae reproduction by approximately 22.2% due to competition, and an additional 12.5% due to competitor–predator interactions. This interaction of predators and competitors supports the rationale of Safranyik et al. (1996) for considering manipulation of I. pini as a tactic for reducing D. ponderosae reproduction.

Direct and indirect effects of competitors appear to be more severe on D. ponderosae when it colonizes dead rather than live hosts. For example, I. pini arrived only 1.3 days after D. ponderosae on logs baited with D. ponderosae pheromones, compared with 3.3 days after D. ponderosae in trees (Table 2). Likewise, E. lecontei arrived only 1.2 days after D. ponderosae in logs, compared with 3.7 days after D. ponderosae in live trees. Furthermore, the total number of predators per D. ponderosae was 0.30 in logs, compared with only 0.17 in trees (Table 1). Predator load can be an important factor in whether species such as D. ponderosae surpass the eruptive threshold beyond which they generate self-amplifying dynamics (Reeve, 1997; Turchin et al., 1999; Raffa et al., 2005). The combined direct competitive effects and indirect predator-attracting effects of I. pini on D. ponderosae reproduction (Fig. 3) support the view that the tree-killing strategy in bark beetles evolved in part as an escape from competition (Raffa & Berryman, 1987; Rankin & Borden, 1991; Raffa et al., 1993). That is, poorly defended trees reduce risks associated with colonization, but can increase risks to competitors and their associates.

These results yield further insights into how predators exploit pheromones and host tree volatiles associated with bark beetles. The host monoterpene myrcene strongly synergized the attraction of mountain pine beetle to its pheromones (Miller & Borden, 2000), but strongly inhibited attraction of I. pini to its pheromones. Predators tended to respond accordingly. That is, attraction of E. sphegeus to the pheromone of D. ponderosae was synergized by myrcene, but its attraction to the pheromones of I. pini was unaffected. Likewise, attraction of E. lecontei to pheromones of I. pini was reduced by myrcene, but its attraction to pheromones of D. ponderosae was unaffected. Predators use of host plant volatiles to track bark beetle pheromones (Erbilgin & Raffa, 2000, 2001) may be advantageous when the prey modifies enantiomeric ratios and synergists (Aukema et al., 2000; Dahlsten et al., 2003).

These results also suggest that predators of secondary bark beetles face risks arising from predators of the tree-killing beetle. Enoclerus lecontei experienced substantially lower reproduction in the presence of T. chlorodia, due to competition, intraguild predation, or both. Evidence of intraguild predation (i.e. partially consumed E. lecontei) was observed when logs were dissected, but it is unknown when this occurred or which species consumed them. This unidirectional relationship is likely due to the earlier introduction of T. chlorodia in our laboratory manipulations, that likewise emulated the earlier arrival of T. chlorodia in the sequential arrival experiment (Table 2), and its attraction to D. ponderosae pheromones (Fig. 1) in the field. Selective pressures resulting from the sequence of arrival could partially explain why E. lecontei prefers logs to trees with I. pini lures when given a choice (Table 1). In logs, there were only 0.15 T. chlorodia per E. lecontei but, in trees, this rose to 0.57.

The mechanism by which I. pini locates trees colonized by D. ponderosae is not known (Amman & Safranyik, 1985; Rankin & Borden, 1991; Safranyik et al., 2000). We did not find evidence for attraction to pheromones of D. ponderosae (Fig. 2). Post-landing detection of chemical cues associated with weakened trees is known to be an important component of host selection by I. pini (Klepzig et al., 1996; Wallin & Raffa, 2000), and so may be involved in detecting trees successfully attacked by D. ponderosae.

We currently have little data on how competitors and predators interact with important abiotic factors such as temperature to affect bark beetle populations. However, the available information suggests they may be augmentative. Dendroctonus ponderosae has a facultative diapause, requiring 1 year for development under warm conditions but 2 years under cooler conditions (Bentz & Mullins, 1999; Bentz et al., 2001). In cases with semivoltinism, trees containing D. ponderosae brood can be colonized by I. pini not only during the year of attack, but also the next year (Rankin & Borden, 1991; Safranyik et al., 1999), thus adding further to the predator load of D. ponderosae. However, with recent warming trends, D. ponderosae is now univoltine in many regions with historically semivoltine development (Carroll et al., 2004; Hicke et al., 2006). In some of these areas, the mountain pine beetle is now causing unprecedented levels of mortality. A shortened window of opportunity for predators responding to secondary beetles could potentially benefit D. ponderosae survival by adding to the effects of reduced freezing mortality and accelerated development in temperature-driven population releases (Bentz & Mullins, 1999; Hicke et al., 2006).

Acknowledgements

The generosity and assistance of the staff of Lubrecht Experimental Forest, University of Montana, especially Hank Goetz and Frank Maus are deeply appreciated. Glenn Koepke (USDA Forest Sevice, St Regis, Montana), Rich Stocker (DNRC-Forestry Division, Missoula, Montana), Bob Rich (DNRC-Forestry Division, Missoula, Montana) and Julie Hacker provided access to additional land for experiments and insect trapping. Ken Gibson (USDA Forest Service, Missoula, Montana) and Mike Albers (DNR-MN, Grand Rapids, Minnesota) provided supplemental funnel traps and Darek Czokajlo (IPM Tech, Portland, Oregon) provided panel traps. Jennifer (Shaw) Rackley and Kendall Crawford (Dept. Ecosystem and Conservation Sciences, University of Montana, Missoula, Montana), David Patton, Regan Patton and Chris Maus provided field assistance and Katie Kieler (Department of Entomology, University of Wisconsin, Madison) provided technical assistance. Steven Krauth (Department of Entomology, University of Wisconsin, Madison) assisted with insect identification. The critical reviews of Aaron Adams (Department of Entomology, University of Wisconsin, Madison), Brian Aukema (Canadian Forest Service) and four anonymous reviewers are greatly appreciated. This work was supported by USDA-NRI and the University of Wisconsin, Madison College of Agricultural and Life Sciences.