Do Native Insects and Associated Fungi Limit Non-Native Woodwasp, Sirex noctilio, Survival in a Newly Invaded Environment?

Sirex noctilio F. (Hymenoptera: Siricidae) is an introduced pest of pines (Pinus spp.) in several countries in the Southern Hemisphere. Although S. noctilio is established in North America (first discovered in 2004), it has not been a destructive pest there so far, where forest communities more closely resemble those in its native Eurasian range—where it is not a pest. To investigate the influence of the existing community of associated insects (competitors + natural enemies) and fungi (vectored by insects) on S. noctilio survival in North America, we examined stage-specific mortality factors and their relative importance, generating life tables drawn from experimentally-manipulated and natural cohorts of Sirex spp. (mostly S. noctilio, but some native S. nigricornis F.). For both natural and experimentally-manipulated cohorts, factors which acted during the earliest Sirex life stages, most likely tree resistance and/or competition among fungal associates, were paramount in dictating woodwasp survival. Experimentally-manipulated life tables revealed that protection from the community of associates resulted in a significantly, and substantially larger (>15x) S. noctilio F1 generation than exposure to it. Seventy percent of generation mortality in the exposed cohort was due to tree resistance or unknown causes early in larval development, which could have included competition among other bark- or wood-inhabiting insects and/or their fungal associates. Only 46% of generation mortality in the protected cohort was due to tree resistance and/or unknown causes. Parasitoids, particularly endoparasitoids (Ibalia spp.), showed limited ability to control S. noctilio, and reduced the experimentally-established cohort by only 11%, and natural cohorts an average of 3.4%. The relative importance of tree resistance vs. competition with bark- and wood-borers in reducing S. noctilio survival remains unclear. Tree resistance and/or competition likely contribute more than natural enemies in maintaining the S. noctilio population in North America below damaging levels.

Introduction quality of each potential oviposition site and drill one to five chambers in a cluster, drilling more chambers and inserting more eggs if tree condition is favorable [26,27]. Eggs hatch inside these chambers, and larvae feed on A. areolatum, using the fungus to externally digest xylem [28], while creating meandering tunnels through the wood [29]. Larvae complete a variable number of instars (6-12, [26]) prior to pupation in late spring, usually of the following year.
The most important natural enemy of S. noctilio is a parasitic nematode, followed by two parasitoids, each acting at different times during S. noctilio development. Ibalia leucospoides ensiger (Norton) (Hymenoptera: Ibaliidae) and Rhyssa persuasoria (L.) (Hymenoptera: Ichneumonidae) are ideal natural enemies for S. noctilio in North America, since they are two of the most widely distributed parasitoids of the Siricinae in North America [30]. Ibalia spp. are endoparasitic kionobionts that attack eggs and early-instar siricids, have a similar seasonal phenology as Sirex spp., and generally complete one generation per year in North America (reviewed in [30]) [21]. Rhyssa spp. are ectoparasitic idiobionts that attack late-instar siricids [16,30]. Rhyssa spp. may undergo a short or a long life cycle, with potentially two generations per year, whereby adults attack late-instar Sirex spp. (most likely S. noctilio) in the fall, and emerge the following spring to attack the same generation of late-instar Sirex spp. still developing inside trees [16,21]. In other words, Sirex spp. larvae are vulnerable to attack from Rhyssa spp. for a relatively longer period of time than they are vulnerable to attack from Ibalia spp. The parasitic nematode, Deladenus (= Beddingia) siricidicola Bedding, has been widely used in biological control programs with varying success from very little to near complete control through sterilization of female S. noctilio, largely due to variation in virulence among different D. siricidicola strains [20,31]. Deladanus siricidicola is present in North America, and was likely introduced along with S. noctilio, yet it remains in the wasp's body cavity and does not penetrate eggs, which suggests that it is not capable of effectively sterilizing female wasps in North America [21,32].
Some evidence exists that the native community of associated insects, and the fungi that they vector, affects S. noctilio in North America. Sirex noctilio commonly (90% of the time) shared habitat (pine boles) with several other subcortical phytophagous species (bark beetles and other wood borers, including S. nigricornis F.) [33], and a wide array of coniferous-inhabiting species have been captured arriving at S. noctilio-infested trees [34]. Sirex nigricornis, native to North America, may at times co-habit the same pine trees as S. noctilio. The two species share parasitoids; compete with other wood borers for habitat; and attack stressed pinesthough S. nigricornis appears to require host material in a more advanced weakened condition than S. noctilio [21,33,[35][36][37]. Indirect competition among bark beetles and S. noctilio through respective obligate symbiotic fungi likely occurs in North American forests, and fungi associated with these bark beetles out-competed Amylostereum spp. in the laboratory [38]. Ibalia spp. (mostly leucospoides ensiger), are the most abundant natural enemies recovered from Sirex-infested pine in North America, with reports of 10-20% parasitism, although Rhyssa spp. were also commonly collected [21,35,36,39].
We investigated the influence of the existing community of insects, and the fungi that they vector, in recently-invaded pine forests on S. noctilio survival with two goals in mind. First, our results would indicate the importance of associates in limiting S. noctilio population growth in northeastern North America, which will aid in predictions of its behavior as it spreads further into eastern North America. And second, it may provide insight into characteristics of competitors or natural enemies in native communities with the capacity to mitigate impacts of potentially destructive non-native pests. Considerable attention has been paid to species that have become pests in their introduced environments (e.g. [40,41]) compared with species that have not caused excessive damage. This study took advantage of the fact that S. noctilio has had little impact on North American pine forests to investigate the resistance of the native, existing insect and fungal community to a new invader.
Our objective was to determine whether the existing community of associates can or already does have a negative impact on S. noctilio survival, and to quantify the importance of specific mortality factors throughout woodwasp development. We employed a life table approach to examine factors involved in life stage-specific mortality throughout one year (one generation) in the woodwasp's co-evolved host, P. sylvestris. We (1) examined the natural population of Sirex spp., sampling from cohorts at several sites throughout Ontario, and (2) experimentally manipulated S. noctilio cohorts that were exposed to, or protected from the natural community of associates. We predicted that Sirex mortality would be greatest in the earliest life stages, as is true of most other endophytic insects [42], probably due to tree resistance, which may prevent A. areolatum from establishing and creating a suitable habitat for larval development. Also, we expected that native parasitoids would limit S. noctilio survival.

Field methods: experimentally-manipulated cohorts
We selected an environment in which to establish manipulated S. noctilio cohorts that was representative of pine forests in Ontario with limited management. It was an unthinned, mixed Pinus sylvestris/P. resinosa plantation, near Angus, Ontario, where S. noctilio was present, but not in high densities (0.4 m 2 ha -1 of pine basal area attacked by S. noctilio in 2007 [14]). We obtained permission to conduct this study On the Brentwood tract from Simcoe County Forests. We felled 20 healthy, but small P. sylvestris (7-11 cm in diameter at breast height, dbh, and > 10 m tall) over a three-week period in July 2013. We cut a 3-4 m section from the midbole of each pine and placed each end of these logs onto a cinder block (25x10x10 cm). On the same day, we then secured two wire mesh (2x2 mm) screen cages (1 m long) to each horizontal log with heavy duty, plastic zip-ties at each end. The cages were equipped with Velcro 1 closures, sewn directly onto the wire mesh. To prevent cages from collapsing in on themselves, we screwed three wooden struts (10-15 cm long, cut from 5x5 cm lumber) into logs, equidistant apart (e.g. at 0°, 120°, and 240°), near both ends of cages, and encircled the struts in a hoop of plastic tubing (1.75 cm in diameter), screwed to the struts.
To artificially infest manipulated cohorts with S. noctilio, we obtained adults from infested pines in Innisfil, Ontario, identified by characteristic resin beading (resinosis) on the main bole [43]. These pines were cut in late June 2013, transported to the laboratory, and placed in large outdoor tents for collection of emerging adults. Each cage received two male/female pairs of S. noctilio, 1.5-3 weeks after pines were cut and caged, in an attempt to create a physically suitable (i.e. stressed, weakened) host tree for optimal survival of the F1 Sirex cohort (see [44]). Resin beading in response to S. noctilio attack was visible on most logs. To create the exposed cohort (n = 20), we removed one cage from each log (top or bottom section decided randomly by a coin flip) 1-2 weeks after wasps were inserted into cages. At this time all females were dead, and had presumably oviposited on logs. The remaining cages comprised the protected cohort (n = 20), and were left on logs throughout Sirex F1 development (or until October, see Sampling different life stages below).

Field methods: natural cohorts
Because Sirex-infested pines were difficult to locate throughout much of Ontario, and because sites used in this study were also used to conduct other Sirex-related studies, we were only able to remove one Sirex-infested tree per sampling date (see Sampling different life stages below) at each site. We selected six sites, representing a range of S. noctilio activity levels ( Table 1): S. noctilio was clearly dominant, both S. noctilio and S. nigricornis were common, or S. noctilio was newly established (first detected in 2013) and not yet dominant. It was not possible to determine whether a Sirex-attacked pine was attacked by S. noctilio or S. nigricornis until adults emerged from it. We identified Sirex-infested P. sylvestris as described above [43]. We obtained permission to removed trees from these sites from five private landowners, and the York Region Forests. We felled trees, cut the infested boles into~0.5 m sections, and transported them to the laboratory to sample the different life stages of Sirex.

Sampling different life stages
To estimate the number of Sirex in each life stage and quantify mortality factors throughout Sirex development, we cut, collected, and destructively sampled infested trees (natural cohorts) and logs (experimentally-manipulated cohorts) in October 2013 and again in June 2014 ( Table 2). We collected trees and logs in late June, just prior to adult S. noctilio emergence, to provide Rhyssa spp. a maximal amount of time to attack Sirex larvae.
From infested material collected in October, we quantified the number of Sirex surviving and mortality factors that acted in egg/neonate, small-, and mid-sized larval life stages. We first peeled the bark from logs with a drawknife to reveal the sapwood surface. Then, to estimate egg/neonate densities, we used a dissecting microscope (6.4x), positioned over the log surface, to count the number and type (single, double, triple, etc.) of oviposition drills. Many oviposition drills were flooded with resin. We used an equation developed by Madden [27] to estimate the number of eggs oviposited by females (i.e. realized fecundity) based on drill type (Total no. To count the number of developing Sirex larvae, identifiable by a characteristic spine protruding from the terminal abdominal segment [45], we used a hammer and chisel to carefully pull layers of wood away beneath oviposition drills, and followed frass-filled galleries, which revealed larvae at various depths in the wood. We assumed that early instars (neonates) would not be visible via this sampling technique (without magnification), so estimates of the first life stage encompassed both the egg and early-instar (neonate) stages.
Larvae were either dead (yellow in color and deflated, and mortality was assumed to be treerelated-either resistance or nutrition) or alive and apparently healthy. We dissected all live larvae in search of endoparasitoids (most likely Ibalia spp.). Ectoparasitoids were also found, and we reared a subsample (~10) to the adult stage, all of which were Rhyssa spp. Some larvae were damaged during log dissections and could not be dissected. We estimated the number of damaged larvae presumed to be harboring an endoparasitoid based on the level of parasitism observed from the remaining intact, live larvae recovered from the same log sample. The total number of wasp larvae (ectoparasitoids + dead Sirex larvae + live, apparently-healthy Sirex larvae) recovered was the number of Sirex entering the small larval stage, and the number of healthy, un-parasitized larvae was the number of Sirex entering the mid-sized larval stage.
Logs and trees collected in late June 2014 were placed in cardboard rearing tubes located in a covered outdoor shed in Angus, Ontario (see [33] for a description). Rearing tubes were checked 5x per week, and adult Sirex and ectoparastoids were collected and counted from early July until early October, 2014. Competitors (bark beetles and cerambycid wood-borers) may emerge from and/or colonize pines prior to Sirex, so we were not able to quantify their direct or indirect effects on Sirex survival during development, or differentiate their effects from those of tree resistance. This mortality was categorized as "unknown" in life tables. Unknown mortality during the egg/neonate and small larval stages was likely due to tree resistance and/ or competition, whereas unknown mortality during the mid-larval stage was likely due to overwintering. In the October log and tree collection, five of nine (55%) logs from the exposed cohort and two of the five trees from natural cohorts contained evidence of bark beetles and/or wood borers. Also, we were not able to identify whether Sirex were S. noctilio or S. nigricornis until the adult stage; when we were unsure, we referred to them as Sirex.

Data analyses
Using the estimated numbers of Sirex entering each life stage (egg/neonate, small larval, midsized larval, and adult), we constructed life tables, and expressed wasp densities as numbers per unit volume of wood, so that comparisons could be made among all cohorts (see Table 2 for actual numbers). The notation that we used for life tables was as follows: x = life stage (October log/tree collection included egg-mid-larval stages; June included adult stage); n = volume of pine wood sampled; lx = number entering life stage, estimated per m 3 of wood; dFx = mortality factor; dx = number dying during life stage.
All data analyses were conducted in R, version 3.1.1 [46]. We used generalized linear models (function = glm) to compare survivorship and mortality among and within life tables, fitted with appropriate error distributions. For all post-hoc testing, we used Tukey's HSD in package "multcomp" [47]. Significant differences were determined from the z-statistic. Statistical significance was set at P < 0.05. To compare survivorship (no. of Sirex entering each life stage) between the two experimentally-manipulated cohorts, and with log as the experimental unit, we tested exposure to the associate community (exposed vs. protected) with a glm, and fit a poisson error distribution, because the response data were counts, with an added term in the model to account for overdispersion in the data (family = quasipoisson). Statistical comparisons in survivorship among sites were not made for natural cohorts because so few trees were collected from each site. To search for a relationship between Sirex density and mortality (i.e. the possibility of density dependent mortality), we employed Pearson's correlations, with log as the experimental unit for experimentally-manipulated cohorts and site as the experimental unit for natural cohorts. For natural cohorts and with site as the experimental unit, we also used Pearson's correlations to search for relationships between stage-specific mortality and site-level Sirex activity and parasitism, as well as the relationship between estimated small larval and parasitoid densities from life tables. Patton was excluded from analyses because mortality was 100% in the egg/neonate stage.
We examined both real and apparent mortality. Real mortality is a measure of the mortality incurred during a particular life stage relative to the number of individuals entering the life table (no. of eggs/neonates), and is used to determine, within a life table, in which life stage mortality is greatest [48]. In contrast, apparent mortality is mortality in a particular life stage relative to the number of individuals entering that life stage, and is useful for comparing relative levels of mortality among life tables, for a particular life stage [48]. To compare apparent mortality among cohorts, but separately for each life stage, and real mortality among life stages, but separately for each cohort, we used glm (family = binomial (logit)). For experimental cohorts, the number of "successes" (no. of Sirex that died) relative to the total number of trials (no. of Sirex entering a life stage or the life table for apparent or real mortality, respectively) was the true number, and for natural cohorts it was an estimate, adjusted for the volume of wood sampled. This was necessary for natural cohorts, because a different volume of wood was sampled among sites and at each site in October compared with June (range = 0.006-0.21 m 3 of wood), whereas the total volume of wood that was sampled for each experimental cohort was similar between the two cohorts and log collection dates (0.06-0.07 m 3 ).
For the experimentally-exposed cohort, we explored the relationship between S. noctilio larval density and total parasitoid (ecto-+ endoparasitoids) as well as endoparasitoid density, first with a linear model (function = lm) and then with a polynomial model (function = lm(poly)), which provided a better fit (i.e. higher adjusted-R 2 ). We examined residual and normal probability plots for violation of linear model assumptions (function = plot); no violations were detected.

Life tables
In the experimentally-manipulated life tables (Table 3), the majority of mortality occurred during the egg/neonate stage (44% and 63% for the protected and exposed cohorts, respectively; real = apparent mortality for first life stage). Unknown mortality during the mid-larval stage (likely overwintering mortality) was also very high, accounting for 30% and 17% of real mortality in protected and exposed cohorts, respectively. In the exposed cohort, endo-and ectoparasitoids accounted for 9% and 2% of real mortality, during development (egg-mid-larval). In the exposed cohort, endo-and ectoparasitoids accounted for 24% and 4% of apparent mortality, in the small larval stage, and ectoparasitoids accounted for 8% of apparent mortality in the mid-sized larval stage. The protected cohort was slightly smaller (about three-quarters as large) than the exposed cohort initially; yet it was more than 15 times greater by the time adults emerged ( Table 3, Fig 1).
Only S. noctilio, no S. nigricornis, adults emerged from natural cohorts. Similar to the experimentally-manipulated life tables, mortality in the natural life tables was greatest during the egg/neonate stage (44-100%; Table 4). Mortality during this earliest life stage was especially high at sites where S. nigricornis was the dominant species and S. noctilio was new (Iron Bridge = 94%, Patton = 100%) or both species were present (Zephyr = 87%) ( Table 4). Real mortality due to unknown factors in the mid-sized larval stage, which was most likely due to overwintering, was also high, especially among cohorts that experienced relatively less mortality during the egg/neonate stage (range among Innisfil, Thames, and Little Lake = 23-37%). Among natural Sirex life tables, endo-and ectoparasitoids accounted for very little, 0-6% and 0-2%, of real mortality during development. Within the small larval stage, endo-and ectoparasitoids accounted for 0-14% and 0-4% of apparent mortality, which was much less than in the experimentally-exposed life table. At Thames, ectoparasitoids accounted for < 1% of apparent mortality during the mid-larval stage. No ectoparasitoids were recovered from the mid-larval stage in any of the other natural cohorts.

Survivorship
All life tables exhibited a Type IV survivorship curve (according to Slobodkin 1962), wherein mortality acted more severely on the earliest life stage, as evidenced by the steepest portion of Table 3. Partial age-specific life tables (2013-2014), including mortality factors, from experimentally-established Sirex noctilio populations, which were protected from or exposed to the naturally-occurring community of associate insects and fungi in southern Ontario.
Life  Associates Limit S. noctilio the curves existing between the first two life stages (Fig 1). Densities of surviving S. noctilio were not different between the experimentally-manipulated cohorts in the first two life stages, but they were significantly greater in the protected compared with the exposed cohort in the final two life stages (z = 2.49; df = 1,16; P = 0.012 and z = 3.64; df = 1,16; P < 0.001 for midsized larval and adult stages, respectively; Fig 2). The number of mid-sized larvae and adults were, on average, 2.5 and 16.5 times larger in the protected compared with the exposed cohort. At most sites, the shape of survivorship curves for natural cohorts (Fig 1b) was similar to that of the experimentally-protected cohort (Fig 1a); Thames was most similar to the experimentally-exposed cohort.

Mortality
Apparent mortality was significantly greater in the experimentally-exposed than in the protected cohort during all life stages (z = 9.91; df = 1; P < 0.001, z = 14.14; df = 1; P <0.001, and z = 9.32; df = 1; P <0.001 for egg/neonate, small larval, and mid-sized larval stages, respectively, Table 5). Apparent mortality in the small larval stage was more than 15 times greater in the experimentally-exposed (47%) than in the protected cohort (3%). In natural cohorts (Table 5), there was more variability among sites in apparent mortality during the egg/neonate than during the small-and mid-larval stages (44-100% vs. 11-23% and 58-96%, for egg/neonate vs. the small-and mid-sized larval stages, respectively). Comparisons of real mortality by life stage revealed that the egg/neonate stage incurred significantly more mortality than all other life stages, which was a consistent pattern for all life tables (44-100%; Table 6). Real mortality during the two larval stages was not different in the exposed cohort (Table 6), but was greater in the mid-larval than the small larval stage in the protected cohort (30% vs. 2%; Table 6). Similar to the protected cohort, real mortality during the mid-sized larval stage in natural cohorts was intermediate (4-37%), and lowest during the small larval stage (1-12%; Table 6).  For each treatment, dots represent the mean number of S. noctilio recovered from 9 logs (~0.07 m 3 wood) that were collected in October (egg/neonate-mid-sized larval), or from a second set of 9 logs (0.07 m 3 wood) that were collected in June (adult), and error bars represent standard errors. Cohorts were either protected from (filled dots) or exposed to (open dots) the naturally-occurring community of associates. The effect of exposure to associates was tested separately for each life stage. *P < 0.01 according to Tukey's HSD. doi:10.1371/journal.pone.0138516.g002

Density relationships
There was no evidence for density dependent mortality in either the experimentally-manipulated cohorts or in the natural cohorts. For the natural cohorts, comparisons between site-level Sirex activity (current and cumulative) and parasitism (Table 1), and Sirex density and apparent mortality at different life stages, revealed that density of eggs/neonates was significantly positively correlated with cumulative basal area of pine attacked by Sirex at a site (r = 0.857; P = 0.029, not shown), and apparent mortality in the small larval stage was negatively correlated with mean percentage of Ibalia parasitism at a site (r = -0.903; P = 0.036, not shown). From natural cohorts, estimated densities of small larvae were significantly, positively correlated with estimated densities of total parasitoids (r = 0.950; P = 0.004, not shown) and endoparasitoids (r = 0.948; P = 0.004, not shown), but not ectoparasitoids. All other pairwise correlations were not significant. Endoparasitiods in the experimentally-exposed cohort displayed a Type III functional response (Fig 3). As the number of small Sirex larvae per log increased, so did the number of Sirex that were parasitized, and at an accelerating rate (y = 0.2906 + 0.1365x + 0.0016x 2 ; F = 62; df = 2,6; P < 0.0001; adjusted R 2 = 0.94). High enough densities of Sirex to induce a saturation point (i.e. a sigmoidal curve with a clear upper asymptote) were not present in our study. Number of all parasitioids (endo-+ ectoparasitoids) exhibited a similar significant response to number of small Sirex larvae (y = 1.0478 + 0.0798x + 0.0025x 2 ; F = 80; df = 2,6; P < 0.0001; adjusted R 2 = 0.95, not shown). Ectoparasitoids were only recovered from four logs, and so were not analyzed separately.

Community resistance: trees
High levels of early-stage mortality indicate that initial conditions in pines were very important for Sirex survival. This same trend was reported from life tables of other wood-inhabiting insects [49][50][51][52]. It follows that mechanisms of tree resistance would be most effective at the early stages of insect attack: female oviposition or larval establishment. Indeed, tree resistance appears to be the most important factor limiting populations of many bark-and wood-boring insects (e.g. [52][53][54][55][56]).
Tree resistance is clearly an important obstacle for Sirex to overcome. The combined action of a symbiotic fungus and phytotoxic mucus represent a co-evolved mechanism for Sirex to combat a well-defended host habitat. Sirex noctilio oviposition drills are often flooded with resin when a tree is unsuccessfully attacked, killing eggs and small larvae [57][58][59]. We observed this constitutive resistance in many of the trees that we sampled-even on pine logs two weeks after cutting. In fact, all oviposition drills (~2900 eggs, in~1 m 3 of wood) were flooded with resin in our first attempt at this study in 2012, when we conditioned pines by girdling two weeks prior to infestation with S. noctilio. Pines also produce polyphenols in response to S. noctilio attack, in the sapwood surrounding the attack site [60,61]. This induced defensive response contains a high amount of pinosylvin, a fungicidal compound that likely prevents Amylostereum spp. from establishing [61]. Female S. noctilio will condition pines prior to oviposition, by injecting noctilisin and A. areolatum only, rendering trees susceptible to successful attack (i.e. larval development) in succeeding years [58]. In North America, S. noctilio (and S. nigricornis) usually kill, and develop within, unhealthy or suppressed pines [14]-those trees that are likely to have compromised constitutive and induced resistance capacity.

Community resistance: competitors
We were not able to separate the negative effects of tree resistance or overwintering from those of competitors (direct or indirect) on Sirex survival. Interspecific competition could have been important, given that apparent mortality was significantly higher in the experimentallyexposed compared with the protected cohort during all life stages (Table 5), and real mortality not attributable to parasitism was very high (76-100%) in natural cohorts, although some of that mortality was undoubtedly due to tree resistance and overwintering. Bark-beetle vectored fungi, Leptographium wingfieldii Morelet and Ophiostoma minus (Hedgcock) H. and P. Sydow, were more aggressive colonizers than A. areolatum of artificial media in the laboratory [38].
Amylostereum areolatum was not able to colonize media already occupied by bark-beetle fungi [38]. Bark beetles colonize pines earlier in the season than S. noctilio [33], which may afford their fungal associates an advantage over A. areolatum. General patterns among endophytic insects indicate that they suffer the most mortality from natural enemies and the least from competitors, with tree resistance being intermediate [42]. Our results indicate that natural enemies appear to be less important than other factors, such as tree resistance, and possibly competition, in limiting S. noctilio in North America.

Community resistance: parasitoids
Parasitoids, particularly endoparasitoids (Ibalia spp.), demonstrated limited ability to control S. noctilio, and reduced the experimentally-exposed S. noctilio cohort by only 11%. However, real mortality during the small-larval stage was significantly, and substantially smaller in the protected (2%) compared with the exposed cohort (17%), indicating that parasitism, especially by Ibalia, limited S. noctilio survival. Interestingly, life tables constructed from the natural Sirex population were more similar to the protected cohort in terms of mortality; extremely little mortality occurred during the small-larval stage (1-12%), when Ibalia parasitism became evident. In our experimental setting, Ibalia had a measureable, albeit small, negative effect on S. noctilio survival; yet this was not the case in the natural population, where parasitism reduced Sirex survival by a mean of 3.4%. Similar to our results, parasitism has been noted for other bark-and wood-boring insects ( [53,62], [55]; reviewed by [56]), including S. noctilio in North America [21,35,36,39], but these studies indicated that parasitoids do not exert enough impact ( 50% parasitism) to control borer populations. There are many possible explanations for this. For example, parasitoids may be generalists, and thus not explicitly co-evolved with the biology of the specific subcortical insect (e.g. Megarhyssa spp.). But most importantly, whether a specialist or a generalist, and for whatever reason, parasitoids may not have the ability to respond functionally or numerically to increases in their host populations.
The life history of Ibalia spp. and Rhyssa spp., and their efficacy as biological control agents in other areas, helps to interpret why they had such a low impact on Sirex in our study. Life history traits of I. leucospoides have facilitated its widespread establishment in areas where it has been introduced as a biological control agent (see Table 3, [63]). For example, I. leucospoides is phenologically synchronized with S. noctilio; it has a high potential fecundity (~600 eggs, 75% of which are mature upon adult emergence, [64]); it exhibits a functional response [65] and is a good forager; it can detect host patch richness from a distance and adjust foraging time accordingly [66,67]. However, other traits limit the success of I. leucospoides in controlling S. noctilio populations. Ibalia leucospoides has a long handling time (5-20 minutes per oviposition), which results in a sigmoidal functional response with a saturation point [65,66]. Local Ibalia spp. densities may be slow to respond to increasing Sirex densities due to their solitary lifestyle. Our observations from natural cohorts, that endoparasitism occurred at most sites, but at low frequency, were generally consistent with these ideas.
In our study, Ibalia spp. were likely more successful when confronted with larger initial densities of Sirex. The natural population of S. noctilio at the site we used to experimentally manipulate cohorts was low. Consequently, we expected that the parasitoid population was also low; yet, when presented with essentially a Sirex-larvae-buffet in our artificially-infested logs (mean = 56, maximum = 154 larvae per log), endoparasitoids exhibited a functional response, with higher numbers of parasitoids recovered from logs that also had higher numbers of Sirex larvae. We did not observe a saturation point, which suggests that the existing population of Ibalia spp., at least in that forest, was not limited by handling time. Among natural cohorts, sites with the largest densities of small larvae also had greater incidence of parasitism (Tables 1  and 4). Finally, endoparasitism could have been greater than what we reported, because we did not recover and dissect eggs and neonates. Ibalia spp. could have been responsible for some Sirex mortality during the egg/neonate stage that did not result in successful parasitoid development.
Ectoparasitoids, most likely Rhyssa spp., were substantially less common in our study than endoparasitoids. An intensive life table, constructed in Australia before D. siricidicola was introduced, provided long-term evidence (10 years) that Rhyssines (M. nortoni (Cresson) + R. persuasoria (L.)) were able to limit a S. noctilio population in a delayed density-dependent manner [68]. Taylor's [68] study also determined that R. persuasoria emergence was greatest during the second spring, more than one year after pines were attacked by S. noctilio. In contrast, Ryan et al. [33] found that only a very small amount ( 5%) of the wasp population (Sirex + parasitoids) required more than one year to complete development in Ontario. Further, re-infestation of the same (natal) logs upon adult emergence could not likely be ruled out for either study. These differences could be due to environment; Taylor [68] used emergence cages to capture adult insects on live/dying pines, whereas we (and Ryan et al. [21,33]) stored cut logs in a covered shed, which likely reduced insect development time through heating and drying of host material. Removal of Sirex-infested pines from Ontario forests in late spring, when at least a portion of the Rhyssa spp. population were actively seeking hosts, further complicates our results, and may have led to an underestimate of the impact of Rhyssa spp. on S. noctilio in North America.

Predictions for S. noctilio in North American pine forests
The native ecology in North American pine forests, including the community of associated insects and fungi that they vector, and pine resistance, reduced S. noctilio survival. However, the relative importance of tree resistance vs. competition among bark-and wood-borers and their fungal associates remains unclear. Future research should focus on this as well as improving tree resistance, and understanding what factors prevent parasitoids from effectively controlling S. noctilio. For instance, silvicultural treatments can improve pine health and reduce the number of S. noctilio-attacked trees by 75% [69]. And, Ibalia spp. may be more successful in warmer climates, where Sirex diapause is short, and Rhyssa spp. may be more successful in colder climates, where Sirex diapause is long, allowing Rhyssa spp. more search time [16,68,70]. Such variability in parasitoid success may have different implications for pine ecosystems in the southeastern U.S. vs. those in Canada.

Conclusions
Life table analyses indicated that factors which acted during the earliest Sirex life stages, most likely tree resistance and/or competition among fungal associates, were paramount in dictating woodwasp survival. Natural enemies, which included Ibalia spp. and Rhyssa spp., were not important mortality factors for Sirex, especially in cohorts drawn from the natural population. Experimentally-manipulated life tables revealed that protection from the community of associates resulted in a substantially larger (~15x) F1 generation than exposure to it. Seventy percent of generation mortality in the experimentally-exposed cohort was due to tree resistance (resin flooding the oviposition drills, and/or polyphenols that prevented A. aereolatum from establishing) or unknown causes early in larval development, which could have included competition among other subcortical insects and/or their fungal associates. Only 46% of generation mortality in the experimentally-protected cohort was due to tree resistance and/or unknown causes. In natural cohorts, generation mortality due to tree resistance, fungal competition, and/ or unknown causes early in larval development was more variable, and ranged from 47 to 100% among sites.