Seasonal Captures of Trissolcus japonicus (Ashmead) (Hymenoptera: Scelionidae) and the Effects of Habitat Type and Tree Species on Detection Frequency

Simple Summary Trissolcus japonicus, an important natural enemy of brown marmorated stink bug in Asia, was first detected in the USA in 2014. To investigate when and where T. japonicus is found in the field, yellow sticky traps were deployed in the canopy of tree of heaven growing at the edge of small isolated patches, windbreaks, and woodlots in 2018 and 2019. In both years, captures occurred from May to September, with peaks in July and August. Captures of T. japonicus were recorded from all three habitats but were not consistently associated with a particular habit. In 2017 and 2018, T. japonicus captures were compared between tree of heaven paired with several other H. halys host trees growing at the woods edge, and in 2019, captures in tree of heaven, black walnut, and black locust growing in the same windbreaks were compared. Trissolcus japonicus and several native H. halys parasitoids were captured in all hosts, but there was not a consistent effect of host tree species on T. japonicus captures. These results can be used to inform and optimize future surveillance efforts for detecting T. japonicus as it continues to expand its range in the USA. Abstract Trissolcus japonicus, an important egg parasitoid of Halyomorpha halys in Asia, was first detected in the USA in 2014. To evaluate the effect of habitat and the seasonality of T. japonicus detections in the USA, yellow sticky traps were placed in the canopy of Ailanthus altissima growing at the edge of isolated patches of trees, windbreaks, and woodlots in northern Virginia in 2018 and 2019. In both years, captures occurred from May to September, and peaked in July and August. While T. japonicus was detected in all habitats, there was not a consistent effect of habitat type on capture frequency. To evaluate tree species effects on T. japonicus captures, in 2017 and 2018, yellow sticky traps deployed in the canopy of A. altissima bordering apple orchards were paired with a nearby trap in one of several wild tree species along a common woods edge. In 2019, these traps were deployed in A. altissima, black walnut, and black locust growing in the same windbreaks. No consistent association between captures of T. japonicus or native parasitoids of H. halys and the tree species sampled was observed among years. Results are discussed in relation to the ecology and sampling optimization of T. japonicus.


Introduction
Halyomorpha halys (Stål) (Hemiptera: Pentatomidae), is a polyphagous invasive stink bug from Asia that has been a severe agricultural and nuisance pest in many parts of the USA since the late 2000s [1]. A widespread outbreak in 2010 resulted in major losses to the apple and peach crops in the Mid-Atlantic USA [2]. To manage H. halys, many American Insects 2021, 12, 118 2 of 12 tree fruit producers increased their use of broad-spectrum insecticides [3], but this is certainly not considered a sustainable, long-term solution. Ultimately, biological control may play an important role in sustainable H. halys management [1]. Surveys of native H. halys parasitoids in North America [4] revealed that the species detected most commonly from sentinel egg masses included members of the genera Anastatus Motschulsky (Hymenoptera: Eupelmidae), Telenomus Haliday (Hymenoptera: Scelionidae), and Trissolcus Ashmead (Hymenoptera: Scelionidae). However, field studies have suggested that endemic parasitoids and predators in the USA are not yet regulating H. halys populations adequately [4,5].
Detections of North American H. halys parasitoids have been somewhat habitat dependent [4]. Trissolcus species tended to be more prevalent in ornamental, semi-natural/urban, and forest habitats than in other systems [4], and Herlihy et al. [6] found that this genus predominated in wooded habitats. Parasitization of sentinel H. halys eggs did not differ between exotic or native host plants but was greatest at the edge of unmanaged woods [7]. Anastatus species were among the most common parasitoids in ornamental systems [8].
Trissolcus japonicus (Ashmead) (Hymenoptera: Scelionidae) is one of the most important natural enemies of H. halys in Asia [9,10]. The discovery of an adventive population of T. japonicus in Maryland in 2014 [11] prompted more intensive and extensive surveillance for H. halys parasitoids in North America and Europe, resulting in T. japonicus detections in 13 states, Washington, DC [6,[12][13][14], Canada, Italy, and Switzerland [15][16][17][18]. Although Abram et al. [19] noted that the effects of egg parasitoids, including T. japonicus, on H. halys populations have not been quantified, intensified sampling for T. japonicus in the USA to track changes in its spread and abundance is warranted and would benefit from a better understanding of its temporal and spatial distributions.
Zhang et al. [10] documented the seasonal parasitism of H. halys eggs by T. japonicus in Beijing, China, although its seasonal phenology elsewhere has not been reported. In the USA, it has been detected in woodland [6,20,21] and lightly wooded residential habitats [13], and Herlihy et al. [6] reported that it parasitized significantly more H. halys sentinel eggs in wooded habitats than in soybean or apple plantings. In both the USA [20] and China [10], T. japonicus has also been detected in peach orchards. However, there have not been systematic comparisons of T. japonicus detection frequency and relative abundance among the wooded habitats in which it has been found, particularly those adjacent to crops at risk from H. halys attack. Moreover, although T. japonicus detections in the USA have been most common in arboreal habitats [6,11,[21][22][23], its presence and abundance among different H. halys tree hosts has not been examined.
Optimizing the efficiency of sampling to track changes in the range and abundance of T. japonicus in the USA can be achieved and informed by a greater understanding of potential habitat and plant species effects on its detection frequency, and by documenting its seasonal phenology. Here, studies conducted in northern Virginia, USA, where T. japonicus has been present since at least 2015 [21], address adventive T. japonicus captures in relation to habitat type, plant species, and seasonality.

Seasonal Phenology and Habitat Type
Study sites were unmanaged, wooded habitats adjacent to commercial and experimental orchards in the counties of Frederick (14 sites)  Tree of heaven, Ailanthus altissima (Mill.) Swingle, is an invasive Asian species [24] that often grows prolifically in disturbed or semi-disturbed locations in Virginia and elsewhere in the USA [25]. In this area, it was the most abundant deciduous tree species at the edge of woodlands bordering tree fruit orchards [26] and supports all H. halys life stages [27]. In addition, inspection of the foliage of felled tree of heaven yielded H. halys egg masses Insects 2021, 12, 118 3 of 12 parasitized by T. japonicus [20]. The prevalence of this tree and the occurrence of both H. halys and T. japonicus on it led to the selection of female tree of heaven for standardized sampling.
In this area, three common habitat types in which tree of heaven grows are: (1) spatially isolated, often roughly circular patches typically associated with rock breaks in otherwise cultivated or fallow fields; (2) windbreaks or hedgerows; (3) the edge of woodlands ( Figure 1). All of these habitat types are commonly associated with commercial tree fruit orchards and thus were selected for sampling (n = 5 sites per type). The mean (± SE) distance (km) between isolated patch, windbreak, and woodlot sites was, 3.8 ± 0.5, 6.5 ± 0.7, and 4.2 ± 0.5, respectively, and the mean (± SE) distance (km) between a given site and the site nearest to it was 0.84 ± 0.24. The mean (± SE) area (m 2 ) comprised of trees at these sites was; (1) isolated patches 278.6 ± 88.1 (17.3  Tree of heaven, Ailanthus altissima (Mill.) Swingle, is an invasive Asian species [24] that often grows prolifically in disturbed or semi-disturbed locations in Virginia and elsewhere in the USA [25]. In this area, it was the most abundant deciduous tree species at the edge of woodlands bordering tree fruit orchards [26] and supports all H. halys life stages [27]. In addition, inspection of the foliage of felled tree of heaven yielded H. halys egg masses parasitized by T. japonicus [20]. The prevalence of this tree and the occurrence of both H. halys and T. japonicus on it led to the selection of female tree of heaven for standardized sampling.
In this area, three common habitat types in which tree of heaven grows are: 1) spatially isolated, often roughly circular patches typically associated with rock breaks in otherwise cultivated or fallow fields; 2) windbreaks or hedgerows; 3) the edge of woodlands ( Figure 1). All of these habitat types are commonly associated with commercial tree fruit orchards and thus were selected for sampling (n = 5 sites per type). The mean (± SE) distance (km) between isolated patch, windbreak, and woodlot sites was, 3.8 ± 0.5, 6.5 ± 0.7, and 4.2 ± 0.5, respectively, and the mean (± SE) distance (km) between a given site and the site nearest to it was 0.84 ± 0.24. The mean (± SE) area (m 2 ) comprised of trees at these sites was; 1) isolated patches 278.6 ± 88.1 (17.  As described in Quinn et al. [22], backfolding yellow sticky traps (23 × 28 cm, Alpha Scents, West Linn, OR) deployed atop 4.8 m bamboo poles were used for sampling. Holes were punched through both halves of folded traps, at about 2.54 cm on either side of the center point and about 1.9 cm from the edge. A 45.7 cm length of twist tie was inserted through the holes, the top of the pole was inserted between the trap sides, and the twist tie was used to secure the trap by wrapping and twisting it around the pole and finally to the shank of a wire hook affixed to the pole below the trap. Interlocking tabs on the trap corners ensured that both sides were closely appressed to the pole when deployed. One trap was deployed in the mid-canopy of a mature female tree of heaven growing at the habitat edge at each site by suspending the pole from a lateral branch via the wire hook. Traps were deployed on 3 May and 20 April in 2018 and 2019, respectively, and replaced at 7 ± 2-day intervals through 21 or 30 September in the respective years.

Host Plant Comparisons
Mature trees at the edge of woodlands (2017 and 2018) and windbreaks (2019) adjacent to tree fruit orchards within 10 km of the AHSAREC were used for trapping. The height and diameter at breast height (DBH) of each sample tree was recorded using a Nikon Forestry Pro Hypsometer (Nikon Corporation, Tokyo, Japan) and measuring tape, respectively. As in the previous study, female tree of heaven (11.2 ± 1.3 m tall, 0.2 ± 0.02 m DBH) was the standard species used in all paired host comparisons. The endemic species As described in Quinn et al. [22], backfolding yellow sticky traps (23 × 28 cm, Alpha Scents, West Linn, OR, USA) deployed atop 4.8 m bamboo poles were used for sampling. Holes were punched through both halves of folded traps, at about 2.54 cm on either side of the center point and about 1.9 cm from the edge. A 45.7 cm length of twist tie was inserted through the holes, the top of the pole was inserted between the trap sides, and the twist tie was used to secure the trap by wrapping and twisting it around the pole and finally to the shank of a wire hook affixed to the pole below the trap. Interlocking tabs on the trap corners ensured that both sides were closely appressed to the pole when deployed. One trap was deployed in the mid-canopy of a mature female tree of heaven growing at the habitat edge at each site by suspending the pole from a lateral branch via the wire hook. Traps were deployed on 3 May and 20 April in 2018 and 2019, respectively, and replaced at 7 ± 2-day intervals through 21 or 30 September in the respective years.

Host Plant Comparisons
Mature trees at the edge of woodlands (2017 and 2018) and windbreaks (2019) adjacent to tree fruit orchards within 10 km of the AHSAREC were used for trapping. The height and diameter at breast height (DBH) of each sample tree was recorded using a Nikon Forestry Pro Hypsometer (Nikon Corporation, Tokyo, Japan) and measuring tape, respectively. As in the previous study, female tree of heaven (11.2 ± 1.3 m tall, 0.2 ± 0.02 m DBH) was the standard species used in all paired host comparisons. The endemic species used were black walnut, Juglans nigra, L. , all of which were also among the most common wild trees recorded in this region [26] and are known hosts of H. halys [27]. In addition to representing a diversity of plant families, these trees differ in leaf structure (i.e., simple versus complex). The same tree of heaven, black walnut, black locust, and hackberry trees were used in 2017 and 2018, and black cherry was added in 2018.
Backfolding yellow sticky traps were deployed as described previously. At each site, a trap in female tree of heaven was paired with a trap in one of the aforementioned species (n = 5 per species pairing). Trees within pairs were 10 to 25 m apart and the mean distance between pairs was 3.4 ± 0.2 km. Traps were replaced at 7 ± 2-day intervals from 31 July until 29 August 2017 and 13 June until 20 September 2018.
In 2019, sampling was conducted at five windbreaks, separated by 5.6 ± 0.7 km. At each site, a single yellow sticky trap was deployed as described previously in one tree of heaven (8.5 ± 1.0 m tall, 0.1 ± 0.01 m DBH), one black walnut (10.1 ± 1.02 m tall, 0.4 ± 0.01 m DBH), and one black locust (7.1 ± 1.6 m tall, 0.1 ± 0.0.2 m DBH). Adjacent sampling trees were 23.7 ± 8.6 m apart and the distance between trees at the ends of the sampling area was 47.4 ± 15.9 m. Traps were replaced at 7 ± 1-day intervals from 17 June until 11 August.

Parasitoid Identification
For all studies, all parasitoids of interest captured (i.e., those considered to be potential H. halys parasitoids) were tentatively identified in situ in the laboratory, following Talamas et al. [28]. With the exception of Anastatus spp., all specimens were sent in situ on a small piece of the trap to E.J. Talamas for species confirmation. Male and female T. japonicus captured in the 2018 and 2019 habitat type study were differentiated based on antennal morphology [29].

Statistical Analysis
For each year, seasonal detections of T. japonicus are presented as total male and female T. japonicus from pooled captures across all habitat types by week. To compare T. japonicus captures among habitat types, data from each year were pooled across sample dates and analyzed using the Kruskal-Wallis test followed by the Bonferroni corrected Dunn's test (SAS Institute, Cary, NC, USA; SAS Institute Inc. 2018). For the 2017 and 2018 paired host study, captures of T. japonicus were compared by tree species pair using the Wilcoxon signed-rank test. In 2019, captures were compared among the three host species using the Kruskal-Wallis test followed by the Bonferroni corrected Dunn's test. All statistical comparisons used SAS 9.4 [30] and were considered significant at p < 0.05.

Seasonal Captures of Trissolcus japonicus
In 2018 and 2019, respectively, 101 (83.2% female) and 104 (95.2% female) T. japonicus were captured across all habitats sampled (Tables 1 and 2). In 2018, the first capture was recorded on 18 May (Figure 2A) and captures occurred on most weeks through 14 September, with peak captures on 13 July and 10 August. The last detection of T. japonicus was recorded on 14 September. Interestingly, males were captured only between mid-June and late August. In 2019, the date of first T. japonicus capture, 13 May ( Figure 2B), was similar to that in 2018. Again, T. japonicus was captured on most weeks between mid-May and early September, but males were recorded only on 11 August. Additionally, similar to 2018, peak captures in 2019 occurred on 22 July and between 5 and 11 August, and the last capture was recorded on 2 September.

Trissolcus japonicus Captures in Paired Host Trees at the Woods Edge
In 2017, 24 T. japonicus were captured during the four weeks of sampling in July and August (Table 3). All captures of T. japonicus were from tree of heaven, although captures did not differ significantly among the native tree species; tree of heaven vs black locust, S = −3, p > 0.05, tree of heaven vs black walnut, S = −5, p > 0.5, tree of heaven vs hackberry, S = −3, p > 0.05 ( Figure 4A).

Trissolcus japonicus Captures in Paired Host Trees at the Woods Edge
In 2017, 24 T. japonicus were captured during the four weeks of sampling in July and August (Table 3). All captures of T. japonicus were from tree of heaven, although captures did not differ significantly among the native tree species; tree of heaven vs black locust, S = −3, p > 0.05, tree of heaven vs black walnut, S = −5, p > 0.5, tree of heaven vs hackberry, S = −3, p > 0.05 ( Figure 4A).   (Table 4). Of all T. japonicus captures, 66.7% were from tree of heaven, and captures among the native tree species did not differ significantly; tree of heaven vs black locust, S = −1.5, p > 0.05, tree of heaven vs black walnut, S = −10, p > 0.05, tree of heaven vs hackberry S = −11, p > 0.05, tree of heaven vs black cherry, S = −1, p > 0.05 ( Figure  4B).

Trissolcus japonicus Captures Among Three Tree Species in Windbreaks
During the 8-week sampling period between June and August, 2019, only 13 T. japonicus were captured (Table 5), 61.5, 38.5, and 0.0% of which were from black locust, black walnut, and tree of heaven, respectively. Captures of T. japonicus differed significantly among the tree species in which traps were deployed (χ 2 = 6.32, df = 2, p < 0.05); significantly more were captured in black locust than in tree of heaven (z black locust = 8.4, z tree of heaven = 7.3, p < 0.05) (Figure 5), while captures in black walnut were not significantly different from those in black locust (z black locust = 8.4, z black walnut = 7.8, p > 0.05) or tree of heaven (z black walnut = 7.8, z tree of heaven = 7.3, p > 0.05).  (Table 4). Of all T. japonicus captures, 66.7% were from tree of heaven, and captures among the native tree species did not differ significantly; tree of heaven vs black locust, S = −1.5, p > 0.05, tree of heaven vs black walnut, S = −10, p > 0.05, tree of heaven vs hackberry S = −11, p > 0.05, tree of heaven vs black cherry, S = −1, p > 0.05 ( Figure 4B).

Trissolcus japonicus Captures Among Three Tree Species in Windbreaks
During the 8-week sampling period between June and August 2019, only 13 T. japonicus were captured (Table 5), 61.5, 38.5, and 0.0% of which were from black locust, black walnut, and tree of heaven, respectively. Captures of T. japonicus differed significantly among the tree species in which traps were deployed (χ 2 = 6.32, df = 2, p < 0.05); significantly more were captured in black locust than in tree of heaven (z black locust = 8.4, z tree of heaven = 7.3, p < 0.05) (Figure 5), while captures in black walnut were not significantly different from those in black locust (z black locust = 8.4, z black walnut = 7.8, p > 0.05) or tree of heaven (z black walnut = 7.8, z tree of heaven = 7.3, p > 0.05).

Discussion
Phenological synchrony of parasitoids with their hosts influences parasitoid population size and rate of colonization, with phenological mismatch potentially reducing observed parasitism of H. halys eggs by T. japonicus in Beijing, China [31]. Sampling in two consecutive years revealed first detections of T. japonicus in mid-May, peak captures in mid-July and early August, a marked decline in late August, and last detections in early

Discussion
Phenological synchrony of parasitoids with their hosts influences parasitoid population size and rate of colonization, with phenological mismatch potentially reducing observed parasitism of H. halys eggs by T. japonicus in Beijing, China [31]. Sampling in two consecutive years revealed first detections of T. japonicus in mid-May, peak captures in mid-July and early August, a marked decline in late August, and last detections in early to mid-September. The onset of captures in May coincided with the period of peak H. halys emergence from overwintering sites in the eastern USA [32] and peak captures followed predicted periods of peak H. halys oviposition [33]. As it does for other species [34], this synchrony should increase the likelihood of the persistence of adventive T. japonicus, supported by consistent indications of its range expansion in the USA in recent years [1]. Importantly, the seasonality of T. japonicus captures aligned well with the seasonal parasitism of H. halys eggs by T. japonicus in China [10], based on sentinel egg mass deployments at regular intervals from May or June through August or September. Moreover, declining detections in late August aligned with the cessation of H. halys oviposition by approximately mid-August [35], despite highest annual H. halys populations from late August through much of September [36,37]. While the overwintering biology of T. japonicus remains poorly understood, declining captures starting in late August may indicate that T. japonicus enters overwintering sites during this period. Similar total captures in 2018 and 2019 were notable given that 2019 was much drier than 2018 [38], suggesting that the abundance of T. japonicus remained stable despite annual climate variation, also boding well for its persistence. Our documentation of seasonal changes in captures of adventive T. japonicus in the Mid-Atlantic USA can inform the timing and efficiency of T. japonicus surveillance in H. halys host trees, particularly given the concurrence with results from China [10].
It is generally believed that decreased habitat size and increased fragmentation reduce parasitoid abundance [39] and parasitism [40]. Thus, highest captures of T. japonicus might have been expected in our largest and most contiguous habitat type, woods edge. However, in 2018, the fewest T. japonicus were captured at the edge of woodlots and most were captured in windbreaks, with no significant differences in captures among all habitats in 2019. These findings suggest site-specific variability in T. japonicus abundance. Positive density-dependent responses of parasitoids to hosts have been documented for other Trissolcus species [41,42], and T. japonicus abundance was likely influenced by H. halys density. While we did not monitor H. halys populations at the study sites, simultaneous sampling of H. halys and T. japonicus in future research may prove instructive.
Given the broad host range of H. halys [27] and the diversity of feeding and reproductive hosts available in this region [26], in theory, T. japonicus foraging for egg masses should not be limited by tree species. Rather, T. japonicus foraging may be most strongly associated with the presence or density of H. halys and its egg masses. Counts of H. halys egg masses on ornamental trees in an urban landscape by Formella et al. [43] revealed no significant differences among hosts in egg mass numbers, confirming H. halys oviposition on many plant species. However, their ground-based counts via visual observations may have underestimated egg mass density, based on data [20] showing greater numbers of H. halys egg masses producing T. japonicus from those collected at mid-canopy compared with other tree strata. In ornamental tree nurseries, more H. halys egg masses were found on angiosperms than gymnosperms, spanning numerous plant species and families [44]. Boyle et al. [45] suggested that T. japonicus may respond primarily to kairomones left by gravid H. halys walking on plant surfaces. If the distribution of H. halys populations are stochastic and show spatial and temporal changes in relative density, the distribution of T. japonicus might be expected to show the same trend, thereby resulting in changing parasitoid densities in a given area over time [46]. The curious differences in T. japonicus captures between native hosts and tree of heaven in 2017-2018 and 2019 may reflect this stochastic process.
Our studies using tree of heaven for standardized sampling have shown it to be a productive species for detecting T. japonicus. Across several studies in 2020, > 500 T. japonicus were captured in yellow sticky traps in female tree of heaven (Dyer, unpublished data). However, these data indicate that surveillance for T. japonicus need not be limited to specific H. halys host trees, thus enabling greater sampling flexibility. In other parts of the USA, sentinel H. halys eggs yielded T. japonicus detections from vine maple (Acer circinatum), Catalpa sp. [23], and English holly (Ilex aquifolium) [12]. Native parasitoids that attack H. halys eggs were captured, and in most studies, captures of T. japonicus were much higher than those of any other species. The community of native parasitoids observed was similar in composition to that reported by Tillman [47] in GA, USA, where T. japonicus has not yet been detected. Additionally, T. japonicus attacked more H. halys eggs than native stink bug eggs in field choice trials [23], suggesting that, as Konopka et al. [48] concluded, T. japonicus may be able to successfully coexist with native species in the biological control of H. halys. However, long-term effects of the addition of T. japonicus to the community of pentatomid parasitoids remain to be determined and warrant continued monitoring. Longitudinal studies using sticky traps alone or in concert with or other sampling methods, such as sentinel egg masses, may provide an indication of the impact of T. japonicus on the relative abundance of native H. halys parasitoids.

Conclusions
These studies further validate previous results [21] by showing that yellow sticky traps were effective for T. japonicus monitoring and surveillance. Detections of T. japonicus in several different habitats and BMSB host tree species indicated flexibility in the spatial component of T. japonicus sampling. Temporally, the consistently greatest captures from mid-June through early August can inform the timing of sampling and increase the likelihood of its detection. As discussed previously [21], yellow sticky traps are effective for addressing questions about the presence of T. japonicus, communities of native H. halys parasitoids, and the spread of T. japonicus in the invaded range of H. halys, but do not replace the use of sentinel or wild egg masses to assess H. halys egg parasitism or the impacts of parasitism on its populations.