Southern pine beetle (Dendroctonus frontalis Zimmermann) (Coleoptera: Curculionidae: Scolytinae) was detected on Long Island, New York, United States of America, in 2014, attacking and killing pitch pine, Pinus rigida Miller (Pinaceae). Since then, annual damage has occurred across the Central Pine Barrens and adjacent areas on Long Island. Dendroctonus frontalis impacts in these stands have been severe and compounded by mesophication, and they threaten the persistence of P. rigida in these forests (Nowacki and Abrams Reference Nowacki and Abrams2008; Heuss et al. Reference Heuss, D’Amato and Dodds2019). Unfortunately, D. frontalis has not been limited to Long Island and was found infesting trees elsewhere in the northeast United States of America – in Connecticut in 2015, and on Martha’s Vineyard and Nantucket, Massachusetts, during summer of 2023. In addition, adult D. frontalis have been collected in pheromone-baited detection traps in Maine, New Hampshire, Rhode Island, and upstate New York, where infestations have yet to be detected (Dodds et al. Reference Dodds, Aoki, Arango-Velez, Cancelliere, D’Amato, DiGirolomo and Rabaglia2018; Kanaskie et al. Reference Kanaskie, Schmeelk, Cancelliere and Garnas2023; Fig. 1).

Figure 1.

Map depicting Dendroctonus frontalis positive trap locations in New York and New England, United States of America, along with the distribution of Pinus strobus in North America.

Figure 2.

Size class distribution of Pinus strobus (≥ 7.5 cm dbh) attacked and unattacked by Dendroctonus frontalis in A, Brookhaven National Lab 1 (BNL1), B, Brookhaven National Lab 2 (BNL2), and C, Southaven County Park 1 (SH1) stands on Long Island, New York, United States of America.

The presence of D. frontalis on Long Island and in portions of New England represents a range expansion and is related to warming that is projected to continue over the coming decades (Intergovernmental Panel on Climate Change 2023). Warming temperatures are expected to further release climate constraints on D. frontalis, likely resulting in greater access to stands of P. rigida farther north, as well as to stands of jack, Pinus banksiana Lambert, and red pine, Pinus resinosa Aiton, (both Pinaceae) (Lesk et al. Reference Lesk, Coffel, D’Amato, Dodds and Horton2017). In much of the region, eastern white pine, Pinus strobus Linnaeus (Pinaceae), is the most common pine species on the landscape (Fig. 1). Pinus strobus is a host of D. frontalis in southern Appalachia (Hopkins Reference Hopkins1899; Hain et al. Reference Hain, Duehl, Gardner, Payne, Coulson and Klepzig2011); however, there is little data about stand- or tree-level interactions, particularly in northern regions of the United States of America.

The distributions of P. strobus and D. frontalis have historically overlapped in the mid-Atlantic and along portions of the Appalachian Mountains in the southeastern United States of America. There is evidence that D. frontalis can cause mortality in P. strobus stands in southern Appalachia and that infestations can be sustained in these stands (i.e., produce enough brood to maintain spot growth through a season; Hain et al. Reference Hain, Duehl, Gardner, Payne, Coulson and Klepzig2011). For example, P. strobus were attacked and killed in pine-dominated forests that also contained P. rigida and other hard pine species in North Carolina (Knebel and Wentworth Reference Knebel and Wentworth2007). In a separate study in North Carolina, infestations mapped from the air were used to assess D. frontalis host preference across multiple years and suggested that P. strobus is much less preferred compared to most common hard pines in the area (Anderson and Doggett Reference Anderson and Doggett1993). Pinus strobus mortality caused by D. frontalis has also been noted in portions of West Virginia (Hopkins Reference Hopkins1899). Attack and within-tree brood estimates for D. frontalis in P. strobus are more limited. One lab study suggested that P. strobus was a less suitable host than the more commonly used loblolly pine, Pinus taeda Linnaeus (Pinaceae), based on within-phloem life history estimates (Gardner Reference Gardner2011).

Although limited in distribution on Long Island, P. strobus has been intermittently infested by D. frontalis since 2015, providing opportunities to study interactions between the beetle and an uncommon host tree. Knowledge of D. frontalis behaviour in P. strobus is lacking but important for assessing risk to an important timber species. We evaluated D. frontalis impact at the stand level in forests dominated by P. strobus and compared attack density and within-tree estimates in P. strobus and P. rigida on Long Island.

We visited and sampled P. strobus stands on Long Island that were infested during summer and fall 2022. Two stands (each ∼1 ha) on Brookhaven National Lab lands (BNL1, BNL2) and one stand (∼1.5 ha) in Southaven County Park (SH1) were sampled using 11.3-m fixed-radius overstorey plots to record tree species, diameter at breast height, canopy class, and D. frontalis attacks. Sampling occurred in early February 2023. A minimum of three plots were established across each stand.

To compare attack density between P. strobus and P. rigida, we used BNL1 and an adjacent P. rigida stand. The P. rigida stand was located directly across a small road from BNL1, and both stands were surrounded by a larger forested and urbanised area where D. frontalis populations were high during the previous two years and were causing widespread tree mortality evidenced from ground and aerial surveys during that time. From the attacked trees in each area, trees with brood at the mid-larvae stages or later present at breast height were considered for sampling. Trees were selected as randomly as possible, accounting for logistics of safe tree felling in dense stands and adjacency to active roads and trails. Six P. strobus and five P. rigida were felled on 29–30 March 2023. Tree height, diameter at breast height, total infested bole length, height to base of live crown, and diameter at top and bottom of infestation were recorded. Each infested bole was then partitioned into four equal sections, with length relative to infested bole length. Three 81-cm² bark samples were taken from each section using a 10.2-cm-diameter hole saw and returned to the lab for processing.

Bark samples were dissected under a stereo microscope (Olympus SZ61, Tokyo, Japan). Number of attacks were counted on the outside of each bark sample. Attacks were differentiated from exit holes or ventilation holes by the angle, presence of resin and frass, and position at the base of a gallery (Stephen and Taha Reference Stephen and Taha1976). On the phloem side of bark samples, successful and unsuccessful brood galleries were differentiated, based on the presence or absence of resin and larval galleries, and then tabulated. Total successful brood gallery lengths were measured using a chartometer (Map Measurer Classic, Kasper & Richter GmbH & Co. KG, Uttenreuth, Germany). The presence of early- and mid-instar larvae in the phloem and of late-stage larvae or pupae in the outer bark were documented for each sample. Trees were considered successfully attacked if they had either living or dead mid- or late-instar larvae. Brood emergence densities were not used for analyses due to the following concerns: (1) asynchronous development times among trees (i.e., some trees had emergence holes in some portions and late-stage brood in other portions); (2) multiple brood emerging from one exit hole; and (3) potential destruction of brood during dissection.

T-tests were used to compare infested tree variables between P. rigida and P. strobus. For comparison of attacks, successful galleries, unsuccessful galleries, and gallery length (cm), two-way analysis of variance was used, with tree species and sample position as main effects. Individual bark samples at each height were pooled for analyses, and data were transformed when necessary to meet assumptions of normality and homoskedasticity. Tukey’s honestly significant difference tests were used to differentiate significant treatments.

Each P. strobus stand was nearly pure and overstocked, with basal area estimates between 48.5 and 53.1 m²/ha (Table 1). Only P. rigida and red maple, Acer rubrum Linnaeus (Sapindaceae), co-occurred with P. strobus in the stands. The average tree diameters were larger at the two Brookhaven National Lab sites than at the Southaven County Park site (Table 1). Dendroctonus frontalis attacks killed 36.7–54.7% of P. strobus trees (34.4–55% of basal area) across the three stands. Mortality occurred across all tree size classes available (Fig. 2A, B, C) and did not continue into a second year, even though hosts were still plentiful in each stand. This pattern is similar to the impacts in P. rigida and P. rigida–Quercus (Fagaceae) forests on Long Island but without the presence of other tree species in the canopy (Heuss et al. Reference Heuss, D’Amato and Dodds2019). High basal area in P. rigida stands has been linked to increased susceptibility to D. frontalis on Long Island (Jamison et al. Reference Jamison, D’Amato and Dodds2022), and it is likely also important in P. strobus stands. Although brood stage was not sampled for every tree during overstorey sampling in the present study, a subset of trees was assessed in each stand. Many trees had unsuccessful attacks at breast height, and when brood was present, development stages were similar.

Table 1.

Stand descriptors, including mean diameter at breast height (dbh), stand basal area (BA), Pinus strobus basal area, percent of pine stems killed by Dendroctonus frontalis, and percent of P. strobus basal area infested in each stand calculated from overstorey plots

Trees selected for sampling were estimated to have been attacked by D. frontalis during late summer and early fall 2022, based on phloem condition, presence or absence of brood, and crown condition. Pinus strobus selected for bark sampling were significantly taller, had longer infested bole lengths, and had more infested surface area than P. rigida did (Table 2). Percent surface area infested was equal between the two species. All sampled P. rigida were successfully attacked (i.e., produced brood), whereas only 50% of the P. strobus were successfully attacked (i.e., had exit holes or late-stage brood in bark). Unsuccessfully attacked P. strobus had early- or mid-stage larvae galleries present but no living larval life stages found in the phloem. In most cases, phloem tissue around larval mines was flooded with resin and crystalised resin.

Table 2.

Average diameter at breast height (dbh), average tree height, average infested bole length, average infested surface area, and percent of surface area infested by Dendroctonus frontalis in Pinus rigida and Pinus strobus sampled for brood estimates

* Significant difference, P < 0.05.

There was no interaction between the main factors for attacks (F _3,35 = 2.2, P = 0.1). Average number of attacks (per 81 cm²) were significantly higher in P. rigida (1.5 ± 0.1) than in P. strobus (0.6 ± 0.1; F _1,35 = 28.3, P < 0.0001), suggesting the former is a more attractive host than the latter. No differences in attacks from the four heights were found (Table 3), which contrasts with attack patterns on P. taeda (Coulson et al. Reference Coulson, Foltz, Hain, Martin and Mayyasi1976) and landing rates on shortleaf pines, Pinus echinata Miller (Pinaceae) (Coster et al. Reference Coster, Payne, Hart and Edson1977), in the southeastern United States of America. Understanding suitability for D. frontalis between the two hosts was difficult to assess without brood emergence estimates. However, the numbers of successful and unsuccessful galleries and brood gallery lengths provide some indication of suitability. There were no interactions between main factors for successful (F _3,35 = 0.8, P = 0.5) or unsuccessful (F _3,35 = 1.9, P = 0.2) galleries or brood gallery length (F _3,35 = 0.4, P = 0.7). The average number of successful galleries was significantly higher in P. rigida (4.6 ± 0.4) than in P. strobus (1.1 ± 0.3; F _1,35 = 57.1, P < 0.0001) and higher in the lower middle bole section compared to the top section (Table 3). There was no difference between average numbers of unsuccessful galleries for P. rigida (0.4 ± 0.1) and P. strobus (0.4 ± 0.1; F _1,35 = 0.05, P = 0.8), but the average numbers in bottom and upper middle bole sections differed significantly (Table 3). Given successful galleries were four times less abundant in P. strobus than in P. rigida, the latter appears to be a more suitable host, at least during initial attack and egg laying. Gallery length was also substantially longer in P. rigida (23.7 ± 1.7) than in P. strobus (7.2 ± 1.5; F _1,35 = 52.0, P < 0.0001). This result is similar to Gardner’s (Reference Gardner2011) work comparing gallery length in P. taeda and P. strobus. Longer gallery lengths were found in the lower middle bole sections than in the tops of trees in the present study (Table 3), whereas Fargo et al. (Reference Fargo, Coulson, Pulley, Pope and Kelley1978) reported that the longest gallery lengths in southeastern P. taeda were found at 3.5 m. This height would equate to bottom sections in the present study, where estimates were also higher across bottom to upper middle sections.

Table 3.

Average number of attacks, successful galleries, unsuccessful galleries, and gallery length from Pinus rigida and Pinus strobus bark samples (81 cm²). Different letters within a row indicate significant differences (P < 0.05)

Pinus rigida was attacked at a higher density, had more successful brood galleries, and had longer brood gallery lengths than P. strobus did, strongly suggesting the former is a better host than the latter. In addition, all attacked P. rigida produced living brood. The lack of brood production in several sampled P. strobus, coupled with observations from stand inventories that many P. strobus were unsuccessfully attacked, further provide evidence that P. strobus is not an optimal D. frontalis host. Variation in host selection and brood success is not uncommon and has been documented previously for D. frontalis for hard pines in the southeastern United States of America (Veysey et al. Reference Veysey, Ayres, Lombardero, Hofstetter and Klepzig2003), as well as in other conifer-infesting bark beetles (Amman Reference Amman1982; Švihra and Volney Reference Švihra and Volney1983; Siegert and McCullough Reference Siegert and McCullough2003).

Understanding D. frontalis behaviour in P. strobus is important for understanding the risk the beetle poses beyond hard pine and to the most abundant and widely distributed pine in the northeastern United States of America. Opportunities to evaluate D. frontalis in P. strobus have thus far been limited, but focus should be on collecting tree- and stand-level data in these forests whenever opportunities arise.