The freeze-avoiding mountain pine beetle (Dendroctonus ponderosae) survives prolonged exposure to stressful cold by mitigating ionoregulatory collapse Open Access

Most insects have limited capacity for thermoregulation, so their body temperature closely matches that of the environment (Heinrich, 2013; Lahondère, 2023). Their physiology is therefore intimately tied to external temperature, making performance and survival in extreme or variable thermal environments a key challenge (Harrison et al., 2012). This is particularly true for temperate, alpine and polar insect species, which often need to survive through extended periods of sub-zero temperatures during winter, a period that can represent a large proportion of their life span (Sinclair et al., 2003; Williams et al., 2015). For these reasons, low temperature survival and overwintering success are continuously highlighted as key factors in determining insect abundance and distribution (Bale, 1987; Addo-Bediako et al., 2000; Ramløv, 2000; Bale, 2002; Marshall et al., 2020). This is also true for the mountain pine beetle, Dendroctonus ponderosae, a pest native to western North America, which has spread beyond its native range and invaded regions east of the Rocky Mountains that experience more severe winter conditions than its native range (Safranyik et al., 2010; de la Giroday et al., 2012; Creeden et al., 2014). Thus, exploring the cold tolerance physiology of D. ponderosae is critical to understanding the processes that promote or limit its continued invasion into the eastern boreal forest.

Insects have acquired a wide range of adaptations that allow them to adjust to, thrive in, or at the very least survive low temperature environments. However, the majority of insects are characterized as chill-susceptible, meaning they succumb to relatively mild cold at temperatures above those that cause extracellular freezing (Overgaard and MacMillan, 2017). For these insects, survival depends on overcoming the direct, disruptive effects of low temperature on homeostasis, cellular integrity and molecular processes. Central to the current conceptual model of chill tolerance physiology is a substantial, cold-induced disruption of extracellular ion homeostasis, referred to as an ‘ionoregulatory collapse’ (MacMillan, 2019). Under benign conditions, the renal system (Malpighian tubules and hindgut) maintains hemolymph ionic and osmotic balance through a continuous cycle of secretion and reabsorption, which counteracts transepithelial and cellular leakage (Edney, 1977; Phillips, 1981; Dow, 2017). However, at low temperature, the activity of ion-motive transporters responsible for secretion and reabsorption are slowed below the level required for maintenance of homeostasis (MacMillan and Sinclair, 2011; Overgaard et al., 2021). Thus, prolonged exposure to stressful cold leads to a debilitating increase in hemolymph K⁺ concentration (hyperkalemia), which, in combination with cold, depolarizes excitable tissues, triggers an intracellular calcium overload, and leads to cell death and organismal injury (MacMillan et al., 2015c; Bayley et al., 2018). Subsequently, recovery from cold exposure depends on the ability to either restore hemolymph ion concentrations upon return to permissive temperatures or prevent ionoregulatory collapse altogether (MacMillan et al., 2014). However, within the group of chill-susceptible insects, there is substantial variation in the amount of cold that species or populations can handle, which has prompted the establishment of a vaguely defined separation of insects that are truly chill susceptible from those that are chill tolerant, where the latter group is able to either prevent or mitigate the loss of ion homeostasis via a wide range of ionoregulatory adaptations and therefore tolerate slightly more severe cold exposures (Sinclair, 1999; Overgaard and MacMillan, 2017; Overgaard et al., 2021). For example, the chill-tolerant fruit fly Drosophila montana is able to maintain ion balance and renal function under low temperature conditions where less tolerant congeners lose renal function and suffer cold- and hyperkalemia-induced cell depolarization and injury (MacMillan et al., 2015a; Andersen and Overgaard, 2019).

For other insects, cold survival is determined below or at the melting point of their extracellular fluids (Denlinger and Lee, 2010; Lee, 2012). These species are divided into two main strategies: freeze tolerant and freeze avoidant (Bale, 1987; Lee, 1991; Bale, 1996). As the name suggests, freeze-tolerant insects can survive, and even promote, freezing of the extracellular fluids and initiate freezing at relatively high sub-zero temperatures. This slows ice formation and helps limit ice to the extracellular space (Salt, 1961; Zachariassen, 1985). Thus, these insects have a relatively high supercooling point (SCP), the temperature at which freezing occurs spontaneously during a cooling challenge (Sinclair, 1999; Sinclair et al., 2015). Freeze-avoidant insects, in contrast, survive substantial cold by possessing the ability to supercool to extremely low temperatures (often lower than −30°C) (Sinclair et al., 2003). These species have adapted to survive through harsh winters without freezing by, for example, accumulation of cryoprotectants (i.e. compatible osmolites and anti-freeze proteins), removal of ice-nucleating agents and cryoprotective dehydration (Salt, 1961; Sømme, 1982; Graham et al., 1997; Holmstrup et al., 2002). Each of these processes effectively lowers the SCP of the insects but is rarely found alone, meaning that the impressive supercooling capacity of freeze-avoidant insects is achieved via a combination of colligative and non-colligative factors (Zachariassen, 1985; Block et al., 1990; Sinclair et al., 2003; Lee, 2012; Storey and Storey, 2012). For example, freeze-avoidant beetles accumulate low molecular weight polyols in preparation for overwintering and produce a wide range of anti-freeze proteins, which lower the limit for freezing by preventing the formation of nucleating ice crystals (Sømme, 1964; Storey and Storey, 1983; Gehrken, 1984; Graham et al., 2007).

Dendroctonus ponderosae typically overwinter as freeze-avoidant larvae and possess a remarkable ability to lower their SCP in this life stage (Régnière and Bentz, 2007; Rosenberger et al., 2017). This reduction in SCP is mainly achieved through accumulation of substantial amounts of glycerol in response to emerging winter conditions (Bentz and Mullins, 1999; Robert et al., 2016; Thompson et al., 2020). This ability to survive severe cold is thought to contribute to the recent eastward spread of D. ponderosae from its native range across the Rocky Mountains, and into Alberta and northeastern British Columbia (Canada) (Safranyik et al., 1975; de la Giroday et al., 2012; Creeden et al., 2014), where it has devastated substantial areas of coniferous forest and required over $500M to manage its eastern spread (Cullingham et al., 2011; Hodge et al., 2017). However, despite an impressive capacity for freeze avoidance, cold mortality remains a key factor in determining population abundance of D. ponderosae (Renault et al., 2002; Régnière and Bentz, 2007). Thus, D. ponderosae represents an interesting and valuable system to study the physiological mechanisms underlying pre-freeze mortality in freeze-avoidant insects, which are generally thought to survive unless frozen (Salt, 1961).

Little is known about the processes that limit cold survival of freeze-avoidant insects, independent of freezing, during prolonged cold exposures, although everything from dysregulation of metabolism to disrupted development has been suggested (Bale, 1987, 1991; Sømme, 1999). That being said, earlier studies have brought up the notion that ion balance regulation during cold exposure might be important for survival in freeze-avoidant insects (Koštál et al., 2004; Zachariassen et al., 2004). Thus, for freeze avoidance to be a viable cold tolerance strategy, its evolution must have included an ancestral or convergent ability to avoid ionoregulatory collapse or tolerate substantial changes in transmembrane ion concentrations. If this is the case, we expect that freeze-avoidant species can maintain ion balance during exposure to very low temperatures just above the lethal SCP. However, if ion homeostasis is disrupted in a freeze-avoidant insect, this could indicate an unappreciated role for ionoregulatory collapse in limiting cold survival, which could indicate that, much like in chill-susceptible insects, variation in ionoregulatory capacity dictates cold survival. Interestingly, phytophagous bark beetles, such as D. ponderosae, possess unique hemolymph ion compositions with very low concentrations of Na⁺ and elevated concentrations of K⁺, Mg²⁺ and Ca²⁺ (Duchǎteau et al., 1953; Jeuniaux, 1971). In other insects, low extracellular Na⁺ is associated with increased cold tolerance via mitigation of ionoregulatory collapse (MacMillan et al., 2015b; Lebenzon et al., 2020). Transmembrane concentrations of Na⁺ can play a role in freeze tolerance and avoidance; however, the evidence for this remains scarce (Zachariassen et al., 2004). Thus, exploring ion balance and its regulation in the freeze-avoidant D. ponderosae larvae might provide insight into evolutionarily conserved cold tolerance mechanisms.

In the present study, we investigated whether ion balance plays a role in limiting the cold tolerance of the freeze-avoidant D. ponderosae and asked the following main question: do D. ponderosae suffer from ionoregulatory collapse when exposed to prolonged periods of stressful cold at temperatures immediately above those that cause extracellular freezing? If this extremely cold-hardy insect suffers from ionoregulatory collapse in the cold, renal mechanisms of failure may be critical to fully understanding overwintering success of this and other freeze-avoidant species. If, in contrast, D. ponderosae do not suffer from ionoregulatory collapse, mechanisms that serve to defend against this disruption in the cold represent undescribed means by which freeze-avoidant insects acquire cold tolerance (Overgaard and MacMillan, 2017; MacMillan, 2019; Overgaard et al., 2021). Given that non-freezing cold mortality occurs during overwintering in this species (Régnière and Bentz, 2007), we hypothesized that this could be linked to a gradual loss of ion balance (i.e. hemolymph hyperkalemia). Here, we exposed D. ponderosae that were (1) in preparation for overwintering, (2) in the middle of overwintering and (3) emerging from overwintering to extended periods of cold stress (above their SCP). We obtained parallel measurements of cold tolerance phenotypes and hemolymph ion concentrations at select time points that yielded a dataset that demonstrates mountain pine beetles can partly, but not completely, mitigate ionoregulatory collapse in the cold during overwintering.

All specimens of Dendroctonus ponderosae (Hopkins 1902) used in this study were collected from a population from its expanded range in central Alberta. To collect experimental animals, adult D. ponderosae were lured to lodgepole pine trees (Pinus contorta var. latifolia) at a field site in the expanded range of D. ponderosae outside of Cynthia, Alberta, Canada (53.34 N, −115.48W; see Fig. 1). Beetles were attracted to trees at the field site by four funnel traps containing aggregation pheromones (trans-verbenol and exo-brevicomin) and host tree volatiles (myrcene and terpinolene) (product 3093, Synergy Semiochemical, Delta, Canada) and trees were mass attacked over a 5-week period during July and August 2022. Mass-attacked trees were left on the landscape for egg laying and brood development until harvest. In October 2022, seven trees were harvested and cut into 50 logs (∼40–­50 cm long and ∼30–­50 cm in diameter), and the ends were sealed with paraffin wax to prevent desiccation. Infested logs were transported to the Department of Biological Sciences at the University of Alberta for outdoor storage and overwintering. Temperature loggers (HOBO 2x External Temperature Data Logger, Onset, Bourne, MA, USA) were implanted underneath the bark of three separate logs to monitor the thermal environment of the larvae during the overwintering period. Under-bark recordings could only be started after trees were harvested and the logs positioned for storage overwinter, so the prior thermal history of the trees was approximated by obtaining and averaging temperature data from three weather stations located near the field site (https://climate.weather.gc.ca/index_e.html, see Fig. 1 and Table S1 for locations). At three points during the overwintering period, logs were transported (by air, overnight and in double-walled secure containers with export authorization from Alberta Agriculture, Forestry and Rural Economic Development, permit #UofA-03-2022) to the Insect Production and Quarantine Laboratory at the Great Lakes Forestry Centre (GLFC) in Sault Ste. Marie, Ontario (part of Natural Resources Canada, Canada), where experiments on live animals were carried out in a plant pest containment and quarantine facility (level 2A, see Canadian Food Inspection Agency, 2014). The time points were chosen such that they included: (1) larvae from mid-autumn at the very beginning of overwintering (25 October 2022), (2) overwintering larvae from mid-winter (24 January 2023) and (3) larvae towards the end of overwintering in spring (12 April 2023). Ten, 17 and 23 logs were transported to GLFC in October, January and April, respectively, and after their arrival they were stored in the dark at 6°C until extraction for experiments (within 9 days). Extraction was carried out at room temperature (21–22°C) and larvae (only 4th instar) were used in experiments within an hour of being removed from the log.

At each of the three seasonal sampling points, we started by measuring SCPs of a subset of extracted D. ponderosae larvae before sampling larvae for all other experiments. SCPs were measured following the approach outlined by Sinclair et al. (2015). First, type-K thermocouples were gently attached to the outside of each larva with a small amount of vacuum grease, after which individual larvae were put into 200 µl microcentrifuge tubes. These were then gently stopped with cotton and placed into a custom-built aluminum casing through which a cooling bath (Proline RP855; Lauda, Würzburg, Germany) circulated a 1:1 water:methanol solution. From here, larvae were kept at 6°C for 10 min, after which the temperature was lowered by 0.5°C min⁻¹ until the temperature reached −35°C to ensure that all animals froze. At the end of the experiment, temperature traces were analyzed for each animal and the temperature at which the freezing exotherm occurred was recorded as the SCP. In this and subsequent experiments, all frozen animals were confirmed dead. A total of 53 larvae were used for this experiment, with 19, 22 and 12 in autumn, winter and spring, respectively.

After establishing the supercooling capacity of the larvae, we chose exposure temperatures that we expected to be stressful, but also minimized the risk of freezing. We chose (1) −10°C in the autumn, (2) −10°C and −20°C in the winter and (3) −10°C in the spring, which represented 0.07%, 0.01%, 7.33% and 7.59% risk of reaching the SCP, respectively (based on standard deviations and assumed normally distributed datasets). This approach does not account for the stochastic nature of freezing in cold-exposed insects (Ditrich, 2018); however, frozen larvae were easy to detect and were discarded (0, 0 and 17 larvae froze in the autumn, winter and spring, respectively, representing 0, 0 and ∼6.9% of larvae in these experiments). The −10°C exposure in the winter-acclimatized animals was included to allow for direct comparisons across seasons. From here, larvae were exposed to their respective temperature treatments by first holding them at 6°C for 10 min, before lowering the temperature by 0.5°C min⁻¹ until −10°C or −20°C was reached and holding at that temperature for up to 168 h (7 days). During this cold exposure, larvae were removed from the cold at pre-determined time points at 0 (immediately upon reaching the exposure temperature), 4, 16, 24, 48, 96 and 168 h.

At the specified time points, larvae were removed from the cold treatments and returned to room temperature (21–22°C) to estimate chill coma recovery time and survival as described for other insect larvae (Sinclair et al., 2015); any frozen larvae were discarded. Briefly, we removed larvae rapidly from tubes and placed them on a piece of paper at room temperature to observe and time the onset of movement for each larva. During this time, larvae were encouraged to move by gently poking them with a fine brush. Larvae were omitted from the data analyses if they did not move within 90 min. After the 90 min observation period, larvae were moved back into their tubes, which were gently stopped with cotton, and placed in the dark for 24 h at room temperature. Subsequently, their survival was estimated after the 24 h period by again placing them on a piece of paper and scoring them based on the ability to move as follows: 2=spontaneous movements (alive, uninjured), 1=only moving when stimulated (alive, injured) and 0=no movements (dead). A total of 390 larvae were used in these experiments (see Dataset 1 for distribution between seasons and treatments).

In parallel with the experiment described above, larvae exposed to the same conditions were sampled for hemolymph after 0, 4, 16, 24, 48, 96 and 168 h after being held at either −10°C or −20°C. Additional hemolymph samples were collected immediately after beetle extraction from the logs at room temperature (21–22°C) and after 24 h at 6°C following extraction. To sample hemolymph, we followed the procedure described for larvae by Campbell et al. (2018). Briefly, after removal from the cold treatments, larvae were rapidly submerged in hydrated paraffin oil in a Petri dish. A 26G needle or a pair of micro scissors was used to puncture the dorsal cuticle of the prothorax or metathorax. The extruding hemolymph droplets from individual larvae were collected with a micropipette, transferred to a round-bottomed 96-well PCR plate, and stored under hydrated paraffin oil at −80°C. Hemolymph samples were shipped to the Department of Biology at Carleton University in Ottawa on dry ice and stored at −80°C until hemolymph ion concentrations were measured. A total of 413 hemolymph samples were collected in these experiments (see Dataset 1 for distribution between seasons and treatments).

All data analyses were performed in R software (v4.2.2, https://www.r-project.org/). Data normality was tested with Shapiro–Wilk tests, and homoscedasticity was tested for normally distributed datasets with F-tests. SCPs were compared between seasonal sampling points with a Kruskal–Wallis rank sum test followed by pairwise Mann–Whitney–Wilcoxon tests with a Benjamini–Hochberg correction. The effects of season, exposure time and exposure temperature on chill coma recovery times and survival were analyzed using linear models. Concentrations of Na⁺ and K⁺ were first compared with control groups (24 h at 22°C, 24 h at 6°C and 0 h at each exposure temperature) with one-way ANOVAs. These analyses found no differences, so the effects of season, exposure time and exposure temperature were analyzed with linear models while excluding the room temperature and 6°C controls for each ion. The relationships between chill coma recovery time, survival and hemolymph K⁺ concentration were all analyzed with linear regression, linear regression on log-transformed data (i.e. an exponential relationship), or non-linear regression to a sigmoidal model using the nls() function. The best fitting models were chosen based on AIC scores. For all analyses, the level of statistical significance was 0.05, and values reported represent means±s.e.m. unless stated otherwise. For sample sizes, see Tables S2–S4.

To simulate overwintering, tree logs containing D. ponderosae larvae were kept in an outdoor enclosure at the Department of Biological Sciences at University of Alberta. Here, they experienced a natural winter and a range of temperatures from ∼10°C in mid-autumn to a minimum temperature of −26.3°C in mid-late December and a gradual climb back to ∼20°C towards mid-April with several cold snaps in between (Fig. 2A). During this time, larval SCPs changed significantly (χ²=40.4, d.f.=2, P<0.001) from −15.7±0.4°C in the autumn, down to −26.7±1.0°C in mid-winter and back to −12.6±0.5°C in the spring (Fig. 2B).

At the same time, we estimated the ability of larvae to tolerate and survive prolonged exposure to extreme cold by measuring the chill coma recovery times and subsequent survival scores after each exposure (Fig. 3). The exposure temperatures were designed to be stressful while minimizing the risk of freezing; however, a few animals did freeze during the experiments performed in the spring, and these were removed from later analyses. We found that the recovery times increased with time exposed to cold stress (F_1,357=616.2, P<0.001), decreasing temperatures (estimated only in winter; F_1,357=4.7, P=0.031) and their interaction (F_1,357=7.6, P=0.006) (Fig. 3A–C). At the same time, we found a clear effect of season on recovery times (i.e. autumn versus winter versus spring; F_2,358=14.2, P<0.001), which also shared a significant interaction with exposure time (F_2,357=45.5, P<0.001). Specifically, after having just reached their exposure temperatures at the end of the ramp-down (i.e. the 0 h exposure), the recovery time was similar between all larvae exposed to −10°C (2.2±0.1, 1.9±0.3 and 2.4±0.1 min, for autumn-, winter- and spring-acclimatized larvae) with winter-acclimatized larvae exposed to −20°C taking slightly longer (4.8±0.8 min). From here, recovery times increased with exposure time, and more so at lower temperatures (measured in winter) and in the spring, such that after 168 h these values had grown to 26.7±3.5, 24.2±3.8 and 51.3±2.7 min for −10°C exposed autumn-, winter- and spring-acclimatized larvae, respectively, and to 44.1±5.9 for winter-acclimatized larvae exposed to −20°C. The proportion of larvae that did not start moving within the 90 min time frame was generally low but tended to be higher in the spring group compared with the autumn and winter groups (Table 1, data not analyzed).

After recovery from the cold exposure, larvae were left to recover at room temperature for 24 h, after which their survival scores were estimated based on their ability to move. We saw very little evidence of injury in the larvae left to recover at room temperature for 24 h (Fig. 3D–F). Nonetheless, we identified a significant effect of exposure time on the survival score, as longer exposures slightly reduced the survival score (F_1,382=6.6, P=0.011). In addition, there was a marginally significant trend for exposure to lower temperatures to further reduce survival (F_1,382=3.5, P=0.061). Seasonality did not affect survival (F_2,382=2.2, P=0.107), and there were no interactions between exposure time and other fixed factors (exposure time×temperature: F_1,382=2.6, P=0.107; exposure time×season: F_2,382=1.5, P=0.219).

To investigate whether the observed increase in chill coma recovery time and reduction in survival score could be the result of an ionoregulatory collapse, as seen in other insects, we took hemolymph samples from animals exposed to the same time and temperature treatments and measured the concentrations of K⁺ and Na⁺ with ion-sensitive electrodes (Fig. 4).

At each seasonal sampling point (autumn, winter and spring), we first compared the hemolymph ion concentration in samples from control animals that were kept at room temperature (21–22°C), at 6°C for 24 h, and immediately upon reaching the exposure temperature (either −10 or −20°C for 0 h). For Na⁺, however, we did not include the room temperature control. For hemolymph K⁺ concentration, the values never differed between control groups (autumn: F_2,30=0.8, P=0.464; winter: F_2,37=1.9, P=0.148; spring: F_2,43=0.5, P=0.583) and always stayed at ∼32 mmol l⁻¹ across seasons (mean±s.d.=32.3±5.0 mmol l⁻¹). The same was true for Na⁺ concentrations (autumn: F_1,19=0.4, P=0.545; winter: F_2,26=0.1, P=0.889; spring: F_1,29=0.1, P=0.709), where the concentration stayed constant at ∼6 mmol l⁻¹ (mean±s.d.=5.8±1.1 mmol l⁻¹).

After confirming that our control conditions and very short cold exposure resulted in no change to hemolymph K⁺ and Na⁺ concentrations, we tested whether these were altered by prolonged exposure to stressful, non-freezing cold at −10°C, and at −20°C in winter. Hemolymph K⁺ concentration (Fig. 4A–C) increased with exposure time (F_1,335=257.0, P<0.001), a lower exposure temperature (F_1,335=35.4, P<0.001) and their interaction (F_1,335=9.7, P=0.002), with no effect of seasonality (F_2,335=2.1, P=0.126) or interaction between seasonality and exposure time (F_2,335=0.4, P=0.646). Specifically, after 168 h at −10°C it increased from the control value of ∼32 mmol l⁻¹ to 56.5±1.5, 56.2±3.5 and 58.9±1.5 mmol l⁻¹ in the autumn, winter and spring, respectively, in a seemingly asymptotical manner, whereas it increased further to 69.9±3.2 mmol l⁻¹ in the winter group exposed to −20°C. Conversely, the hemolymph Na⁺ concentration (Fig. 4D–F) decreased as exposure time increased (F_1,219=63.3, P<0.001), with no effects of exposure temperature (F_1,219=2.3, P=0.130), season (F_2,219=1.1, P=0.320) or any interactions (time×temperature: F_1,219=0.1 and P=0.810, time×season: F_2,219=0.5, P=0.606). Specifically, over the time course of 168 h, it decreased from the control value of ∼6 mmol l⁻¹ and stabilized at 4.0±0.1, 4.3±0.2 and 3.8±0.1 mmol l⁻¹, in the autumn, winter and spring, respectively, and 3.6±0.2 in the winter group exposed to −20°C.

Because we collected measurements of cold tolerance and hemolymph ion concentrations at the same time points throughout seasons and exposures, we can investigate the putative relationship between these parameters based on group means to investigate whether this matches patterns observed for less cold tolerant, chill-susceptible insects (Fig. 5).

We found a strong link between chill coma recovery time and hemolymph hyperkalemia, which closely matched an exponential relationship in the autumn (F_1,5=67.8, P<0.001), winter (F_1,10=94.0, P<0.001) and spring (F_1,5=57.9, P<0.001), as well as throughout the overwintering season when combined into one dataset (F_1,24=227.6, P<0.001) (Fig. 5A–D). Interestingly, we also found that a lower exposure temperature (in mid-winter) resulted in improved recovery rates, but only at higher degrees of hyperkalemia (hyperkalemia×temperature interaction: F_1,10=5.2, P=0.045) (effect of temperature: F_1,24=2.6, P=0.135; Fig. 5B). When datasets were combined, the faster recovery rate after exposure to lower temperature became statistically significant (F_1,24=10.8, P=0.003), and the effect remained larger at more severe hyperkalemia (F_1,24=9.7, P=0.005) (Fig. 4D).

The link between hemolymph hyperkalemia during cold exposure and survival outcome after recovery was less clear (Fig. 5E–H). In the autumn and spring, there were no apparent relationships (autumn: F_1,5=0.2, P=0.655, Fig. 5E; spring: F_1,5=3.4, P=0.123, Fig. 5G); however, a significant relationship emerged in winter (F_1,12=6.9, P=0.022; Fig. 5F). When combining data across seasons, the overall relationship between hemolymph hyperkalemia and survival was supported (F_1,26=9.8, P=0.004; Fig. 5H). The sigmoidal regressions fit to the statistical significant relationships showed that the degree of hemolymph hyperkalemia expected to injure all animals was an increase of ∼170–180% above control (i.e. ∼86–90 mmol l⁻¹; Fig. 4F,H).

Similar significant relationships occurred between cold tolerance measures and hemolymph hyponatremia (Fig. S1). Specifically, we found an exponential relationship between chill coma recovery and hyponatremia (F_1,16=53.0, P<0.001), which was season-specific (effect of season: F_2,16=5.1, P=0.020), and a sigmoidal, dose-response-like relationship with the survival score (F_1,18=5.4, P=0.031) which indicated an onset of injury at a 72.0±19.2% decrease in hemolymph Na⁺ concentration.

In the present study, we examined how the cold tolerance of D. ponderosae larvae changed during overwintering and how potential non-freezing mortality might relate to ion balance disruption as seen in less tolerant insects. We show that overwintering larvae exposed to extremely low temperatures, just above those that cause them to spontaneously freeze, for prolonged periods experience physiological stress and sustain injury which manifested as increased recovery times and reduced survival. Stressed larvae experience a loss of ion balance, characterized by hyperkalemia and hyponatremia, which correlates strongly with the degree of cold-induced injury.

Previous studies demonstrated that overwintering larvae of D. ponderosae are freeze avoiding with the ability to substantially lower their SCP during winter (Régnière and Bentz, 2007; Rosenberger et al., 2017). Here, we reaffirm that overwintering larvae in the expanded range are able to lower their supercooling points, and also show that dysregulation of ion balance and non-freezing chilling injury occur during prolonged chilling at ecologically relevant temperatures in this species.

The SCPs we observed in mid-winter were well below the lowest observed temperature of the stored logs (see Fig. 2). This extensive SCP depression is a defining characteristic of freeze-avoidant species (Zachariassen, 1985), and the lowest SCP observed here is similar to that found in another beetle, the emerald ash borer (Agrilus planipennis; see Crosthwaite et al., 2011; Duell et al., 2022). During our experiments, we noted that fall-collected larvae had food in their guts, whereas those extracted in winter and spring had empty guts. This is a common strategy for freeze-avoidant insects, as this removes most internal ice nucleating agents. We also noted that some fall-collected larvae held for prolonged periods at –10°C evacuated their guts during the experiment.

Mechanisms that suppress the SCP, and thereby reduce the risk of freezing, have historically been considered as the primary means by which freeze-avoidant insects survive winter. The potential for non-freezing injury has received considerably less attention in freeze-avoidant species. For less cold-tolerant (chill-susceptible) insects, survival is determined by physiological effects of low temperature in the absence of freezing. In these species, low temperature tolerance is estimated by assessing the capacity to recover from and survive a non-freezing cold exposure (Sinclair et al., 2015; Overgaard and MacMillan, 2017). Our findings reveal that larvae of D. ponderosae exposed to prolonged periods of low temperatures just above their SCP were cold-stressed; recovery times gradually increased with exposure time and survival decreased slightly after prolonged exposure. These effects of cold exposure were exacerbated at lower temperatures and during seasons of reduced cold tolerance (i.e. spring, see Fig. 3). These trends are remarkably similar to those observed for chill-susceptible insects, where both cold acclimation and milder exposure temperatures improve recovery times and survival outcomes (Andersen et al., 2017a; Yerushalmi et al., 2018; Tarapacki et al., 2021). Similarly, recovery times in other chill-susceptible insects reach an apparent plateau as injury and mortality begin to set in (MacMillan et al., 2012). That being said, the progression of adverse effects in D. ponderosae is markedly slower than that for chill-susceptible insects, even though the larvae were exposed to temperatures that would rapidly kill less tolerant insects (Koštál et al., 2004, 2006; Findsen et al., 2013; Andersen et al., 2015). Interestingly, chill-tolerant Drosophila montana recover rapidly following prolonged cold stress without injury, which could suggest shared physiological mechanisms to survive low temperatures (MacMillan et al., 2015a; Andersen and Overgaard, 2019). However, despite D. ponderosae larvae being able to mitigate the cascade of events that prolongs recovery times, they eventually suffer injury and mortality, indicating that they are unable to completely counteract the negative effects of the cold stress.

In the ‘ionoregulatory collapse’ model of insect cold tolerance, increased recovery times and decreased survival are linked to a progressive, debilitating cold-induced hemolymph hyperkalemia (Overgaard and MacMillan, 2017; MacMillan, 2019; Overgaard et al., 2021). The hemolymph composition of phytophagous coleopterans has previously been described as ‘unconventional’; they often have high concentrations of K⁺ (>30 mmol l⁻¹) and Mg²⁺ (>30 mmol l⁻¹), and unusually low concentrations of Na⁺ (<15 mmol l⁻¹) (Duchǎteau et al., 1953; Jeuniaux, 1971). This is also the case in D. ponderosae. Hemolymph concentrations of K⁺ and Na⁺ were ∼32 and ∼6 mmol l⁻¹, respectively, for larvae kept under benign conditions (Mg²⁺ was not measured) (Fig. 4). As in chill-susceptible species, these values remained unchanged during the cooling process (MacMillan et al., 2014), but prolonged exposure caused hemolymph hyperkalemia and hyponatremia. Further, the degree of change in the concentration of these ions strongly correlated with the severity of the cold exposure and relative cold tolerance of the larvae (compare Figs 3 and 4). Cold-induced hemolymph hyperkalemia is a hallmark of the ionoregulatory collapse model derived mainly from the study of dipteran and orthopteran species that are chill susceptible (Overgaard and MacMillan, 2017; Overgaard et al., 2021). The same phenomenon has occurred in a chill-susceptible coleopteran, Alphitobius diaperinus (Koštál et al., 2007), and chill-susceptible lepidopterans with an ion homeostasis similar to that of D. ponderosae (Natochin and Parnova, 1987; Andersen et al., 2017b). Thus, despite being categorized as a freeze-avoidant insect during short cold exposures with acute endpoints (Régnière and Bentz, 2007; Rosenberger et al., 2017), D. ponderosae appears to be limited in its non-freezing cold tolerance during prolonged exposures via mechanisms that closely resemble those limiting chill-susceptible insects. Indeed, when the degree of hemolymph hyperkalemia is correlated with recovery times and survival outcomes at the different time points, the emerging relationships resemble those found for chill-susceptible insects (Fig. 5). Specifically, there is a strong exponential correlation between the increase in hemolymph [K⁺] and recovery times, whereas the relationship with survival appears sigmoidal in nature (yet we caution that the fitted sigmoidal relationships extrapolate beyond the data). This relationship between cold-induced hemolymph hyperkalemia and injury is remarkably similar to that observed for the fire bug (Pyrrhocoris apterus), which is characterized as chill susceptible despite a substantial capacity for supercooling, survival and mitigation of ionoregulatory collapse during month-long exposures to temperature just above their SCP (Koštál et al., 2004). Thus, the ability to mitigate, or delay, ionoregulatory collapse during prolonged exposure to severe cold is not unique to freeze-avoidant insects, indicating shared cold-tolerance mechanisms between freeze-avoidant and chill-tolerant insects.

In chill-susceptible insects, the exponential relationship between recovery times and hemolymph hyperkalemia has been suggested to stem from the need for water redistribution after the cold-induced osmotic disruption (MacMillan et al., 2015a,b). In the present study, we did not investigate water balance; however, the cold-induced osmotic disruption in fruit flies is caused by a steep hemolymph-to-gut Na⁺ gradient that is unlikely to exist in D. ponderosae larvae as control concentrations are ∼6 mmol l⁻¹. Nonetheless, hemolymph hyperkalemia does occur, and the approximate LD₅₀ found here (∼170–180% increase, see Fig. 5) closely matches that for chill-susceptible insects (∼200% increase; see Overgaard and MacMillan, 2017). Thus, even if different upstream homeostatic disruptions cause hyperkalemia in this species, the mechanism(s) by which increased hemolymph [K⁺] causes injury and cell death during a cold exposure might be shared.

Intriguingly, the level of hemolymph hyperkalemia we observed in D. ponderosae plateaus at a level just below those that start to cause extensive mortality in other chill-susceptible species (i.e. they stabilize below a ∼150% increase whereas a ∼200% increase leads to injury; see Overgaard and MacMillan, 2017). That being said, the absolute concentrations of hemolymph K⁺ experienced by D. ponderosae here (55–70 mmol l⁻¹) would be detrimental to other chill-susceptible insects. Regardless, this suggests that this freeze-avoidant species has previously undescribed mechanisms that help prevent extensive cell depolarization, cell death and organismal mortality in the cold. These mechanisms could already be established through seasonal plasticity before cold stress (i.e. seasonal plasticity or acclimatization), could be induced rapidly during cold stress (i.e. rapid hardening), or a combination of both.

In chill-tolerant Drosophila species or cold-acclimated D. melanogaster, concentration gradients for the prevailing cation in the hemolymph (Na⁺) are reduced, and other compatible osmolytes are more abundant and serve to reduce the driving force for Na⁺ leak (MacMillan et al., 2015b; Olsson et al., 2016). Similar osmolyte activity occurs in some freeze-avoidant beetles (Hanzal et al., 1992). In Drosophila, these changes occur in response to environmental change but before a stressful cold exposure. This modification of the ‘resting’ state is thought to limit the extent of ion balance disruption during the chilling event. The osmolytes that take the place of these extracellular ions include classical cryoprotectants such as trehalose and proline, which led to the hypothesis that their accumulation may protect against ionoregulatory collapse regardless of the risk of freezing (Olsson et al., 2016). This mechanism of protection is compelling, because it would provide an evolutionary stepping-stone toward the role of cryoprotectants in SCP depression and support the notion of chill susceptibility, chill tolerance and freeze avoidance as a continuum of thermal tolerance strategies. Indeed, chill-susceptible insects accumulate cryoprotectants in preparation for overwintering (Koštál and Šimek, 2000; Koštál et al., 2004) or when given mild cold treatments (Wang and Kang, 2005; Overgaard et al., 2007). This putative role for cryoprotectants relies on far lower concentrations of these solutes to provide protection against the cold, and prevention of ionoregulatory collapse could therefore represent a prerequisite for acquiring either freeze avoidance or freeze tolerance as a survival strategy. If cryoprotectants play this role in D. ponderosae, seasonal accumulation of glycerol and other cryoprotectants in the hemolymph may be critical to establishing a favorable equilibrium of water balance during chilling that passively protects against catastrophic hyperkalemia. By contrast, rapid production of cryoprotectants during chilling in this species (Sømme, 1964; Thompson et al., 2020) could actively prevent further progression of ionoregulatory collapse during cold stress and lead to the plateau in [K⁺] we observed here. Most likely, these means of altering osmotic gradients are combined with extensive changes to the renal system that allow for maintenance of ion homeostasis at extreme low temperatures (MacMillan et al., 2015a; Yerushalmi et al., 2018; Andersen and Overgaard, 2020). In less tolerant insects, renal plasticity plays a critical role in chill tolerance, and we expect that maintenance of ion balance in the cold is achieved by D. ponderosae through complementary means.

An alternative explanation for the observed plateau in hemolymph ion concentrations could be that these reflect a change in steady-state conditions for D. ponderosae rather than a physiological failure leading to reduced fitness (i.e. survival). This would switch the survival strategy from one of resisting cold-induced ionoregulatory dysfunction to one of tolerating cold-induced suppression of ionoregulatory mechanisms or similar. In fact, such a strategy has been suggested in terms of abiotic stress tolerance in insects and ectotherms and has been hypothesized to conserve energy during exposure to severely unfavorable conditions via mechanisms such as channel or spike arrest (Boutilier, 2001; Campbell et al., 2018; Robertson et al., 2020) or diapause (Hahn and Denlinger, 2011; Sinclair, 2015). Energy conservation during winter is critical for spring emergence and life history of D. ponderosae; however, whether they enter a true diapause remains a topic of debate (Lester and Irwin, 2012; Bentz and Hansen, 2018). More research is needed to examine which ‘strategy’ is utilized, whether these strategies come with potential trade-offs in terms of energy conservation, and what that means in terms of physiological mechanisms of injury and injury prevention.

In conclusion, this study unveils unique cold-tolerance mechanisms in overwintering D. ponderosae larvae. Specifically, we show that despite possessing an impressive capacity for freeze-avoidance, these cold-hardy larvae experience physiological stress and sustain injury during prolonged exposure to severe, stressful cold, which is linked to a gradual disruption of hemolymph ion balance (hyperkalemia) in a manner similar to that of less tolerant, chill-susceptible insects. The asymptotical nature of the ion balance disruption and the ability of D. ponderosae to stabilize hemolymph ion concentrations to just below levels causing extensive injury suggest previously unknown mechanisms of ion balance regulation and protection, which we speculate might be related to the production and accumulation of cryoprotectants. These mechanisms may stem from seasonal or rapid changes to their physiology, or a combination of both. Further research into renal and epithelial function is needed to elucidate the molecular and cellular mechanisms that promote this physiological capacity. Indeed, gaining a deeper understanding of the complex, integrative biology and physiology of overwintering in D. ponderosae is likely to affect future management efforts. Lastly, our findings challenge the current and historical division of cold-tolerance strategies and mechanisms, and we posit that chill susceptibility, chill tolerance and freeze avoidance represent a continuum of thermal tolerance strategies. Thus, drawing parallels between the physiological mechanisms that limit the cold tolerance of chill-susceptible insects and those that permit winter survival of more tolerant insects, such as D. ponderosae, might promote rapid advances in our understanding of cold-tolerance physiology.

We extend our utmost gratitude to Ashlyn Wardlaw and Dr Anna Turbelin for their help in running experiments in the quarantine facilities at the Great Lakes Forestry Centre in Sault Ste. Marie (where both are affiliated). We furthermore acknowledge Caroline Whitehouse, Glenn Dobransky and Andrea Sharpe for their help with providing permits and guidance in collecting the beetle-infested tree logs, as well as Leanne Petro for fieldwork associated with baiting of adult beetles to the field site and trees. Lastly, we are thankful for the assistance of Colleen Fortier, Taylor Volappi, Marion Mayerhofer and the High Country Arborists for coordinating and assisting with the tree harvest.