Seasonal polyphenism of spotted‐wing Drosophila is affected by variation in local abiotic conditions within its invaded range, likely influencing survival and regional population dynamics

Abstract Overwintering Drosophila often display adaptive phenotypic differences beneficial for survival at low temperatures. However, it is unclear which morphological traits are the best estimators of abiotic conditions, how those traits are correlated with functional outcomes in cold tolerance, and whether there are regional differences in trait expression. We used a combination of controlled laboratory assays, and collaborative field collections of invasive Drosophila suzukii in different areas of the United States, to study the factors affecting phenotype variability of this temperate fruit pest now found globally. Laboratory studies demonstrated that winter morph (WM) trait expression is continuous within the developmental temperature niche of this species (10–25°C) and that wing length and abdominal melanization are the best predictors of the larval abiotic environment. However, the duration and timing of cold exposure also produced significant variation in development time, morphology, and survival at cold temperatures. During a stress test assay conducted at −5°C, although cold tolerance was greater among WM flies, long‐term exposure to cold temperatures as adults significantly improved summer morph (SM) survival, indicating that these traits are not controlled by a single mechanism. Among wild D. suzukii populations, we found that regional variation in abiotic conditions differentially affects the expression of morphological traits, although further research is needed to determine whether these differences are genetic or environmental in origin and whether thermal susceptibility thresholds differ among populations within its invaded range.


| INTRODUC TI ON
Phenotypic plasticity allows organisms within a given genotype to respond adaptively to the challenges posed by environmental variability via beneficial shifts in morphology, physiology, or behavior (Agrawal, 2001;Thompson, 1993;West-Eberhard, 1989). The resulting changes are known to broadly affect patterns of dispersal, diet use, and reproduction and are well documented in a wide diversity of species, most notably arthropods (Fusco & Minelli, 2010;Heidinger, Hein, & Bonte, 2010;Whitman & Agrawal, 2009). Indeed, the propensity for phenotypic variation and plasticity in trait expression among arthropods is considered a key reason for their widespread success and diversity, even in extreme climates (Moczek, 2010;Nijhout, 1999;Pfennig et al., 2010;West-Eberhard, 1989). Seasonal polyphenism, the predictable shift in phenotype expression associated with temporal changes in the environment, is common among overwintering species which require the ability to shift from a foraging/reproductive phase, to one of survival and metabolic dormancy (Hodkinson, Bird, Miles, Bale, & Lennon, 1999;Shapiro, 1976;Sinclair, 1999). The biochemical mechanisms associated with seasonal trait expression are often induced by specific abiotic thresholds (e.g., temperature, photoperiod, state of hydration) early in development and prepare the individual for thermal stress tolerance through changes in carbohydrate metabolism, dietary cryoprotectant sequestration, or the creation of ice-nucleation proteins (Baust, 1981;Ohtsu, Kimura, & Katagiri, 1998;Sinclair, 1999;Strachan, Tarnowski-Garner, Marshall, & Sinclair, 2011). In addition, the external morphology of cold tolerant arthropods often undergoes change as well (Bale, Hansen, & Baust, 1989;Kimura, Awasaki, Ohtsu, & Shimada, 1992;Storey & Storey, 1986). In cool environments, arthropod larvae generally take longer to complete development than those of the same species reared at warmer temperatures (Holloway, Marriot, & Crocker, 1997;Kimura, 1988;Nyamaukondiwa, Terblanche, Marshall, & Sinclair, 2011). Subsequently, those adults are larger and display darker cuticular melanization than those individuals reared at warmer temperatures, traits which are thought to help retain heat (Atkinson & Sibly, 1997;Kingsolver & Wiernasz, 1991;Shearer et al., 2016;Wallingford & Loeb, 2016). Insects displaying these differentially expressed traits are often referred to as winter morphs (WM) or winter-form insects (David et al., 1994;Oldfield, 1970;Pétavy, Moreteau, Gibert, & David, 2002), and are prevalent in cool temperate climates, where organisms have evolved strategies to cope with harsh winter conditions (Danks, 2004;Shapiro, 1976;Strathdee & Bale, 1998;Tauber & Tauber, 1981). Indeed, in addition to predictable, cyclic changes in phenotype expression, there is growing evidence of genetic changes on a population level among some species due to changing climate (Hoffmann & Sgró, 2011;Somero, 2010). This is particularly significant in the case of invasive species because thermal limits and the capacity to adapt to novel environments directly affects the potential geographic distribution and thus, the risk of economic damage associated with an expanding host range (Paini et al., 2016;Terblanche, Deere, Clusella-Trullas, Janion, & Chown, 2007).
Phenotypic variation is well documented among Drosophila, and when reared at cooler temperatures, genetic selection for adults with larger body size occurs within a few generations (Ayrinhac et al., 2004;Hoffmann & Hercus, 2000;Hoffmann, Sorensen, & Loeschchke, 2003;Neat, Fowler, French, & Partridge, 1995;Rako & Hoffmann, 2006). This suggests that the ability to survive novel climates may be heritable within Drosophila populations over time (Hoffmann et al., 2003). Genetic analysis of Drosophila melanogaster has shown that loci associated with wing shape and size are directly affected by thermal selection and that wing morphology has adaptive significance in relation to temperature (Cavicchi, Giorgi, Natali, & Guerra, 1991). This is likely because large wings are more effective at heat absorption, making them advantageous during cool conditions when heat acquisition and retention are critical (Douglas, 1981;Heinrich, 1974;Kingsolver & Koehl, 1985). While this is fundamentally a byproduct of slowed development on an individual level, there may also be population-level effects selecting for improved survival under cool conditions (Gotthard, Nylin, & Nylin, 1995;Hoffmann & Hercus, 2000;Hoffmann et al., 2003;. In this case, a species would be said to have acquired some measure of genetic adaptation in response to selection events, rather than merely an adaptive, plastic response to acute environmental conditions (Gotthard & Nylin, 1995).
There appears to be precedent for both events broadly among Drosophila. Some species such as Drosophila bizonata and Drosophila daruma display distinct strain variations in thermal tolerance despite when reared under similar conditions in the laboratory (Kimura, 2004). Among these species, restricted gene flow between allopatric populations in cool and warm climates has been suggested as a likely mechanism driving these genetic changes (Kimura, 2004).
However, it remains unclear how abiotic conditions affect ontogenetic development, which is particularly important among species displaying multiple seasonal body forms (de Aranzamendi, Martínez, & Sahade, 2010). Studies in Oregon and Michigan have reported WM trait expression using the L4 longitudinal wing vein and found that wing size increases with decreasing temperature both in wildtype and colony populations (Leach, Stone, et al., 2019;Shearer et al., 2016). Using a 0-5 rating scale (5 = darker), differences in seasonal abdominal melanization have also been reported with sex-specific differences on 4th abdominal segment among WM females and on the 3rd segment in WM males (Shearer et al., 2016).
Most recently, regression tree analysis has been used to estimate WM cutoff values for wild D. suzukii collected throughout the year during 2017-2018 in Minnesota (Tran et al., 2020). The authors of that study reported specific WM threshold values for wing length (greater than 2.69 mm) and wing: hind-tibia ratio (greater than 2.17) in female specimens, although the measurements used were not consistent with the morphometric criteria used by other groups, nor did they include a metric of abdominal melanization, making comparisons with previous studies difficult.
Despite these advances, more information is needed to determine the influence of temperature on morphotype expression on a more continuous scale, and it remains unclear how these changes in morphology relate to winter stress tolerance. Given the overlap in morphotype expression observed by Leach, Stone, et al. (2019), the time of year in which D. suzukii develops may influence not only morphology, but also the relative degree of cold tolerance.
Among other cool-temperate Drosophila found in the native range of D. suzukii, the timing of seasonal development and the duration of exposure to cool temperatures is directly linked to triglyceride accumulation and overwintering survival (Ohtsu, Kimura, & Hori, 1995).
Furthermore, previous research on the mechanisms underlying thermal acclimation in D. suzukii indicates that regulation of external morphology and internal physiology may not be directly linked.
Indeed, our previous research has showed that even SM flies can develop cold tolerance if exposed to cool temperatures, as additional functional traits develop during the adult life stage (Stockton et al., 2018). For this reason, it is important that we understand how morphological trait expression and cold tolerance compare among D. suzukii whose exposure to cool temperatures begin early or late in larval development. Lastly, it is unclear whether morphotype expression is variable among regional populations, such as those collected from the northeastern versus southeastern United States.
Currently, most field-based research using wild specimens collected in the United States has focused on local populations in a single state or region (Guédot et al., 2018;Leach, Stone, et al., 2019;Shearer et al., 2016;Tran et al., 2020). While genetic analysis has found little difference in D. suzukii populations occupying climatically different regions of the United States (e.g., New York and North Carolina), well-defined genetic clusters in the Eastern and Western sides of the country indicate limited movement following establishment (Fraimout et al., 2017). Distinct populations also exist between North America and Europe, likely due to similarly isolated invasion events and little secondary trans-Atlantic movement (Rota-Stabelli et al., 2020). Furthermore, at least some significant phenotype differences between populations have been identified, including changes in maternal fecundity, susceptibility to parasitoids, and Wolbachia frequencies (Rota-Stabelli et al., 2020). These and other nongenetic differences require investigation using behavioral and physiological bioassays and are not likely to be identified by genetic analysis alone.
In this study, we aimed to determine the relationship between adaptive plasticity and thermal tolerance in D. suzukii. First, we developed a method for creating and characterizing adult WM flies using controlled bioassays to obtain morphometric measurements of wing, thorax, tibia size, and abdominal melanization to determine which traits were most strongly associated with changes in the larval abiotic environment. Next, we observed how the duration and timing of cold exposure during development affected both WM trait expression and thermal susceptibility. This was important because while internal and external traits associated with cold tolerance often develop concomitantly, expression may vary depending on the life stage at which cold exposure occurs (Stockton et al., 2018).  Hoffmann et al., 2003). By focusing on the factors affecting morphotype and cold tolerance variation among D. suzukii in both laboratory and field-collected samples, we aim to better understand the relative thermal limits of survival in different regions of the invaded range. If significant regional differences in pest phenotype and cold tolerance are detected, a more population-centered approach to future research and management of D. suzukii may be warranted (Reichard et al., 2015;Rota-Stabelli et al., 2020). anti-fungal additive (see Stockton, Brown, et al., 2019;Stockton, Wallingford, et al., 2019). Approximately 100 mixed-sex flies were housed in each bottle, which was replaced once weekly until all adults died or were used in the study. Newly eclosed offspring flies were moved to new bottles to separate the flies by age. The SM colony environmental conditions were set at 25°C with a 16L:8D (light/ dark photoperiodic cycle) at 55% relative humidity (RH). Unless otherwise stated, WM induction began 24 hr after oviposition by moving bottles of eggs (collected from the SM colony) to a 15°C growth chamber with a 12L:12D light cycle. After eclosion, we maintained the WM flies at 15°C until they were used for experiments.

| Defining winter morph traits
We assessed changes in wing length, thorax length, tibia length, and abdominal color score in SM and WM flies from NY (N = 32) and NC (N = 40). Equal numbers of males and females were measured from each state. Five days posteclosion, flies from either location were euthanized in 95% ethanol and stored at −4°C until dissection.
Flies from NC were shipped to NY for evaluation. Morphometric assessments were conducted using a stereo microscope set at 10× The left wing was dissected to obtain accurate wing length measurements and make the other body features more accessible. Two wing measurements were taken to compare how well each predicted WM body forms ( Figure 1b). The first wing length measure was taken along the L3 longitudinal wing vein from the proximal end at the base of the thorax to the distal end of L3 at the wing apex (Gidaszewski, Baylac, & Klingenberg, 2009;Wallingford & Loeb, 2016). A second wing vein measurement was taken from the proximal end of the L4 F I G U R E 1 Morphometric characters assessed included Drosophila suzukii wing, thorax, and tibia length (a). The length of the L3 and L4 longitudinal wing veins was measured along the dissected left wing of each fly (b). Abdominal color score (1-10) was based on the percent melanization of the anterior dorsal abdominal tergites (c) longitudinal wing vein to the posterior crossvein and continued to the distal end of L4 at the wing apex (Leach, Stone, et al., 2019;Shearer et al., 2016). Thorax length was the distance between the anterior margin of the thorax (propleuron) and the posterior tip of the scutellum.
Tibia length was measured as the distance between the distal end of the femur and the proximal end for the tarsus on the left foreleg. Color score assignments were made at the same time that body measurements were taken for each individual fly sample. A color score of 1-10, which indicated the percent melanization that was observed along the dorsal abdominal surface, was assigned to each of the five abdominal segments separately (Figure 1c; Shearer et al., 2016).
We also examined the effect of rearing temperature on color score assignments. All flies from this experiment were sourced from the colony in Geneva, NY. Five days after oviposition, 2-3rd instar D. suzukii larvae were moved to one of four climate-con-

| Winter morph development and survival
Approximately 100  To determine how exposure to cold temperatures at each life stage affects D. suzukii cold tolerance, we conducted additional laboratory-based thermal stress test assays using the remaining flies not used for morphometric assessments. We evaluated survival in 10 treatments. In treatments, 1-6 flies were the same as those in the previous experiment (egg, 1st instar, 2nd instar, 3rd instar, pupa, and no chill). Four additional treatments allowed us to compare survival outcomes among cold tolerance larval-exposed flies, with flies only exposed to cool temperatures as adults. Flies in treatments 7-8 were only subjected to 15°C early in adult maturation for the first 72 hr after eclosion (labeled "early adult"), or for 72 hr beginning when the adult flies were aged 1 week (labeled "late adult"). Treatments 9-10 were flies held at 15°C for 3 weeks after eclosion (labeled "Aged SM") and WM flies (labeled "Aged WM"), respectively. This allowed us to measure the effect of long-term cold exposure on cold tolerance, controlling for larval development conditions.
After each treatment was complete, we measured thermal susceptibility as the number of surviving flies after 72 hr at −5°C in a growth chamber (10L:14D; 25% RH; Kimura, 1988;Stockton et al., 2018

| Regional variation in winter morph expression
In order to determine temporal and spatial variation in D. suzukii mor-  each segment was referred to as a "color score." Mean melanization across all segments was referred to as a "color rating." Interactions between all three factors were included in the models and we did not use statistical blocking. We used Q-Q plots of the residuals to determine if the assumptions of the models were met. Pillai's trace was used to estimate the effect size of each factor in both MANOVA models (Scheiner, 1993).
We used MANOVA to determine the effect of sex and rearing temperature on abdominal melanization, as described previously.
For post hoc analysis, we ran a separate generalized linear model (GLM) for each abdominal segment and determined the effect of TA B L E 1 Sampling site locations, dates of collection, Drosophila suzukii sample sizes, and mean monthly temperature data from each collection site sex and temperature on color score. Pairwise differences between estimated marginal means were calculated using the R package "emmeans."

| Winter morph development and survival
To assess the effect of chill duration on D. suzukii development time, we used a linear mixed model from the package "lme4" and the function lmer. Chill duration treatment was the fixed effect and replicate number (1-4) was the random effect due to variation in eclosion frequency. We used type 3 analysis of variance with Satterthwaite's method to determine goodness of fit using the package "car." We then used the package "emmeans" for post hoc multiple mean com-  Pillai's trace values are reported.
We used linear regression to compare the relationship between temperature and various WM body traits from flies captured at each sampling site (Table 1). Some variables were combination factors of various trait sizes and color, yielding a new value that incorporated data from both traits. Because the relationship was stronger between L4 length and temperature, rather than overall wing size, L4 length was used for interpreting results. A new appearance factor, referred to as the "Appearance Score," was generated by multiplying L4 length by abdominal color score for each sample.
There was a significant interaction between sex and temperature, indicating that with decreasing temperature, abdominal darkening increased more in females than males (Pillai = 0.73, F (1, 76) = 39.82, p < .001). Furthermore, in females, there was a difference at each abdominal segment, while in males there were differences in abdominal segments 1-3, but not 4 and 5 as those segments were uniformly scored 10, or completely pigmented (Figure 3, Table S2). In males, the greatest differences between SM and WM flies occurred on the third abdominal segment (Mean difference = 5.11), while in females the greatest difference in color score occurred on the fourth abdominal segment (Mean difference = 4.55).

| Winter morph development and survival
Larval development time increased as chill duration increased (F = 2,707.1, df = 5, 1,803.5, p < .001; Figure 4, Table S3). Development times increased by approximately 15 days in D. suzukii chilled during the egg stage (Mean = 28.23 ± 0.19 days), compared to flies never chilled (Mean = 13.07 ± 0.07 days). Pairwise comparisons showed significant differences in development time among each group except for "eggs" and "1st instar larvae," although this was expected given the mean difference in development time between these two treatments was less than 1 day (Table 2). Although wing, thorax, and tibia length was similar among all the larval chill treatments ("egg", "1st instar", "2nd instar," "3rd instar"), flies that began the chill period during the pupal stage were generally smaller ( Figure 5). The exception to this pattern occurred in tibia length, for which "3rd instar" tibia length was shorter than among those in the "pupa" group, although this difference was not significant (Figure 5d). While we observed differences in abdominal melanization among treatments (χ 2 = 193.30, df = 6, p < .001; The remaining flies were used to assess cold tolerance among flies reared at 15°C for varying durations. In total, we monitored the survival of 1,002 individual flies during the 72 hr exposure period in our thermal stress test ( Figure 6). Our analysis showed that there was a significant treatment effect of developmental chill duration on adult survival outcomes (χ 2 = 1,063.1, df = 9, p < .001; short-term cold exposure that began during posteclosion was not associated with improved cold tolerance ("Early adult" and "Late adult"), long-term exposure on otherwise SM flies ("SM aged") at 15°C did result in improved survival, although this effect was only significant at 24 hr (Table S4).
While overall temperatures decreased in NY, MI, ME, and WI each month from August-December, samples in some of these locations were larger and darker in color despite similar temperatures ( Figure 7a,b, Tables S6 and S7). We observed that D. suzukii collected in ME were on average, larger than NY flies (Table S6), while in WI they were darker (Figure 7b; Table S7). We also found that the patterns of trait expression in FL did not follow the trends seen in the other states in which we sampled. Drosophila suzukii captured in FL were generally much smaller (Table S6) and lighter in color (Table S7) than samples collected from more northern locations.
In order to determine the strength of the relationship between each morphological trait and temperature, we compared the effect size of each trait using multiple regression models (Table 4). Among the individual traits recorded, color displayed the strongest relationship with mean weekly temperature (R 2 = 0.59). Each of the body size features was also significantly correlated with temperature, but the effect sizes (R 2 values) were smaller compared to color score (Table 4). Both wing measures displayed stronger correlations with temperature than tibia (R 2 = 0.32) or thorax size (R 2 = 0.13), likely indicating that these latter traits do not show large thermal variation. However, the effect size for the L4 wing vein (R 2 = 0.26) was larger than L3 (R 2 = 0.21), suggesting L4 would be a better predictor of abiotic conditions. A combined factor that is the product of L4 length and abdominal color score had the strongest relationship with temperature (R 2 = 0.64), indicating that this is the strongest abiotic predictor and best measure for characterizing flies as winter morph.

| D ISCUSS I ON
Among highly successful invasive species, the capacity to undergo adaptive changes in response to novel environmental stressors is considered one of the most significant indicators of their potential for ecological establishment (Agrawal, 2001;Chown, Slabber, McGeoch, Janion, & Leinaas, 2007;Davidson, Jennions, & Nicotra, 2011). However, to understand how an organism responds to environmental stress, it is critical to first determine which traits display the greatest phenotypic plasticity, and in doing so, define the criteria used for assessing morphotype variability. In the present

F I G U R E 3 Mean (±SEM) color scores of female (a) and male
Drosophila suzukii (b) on each abdominal segment (1-5). The development temperature is indicated by bar color: red = 10°C, blue = 15°C, yellow = 20°C, gray = 25°C. Significant differences among color score at the level of individual abdominal segments are indicated by different letters F I G U R E 4 The effect of differential chill schedule on Drosophila suzukii development time and total eclosion among flies in the no chill control group (gray dashed), and those chilled beginning as pupae (dark blue), 3rd instar larvae (red), 2nd instar larvae (yellow), 1st instar larvae (light blue), or beginning as eggs 24 hr after oviposition (gray solid) study, we first attempted to identify the external traits that showed the greatest degree of variation between D. suzukii SM and WM morphotypes, yet displayed the smallest variation among individuals. Our data show that the reaction norm for larval development, and therefore morphotype trait expression, occurs along a continuum from 25°C to about 10°C, consistent with the niche temperature range observed in most Drosophila (Hoffmann et al., 2003). We observed, as have others, that as larval development temperature decreased, body size and abdominal melanization increased in a predictable manner (Leach, Stone, et al., 2019;Shearer et al., 2016;Wallingford & Loeb, 2016). Among flies reared in the laboratory under controlled environmental conditions, L3 wing vein length measuring greater than 3 mm, or L4 length greater than 2.5 mm was consistently associated with WM flies. While both wing vein measures were highly correlated, indicating that either would be appropriate to use, the strength of the relationship between wing length and temperature was stronger for the L4 measure, consistent with previously reported results (Leach, Stone, et al., 2019;Shearer et al., 2016). In addition to wing length, we also measured thorax and tibia length. Our initial experiments showed that although thorax length was significantly larger among WM flies, the difference was relatively small and showed the least between-group variation between SM and WM flies. However, among the different chill duration treatments in experiment 2, these differences were more pronounced, and in our wild fly assessments thorax length was a highly significant factor. When we assessed tibia length, we found that although our initial experiments did not show large, consistent changes in tibia length due to rearing temperature, the multistate data revealed this as a strong predictor of seasonal change due to lower within-group variation compared to other measures.
Abdominal color score was also significantly affected by changes in development temperature, although the results were less consistent than those using body size alone. This was because the degree of variation among individuals was quite high, despite consistency in differential expression between morphotypes. Our initial data indicated that a color score of 4 or greater (~40% melanization) on the 3rd abdominal segment of female D. suzukii is a conservative threshold value for WM identification. On average, females displayed a color score greater than 5 when reared at 15°C and less than 2 when reared above 20°C ( Figure S3). Although all abdominal segments showed increased melanization with decreasing development temperature, segment 3 may be the most reliable for accurate WM assessment. This is for two reasons: First, we observed that segments 1 and 2 did not darken significantly above 10°C, indicating that these segments may not accurately reflect larval development temperature across the complete range of temperatures sufficient to induce WM traits. Second, among male flies segments 4-5 displayed elevated darkening above 20°C, consistent with what has been reported previously (Shearer et al., 2016). It is important to point out that the error associated with color score assignments was greater than for morphometric characters, indicating that this may be the most difficult WM trait to standardize. Indeed, color score is inherently more subjective than other measures and can be affected by lighting, the age of the samples, and observer perception, although we attempted to standardize each of these factors in our study and all samples were processed on the same equipment by the same observers, all in New York. Because we used a 10-point scale in this study, it is unclear whether this has a benefit over the 5 point scale used in previous studies (Leach, Stone, et al., 2019;Shearer et al., 2016). Indeed, a smaller scale may result in less observer error, thereby reducing individual variation. Furthermore, it is likely that variation in color score can be attributed to additional abiotic factors such as photoperiod (Leach, Stone, et al., 2019;Shearer et al., 2016).
While wing length does not appear to be significantly affected by short versus long day-length independent of temperature (Leach, Stone, et al., 2019), photoperiod may have some effects on melanization. Indeed when flies are kept at relatively warm temperatures (>20°C), melanization decreased among flies kept on a shortened "winter" photoperiod compared to those on a longer "summer" photoperiod (Shearer et al., 2016). This suggests that the processes contributing to melanization are more complex than temperature alone (Ramniwas, Kajla, Dev, & Parkash, 2013). For that reason, abdominal color score may be a less reliable measure of WM trait expression than wing length. It is also critical that beyond the external morphotype assignment given to species displaying seasonal polymorphic variation, we simultaneously understand the functional relevance of those external traits. The stress test assay revealed that while exposure to cool temperatures during larval development was critical to survival outcomes at temperatures below freezing, the timing and duration of that exposure was also a significant factor affecting morphotype expression and cold tolerance. Development time increased as exposure time increased, with egg and first instar treatments taking nearly 2 weeks longer to develop than SM flies. After eclosion, those WM adults were larger and darker in color than WM flies that began induction later in larval development, although this difference was not statistically significant. This indicates that the mechanism for external trait induction likely occurs quite late in larval development or even during pupation. Differences in trait expression and chill timing were more apparent beginning during the pupal stage, which was associated with smaller body size and decreased development time more similar to SM flies. However, the duration of cold exposure, even among those in the pupal treatment group, did not appear to affect cold tolerance. Rather, exposure to cool temperatures as juveniles, anytime from egg to pupa, was associated with increased survival. Interestingly, flies that began the chill window during the third instar stage were more cold tolerant than all other groups, while insects in the egg and 1st instar groups displayed poorer survival, which may have been caused by the stress of such an extended larval development period at cool temperatures. Although the age of the fly itself did not affect SM survival (all flies died within 24 hr), a 3-week acclimation period did improve SM survival, although this was only significant during the first 24 hr. The same long-term acclimation period in WM flies also appeared to improve survival and reduce within-group variation. This is consistent with previous data suggesting the importance of adult acclimation in determining the cold tolerance of D. suzukii (Stockton, Brown, et al., 2019;Stockton et al., 2018;Stockton, Wallingford, et al., 2019;Wallingford & Loeb, 2016). These data indicate that internal physiological processes such as induced cold tolerance appear to be regulated by factors independent of, or at least in addition to, those regulating larval development and external morphology. For that reason, the functional significance of morphotype assignment may be more ambiguous that previously thought, at the very least, in response to short-term cold stress. Unfortunately, there is surprisingly little literature available regarding the mechanisms regulating morphological shifts relative to cold hardening and acclimation, the latter of which have more extensively been investigated (Teets & Denlinger, 2013).
Given the overlap of morphotypes observed in field populations (Guédot et al., 2018;Leach, Stone, et al., 2019), particularly during thermal transition periods in the fall, our data suggest that SM flies may be able to survive brief decreases in temperature below freezing. Future research should investigate whether SM flies continue to lay eggs after such events, as we would expect those offspring to be the primary overwintering population Rossi-Stacconi et al., 2016). U.S. are genetically distinct (Ayrinhac et al., 2004;Hoffmann et al., 2003), despite data indicating these populations likely began from a single introduction in this part of the country (Fraimout et al., 2017). Indeed, wild flies from NY were generally smaller than those from ME, and lighter in color than those from MI, despite adjustments that allowed us to compare within similar temperature ranges. However, in our warmest sampling location (FL), although mean temperatures were theoretically sufficient to induce WM trait expression, few flies met the criteria for such categorization, suggesting that temperature-independent environmental variation may be the likely cause of trait variation observed in our study (Chown, Jumbam, Sørensen, & Terblanche, 2009), which is more consistent with genetic data from allopatric populations of D. suzukii in its native Japan (Gotthard et al., 1995;Kimura, 2004). It is possible that in warmer climates, cool temperatures are often not stable enough below the threshold of 15°C to induce WM development characteristic of what we observe in the laboratory and in our Northern sites.
Additionally, other abiotic factors, such as differential resource availability (Stockton, Brown, et al., 2019;Stockton, Wallingford, et al., 2019) and/or longer day length, may contribute to variation in external WM trait expression, accounting for the variation we observed among our various northern sites (Hoffmann et al., 2003;Hori & Kimura, 1998;Kimura, 2004). Although more research is needed to definitively determine the differential cause of the observed variation (genetic vs. environmental in origin), our data currently suggest that adaptive plasticity, as determined by differences in the abiotic environment, is the most likely driver of regional variation in this species. Given these results, it may be beneficial to employ regionally specific criteria for morphotype assignment that accounts for variation in WM trait expression.
Future research should address whether additional abiotic factors such as thermal stability and photoperiod may affect WM expression and cold tolerance thresholds (Leach, Stone, et al., 2019;Shearer et al., 2016). This is of economic and ecological importance due to the widespread effects of D. suzukii invasion. Furthermore, in the era of climate change there is concern that among invasive ectothermic species, phenotypic plasticity that favors adaptive responses to thermal variation may be the most significant factor predicting range expansion and total ecological impact (Bale & Hayward, 2010;Chown et al., 2007;Valladares et al., 2014 Stockton, Brown, et al., 2019;Stockton, Wallingford, et al., 2019).

Since this species was first detected in the Northeastern United
States and Great Lakes Region, D. suzukii has never been captured between mid-January and May, when the population begins to reemerge (Bal et al., 2017;Guédot et al., 2018;. Although genetic analyses have just begun to address seasonal population stability in North American and Europe (Rota-Stabelli et al., 2020), and there may be evidence of genetically stable pockets within these regions (J. Chiu, personal communication), these data are difficult to interpret given the amount of human-directed movement of these pests, likely with interstate, and even international, fruit shipments. Ultimately, by continuing to study the morphological and genetic variation of various populations of D. suzukii, we may move toward a better, and even predictive, understanding of range expansion in this species, as well as other globally invasive arthropods.

ACK N OWLED G M ENTS
We thank our technical staff who assisted with this project: Stephen Hesler, Karen Wentworth, Rachael Brown, and Molly Cappiello.
Funding was provided by the United States Department of Agriculture, National Institute for Food and Agriculture SCRI award, #2015-51181-24252.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest or competing interests relevant to this study or the science presented.

O PE N R E S E A RCH BA D G E S
This article has earned an Open Data Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://doi.org/10.5061/ dryad.4j0zp c884.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated and analyzed during the current study are publicly available in the data repository Dryad (https://doi.