Invasion and high-elevation acclimation of the red imported fire ant, Solenopsis invicta, in the southern Blue Ridge Escarpment region of North America

The red imported fire ant (Solenopsis invicta) is a non-native invasive species that rapidly spread northward in the United States after its introduction from South America in the 1930s. Researchers predicted that the northward spread of this invasive ant would be limited by cold temperatures with increased latitude and greater elevation in the Blue Ridge Escarpment region of the United States. The presence of S. invicta at relatively high elevations north of their projected limits suggests greater cold tolerance than previously predicted; however, these populations might be ephemeral indications of strong dispersal abilities. In this study, we investigated potential physiological adaptations of S. invicta that would indicate acclimation to high elevation environments. We hypothesized that if S. invicta colonies can persist in colder climates than where they originated, we would find gradients in S. invicta worker cold tolerance along a montane elevational gradient. We also predicted that higher elevation S. invicta ants might incur greater physiological costs to persist in the colder climate, so we measured colony lipid content to assess health status. For comparison, we also collected physiological temperature tolerance data for the co-occurring dominant native woodland ant Aphaenogaster picea. We found that S. invicta occurring at higher elevations exhibited greater physiological tolerance for cold temperatures as compared to lower-elevation conspecifics–a cold tolerance pattern that paralleled of the native A. picea ants along the same gradient. Both S. invicta and A. picea similarly exhibited lower thermal tolerances for colder temperatures when moving up the elevational gradient, with A. picea consistently exhibiting a lower thermal tolerance overall. There was no change in S. invicta colony lipid content with elevation, suggesting that greater metabolic rates were not needed to sustain these ants at high elevations.


Introduction
Researchers have long believed that introduced Solenopsis invicta [1] populations in the southeastern United States would not expand to higher latitudes and elevations because their pre- adaptation to the subtropical climate of their native range would make them unable to survive prolonged cold weather exposure [2][3][4]. For example, Korzukhin et al. [2] predicted that winter temperatures would limit S. invicta alate production through freeze-kills and stunted reproductive output, and therefore the ant would be unlikely to colonize areas with minimum temperatures below -3.7˚C. As such, the southern Blue Ridge Escarpment region in western North Carolina, U.S. (the zone of abrupt change in elevation between the Blue Ridge and Piedmont physiographic provinces with a vertical relief of 400 m to ca. 760 m) was projected as unsuitable habitat for S. invicta due to the cold temperature extremes at high elevations [2]. Within the last five years at least, however, S. invicta colonies have been observed at elevations > 1220 m in the Blue Ridge Escarpment where temperatures reached anomalous minima of -16.3˚C in Macon County, North Carolina, in 2019 [5,6]. The objective of our study was to explore potential mechanisms that might explain the persistence of S. invicta colonies through the winter season and high elevation acclimation. We investigated S. invicta physiological thermal tolerance along an elevation gradient to determine whether the ants were able to physiologically acclimate to colder temperatures. For comparison, we also tested native Aphaenogaster picea [7] thermal tolerance along the same gradients. We hypothesized that S. invicta would have a lower critical thermal minimum (CT min ) at higher elevations and that the shift in thermal tolerance would be similar to that of the native ant, A. picea. We also predicted that S. invicta critical thermal maximum (CT max ) would be higher than A. picea ants because S. invicta originates from a subtropical climate and occurs in open habitats (in contrast to the forest-dwelling A. picea ants).
The lipid content of ant colonies gives an indication of colony health and can be useful in predicting the reproductive success of a colony [3,8,9]. We expected that S. invicta colonies at higher elevations (915 m and above) would be less healthy (and hence have a lower lipid content) due to the potential shorter optimal foraging time and because they may have to expend more metabolic energy to persist in colder temperatures.

Study species
Solenopsis invicta is one of several fire ant species native to South America. Specifically, S. invicta is indigenous to southern Brazil, Paraguay, and northeast Argentina, and has been introduced to many countries over the past two centuries, including the U.S. where it has rapidly spread since its first introduction to Mobile, Alabama, in the 1930s [10]. Habitat disturbance, a perturbation that is caused by either biotic or abiotic forces [11], such as the construction of buildings and roads or the removal of biomass by a natural force is crucial for the establishment of many invasive species including S. invicta. Globalization and the increase of trade has allowed this species to spread successfully to many disturbed environments via shipping and other modes of transport [12].
Aphaenogaster picea ants are native to deciduous forests in the eastern U.S. and are not only one of the most abundant ants in this region, but they are important ecologically as seed dispersers of many native understory herbs such as Sanguinaria, Trillium, and Hepatica spp. [13,14]. Aphaenogaster picea can coexist in the same general area as S. invicta, however, S. invicta primarily inhabits disturbed, full-sun environments whereas A. picea inhabits shady forests [3,14].

Collection sites
Solenopsis invicta workers were collected at three elevational ranges (hereafter, "sites"): low (0- We found S. invicta colonies abundant in disturbed areas along road rights-of-way, highway rest stops, and along the edges of agricultural fields. All the S. invicta colony that we observed and collected were established in anthropogenically disturbed habitats, and we did not find any colonies in undisturbed habitats (personal observation). For study sampling, we collected from active, mature colonies that were at least ten meters from other colonies. If we failed to find a colony after fifteen minutes of searching, we moved to another location. We collected 300 worker ants from fourteen S. invicta colonies at each of the three sites (n = 42 colonies) occurring along a piedmont-to-mountain elevational gradient. Additionally, fifteen workers from seven A. picea colonies were collected at each of the three sites (n = 21 colonies) as close to the S. invicta colonies as possible. All ants were collected in June and July 2017. The GPS location of each ant colony sampled was recorded with a GPS unit (Garmin GPSMAP 64s) and recorded for mapping purposes through ArcMap (ESRI, Inc., Redlands, CA). No permits were required for field collection; we obtained permission to collect ants from private residential areas and the remainder of the ants were collected from public spaces. Solenopsis invicta and A. picea ants from each colony were placed in plastic bags in a cooler and promptly brought back to the Highlands Biological Station laboratory (Macon Co., North Carolina) and stored in a refrigerator until thermal tolerance tests were performed, no more than 24 hours after field collection. Following the thermal tolerance assays, the rest of the collected S. invicta ants were freeze-killed in a -80˚C freezer to use in the lipid analysis tests.

Thermal tolerance
Lower and upper thermal tolerance tests were performed to determine physiological limits for S. invicta and A. picea. We tested 30 ants from each of the 42 S. invicta colonies and 15 ants from each of the 21 A. picea colonies. Solenopsis invicta are polymorphic, so we separated the larger "major" and smaller "minor" workers of S. invicta colonies for the assay. We tested equal numbers of S. invicta majors (n = 15) and minors (n = 15), whereas A. picea only has one worker size, so we tested 15 total workers from each colony. All workers were placed individually in 16 mm labeled glass test tubes with moistened filter paper to prevent desiccation and plugged with cotton to prevent escape. Ten ants from each S. invicta colony (5 majors and 5 minors) were kept in control test tubes at room temperature (~24˚C) for the duration of the thermal testing trials. Five minors and five majors from each S. invicta colony were used for the cold tolerance test. They were placed in the individual test tubes in a Thermo Fisher ARCTIC A40 refrigerated water bath (NesLab, ThermoScientific, Portsmouth, NH, USA) and acclimated for ten minutes at 20˚C. Temperatures were then decreased by 1˚C min -1 . One minute after the water bath reached the desired temperature, the ants were observed. If an ant was moving, it was placed back into the water bath where it remained for another decrease of 1˚C. CT min was measured as the temperature at which the ants displayed a loss of motor control and were unable to right themselves after being flipped on their back. The temperature at which ants could no longer right themselves was recorded and represented the critical thermal minimum limit. A different set of ants from each colony was then tested for their CT max limits following similar methodology as CT min except that temperatures were stepped up from 30˚C. These methods were repeated for the A. picea thermal tolerance assays with five ants for each of CT min , CT max and control.

Lipid analysis
Following field collection, we freeze-killed S. invicta by storing them in a -80˚C freezer for at least three days. Two hundred ants from each colony were haphazardly selected from the samples and dried at 60˚C for 48 hours. The ants were weighed and their dry mass recorded. Lipids were removed using Soxhlet extraction following the protocol of Smith and Tschinkel [15] except that we sampled whole-colony lipids rather than individual ants. We filled a Soxhlet thimble with the entire sample of ants from each colony to get an estimate of wholecolony lipid content. After lipid removal, we dried the ants at 60˚C for 48 hours and recorded the weight.

Data analysis
Differences in temperature limits between size classes (minors and majors) of S. invicta were tested with a Student's two sample t-test. We evaluated thermal tolerance (CT min and CT max ) as a function of species (S. invicta and A. picea) and elevation using multiple regression models. We included a species x elevation interaction term to test whether ant species thermal response differed by elevation. We analyzed S. invicta colony lipid content as a function of elevation with a linear regression. Residual plots were used to verify that the assumptions of normality and homogeneity of variance were correct for all analyses. All statistical analyses were performed with the R statistical program "R Studio" [16].

Thermal tolerance
There was no difference in thermal tolerance temperatures between S. invicta minor and major worker ants for CT max (df = 417, t-value = 1.610, p-value = 0.108) or CT min (df = 418, tvalue = 1.373, p-value = 0.171). The mean (±SE) CT max was 46.5 ± 0.1˚C for major workers and 46.3 ± 0.1˚C for minor workers. The mean CT min was 6.9 ± 0.2˚C for major workers and 6.6 ± 0.2˚C for minor workers. Therefore, we did not separate the two size classes of worker ants in further analyses.

Discussion
Solenopsis invicta has indeed acclimated or adapted to cold temperatures at high elevations in the southern Blue Ridge Escarpment region. Both the cold and heat tolerance thresholds of S. invicta decreased with increasing elevation indicating a physiological ability to tolerate colder temperatures, and the shift in thermal tolerance paralleled that of the native ant A. picea. Solenopsis invicta lipid content did not vary with elevation, suggesting no metabolic effects of cold acclimation for the ants. These results suggest that S. invicta is not as limited by cold temperatures as previously thought and will likely continue to invade higher elevations and latitudes in the southeastern U.S.
The decrease in S. invicta CT min with higher elevation is consistent with other ants across elevation and heat gradients [17][18][19][20]. For example, Bishop et al. [18] found that African ant CT min decreased with decreasing temperatures along a 1500 m elevation gradient. Ant foraging phenology corresponded with CT min , so that ants at high elevations foraged at lower temperatures resulting in similar emergence times across elevations [18]. Similarly, Temnothorax curvispinosus ants in urban heat islands exhibited a higher CT max and CT min than their rural

PLOS ONE
counterparts in colder environments [19]. However, Warren et al. [21] found the opposite pattern in Brachyponera chinensis invasions in the same study region used here. Brachyponera chinensis is a non-native ant from Asia, and has invaded the forests of North and South Carolina and Georgia. Populations found at high elevations in the Southern Appalachian Mountains do not appear to persist, and Warren et al. [21] found that this pattern is explained by the ants' apparent inability to acclimate to colder temperatures (i.e., its cold tolerance does not change with elevation.
Solenopsis invicta exhibited higher heat tolerance than A. picea, and A. picea exhibited greater cold tolerance than S. invicta. These results were somewhat predictable given that S. invicta originates from a subtropical climate and A. picea is temperate. Moreover, the difference likely is exacerbated by the microhabitats that the two species occupy: A. picea is a woodland ant that favors shaded canopy understory environments, whereas S. invicta is a thermophilic species that thrives in highly disturbed environments with full sun [3,14]. What is interesting however is that the heat and cold tolerances of both species decreased similarly with increased elevation, indicating that the non-native S. invicta ants had the same ability to adapt or acclimatize as the native A. picea ants.
We found the mean S. invicta CT max (46.4 ± 0.1˚C) to be similar to that found in other studies: 50.6˚C [22] and 46.5˚C (for large workers [23]). The mean CT min for S. invicta ants in our study (6.8 ± 0.1˚C) was higher than that found in low-elevation ants collected by Bentley et al. [22]: 4.1˚C. The difference may lie in research methodology. Moreover, we found no appreciable difference in thermal tolerance between the smallest (minors) and largest (majors) polymorphic workers, whereas the relationship between S. invicta size and thermal tolerance are mixed for other studies [22,23]. This suggests that there may be regional differences in how body size of S. invicta affects thermal tolerance, or that differences in research methodologies may affect cross-study comparisons.
We found that a subtropical invasive ant can acclimate to colder temperatures than those found in its home range, and global change also may facilitate further spread of S. invicta to still greater latitudes and elevations as temperatures increase, as has occurred in other insects such as some Lepidoptera and Odonata [24,25]. Warren and Chick [25] investigated the relationship between thermal tolerance limits and distribution shifts in two woodland ants, A. picea and A. rudis. The lower elevation A. rudis consistently exhibited temperature tolerance limits 2˚C higher than that of the higher elevation species, A. picea, in both minimum and maximum thermal tolerance testing. With warming, however, the lower elevation species displaced the higher species. Hence, as temperatures rise, more competitive ant species may be able to shift their distribution into novel habitats where they were once unable to persist and displace native and/or less competitive ants [4,26].
Solenopsis invicta preferably invades anthropogenically disturbed habitats [27], and increasing fragmentation in the Blue Ridge Escarpment region [28] may have facilitated its invasion as much as thermal acclimation. As noted, we observed no S. invicta colonies in undisturbed forest, a haven for A. picea. Certainly, anthropogenic disturbance is not limited to the Blue Ridge Escarpment, and our results suggest that the ant's cold temperature tolerance will not limit its ability to utilize disturbed habitats northward of its current range. In conjunction with climate warming, a much greater portion of the eastern U.S. may be subject to S. invicta invasion.