The Distribution and Habitat Affinities of the Invasive Ant Myrmica rubra (Hymenoptera: Formicidae) in Southern New England

The Distribution and Habitat Affinities of the Invasive Ant Myrmica rubra (Hymenoptera:... Abstract The Eurasian ant Myrmica rubra (L.) (Hymenoptera: Formicidae) was first discovered in North America in the early 1900s in Massachusetts. Populations have since appeared in at least seven states within the United States and in seven Canadian provinces. We conducted a systematic search for the ant across southern New England—the states of Connecticut, Massachusetts, and Rhode Island—where M. rubra is spreading from multiple loci. The species occurs in two large regions in Massachusetts, each spanning approximately 75 km, and in several smaller populations in Massachusetts and Rhode Island. No populations were discovered anywhere in Connecticut or across large expanses of central Massachusetts and northern Rhode Island, despite the presence of apparently favorable habitat. This pattern of distribution suggests a combination of long-distance dispersal by human transport coupled with slow local spread. Resurveys of sites previously known to support M. rubra showed that populations persist for decades. Within invaded areas, M. rubra was strongly associated with particular habitats. Colonies were most prevalent in freshwater wetlands and in moist forests near wetlands and water; they were uncommon in drier forests and were rare in open habitats outside of wetlands. The slow rate of spread over the last 110 yr suggests that the ants do not easily disperse between patches of suitable habitat. distribution, habitat preference, invasive ant, European fire ant Because they play important roles at multiple trophic levels, invasive ants can have major impacts on biodiversity (Holway et al. 2002b, Lach and Hooper-Bui 2010) and can disrupt local communities (e.g., Porter and Savignano 1990, O’Dowd et al. 2003). Myrmica rubra (Linnaeus, 1758) (Hymenoptera: Formicidae), a Eurasian species that has become established in North America, has emerged as a potentially important pest in the invaded range due to its aggressive behavior, painful sting, and high local population densities (Groden et al. 2005). At various sites in North America, M. rubra is negatively associated with arboreal arthropods (Verble-Pearson and Pearson 2016) and with native ants (Naumann and Higgins 2015), which it can displace from food sources (Garnas et al. 2014). Experimental studies have shown that its workers alter homopteran and plant communities (McPhee et al. 2012, Prior et al. 2015). Myrmica is a diverse genus that is native to temperate regions of the Northern Hemisphere (Radchenko and Elmes 2010). A Palearctic species, M. rubra occurs across a wide range of latitudes, from 39°N to 70°N in its native range (Wetterer and Radchenko 2011). By 2014, it had been found in six eastern U.S. states (Maine, Massachusetts, New Hampshire, New York, Rhode Island, and Vermont) and one western state, Washington (Wetterer and Radchenko 2011). It is also established in Canada in the provinces of New Brunswick, Nova Scotia, Ontario, Prince Edward Island, and Quebec (Wetterer and Radchenko 2011), Newfoundland (Hicks et al. 2014), and British Columbia (Naumann and Higgins 2015). Myrmica rubra does not have a standardized common name. It has been called the European fire ant (e.g., Naumann and Higgins 2015), the ruby ant (Wetterer and Radchenko 2011), and the European red ant (Ness et al. 2013). We sought to document the current distribution of M. rubra in the region where it was first discovered in North America, namely southern New England, comprising the states of Massachusetts, Rhode Island, and Connecticut. The earliest North American records were from the edges of a rose garden in Woods Hole, Massachusetts, and from the Forest Hills neighborhood of Boston, Massachusetts (Wheeler 1906, 1908). Wheeler (1908) suggested that the species was brought to both places by importation of plants from Europe. Although more than 100 yr had passed since the species’ arrival, by 2005, it had been had been reported from only 13 of the 559 cities and towns that constitute southern New England (Groden et al. 2005). It is a fair question to ask whether the limited known range of M. rubra accurately reflects its distribution or whether it has spread more extensively below the myrmecologists’ radar. Efforts to reconstruct or predict the spread of invasive species are impeded by incomplete or biased sampling. Many studies of invasive species are based on ‘presence-only’ data—records of occurrences that are typically obtained opportunistically (Elith and Leathwick 2009, Franklin 2010). Data on absences, or places the species is not discovered after one or more surveys, are also useful but can be time-consuming to gather. Search effort that is unevenly distributed across geographic and environmental space can result in biased estimates of a species’ niche (Phillips et al. 2009, Anderson and Gonzalez 2011). Museum records and biodiversity inventories are typically weighted toward particular environments, including areas that are easier to reach or that are of particular interest to ecologists (e.g., Reddy and Dávalos 2003, Kadmon et al. 2004, Newbold 2010), and stinging insects may be especially likely to come to light in areas of high human population density. Data on the distribution of M. rubra in North America consist almost entirely of opportunistic records of presences, except on a local scale (e.g., Ouellette et al. 2010). We, therefore, chose to record where the ant can and cannot be found and to spread out our sampling effort across both geographic and environmental space. To characterize the distribution of a species, it is important to know its habitat affinities in addition to its geographic range. Invasive ants often occur within a limited subset of environmental conditions. For example, in pine forests of northern Florida, the fire ant Solenopsis invicta Buren (Hymenoptera: Formicidae) is restricted to open, disturbed habitats including roadsides (Tschinkel 1988). For the Argentine ant, Linepithema humile (Mayr) (Hymenoptera: Formicidae), vegetation characteristics and soil moisture help to predict which sites are occupied within the invaded range (Ward 1987, Holway et al. 2002a, Menke and Holway 2006, Fitzgerald and Gordon 2012). In its native range, M. rubra is variously described as being widespread in open and forested habitats (e.g., Vepsäläinen et al. 2008) or as being especially common in damper areas (e.g., Elmes and Petal 1990). This study was designed to document the extent, spatial structure, and habitat associations of the M. rubra invasion in southern New England, issues that are critical for understanding its spread and its impacts. We implemented an e-mail campaign and a systematic search to identify the current distribution of the ant in Massachusetts, Rhode Island, and Connecticut, preserving a snapshot of its distribution for future comparisons. Within the invaded range, we compared the prevalence of the ant in open, forested, and wetland habitats. Materials and Methods Distribution of M. rubra in Southern New England Because there is little published information on M. rubra in southern New England, we first sought efficient ways to locate possible occurrences. We solicited reports of stinging ants by e-mail sent to selected organizations. Although several native ants in New England can sting, they have smaller colonies than M. rubra and the workers are typically less aggressive to humans (Wheeler 1908, Sturtevant 1931, Groden et al. 2005, Garnas et al. 2007). In the inquiry e-mail, we asked for information on stinging ants and included a brief description of M. rubra with a link to an online photo. From the winter of 2009 to the summer of 2010, requests were sent to the staff of more than 100 wildlife sanctuaries, parks, and nature centers; 62 golf courses; 40 associations of recreational fisherman; and more than 30 other outdoor organizations in Massachusetts, including land trusts and botanical gardens. Our original e-mails were circulated by the recipients to second and third degree contacts including local gardening societies, conservation organizations, professional and amateur entomologists, and nurseries, by e-mail or social media. We gathered possible locations of M. rubra based on replies and visited most of them to identify ants. To distribute search effort more broadly throughout the study area, we divided a map of southern New England, excluding islands, into a grid of 20 × 20-km cells (Fig.1). To distinguish these cells from other spatial designations, we use the following terminology: ‘towns’ are contiguous political entities that completely fill each of the three states; ‘regions’ are areas invaded by M. rubra that span multiple contiguous towns or cells. As described below, we chose particular ‘sites’ (e.g., parks) where we walked along paths stopping periodically at particular ‘positions’ to look for M. rubra. For analysis of habitat effects, ’blocks’ were identified (see the methods for Habitat Affinities) containing several nearby sites spanning different habitat types. The terms ‘location’ and ‘area’ are used more generally. We first visited sites at which M. rubra had been previously reported and those identified through the e-mail campaign. At those sites, M. rubra was most prevalent close to freshwater wetlands and riparian habitat with wetland vegetation. We, therefore, chose publically accessible wetlands and riparian sites within each 20 × 20-km cell as primary search targets. In the state of Massachusetts, we visited a minimum of two such sites in each cell that were at least 4 km apart. (When additional sites were added, some were less than 4 km from the original two sites.) In Connecticut and Rhode Island, where there have been very few reports of M. rubra, we visited at least one site in each cell, choosing sites that were at least 9 km apart for better spatial coverage. As a rough outline of the invaded area was delineated, we added additional sites near the outer extent of occupied regions to provide greater resolution of the current limits of the distribution. We also searched at all sites in southern New England listed as localities of M. rubra by Groden et al. (2005). To detect M. rubra and to estimate prevalence, we walked paths, stopping at positions every 25–40 m to search for workers of M. rubra in an area of approximately 1 × 2 m, looking for workers on the ground and vegetation and for nest sites beneath stones and fallen branches, in the soil, and at the bases of plants. Identification of M. rubra was confirmed in the laboratory using collected samples. Surveys were conducted from May to September, 2009 to 2016, when temperatures were between 21 and 34°C, avoiding rain. During the first field season, we placed baits of sugar water at each position but found that visual searches alone, which do not require a second visit to census and retrieve the baits, were just as effective. Prevalence was quantified by the proportion of positions at which M. rubra was present. We searched an average of 560 m of trail at each site (median = 400 m; 25–75% quartile: 175 to 800 m), usually devoting at least 30 person-minutes to the hunt, sometimes several hours. Shorter searches were conducted at sites with limited access and at some places where the ant was quickly found. Elsewhere, we analyze the effect of search effort on the probability of finding M. rubra, given habitat characteristics (Chen 2016). Here, to minimize misperceptions caused by false absences, we omit sites where the search path was less than 50 m and no M. rubra were found. Results are shown for 518 sites. Habitat Affinities The prevalence of M. rubra was estimated in replicate blocks for three habitat types: open habitat, consisting of old fields and grass, including mown grass; wetlands, consisting of wet meadows dominated by Phalaris arundinacea L. (reed canary grass), freshwater marshes, swamps, and all positions within 30 m of those habitats; and forests, which included all types of forests other than forested wetlands. Habitat type was scored using land use/land cover (LULC) data for 2005 from the Office of Geographic Information of the Commonwealth of Massachusetts (MassGIS). Some LULC designations were corrected by ground truthing; for example, some positions classified as open land proved to have wetland vegetation. As our knowledge of habitat effects developed, we decided to subdivide the wetland habitat into forested wetlands (swamps) and nonforested wetlands (marshes and wet meadows), and we also subdivided forests into drier and wetter forests based on observations made on site. Drier forests included mid-slope and hilltop locations away from water or wetland with relatively thin soils and sparse herbaceous understory vegetation. Common species of drier forests included oaks and hickories, white pine (Pinus strobus L.), and lowbush blueberry (Vaccinium angustifolium Aiton). Wetter forests were wooded areas with noticeably moister soil and denser herbaceous vegetation, which were mostly on lower slopes or in riparian habitats along streams and ponds. Wetter forests supported understory plants associated with moister conditions such as common jewelweed (Impatiens capensis Meerb.) and sensitive fern (Onoclea sensibilis L.). We sampled each type of habitat in 20 non-overlapping rectangular blocks within the main invasions; 18 of the blocks were spread throughout the greater Boston invasion, and two were within the Berkshires invasion. Because we sought to examine only regions through which M. rubra has already spread, we avoided areas around isolated populations, including those on Cape Cod. Each block was selected to include sites with at least 350 m of paths through each of the three habitat types. The blocks varied in area from 3.5 to 35 km2 (mean = 13.4 km2) depending on the distribution and accessibility of habitat patches. As described above, we looked for M. rubra workers and nests at positions separated by approximately 25–40 m along our search paths. Each habitat type was represented by a minimum of 14 such positions (median = 29; 25–75% quartile: 20–38) within each block. Following the post hoc subdivision of forests into wetter and drier sites, and of wetlands into forested and nonforested wetlands, each habitat type was represented in all 20 blocks (median = 19 positions; 25–75% quartile: 13–24), with the exception of one block lacking wetter forest. The R statistical package version 3.3.2 (R Core Team 2016) was used for all analyses. We used the glmer function from the lme4 package (Bates et al. 2015), version 1.1–12, to fit generalized linear mixed models (GLMM) for which the responses are matched pairs of the number of occupied and unoccupied positions and block is the grouping variable, assuming binomially distributed errors and a logit link (Zuur et al. 2009). The analysis accounts for the possibility of spatial dependence in two ways. First, the hierarchical nature of the model, with positions nested within blocks, recognizes that measurements from locations that are within the same block may be correlated (Zuur et al. 2009). Second, the probability of occurrence of M. rubra may vary among positions of a given habitat type within a block, yielding overdispersion relative to predictions from binomial distributions with fixed probabilities. To evaluate this possibility, we used the dispersion_glmer function from the blmeco package (Korner-Nievergelt et al. 2015), version 1.1, to calculate the scale parameter. Zuur et al. (2009) recommend that the standard errors of the coefficients should be adjusted when the dispersion parameter is greater than 1.5, indicating overdispersion. In our analyses, the dispersion parameter was always less than 1.25, so no corrections were applied. Akaike’s Information Criterion (AIC) scores were used for model selection (Burnham and Anderson 1998). For each model, plots of residuals against fitted values showed no unexpected patterns and quantile plots confirmed that the random effects were approximately normally distributed. The lsmeans package (Lenth 2016), version 2.25–5, was used to calculate 95% CIs for mean prevalence in each habitat type and to make multiple comparisons among habitat types by Tukey’s method. Results Distribution of M. rubra in Southern New England In the early stages of our search, e-mail inquiries functioned as important tools for gathering information on possible locations of M. rubra populations. We received 37 responses about stinging ants, most with specific locations. Our requests were forwarded by the recipients to other individuals, organizations, lists, and social media sites; these secondary contacts were the source of most replies and included locations of stinging ants in Connecticut, Massachusetts, and Rhode Island in southern New England as well as New Hampshire and Maine in northern New England. Based on these responses, we visited 32 locations and found M. rubra at 11 new sites in Massachusetts and at two new sites outside our study area (Rye and Hart’s Location, New Hampshire). Groden et al. (2005) listed 13 towns in southern New England where M. rubra had been found. For six of the towns, more specific locations were given by Groden et al. (2005) or by Wheeler (1906, 1908). We confirmed the presence of M. rubra in all 13 towns and at all but one of the more specific locations: we found M. rubra near the former site of Fay’s Woods in Woods Hole (first reported in 1900), in and around the Arnold Arboretum in the Forest Hills neighborhood of Boston (first reported in 1908), in Shaker Glen in Woburn (assumed to be the ‘Shafer Glen’ listed in Table 3 of Groden et al. 2005; reported in 1954), at Fresh Pond in Cambridge (reported in 1967), Rock Meadow in Belmont (reported in 1982), and Eph’s Pond in Williamstown (reported in 2003). We did not search for the ant on the campus of Harvard University in Cambridge, where it was found in 1975 and where the specific location is unknown. Additional populations were discovered by search across the grid of 20 × 20 km cells and by more intensive searches along the edges of the invasions (Fig. 1; Table 1). At some of these sites, there were conspicuous populations, with high densities of workers active along more than 100 m of trail, while at other sites only a few workers could be found. Most occurrences were within two regions (Fig. 1). The largest of these spanned much of the greater Boston area, stretching from Cape Anne in the north to Marshfield in the south and west to Framingham and Sudbury. Within this region, the greatest distance from the site of the original discovery in Boston was 58 km. The second largest expanse was in valleys of the Berkshires range of western Massachusetts, extending from Sheffield in the south to Williamstown in the north and from there into Vermont. The ant was found at scattered locations on Cape Cod, Massachusetts, at distances of ~100 m to 61 km from the site of Wheeler’s first discovery in Woods Hole (Wheeler 1906). A small and apparently isolated population exists in Worcester, in central Massachusetts. In Rhode Island, we found colonies in two parks separated by about 3 km in Newport and at another site 6.6 km away in Middletown. No M. rubra were found in Connecticut. Table 1. Towns in which Mymrica rubra was found State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport View Large Table 1. Towns in which Mymrica rubra was found State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport View Large Fig. 1. View largeDownload slide Sites where Myrmica rubra was found (dark stars) and was not found (light circles). The grid of 20 × 20 km cells is shown, as well as state boundaries. Place names are shown for selected locations referred to in the text. Fig. 1. View largeDownload slide Sites where Myrmica rubra was found (dark stars) and was not found (light circles). The grid of 20 × 20 km cells is shown, as well as state boundaries. Place names are shown for selected locations referred to in the text. As we searched for M. rubra, we spoke to curious onlookers, many of whom were aware of the abundant, stinging ant in their neighborhood. Several of these persons, as well as correspondents responding to the e-mail campaign, told us the years of their first contacts with this ant. Based on the most credible reports, confirmed by our identifications of ants at present-day locations, M. rubra has been in Topsfield, Massachusetts, since at least the 1950s; in Brace Cove (Gloucester, Massachusetts) since at least 1961; in Rockport, Massachusetts, since at least the 1960s; in Chatham, Massachusetts, since at least 1965; and on Marblehead Neck, Massachusetts, since at least 1974. Habitat Affinities The prevalence of M. rubra varied substantially among habitats. For the initial division into open, forested, and wetland habitats, the AIC value was much lower, indicating greater support, when habitat type was included as a set of fixed effects than when habitats were pooled (ΔAIC = −145). The ants were most prevalent in wetlands and were almost never seen in open habitats (Table 2); all post hoc pairwise contrasts were significant (Tukey’s method, P < 0.001). The prevalence of M. rubra in forested and wetland sites was correlated across the 20 blocks (Spearman’s rank correlation coefficient = 0.70, P < 0.001). Table 2. Model coefficients and estimated mean prevalence of M. rubra for three habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) View Large Table 2. Model coefficients and estimated mean prevalence of M. rubra for three habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) View Large Post hoc forests were divided into wetter and drier forests, and wetlands were divided into forested and non-forested sites, as described in the methods. For this classification of habitats too, the AIC value was much lower when habitat type was included as a predictor than when all habitat types were pooled (ΔAIC = −227). M. rubra was more prevalent in wetter forests than in drier forest, where it was rarely seen, and was more prevalent in nonforested wetlands (marshes and wet meadows) than in forested wetlands (swamps; Table 3, Fig. 2). Table 3. Model coefficients and estimated mean prevalence of M. rubra for five habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) View Large Table 3. Model coefficients and estimated mean prevalence of M. rubra for five habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) View Large Fig. 2. View largeDownload slide Box and whiskers plot for the proportion of positions occupied by M. rubra. Forests were divided into drier and wetter sites. Wetlands were divided into forested wetlands (swamps) and nonforested wetland (marshes and wet meadows). Horizontal bars (a, b, c) connect habitat types that did not differ significantly at the 0.05 level (Tukey’s method). Fig. 2. View largeDownload slide Box and whiskers plot for the proportion of positions occupied by M. rubra. Forests were divided into drier and wetter sites. Wetlands were divided into forested wetlands (swamps) and nonforested wetland (marshes and wet meadows). Horizontal bars (a, b, c) connect habitat types that did not differ significantly at the 0.05 level (Tukey’s method). Discussion Our sampling greatly extended the known range of M. rubra in southern New England and revealed that this species has spread through two large regions, each spanning 75 km, as well as occupying several smaller areas (Fig. 1). An earlier compilation of known locations of M. rubra included only 12 towns in Massachusetts and one in Rhode Island (Groden et al. 2005). We found M. rubra in 74 towns, including 56 contiguous towns in the greater Boston area (Fig. 1, Table 1). Given that there were no systematic efforts to search for M. rubra in this region prior to our study, the expansion of the known range since 2005 is probably due more to increased detection than to a recent acceleration of population growth. The rediscovery of populations at nearly every specific site in southern New England mentioned in previous publications (Wheeler 1906, 1908; Groden et al. 2005), mostly after several decades, indicates that local populations are highly persistent. Persistence of invasive ant populations is not guaranteed: resurveys of other species show that local populations may decline and disappear even in the absence of control measures (e.g., Tschinkel and King 2013, Tartally et al. 2016, Cooling and Hoffmann 2015). The mechanisms by which M. rubra spreads are not well understood, but its current distribution, coupled with fragmentary historical information, provides some clues. Exotic ants typically spread by more than one mechanism, combining long-distance ‘jump dispersal’ due to human transport with relatively slow local processes (Suarez et al. 2001, Tschinkel 2006, Espadaler et al. 2007). M. rubra conforms to this pattern. On a continental scale, human activity is undoubtedly responsible for the spotty occurrence of this species in distant regions of North America, including sites in Washington State and British Columbia on the west coast, Ontario, and the Canadian Maritime provinces, and several areas in the northeastern United States (Groden et al. 2005, Wetterer and Radchenko 2011, Hicks et al. 2014). Within southern New England, the two major invasions in eastern and western Massachusetts are separated from each other by more than 130 km and from the smaller populations by tens of kilometers (Fig. 1), gaps that are too large for the ants to cross on their own. Commerce is the main cause of accidental introductions of invasive ants, through movement of plants, soil, and other commodities (Suarez et al. 2001). Because colonies of M. rubra often contain many queens distributed across multiple nests (Elmes and Petal 1990, Walin et al. 2001, Groden et al. 2005), a small colony fragment could easily include enough workers and reproductives to establish a viable population. Myrmica rubra has been intercepted several times in soil or plant material shipped from Europe to North America (Groden et al. 2005); similar transport could occur within North America. Local dispersal is slow. Assuming that Wheeler’s (1906, 1908) first discoveries of the ant were close to the times and places of introduction, the ant has spread approximately 15–58 km in 110 yr, depending on direction, in the greater Boston area and on Cape Cod. M. rubra can spread by budding new nests from existing colonies (Elmes 1980, Seppä and Pamilo 1995) or by mating flights (Boomsma and Leusink 1981, Noordijk et al. 2008) during which inseminated queens could reach new habitat patches. Some invasive ants form vast supercolonies that can expand by as much as 100–200 m/yr along the edges due to budding of new nests. These include L. humile (Suarez et al. 2001), Wasmannia auropunctata (Roger) (Hymenoptera: Formicidae) (Lubin 1984), Anoplolepis longipes (Jerdon) (Hymenoptera: Formicidae) (Haines and Haines 1978), and Lasius neglectus (van Loon, Boomsma and Andrásfalvy 1990) (Hymenoptera: Formicidae) (Espadaler et al. 2007). Colonies of M. rubra in Massachusetts defend territories that may span more than 100 m, but they do not form greatly expanded supercolonies (Chen et al. 2018) and rates of spread by budding may therefore be relatively low. Species with mating flights and queens that are strong fliers can disperse more rapidly. For example, the fire ant S. invicta spread as a dense population across at least 50 km in 20 yr from its entry point in Mobile, Alabama (Tschinkel 2006). The roughly circular shape of the M. rubra invasion around Boston, excepting Massachusetts Bay, suggests gradual jumps in all directions, which could be produced by short flights by mated queens or by human transport. Groden et al. (2005) hypothesized that mating flights are infrequent in Maine, in part because infestations often end abruptly at roads or streams despite the presence of similar habitat on the other side. To date, only male swarms have been observed in North America (Hicks 2012). Determining the habitat associations of invasive species is essential for understanding their impacts and for predicting their spread. Habitat characteristics control where exotic species invade and where they achieve high densities (Peterson 2003, Hastings et al. 2004). On a continental scale, the most important predictors of the presence of invasive ants are climate and land-use data (e.g., Peterson and Nakazawa 2008, Roura-Pascual et al. 2011). On a regional scale, major predictors include habitat type and vegetation indices, largely because vegetation type is an indicator of soil moisture (Holway et al. 2002a, Fitzgerald and Gordon 2012) but also because plants provide food sources and nesting sites (e.g., Ness et al. 2013). The characteristics and landscape configuration of habitats therefore strongly affect local abundance (Tschinkel 1988, LeBrun et al. 2012) and rates of spread (Pitt et al. 2009, Fitzgerald et al. 2012). In our study area, M. rubra was strongly associated with particular habitat types. It was most prevalent within or next to freshwater marshes and wet meadows and was also found in wetter forests, including riparian habitats along streams and ponds, and in forested wetlands. Not all wetlands support M. rubra: we did not find the ant in salt marshes, except at landward edges fed by freshwater runoff, and it was not collected during a systematic survey of ants living in Massachusetts bogs (Ellison et al. 2002). The ant was rare in drier forests with well-drained soils and in open habitats such as old fields and mown grass (Fig. 2). Thus, in southern New England, M. rubra appears to be limited almost entirely to places where the soil does not dry out during the summer. Ant activity is limited by temperature extremes and the threat of desiccation (e.g., Talbot 1943, Kaspari 1993, Azcárate et al. 2007) and humidity in the brood chambers is also important (Potts et al. 1984). Soil moisture apparently limits the spread of other invasive ants, including the Argentine ant, L. humile (Holway and Suarez 2006, Menke and Holway 2006), and the fire ant S. invicta (LeBrun et al. 2012). Thus, rainfall (DiGirolamo and Fox 2006, Heller et al. 2008) and experimental watering (Menke and Holway 2006) can increase ant abundance and rates of colony expansion. M. rubra has an even stronger association with moist soils than L. humile or S. invicta and may be especially susceptible to desiccation. We know of no systematic surveys on a comparable geographic scale from any other part of M. rubra’s range; however, many investigators have provided verbal descriptions of collection sites. Elsewhere in North America, M. rubra is also found in moist habitats but not as exclusively as in southern New England. An ordination of ant species on Mount Desert Island, in Maine, placed M. rubra with species that inhabit wetter habitats (Ouellette et al. 2010). Populations have been found in wetlands in Maine (Groden et al. 2005, Ouellette et al. 2010) and in moist woods and riparian habitat in British Columbia (Naumann and Higgins 2015). But, they also occur in habitats that would be surprising in our study area, such as in grassy fields, lawns, and rocky outcrops, especially in coastal areas (Groden et al. 2005, Ouellette et al. 2010, Naumann and Higgins 2015). Within Europe, Elmes and Petal (1990) describe M. rubra as an ant of ‘damp, cool places; particularly woodland edges and river banks’. However, it is found in diverse habitats within the native range, including deciduous and coniferous forests, residential areas, and open areas such as abandoned fields, pastures, grasslands, and managed lawns (e.g., Brian 1956, Pearson and Child 1980, Seppä and Pamilo 1995, Walin et al. 2001, Vepsäläinen et al. 2008, Leppänen et al. 2013). Thus, the habitat associations of M. rubra appear to vary across its range. Cooler and shorter summers, greater cloud cover, and coastal fog may expand the range of environments that can support this species by preventing desiccation. Wetterer and Radchenko (2011) reported that M. rubra is found mostly at higher elevations in the southern part of its Eurasian range. Brian (1956) hypothesized that the species is more restricted to forested sites, as opposed to open sites, in sunnier and drier climates. In Massachusetts, we had the impression that M. rubra was more abundant in drier forests and scrubland within roughly 500–800 m of the coast than farther inland, possibly due to the moderating influence of the ocean on temperatures and soil moisture levels. To analyze habitat associations, the samples were grouped into 20 geographic blocks distributed throughout the two major invasions. The prevalence of M. rubra in forested and wetland habitats in these blocks was correlated showing that measurements made within a few kilometers of one another are not independent and underscoring the usefulness of hierarchical models. The reasons for the spatial correlations are not known, but they may reflect dispersal of ants from one habitat type to another within the same block or unmeasured environmental influences that span several kilometers. Not all habitats were sampled due to differences in accessibility. In particular, we had few samples in residential and agricultural areas. Most sites were parks; others were cemeteries, campuses, private residences, and margins of roads and commercial sites. We did not find M. rubra across large swaths of southern New England despite the presence of favorable habitat (Fig. 1). Freshwater marshes, riparian habitats, and swamps are common throughout most of this region, including many areas that are no more than a few hundred meters beyond the current limits of M. rubra invasions. Suitable habitat is harder to find in areas that are heavily developed, where sandy soils allow better drainage (such as on Cape Cod and in parts of Plymouth County, Massachusetts), and at higher elevations, including the hills to the east of the Berkshires invasion. Yet, even those areas contain patches of wetland and riparian environments and M. rubra can be expected to continue to spread along multiple fronts until it has penetrated all parts of southern New England. Acknowledgments We thank Áine O’Sullivan for help with field work, Joseph Funk and Erica Norton for conducting the e-mail campaign to find locations of stinging ants, and Elizabeth Clifton for comments on an earlier draft. We gratefully acknowledge receiving helpful information on locations of this species from Gary Alpert, Lynn Atkinson, Stefan Cover, Frank Drummond, Aaron Ellison, Eleanor Groden, Richard Haradon, Adam Lazarus, Dave Lubertazzi, Manuel Morales, and Thomas Palmer. We thank the Department of Conservation and Recreation of the Commonwealth of Massachusetts, the Audubon Society, the Trustees of Reservations, the Connecticut Department of Energy and Environmental Protection, the Department of Environmental Management of Rhode Island, and numerous other organizations, municipalities, and individuals for access to study sites. References Cited Anderson , R. P. , and I. Gonzalez . 2011 . Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with Maxent . Ecol. Model . 222 : 2796 – 2811 . Google Scholar CrossRef Search ADS Azcárate , F. M. , E. Kovacs , and B. Peco . 2007 . Microclimatic conditions regulate surface activity in harvester ants Messor barbarus . J. Insect Behav . 20 : 315 – 329 . Google Scholar CrossRef Search ADS Bates , D. , M. Mächler , B. Bolker , and S. Walker . 2015 . Fitting linear mixed- effects models using lme4 . J. Stat. Softw . 67 : 1 – 48 . Google Scholar CrossRef Search ADS Boomsma , J. J. , and A. Leusink . 1981 . Weather conditions during nuptial flights of four European ant species . Oecologia 50 : 236 – 241 . Google Scholar CrossRef Search ADS PubMed Brian , M. V . 1956 . Segregation of species of the ant genus Myrmica . J. Anim. Ecol . 25 : 319 – 337 . Google Scholar CrossRef Search ADS Burnham , K. P. , and D. R. Anderson . 1998 . Model selection and multimodel inference . 2nd ed . Springer , New York . Google Scholar CrossRef Search ADS Chen , W . 2016 . The distribution of the invasive ant Myrmica rubra L. in southern New England: population structure, habitat suitability, and spatial prediction . Ph.D dissertation, University of Connecticut , Storrs, CT . Chen , W. , Á. O’Sullivan , and E. S. Adams . 2018 . Intraspecific aggression and the colony structure of the invasive ant Myrmica rubra . Ecol. Entomol . 43: 263–272. Cooling , M. , and B. D. Hoffmann . 2015 . Here today, gone tomorrow: declines and local extinctions of invasive ant populations in the absence of intervention . Biol. Invasions . 17 : 3351 – 3357 . Google Scholar CrossRef Search ADS DiGirolamo , L. A. , and L. R. Fox . 2006 . The influence of abiotic factors and temporal variation on local invasion patterns of the Argentine ant (Linepithema humile) . Biol. Invasions 8 : 125 – 135 . Google Scholar CrossRef Search ADS Elith , J. , and J. R. Leathwick . 2009 . Species distribution models: ecological explanation and prediction across space and time . Annu. Rev. Ecol. Evol. S . 40 : 677 – 697 . Google Scholar CrossRef Search ADS Ellison , A. M. , E. J. Farnsworth , and N. J. Gotelli . 2002 . Ant diversity in pitcher-plant bogs of Massachusetts . Northeast. Nat . 9 : 267 – 284 . Google Scholar CrossRef Search ADS Elmes , G. W . 1980 . Queen number in colonies of ants of the genus Myrmica . Insect. Soc . 27 : 43 – 60 . Google Scholar CrossRef Search ADS Elmes , G. W. , and J. Petal . 1990 . Queen number as an adaptable trait: evidence from wild populations of two red ant species (genus Myrmica) . J. Anim. Ecol . 59 : 675 – 690 . Google Scholar CrossRef Search ADS Espadaler , X. , A. Tartally , R. Schultz , B. Seifert , and C. Nagy . 2007 . Regional trends and preliminary results on the local expansion rate in the invasive garden ant, Lasius neglectus (Hymenoptera, Formicidae) . Insect. Soc . 54 : 293 – 301 . Google Scholar CrossRef Search ADS Fitzgerald , K. , and D. M. Gordon . 2012 . Effects of vegetation cover, presence of a native ant species, and human disturbance on colonization by Argentine ants . Conserv. Biol . 26 : 525 – 538 . Google Scholar CrossRef Search ADS PubMed Fitzgerald , K. , N. Heller , and D. M. Gordon . 2012 . Modeling the spread of the Argentine ant into natural areas: habitat suitability and spread from neighboring sites . Ecol. Model . 247 : 262 – 272 . Google Scholar CrossRef Search ADS Franklin , J . 2010 . Mapping species distributions: spatial inference and prediction . Cambridge University Press , Cambridge, United Kingdom . Google Scholar CrossRef Search ADS Garnas , J. R. , F. A. Drummond , and E. Groden . 2007 . Intercolony aggression within and among local populations of the invasive ant, Myrmica rubra (Hymenoptera: Formicidae), in coastal Maine . Environ. Entomol . 36 : 105 – 113 . Google Scholar CrossRef Search ADS PubMed Garnas , J. , E. Groden , and F. A. Drummond . 2014 . Mechanisms of competitive displacement of native ant fauna by invading Myrmica rubra (Hymenoptera: Formicidae) populations . Environ. Entomol . 43 : 1496 – 1506 . Google Scholar CrossRef Search ADS PubMed Groden , E. , F. A. Drummond , J. Garnas , and A. Franceour . 2005 . Distribution of an invasive ant, Myrmica rubra (Hymenoptera: Formicidae), in Maine . J. Econ. Entomol . 98 : 1774 – 1784 . Google Scholar CrossRef Search ADS PubMed Haines , I. H. , and J. B. Haines . 1978 . Colony structure, seasonality and food requirements of the crazy ant, Anoplolepis longipes (Jerd.), in the Seychelles . Ecol. Entomol . 3 : 109 – 118 . Google Scholar CrossRef Search ADS Hastings , A. , K. Cuddington , K. F. Davies , C. J. Dugaw , S. Elmendorf , A. Freestone , S. Harrison , M. Holland , J. Lambrinos , U. Malvadkar , et al. 2004 . The spatial spread of invasions: new developments in theory and evidence . Ecol. Lett . 8 : 91 – 101 . Google Scholar CrossRef Search ADS Heller , N. E. , N. J. Sanders , J. W. Shors , and D. M. Gordon . 2008 . Rainfall facilitates the spread, and time alters the impact, of the invasive Argentine ant . Oecologia 155 : 385 – 395 . Google Scholar CrossRef Search ADS PubMed Hicks , B. J . 2012 . How does Myrmica rubra (Hymenoptera: Formicidae) disperse in its invasive range? Record of male-only swarming flights from Newfoundland . Myrmecol. News 16 : 31 – 34 . Hicks , B. J. , B. L. Pilgrim , and H. D. Marshall . 2014 . Origins and genetic composition of the European fire ant (Hymenoptera: Formicidae) in Newfoundland, Canada . Can. Entomol . 146 : 457 – 464 . Google Scholar CrossRef Search ADS Holway , D. A. , and A. V. Suarez . 2006 . Homogenization of ant communities in mediterranean California: the effects of urbanization and invasion . Biol. Conserv . 127 : 319 – 326 . Google Scholar CrossRef Search ADS Holway , D. A. , A. V. Suarez , and T. J. Case . 2002a . Role of abiotic factors in governing susceptibility to invasion: a test with Argentine ants . Ecology 83 : 1610 – 1619 . Google Scholar CrossRef Search ADS Holway , D. A. , L. Lach , A. V. Suarez , N. D. Tsutsui , and T. J. Case . 2002b . The causes and consequences of ant invasions . Annu. Rev. Ecol. Evol. S . 33 : 181 – 233 . Google Scholar CrossRef Search ADS Kadmon , R. , O. Farber , and A. Danin . 2004 . Effect of roadside bias on the accuracy of predictive maps produced by bioclimatic models . Ecol. Appl . 14 : 401 – 413 . Google Scholar CrossRef Search ADS Kaspari , M . 1993 . Body size and microclimate use in Neotropical granivorous ants . Oecologia 96 : 500 – 507 . Google Scholar CrossRef Search ADS PubMed Korner-Nievergelt , F. , T. Roth , S. von Felten , J. Guélat , B. Almasi , and P. Korner-Nievergelt . 2015 . Bayesian data analysis in ecology using linear models with R, BUGS, and Stan . Academic Press , London, United Kingdom . Lach , L. , and L. M. Hooper-Bui . 2010 . Consequences of ant invasions , pp. 261 – 286 . In L. Lach , C. L. Parr and K. L. Abbott (eds.), Ant ecology . Oxford University Press , New York . Google Scholar CrossRef Search ADS LeBrun , E. G. , R. M. Plowes , and L. E. Gilbert . 2012 . Imported fire ants near the edge of their range: disturbance and moisture determine prevalence and impact of an invasive social insect . J. Anim. Ecol . 81 : 884 – 895 . Google Scholar CrossRef Search ADS PubMed Lenth , R. V . 2016 . Least-squares means: the R package lsmeans . J. Stat. Softw . 69 : 1 – 33 . Google Scholar CrossRef Search ADS Leppänen , J. , K. Vepsäläinen , H. Anthoni , and R. Savolainen . 2013 . Comparative phylogeography of the ants Myrmica ruginodis and Myrmica rubra . J. Biogeog . 40 : 479 – 491 . Google Scholar CrossRef Search ADS Lubin , Y. D . 1984 . Changes in the native fauna of the Galápagos Islands following invasion by the little red fire ant, Wasmannia auropunctata . Biol. J. Linn. Soc . 21 : 229 – 242 . Google Scholar CrossRef Search ADS McPhee , K. , J. Garnas , F. Drummond , and E. Groden . 2012 . Homopterans and an invasive red ant, Myrmica rubra (L.), in Maine . Environ. Entomol . 41 : 59 – 71 . Google Scholar CrossRef Search ADS PubMed Menke , S. B. , and D. A. Holway . 2006 . Abiotic factors control invasion by Argentine ants at the community scale . J. Anim. Ecol . 75 : 368 – 376 . Google Scholar CrossRef Search ADS PubMed Naumann , K. , and R. J. Higgins . 2015 . The European fire ant (Hymenoptera: Formicidae) as an invasive species: impact on local ant species and other epigaeic arthropods . Can. Entomol . 147 : 592 – 601 . Google Scholar CrossRef Search ADS Ness , J. H. , M. A. Morales , E. Kenison , E. Leduc , P. Leipzig‐Scott , E. Rollinson , B. J. Swimm , and D. R. von Allmen . 2013 . Reciprocally beneficial interactions between introduced plants and ants are induced by the presence of a third introduced species . Oikos . 122 : 695 – 704 . Google Scholar CrossRef Search ADS Newbold , T . 2010 . Applications and limitations of museum data for conservation and ecology, with particular attention to species distribution models . Prog. Phys. Geog . 34 : 3 – 22 . Google Scholar CrossRef Search ADS Noordijk , J. , R. Morssinkhof , P. Boer , A. P. Schaffers , T. Heijerman , and K. V. Sýkora . 2008 . How ants find each other; temporal and spatial patterns in nuptial flights . Insect. Soc . 55 : 266 – 273 . Google Scholar CrossRef Search ADS O’Dowd , D. J. , P. T. Green , and P. S. Lake . 2003 . Invasional ‘meltdown’ on an oceanic island . Ecol. Lett . 6 : 812 – 817 . Google Scholar CrossRef Search ADS Ouellette , G. D. , F. A. Drummond , B. Choate , and E. Groden . 2010 . Ant diversity and distribution in Acadia National Park, Maine . Environ. Entomol . 39 : 1447 – 1456 . Google Scholar CrossRef Search ADS PubMed Pearson , B. , and A. R. Child . 1980 . The distribution of an esterase polymorphism in macrogynes and microgynes of Myrmica rubra Latreille . Evolution . 34 : 105 – 109 . Google Scholar CrossRef Search ADS PubMed Peterson , A. T . 2003 . Predicting the geography of species’ invasions via ecological niche modeling . Q. Rev. Biol . 78 : 419 – 433 . Google Scholar CrossRef Search ADS PubMed Peterson , A. T. , and Y. Nakazawa . 2008 . Environmental data sets matter in ecological niche modelling: an example with Solenopsis invicta and Solenopsis richteri . Global Ecol. Biogeogr . 17 : 135 – 144 . Phillips , S. J. , M. Dudík , J. Elith , C. H. Graham , A. Lehmann , J. Leathwick , and S. Ferrier . 2009 . Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data . Ecol. Appl . 19 : 181 – 197 . Google Scholar CrossRef Search ADS PubMed Pitt , J. P. , S. P. Worner , and A. V. Suarez . 2009 . Predicting Argentine ant spread over the heterogeneous landscape using a spatially explicit stochastic model . Ecol. Appl . 19 : 1176 – 1186 . Google Scholar CrossRef Search ADS PubMed Porter , S. D. , and D. A. Savignano . 1990 . Invasion of polygyne fire ants decimates native ants and disrupts arthropod community . Ecology . 71 : 2095 – 2106 . Google Scholar CrossRef Search ADS Potts , L. R. , O. F. Francke , and J. C. Cokendolpher . 1984 . Humidity preferences of four species of fire ants (Hymenoptera: Formicidae: Solenopsis) . Insect. Soc . 31 : 335 – 340 . Google Scholar CrossRef Search ADS Prior , K. M. , J. M. Robinson , S. A. Meadley Dunphy , and M. E. Frederickson . 2015 . Mutualism between co-introduced species facilitates invasion and alters plant community structure . Proc. Biol. Sci . 282 : 20142846 . Google Scholar CrossRef Search ADS PubMed R Core Team . 2016 . R: a language and environment for statistical computing. R version 3.3.2 . R Foundation for Statistical Computing, Vienna . Radchenko , A. , and G. W. Elmes . 2010 . Myrmica ants (Hymenoptera: Formicidae) of the Old World . Natura Optima Dux Foundation , Warrzawa, Poland . Reddy , S. , and L. M. Dávalos . 2003 . Geographical sampling bias and its implications for conservation priorities in Africa . J. Biogeogr . 30 : 1719 – 1727 . Google Scholar CrossRef Search ADS Roura-Pascual , N. , C. Hui , T. Ikeda , G. Leday , D. M. Richardson , S. Carpintero , X. Espadaler , C. Gómez , B. Guénard , S. Hartley , et al. 2011 . Relative roles of climatic suitability and anthropogenic influence in determining the pattern of spread in a global invader . Proc. Natl. Acad. Sci. USA 108 : 220 – 225 . Google Scholar CrossRef Search ADS Seppä , P. , and P. Pamilo . 1995 . Gene flow and population viscosity in Myrmica ants . Heredity . 74 : 200 – 209 . Google Scholar CrossRef Search ADS Sturtevant , A. H . 1931 . Ants collected on Cape Cod, Massachusetts . Psyche . 38 : 73 – 79 . Google Scholar CrossRef Search ADS Suarez , A. V. , D. A. Holway , and T. J. Case . 2001 . Patterns of spread in biological invasions dominated by long-distance jump dispersal: insights from Argentine ants . Proc. Natl. Acad. Sci. USA 98 : 1095 – 1100 . Google Scholar CrossRef Search ADS Talbot , M . 1943 . Response of the ant Prenolepis imparis Say to temperature and humidity changes . Ecology . 24 : 345 – 352 . Google Scholar CrossRef Search ADS Tartally , A. , V. Antonova , X. Espadaler , S. Csősz , and W. Czechowski . 2016 . Collapse of the invasive garden ant, Lasius neglectus, populations in four European countries . Biol. Invasions . 18 : 3127 – 3131 . Google Scholar CrossRef Search ADS Tschinkel , W. R . 1988 . Distribution of the fire ants Solenopsis invicta and S. geminata (Hymenoptera: Formicidae) in northern Florida in relation to habitat and disturbance . Ann. Entomol. Soc. Am . 81 : 76 – 81 . Google Scholar CrossRef Search ADS Tschinkel , W. R . 2006 . The fire ants . Harvard University Press , Cambridge, MA . Tschinkel , W. R. , and J. R. King . 2013 . The role of habitat in the persistence of fire ant populations . Plos One . 8 : e78580 . Google Scholar CrossRef Search ADS PubMed Vepsäläinen , K. , H. Ikonen , and M. J. Koivula . 2008 . The structure of ant assemblages in an urban area of Helsinki, southern Finland . Ann. Zool. Fenn . 45 : 109 – 127 . Google Scholar CrossRef Search ADS Verble-Pearson , R. , and S. Pearson . 2016 . European fire ant presence decreases native arboreal insect abundance in Acadia National Park, Maine, USA . Nat. Areas J . 36 : 162 – 165 . Google Scholar CrossRef Search ADS Walin , L. , P. Seppä , and L. Sundström . 2001 . Reproductive allocation within a polygyne, polydomous colony of the ant Myrmica rubra . Ecol. Entomol . 26 : 537 – 546 . Google Scholar CrossRef Search ADS Ward , P. S . 1987 . Distribution of the introduced Argentine ant (Iridomyrmex humilis) in natural habitats of the lower Sacramento Valley and its effects on the indigenous ant fauna . Hilgardia . 55 : 1 – 16 . Google Scholar CrossRef Search ADS Wetterer , J. K. , and A. G. Radchenko . 2011 . Worldwide spread of the ruby ant, Myrmica rubra (Hymenoptera: Formicidae) . Myrmecol. News 14 : 87 – 96 . Wheeler , W. M . 1906 . New ants from New England . Psyche . 13 : 38 – 41 . Google Scholar CrossRef Search ADS Wheeler , W. M . 1908 . A European ant (Myrmica levinodis) introduced into Massachusetts . J. Econ. Entomol . 1 : 337 – 339 . Google Scholar CrossRef Search ADS Zuur , A. , E. N. Ieno , N. Walker , A. A. Saveliev , and G. M. Smith . 2009 . Mixed effects models and extensions in ecology with R . Springer , New York . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Entomology Oxford University Press

The Distribution and Habitat Affinities of the Invasive Ant Myrmica rubra (Hymenoptera: Formicidae) in Southern New England

Loading next page...
 
/lp/ou_press/the-distribution-and-habitat-affinities-of-the-invasive-ant-myrmica-uQaAB0L5VJ
Publisher
Entomological Society of America
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0046-225X
eISSN
1938-2936
D.O.I.
10.1093/ee/nvy042
Publisher site
See Article on Publisher Site

Abstract

Abstract The Eurasian ant Myrmica rubra (L.) (Hymenoptera: Formicidae) was first discovered in North America in the early 1900s in Massachusetts. Populations have since appeared in at least seven states within the United States and in seven Canadian provinces. We conducted a systematic search for the ant across southern New England—the states of Connecticut, Massachusetts, and Rhode Island—where M. rubra is spreading from multiple loci. The species occurs in two large regions in Massachusetts, each spanning approximately 75 km, and in several smaller populations in Massachusetts and Rhode Island. No populations were discovered anywhere in Connecticut or across large expanses of central Massachusetts and northern Rhode Island, despite the presence of apparently favorable habitat. This pattern of distribution suggests a combination of long-distance dispersal by human transport coupled with slow local spread. Resurveys of sites previously known to support M. rubra showed that populations persist for decades. Within invaded areas, M. rubra was strongly associated with particular habitats. Colonies were most prevalent in freshwater wetlands and in moist forests near wetlands and water; they were uncommon in drier forests and were rare in open habitats outside of wetlands. The slow rate of spread over the last 110 yr suggests that the ants do not easily disperse between patches of suitable habitat. distribution, habitat preference, invasive ant, European fire ant Because they play important roles at multiple trophic levels, invasive ants can have major impacts on biodiversity (Holway et al. 2002b, Lach and Hooper-Bui 2010) and can disrupt local communities (e.g., Porter and Savignano 1990, O’Dowd et al. 2003). Myrmica rubra (Linnaeus, 1758) (Hymenoptera: Formicidae), a Eurasian species that has become established in North America, has emerged as a potentially important pest in the invaded range due to its aggressive behavior, painful sting, and high local population densities (Groden et al. 2005). At various sites in North America, M. rubra is negatively associated with arboreal arthropods (Verble-Pearson and Pearson 2016) and with native ants (Naumann and Higgins 2015), which it can displace from food sources (Garnas et al. 2014). Experimental studies have shown that its workers alter homopteran and plant communities (McPhee et al. 2012, Prior et al. 2015). Myrmica is a diverse genus that is native to temperate regions of the Northern Hemisphere (Radchenko and Elmes 2010). A Palearctic species, M. rubra occurs across a wide range of latitudes, from 39°N to 70°N in its native range (Wetterer and Radchenko 2011). By 2014, it had been found in six eastern U.S. states (Maine, Massachusetts, New Hampshire, New York, Rhode Island, and Vermont) and one western state, Washington (Wetterer and Radchenko 2011). It is also established in Canada in the provinces of New Brunswick, Nova Scotia, Ontario, Prince Edward Island, and Quebec (Wetterer and Radchenko 2011), Newfoundland (Hicks et al. 2014), and British Columbia (Naumann and Higgins 2015). Myrmica rubra does not have a standardized common name. It has been called the European fire ant (e.g., Naumann and Higgins 2015), the ruby ant (Wetterer and Radchenko 2011), and the European red ant (Ness et al. 2013). We sought to document the current distribution of M. rubra in the region where it was first discovered in North America, namely southern New England, comprising the states of Massachusetts, Rhode Island, and Connecticut. The earliest North American records were from the edges of a rose garden in Woods Hole, Massachusetts, and from the Forest Hills neighborhood of Boston, Massachusetts (Wheeler 1906, 1908). Wheeler (1908) suggested that the species was brought to both places by importation of plants from Europe. Although more than 100 yr had passed since the species’ arrival, by 2005, it had been had been reported from only 13 of the 559 cities and towns that constitute southern New England (Groden et al. 2005). It is a fair question to ask whether the limited known range of M. rubra accurately reflects its distribution or whether it has spread more extensively below the myrmecologists’ radar. Efforts to reconstruct or predict the spread of invasive species are impeded by incomplete or biased sampling. Many studies of invasive species are based on ‘presence-only’ data—records of occurrences that are typically obtained opportunistically (Elith and Leathwick 2009, Franklin 2010). Data on absences, or places the species is not discovered after one or more surveys, are also useful but can be time-consuming to gather. Search effort that is unevenly distributed across geographic and environmental space can result in biased estimates of a species’ niche (Phillips et al. 2009, Anderson and Gonzalez 2011). Museum records and biodiversity inventories are typically weighted toward particular environments, including areas that are easier to reach or that are of particular interest to ecologists (e.g., Reddy and Dávalos 2003, Kadmon et al. 2004, Newbold 2010), and stinging insects may be especially likely to come to light in areas of high human population density. Data on the distribution of M. rubra in North America consist almost entirely of opportunistic records of presences, except on a local scale (e.g., Ouellette et al. 2010). We, therefore, chose to record where the ant can and cannot be found and to spread out our sampling effort across both geographic and environmental space. To characterize the distribution of a species, it is important to know its habitat affinities in addition to its geographic range. Invasive ants often occur within a limited subset of environmental conditions. For example, in pine forests of northern Florida, the fire ant Solenopsis invicta Buren (Hymenoptera: Formicidae) is restricted to open, disturbed habitats including roadsides (Tschinkel 1988). For the Argentine ant, Linepithema humile (Mayr) (Hymenoptera: Formicidae), vegetation characteristics and soil moisture help to predict which sites are occupied within the invaded range (Ward 1987, Holway et al. 2002a, Menke and Holway 2006, Fitzgerald and Gordon 2012). In its native range, M. rubra is variously described as being widespread in open and forested habitats (e.g., Vepsäläinen et al. 2008) or as being especially common in damper areas (e.g., Elmes and Petal 1990). This study was designed to document the extent, spatial structure, and habitat associations of the M. rubra invasion in southern New England, issues that are critical for understanding its spread and its impacts. We implemented an e-mail campaign and a systematic search to identify the current distribution of the ant in Massachusetts, Rhode Island, and Connecticut, preserving a snapshot of its distribution for future comparisons. Within the invaded range, we compared the prevalence of the ant in open, forested, and wetland habitats. Materials and Methods Distribution of M. rubra in Southern New England Because there is little published information on M. rubra in southern New England, we first sought efficient ways to locate possible occurrences. We solicited reports of stinging ants by e-mail sent to selected organizations. Although several native ants in New England can sting, they have smaller colonies than M. rubra and the workers are typically less aggressive to humans (Wheeler 1908, Sturtevant 1931, Groden et al. 2005, Garnas et al. 2007). In the inquiry e-mail, we asked for information on stinging ants and included a brief description of M. rubra with a link to an online photo. From the winter of 2009 to the summer of 2010, requests were sent to the staff of more than 100 wildlife sanctuaries, parks, and nature centers; 62 golf courses; 40 associations of recreational fisherman; and more than 30 other outdoor organizations in Massachusetts, including land trusts and botanical gardens. Our original e-mails were circulated by the recipients to second and third degree contacts including local gardening societies, conservation organizations, professional and amateur entomologists, and nurseries, by e-mail or social media. We gathered possible locations of M. rubra based on replies and visited most of them to identify ants. To distribute search effort more broadly throughout the study area, we divided a map of southern New England, excluding islands, into a grid of 20 × 20-km cells (Fig.1). To distinguish these cells from other spatial designations, we use the following terminology: ‘towns’ are contiguous political entities that completely fill each of the three states; ‘regions’ are areas invaded by M. rubra that span multiple contiguous towns or cells. As described below, we chose particular ‘sites’ (e.g., parks) where we walked along paths stopping periodically at particular ‘positions’ to look for M. rubra. For analysis of habitat effects, ’blocks’ were identified (see the methods for Habitat Affinities) containing several nearby sites spanning different habitat types. The terms ‘location’ and ‘area’ are used more generally. We first visited sites at which M. rubra had been previously reported and those identified through the e-mail campaign. At those sites, M. rubra was most prevalent close to freshwater wetlands and riparian habitat with wetland vegetation. We, therefore, chose publically accessible wetlands and riparian sites within each 20 × 20-km cell as primary search targets. In the state of Massachusetts, we visited a minimum of two such sites in each cell that were at least 4 km apart. (When additional sites were added, some were less than 4 km from the original two sites.) In Connecticut and Rhode Island, where there have been very few reports of M. rubra, we visited at least one site in each cell, choosing sites that were at least 9 km apart for better spatial coverage. As a rough outline of the invaded area was delineated, we added additional sites near the outer extent of occupied regions to provide greater resolution of the current limits of the distribution. We also searched at all sites in southern New England listed as localities of M. rubra by Groden et al. (2005). To detect M. rubra and to estimate prevalence, we walked paths, stopping at positions every 25–40 m to search for workers of M. rubra in an area of approximately 1 × 2 m, looking for workers on the ground and vegetation and for nest sites beneath stones and fallen branches, in the soil, and at the bases of plants. Identification of M. rubra was confirmed in the laboratory using collected samples. Surveys were conducted from May to September, 2009 to 2016, when temperatures were between 21 and 34°C, avoiding rain. During the first field season, we placed baits of sugar water at each position but found that visual searches alone, which do not require a second visit to census and retrieve the baits, were just as effective. Prevalence was quantified by the proportion of positions at which M. rubra was present. We searched an average of 560 m of trail at each site (median = 400 m; 25–75% quartile: 175 to 800 m), usually devoting at least 30 person-minutes to the hunt, sometimes several hours. Shorter searches were conducted at sites with limited access and at some places where the ant was quickly found. Elsewhere, we analyze the effect of search effort on the probability of finding M. rubra, given habitat characteristics (Chen 2016). Here, to minimize misperceptions caused by false absences, we omit sites where the search path was less than 50 m and no M. rubra were found. Results are shown for 518 sites. Habitat Affinities The prevalence of M. rubra was estimated in replicate blocks for three habitat types: open habitat, consisting of old fields and grass, including mown grass; wetlands, consisting of wet meadows dominated by Phalaris arundinacea L. (reed canary grass), freshwater marshes, swamps, and all positions within 30 m of those habitats; and forests, which included all types of forests other than forested wetlands. Habitat type was scored using land use/land cover (LULC) data for 2005 from the Office of Geographic Information of the Commonwealth of Massachusetts (MassGIS). Some LULC designations were corrected by ground truthing; for example, some positions classified as open land proved to have wetland vegetation. As our knowledge of habitat effects developed, we decided to subdivide the wetland habitat into forested wetlands (swamps) and nonforested wetlands (marshes and wet meadows), and we also subdivided forests into drier and wetter forests based on observations made on site. Drier forests included mid-slope and hilltop locations away from water or wetland with relatively thin soils and sparse herbaceous understory vegetation. Common species of drier forests included oaks and hickories, white pine (Pinus strobus L.), and lowbush blueberry (Vaccinium angustifolium Aiton). Wetter forests were wooded areas with noticeably moister soil and denser herbaceous vegetation, which were mostly on lower slopes or in riparian habitats along streams and ponds. Wetter forests supported understory plants associated with moister conditions such as common jewelweed (Impatiens capensis Meerb.) and sensitive fern (Onoclea sensibilis L.). We sampled each type of habitat in 20 non-overlapping rectangular blocks within the main invasions; 18 of the blocks were spread throughout the greater Boston invasion, and two were within the Berkshires invasion. Because we sought to examine only regions through which M. rubra has already spread, we avoided areas around isolated populations, including those on Cape Cod. Each block was selected to include sites with at least 350 m of paths through each of the three habitat types. The blocks varied in area from 3.5 to 35 km2 (mean = 13.4 km2) depending on the distribution and accessibility of habitat patches. As described above, we looked for M. rubra workers and nests at positions separated by approximately 25–40 m along our search paths. Each habitat type was represented by a minimum of 14 such positions (median = 29; 25–75% quartile: 20–38) within each block. Following the post hoc subdivision of forests into wetter and drier sites, and of wetlands into forested and nonforested wetlands, each habitat type was represented in all 20 blocks (median = 19 positions; 25–75% quartile: 13–24), with the exception of one block lacking wetter forest. The R statistical package version 3.3.2 (R Core Team 2016) was used for all analyses. We used the glmer function from the lme4 package (Bates et al. 2015), version 1.1–12, to fit generalized linear mixed models (GLMM) for which the responses are matched pairs of the number of occupied and unoccupied positions and block is the grouping variable, assuming binomially distributed errors and a logit link (Zuur et al. 2009). The analysis accounts for the possibility of spatial dependence in two ways. First, the hierarchical nature of the model, with positions nested within blocks, recognizes that measurements from locations that are within the same block may be correlated (Zuur et al. 2009). Second, the probability of occurrence of M. rubra may vary among positions of a given habitat type within a block, yielding overdispersion relative to predictions from binomial distributions with fixed probabilities. To evaluate this possibility, we used the dispersion_glmer function from the blmeco package (Korner-Nievergelt et al. 2015), version 1.1, to calculate the scale parameter. Zuur et al. (2009) recommend that the standard errors of the coefficients should be adjusted when the dispersion parameter is greater than 1.5, indicating overdispersion. In our analyses, the dispersion parameter was always less than 1.25, so no corrections were applied. Akaike’s Information Criterion (AIC) scores were used for model selection (Burnham and Anderson 1998). For each model, plots of residuals against fitted values showed no unexpected patterns and quantile plots confirmed that the random effects were approximately normally distributed. The lsmeans package (Lenth 2016), version 2.25–5, was used to calculate 95% CIs for mean prevalence in each habitat type and to make multiple comparisons among habitat types by Tukey’s method. Results Distribution of M. rubra in Southern New England In the early stages of our search, e-mail inquiries functioned as important tools for gathering information on possible locations of M. rubra populations. We received 37 responses about stinging ants, most with specific locations. Our requests were forwarded by the recipients to other individuals, organizations, lists, and social media sites; these secondary contacts were the source of most replies and included locations of stinging ants in Connecticut, Massachusetts, and Rhode Island in southern New England as well as New Hampshire and Maine in northern New England. Based on these responses, we visited 32 locations and found M. rubra at 11 new sites in Massachusetts and at two new sites outside our study area (Rye and Hart’s Location, New Hampshire). Groden et al. (2005) listed 13 towns in southern New England where M. rubra had been found. For six of the towns, more specific locations were given by Groden et al. (2005) or by Wheeler (1906, 1908). We confirmed the presence of M. rubra in all 13 towns and at all but one of the more specific locations: we found M. rubra near the former site of Fay’s Woods in Woods Hole (first reported in 1900), in and around the Arnold Arboretum in the Forest Hills neighborhood of Boston (first reported in 1908), in Shaker Glen in Woburn (assumed to be the ‘Shafer Glen’ listed in Table 3 of Groden et al. 2005; reported in 1954), at Fresh Pond in Cambridge (reported in 1967), Rock Meadow in Belmont (reported in 1982), and Eph’s Pond in Williamstown (reported in 2003). We did not search for the ant on the campus of Harvard University in Cambridge, where it was found in 1975 and where the specific location is unknown. Additional populations were discovered by search across the grid of 20 × 20 km cells and by more intensive searches along the edges of the invasions (Fig. 1; Table 1). At some of these sites, there were conspicuous populations, with high densities of workers active along more than 100 m of trail, while at other sites only a few workers could be found. Most occurrences were within two regions (Fig. 1). The largest of these spanned much of the greater Boston area, stretching from Cape Anne in the north to Marshfield in the south and west to Framingham and Sudbury. Within this region, the greatest distance from the site of the original discovery in Boston was 58 km. The second largest expanse was in valleys of the Berkshires range of western Massachusetts, extending from Sheffield in the south to Williamstown in the north and from there into Vermont. The ant was found at scattered locations on Cape Cod, Massachusetts, at distances of ~100 m to 61 km from the site of Wheeler’s first discovery in Woods Hole (Wheeler 1906). A small and apparently isolated population exists in Worcester, in central Massachusetts. In Rhode Island, we found colonies in two parks separated by about 3 km in Newport and at another site 6.6 km away in Middletown. No M. rubra were found in Connecticut. Table 1. Towns in which Mymrica rubra was found State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport View Large Table 1. Towns in which Mymrica rubra was found State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport State County Town Massachusetts Barnstable Barnstable Brewster Chatham Falmouth (includes Woods Hole) Massachusetts Berkshire Dalton Great Barrington Lanesborough Lee Lenox North Adams Sheffield Stockbridge Washington Williamstown Massachusetts Essex Andover Beverly Danvers Essex Gloucester Hamilton Ipswich Lawrence Lynn Lynnfield Manchester-by-the-Sea Marblehead Peabody Rockport Salem Saugus Swampscott Topsfield Wenham West Newbury Massachusetts Middlesex Arlington Bedford Belmont Cambridge Everett Framingham Lexington Malden Medford Melrose Natick Newton North Reading Reading Stoneham Sudbury Wakefield Watertown Woburn Massachusetts Norfolk Braintree Canton Cohasset Dedham Milton Needham Norwood Quincy Stoughton Wellesley Weymouth Massachusetts Plymouth Brockton Hingham Marshfield Scituate Massachusetts Suffolk Boston (includes Forest Hills) Revere Winthrop Massachusetts Worcester Worcester Rhode Island Newport Middletown Newport View Large Fig. 1. View largeDownload slide Sites where Myrmica rubra was found (dark stars) and was not found (light circles). The grid of 20 × 20 km cells is shown, as well as state boundaries. Place names are shown for selected locations referred to in the text. Fig. 1. View largeDownload slide Sites where Myrmica rubra was found (dark stars) and was not found (light circles). The grid of 20 × 20 km cells is shown, as well as state boundaries. Place names are shown for selected locations referred to in the text. As we searched for M. rubra, we spoke to curious onlookers, many of whom were aware of the abundant, stinging ant in their neighborhood. Several of these persons, as well as correspondents responding to the e-mail campaign, told us the years of their first contacts with this ant. Based on the most credible reports, confirmed by our identifications of ants at present-day locations, M. rubra has been in Topsfield, Massachusetts, since at least the 1950s; in Brace Cove (Gloucester, Massachusetts) since at least 1961; in Rockport, Massachusetts, since at least the 1960s; in Chatham, Massachusetts, since at least 1965; and on Marblehead Neck, Massachusetts, since at least 1974. Habitat Affinities The prevalence of M. rubra varied substantially among habitats. For the initial division into open, forested, and wetland habitats, the AIC value was much lower, indicating greater support, when habitat type was included as a set of fixed effects than when habitats were pooled (ΔAIC = −145). The ants were most prevalent in wetlands and were almost never seen in open habitats (Table 2); all post hoc pairwise contrasts were significant (Tukey’s method, P < 0.001). The prevalence of M. rubra in forested and wetland sites was correlated across the 20 blocks (Spearman’s rank correlation coefficient = 0.70, P < 0.001). Table 2. Model coefficients and estimated mean prevalence of M. rubra for three habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) View Large Table 2. Model coefficients and estimated mean prevalence of M. rubra for three habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −6.54 1.01 −6.49 <0.0001 0.001 (0 – 0.01) Forested habitat 3.63 1.00 3.62 <0.001 0.05 (0.03 – 0.08) Wetland habitat 4.88 0.99 4.90 <0.001 0.16 (0.11 – 0.22) View Large Post hoc forests were divided into wetter and drier forests, and wetlands were divided into forested and non-forested sites, as described in the methods. For this classification of habitats too, the AIC value was much lower when habitat type was included as a predictor than when all habitat types were pooled (ΔAIC = −227). M. rubra was more prevalent in wetter forests than in drier forest, where it was rarely seen, and was more prevalent in nonforested wetlands (marshes and wet meadows) than in forested wetlands (swamps; Table 3, Fig. 2). Table 3. Model coefficients and estimated mean prevalence of M. rubra for five habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) View Large Table 3. Model coefficients and estimated mean prevalence of M. rubra for five habitat types, with open habitat as the baseline level, within the invaded range Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) Random effects: Block. SD = 0.76 Fixed effects: Estimate SE z-value P Prevalence (95% CI) Intercept (open habitat) −.52 1.01 −6.46 <0.0001 0.001 (0.00–0.01) Drier forest 0.19 1.40 0.13 0.89 0.002 (0.00–0.01) Swamp 4.35 1.01 4.30 <0.0001 0.10 (0.06–0.16) Wetter forest 4.55 1.00 4.53 <0.0001 0.12 (0.08–0.18) Marsh and wet meadow 5.17 1.00 5.16 <0.0001 0.21 (0.14–0.29) View Large Fig. 2. View largeDownload slide Box and whiskers plot for the proportion of positions occupied by M. rubra. Forests were divided into drier and wetter sites. Wetlands were divided into forested wetlands (swamps) and nonforested wetland (marshes and wet meadows). Horizontal bars (a, b, c) connect habitat types that did not differ significantly at the 0.05 level (Tukey’s method). Fig. 2. View largeDownload slide Box and whiskers plot for the proportion of positions occupied by M. rubra. Forests were divided into drier and wetter sites. Wetlands were divided into forested wetlands (swamps) and nonforested wetland (marshes and wet meadows). Horizontal bars (a, b, c) connect habitat types that did not differ significantly at the 0.05 level (Tukey’s method). Discussion Our sampling greatly extended the known range of M. rubra in southern New England and revealed that this species has spread through two large regions, each spanning 75 km, as well as occupying several smaller areas (Fig. 1). An earlier compilation of known locations of M. rubra included only 12 towns in Massachusetts and one in Rhode Island (Groden et al. 2005). We found M. rubra in 74 towns, including 56 contiguous towns in the greater Boston area (Fig. 1, Table 1). Given that there were no systematic efforts to search for M. rubra in this region prior to our study, the expansion of the known range since 2005 is probably due more to increased detection than to a recent acceleration of population growth. The rediscovery of populations at nearly every specific site in southern New England mentioned in previous publications (Wheeler 1906, 1908; Groden et al. 2005), mostly after several decades, indicates that local populations are highly persistent. Persistence of invasive ant populations is not guaranteed: resurveys of other species show that local populations may decline and disappear even in the absence of control measures (e.g., Tschinkel and King 2013, Tartally et al. 2016, Cooling and Hoffmann 2015). The mechanisms by which M. rubra spreads are not well understood, but its current distribution, coupled with fragmentary historical information, provides some clues. Exotic ants typically spread by more than one mechanism, combining long-distance ‘jump dispersal’ due to human transport with relatively slow local processes (Suarez et al. 2001, Tschinkel 2006, Espadaler et al. 2007). M. rubra conforms to this pattern. On a continental scale, human activity is undoubtedly responsible for the spotty occurrence of this species in distant regions of North America, including sites in Washington State and British Columbia on the west coast, Ontario, and the Canadian Maritime provinces, and several areas in the northeastern United States (Groden et al. 2005, Wetterer and Radchenko 2011, Hicks et al. 2014). Within southern New England, the two major invasions in eastern and western Massachusetts are separated from each other by more than 130 km and from the smaller populations by tens of kilometers (Fig. 1), gaps that are too large for the ants to cross on their own. Commerce is the main cause of accidental introductions of invasive ants, through movement of plants, soil, and other commodities (Suarez et al. 2001). Because colonies of M. rubra often contain many queens distributed across multiple nests (Elmes and Petal 1990, Walin et al. 2001, Groden et al. 2005), a small colony fragment could easily include enough workers and reproductives to establish a viable population. Myrmica rubra has been intercepted several times in soil or plant material shipped from Europe to North America (Groden et al. 2005); similar transport could occur within North America. Local dispersal is slow. Assuming that Wheeler’s (1906, 1908) first discoveries of the ant were close to the times and places of introduction, the ant has spread approximately 15–58 km in 110 yr, depending on direction, in the greater Boston area and on Cape Cod. M. rubra can spread by budding new nests from existing colonies (Elmes 1980, Seppä and Pamilo 1995) or by mating flights (Boomsma and Leusink 1981, Noordijk et al. 2008) during which inseminated queens could reach new habitat patches. Some invasive ants form vast supercolonies that can expand by as much as 100–200 m/yr along the edges due to budding of new nests. These include L. humile (Suarez et al. 2001), Wasmannia auropunctata (Roger) (Hymenoptera: Formicidae) (Lubin 1984), Anoplolepis longipes (Jerdon) (Hymenoptera: Formicidae) (Haines and Haines 1978), and Lasius neglectus (van Loon, Boomsma and Andrásfalvy 1990) (Hymenoptera: Formicidae) (Espadaler et al. 2007). Colonies of M. rubra in Massachusetts defend territories that may span more than 100 m, but they do not form greatly expanded supercolonies (Chen et al. 2018) and rates of spread by budding may therefore be relatively low. Species with mating flights and queens that are strong fliers can disperse more rapidly. For example, the fire ant S. invicta spread as a dense population across at least 50 km in 20 yr from its entry point in Mobile, Alabama (Tschinkel 2006). The roughly circular shape of the M. rubra invasion around Boston, excepting Massachusetts Bay, suggests gradual jumps in all directions, which could be produced by short flights by mated queens or by human transport. Groden et al. (2005) hypothesized that mating flights are infrequent in Maine, in part because infestations often end abruptly at roads or streams despite the presence of similar habitat on the other side. To date, only male swarms have been observed in North America (Hicks 2012). Determining the habitat associations of invasive species is essential for understanding their impacts and for predicting their spread. Habitat characteristics control where exotic species invade and where they achieve high densities (Peterson 2003, Hastings et al. 2004). On a continental scale, the most important predictors of the presence of invasive ants are climate and land-use data (e.g., Peterson and Nakazawa 2008, Roura-Pascual et al. 2011). On a regional scale, major predictors include habitat type and vegetation indices, largely because vegetation type is an indicator of soil moisture (Holway et al. 2002a, Fitzgerald and Gordon 2012) but also because plants provide food sources and nesting sites (e.g., Ness et al. 2013). The characteristics and landscape configuration of habitats therefore strongly affect local abundance (Tschinkel 1988, LeBrun et al. 2012) and rates of spread (Pitt et al. 2009, Fitzgerald et al. 2012). In our study area, M. rubra was strongly associated with particular habitat types. It was most prevalent within or next to freshwater marshes and wet meadows and was also found in wetter forests, including riparian habitats along streams and ponds, and in forested wetlands. Not all wetlands support M. rubra: we did not find the ant in salt marshes, except at landward edges fed by freshwater runoff, and it was not collected during a systematic survey of ants living in Massachusetts bogs (Ellison et al. 2002). The ant was rare in drier forests with well-drained soils and in open habitats such as old fields and mown grass (Fig. 2). Thus, in southern New England, M. rubra appears to be limited almost entirely to places where the soil does not dry out during the summer. Ant activity is limited by temperature extremes and the threat of desiccation (e.g., Talbot 1943, Kaspari 1993, Azcárate et al. 2007) and humidity in the brood chambers is also important (Potts et al. 1984). Soil moisture apparently limits the spread of other invasive ants, including the Argentine ant, L. humile (Holway and Suarez 2006, Menke and Holway 2006), and the fire ant S. invicta (LeBrun et al. 2012). Thus, rainfall (DiGirolamo and Fox 2006, Heller et al. 2008) and experimental watering (Menke and Holway 2006) can increase ant abundance and rates of colony expansion. M. rubra has an even stronger association with moist soils than L. humile or S. invicta and may be especially susceptible to desiccation. We know of no systematic surveys on a comparable geographic scale from any other part of M. rubra’s range; however, many investigators have provided verbal descriptions of collection sites. Elsewhere in North America, M. rubra is also found in moist habitats but not as exclusively as in southern New England. An ordination of ant species on Mount Desert Island, in Maine, placed M. rubra with species that inhabit wetter habitats (Ouellette et al. 2010). Populations have been found in wetlands in Maine (Groden et al. 2005, Ouellette et al. 2010) and in moist woods and riparian habitat in British Columbia (Naumann and Higgins 2015). But, they also occur in habitats that would be surprising in our study area, such as in grassy fields, lawns, and rocky outcrops, especially in coastal areas (Groden et al. 2005, Ouellette et al. 2010, Naumann and Higgins 2015). Within Europe, Elmes and Petal (1990) describe M. rubra as an ant of ‘damp, cool places; particularly woodland edges and river banks’. However, it is found in diverse habitats within the native range, including deciduous and coniferous forests, residential areas, and open areas such as abandoned fields, pastures, grasslands, and managed lawns (e.g., Brian 1956, Pearson and Child 1980, Seppä and Pamilo 1995, Walin et al. 2001, Vepsäläinen et al. 2008, Leppänen et al. 2013). Thus, the habitat associations of M. rubra appear to vary across its range. Cooler and shorter summers, greater cloud cover, and coastal fog may expand the range of environments that can support this species by preventing desiccation. Wetterer and Radchenko (2011) reported that M. rubra is found mostly at higher elevations in the southern part of its Eurasian range. Brian (1956) hypothesized that the species is more restricted to forested sites, as opposed to open sites, in sunnier and drier climates. In Massachusetts, we had the impression that M. rubra was more abundant in drier forests and scrubland within roughly 500–800 m of the coast than farther inland, possibly due to the moderating influence of the ocean on temperatures and soil moisture levels. To analyze habitat associations, the samples were grouped into 20 geographic blocks distributed throughout the two major invasions. The prevalence of M. rubra in forested and wetland habitats in these blocks was correlated showing that measurements made within a few kilometers of one another are not independent and underscoring the usefulness of hierarchical models. The reasons for the spatial correlations are not known, but they may reflect dispersal of ants from one habitat type to another within the same block or unmeasured environmental influences that span several kilometers. Not all habitats were sampled due to differences in accessibility. In particular, we had few samples in residential and agricultural areas. Most sites were parks; others were cemeteries, campuses, private residences, and margins of roads and commercial sites. We did not find M. rubra across large swaths of southern New England despite the presence of favorable habitat (Fig. 1). Freshwater marshes, riparian habitats, and swamps are common throughout most of this region, including many areas that are no more than a few hundred meters beyond the current limits of M. rubra invasions. Suitable habitat is harder to find in areas that are heavily developed, where sandy soils allow better drainage (such as on Cape Cod and in parts of Plymouth County, Massachusetts), and at higher elevations, including the hills to the east of the Berkshires invasion. Yet, even those areas contain patches of wetland and riparian environments and M. rubra can be expected to continue to spread along multiple fronts until it has penetrated all parts of southern New England. Acknowledgments We thank Áine O’Sullivan for help with field work, Joseph Funk and Erica Norton for conducting the e-mail campaign to find locations of stinging ants, and Elizabeth Clifton for comments on an earlier draft. We gratefully acknowledge receiving helpful information on locations of this species from Gary Alpert, Lynn Atkinson, Stefan Cover, Frank Drummond, Aaron Ellison, Eleanor Groden, Richard Haradon, Adam Lazarus, Dave Lubertazzi, Manuel Morales, and Thomas Palmer. We thank the Department of Conservation and Recreation of the Commonwealth of Massachusetts, the Audubon Society, the Trustees of Reservations, the Connecticut Department of Energy and Environmental Protection, the Department of Environmental Management of Rhode Island, and numerous other organizations, municipalities, and individuals for access to study sites. References Cited Anderson , R. P. , and I. Gonzalez . 2011 . Species-specific tuning increases robustness to sampling bias in models of species distributions: an implementation with Maxent . Ecol. Model . 222 : 2796 – 2811 . Google Scholar CrossRef Search ADS Azcárate , F. M. , E. Kovacs , and B. Peco . 2007 . Microclimatic conditions regulate surface activity in harvester ants Messor barbarus . J. Insect Behav . 20 : 315 – 329 . Google Scholar CrossRef Search ADS Bates , D. , M. Mächler , B. Bolker , and S. Walker . 2015 . Fitting linear mixed- effects models using lme4 . J. Stat. Softw . 67 : 1 – 48 . Google Scholar CrossRef Search ADS Boomsma , J. J. , and A. Leusink . 1981 . Weather conditions during nuptial flights of four European ant species . Oecologia 50 : 236 – 241 . Google Scholar CrossRef Search ADS PubMed Brian , M. V . 1956 . Segregation of species of the ant genus Myrmica . J. Anim. Ecol . 25 : 319 – 337 . Google Scholar CrossRef Search ADS Burnham , K. P. , and D. R. Anderson . 1998 . Model selection and multimodel inference . 2nd ed . Springer , New York . Google Scholar CrossRef Search ADS Chen , W . 2016 . The distribution of the invasive ant Myrmica rubra L. in southern New England: population structure, habitat suitability, and spatial prediction . Ph.D dissertation, University of Connecticut , Storrs, CT . Chen , W. , Á. O’Sullivan , and E. S. Adams . 2018 . Intraspecific aggression and the colony structure of the invasive ant Myrmica rubra . Ecol. Entomol . 43: 263–272. Cooling , M. , and B. D. Hoffmann . 2015 . Here today, gone tomorrow: declines and local extinctions of invasive ant populations in the absence of intervention . Biol. Invasions . 17 : 3351 – 3357 . Google Scholar CrossRef Search ADS DiGirolamo , L. A. , and L. R. Fox . 2006 . The influence of abiotic factors and temporal variation on local invasion patterns of the Argentine ant (Linepithema humile) . Biol. Invasions 8 : 125 – 135 . Google Scholar CrossRef Search ADS Elith , J. , and J. R. Leathwick . 2009 . Species distribution models: ecological explanation and prediction across space and time . Annu. Rev. Ecol. Evol. S . 40 : 677 – 697 . Google Scholar CrossRef Search ADS Ellison , A. M. , E. J. Farnsworth , and N. J. Gotelli . 2002 . Ant diversity in pitcher-plant bogs of Massachusetts . Northeast. Nat . 9 : 267 – 284 . Google Scholar CrossRef Search ADS Elmes , G. W . 1980 . Queen number in colonies of ants of the genus Myrmica . Insect. Soc . 27 : 43 – 60 . Google Scholar CrossRef Search ADS Elmes , G. W. , and J. Petal . 1990 . Queen number as an adaptable trait: evidence from wild populations of two red ant species (genus Myrmica) . J. Anim. Ecol . 59 : 675 – 690 . Google Scholar CrossRef Search ADS Espadaler , X. , A. Tartally , R. Schultz , B. Seifert , and C. Nagy . 2007 . Regional trends and preliminary results on the local expansion rate in the invasive garden ant, Lasius neglectus (Hymenoptera, Formicidae) . Insect. Soc . 54 : 293 – 301 . Google Scholar CrossRef Search ADS Fitzgerald , K. , and D. M. Gordon . 2012 . Effects of vegetation cover, presence of a native ant species, and human disturbance on colonization by Argentine ants . Conserv. Biol . 26 : 525 – 538 . Google Scholar CrossRef Search ADS PubMed Fitzgerald , K. , N. Heller , and D. M. Gordon . 2012 . Modeling the spread of the Argentine ant into natural areas: habitat suitability and spread from neighboring sites . Ecol. Model . 247 : 262 – 272 . Google Scholar CrossRef Search ADS Franklin , J . 2010 . Mapping species distributions: spatial inference and prediction . Cambridge University Press , Cambridge, United Kingdom . Google Scholar CrossRef Search ADS Garnas , J. R. , F. A. Drummond , and E. Groden . 2007 . Intercolony aggression within and among local populations of the invasive ant, Myrmica rubra (Hymenoptera: Formicidae), in coastal Maine . Environ. Entomol . 36 : 105 – 113 . Google Scholar CrossRef Search ADS PubMed Garnas , J. , E. Groden , and F. A. Drummond . 2014 . Mechanisms of competitive displacement of native ant fauna by invading Myrmica rubra (Hymenoptera: Formicidae) populations . Environ. Entomol . 43 : 1496 – 1506 . Google Scholar CrossRef Search ADS PubMed Groden , E. , F. A. Drummond , J. Garnas , and A. Franceour . 2005 . Distribution of an invasive ant, Myrmica rubra (Hymenoptera: Formicidae), in Maine . J. Econ. Entomol . 98 : 1774 – 1784 . Google Scholar CrossRef Search ADS PubMed Haines , I. H. , and J. B. Haines . 1978 . Colony structure, seasonality and food requirements of the crazy ant, Anoplolepis longipes (Jerd.), in the Seychelles . Ecol. Entomol . 3 : 109 – 118 . Google Scholar CrossRef Search ADS Hastings , A. , K. Cuddington , K. F. Davies , C. J. Dugaw , S. Elmendorf , A. Freestone , S. Harrison , M. Holland , J. Lambrinos , U. Malvadkar , et al. 2004 . The spatial spread of invasions: new developments in theory and evidence . Ecol. Lett . 8 : 91 – 101 . Google Scholar CrossRef Search ADS Heller , N. E. , N. J. Sanders , J. W. Shors , and D. M. Gordon . 2008 . Rainfall facilitates the spread, and time alters the impact, of the invasive Argentine ant . Oecologia 155 : 385 – 395 . Google Scholar CrossRef Search ADS PubMed Hicks , B. J . 2012 . How does Myrmica rubra (Hymenoptera: Formicidae) disperse in its invasive range? Record of male-only swarming flights from Newfoundland . Myrmecol. News 16 : 31 – 34 . Hicks , B. J. , B. L. Pilgrim , and H. D. Marshall . 2014 . Origins and genetic composition of the European fire ant (Hymenoptera: Formicidae) in Newfoundland, Canada . Can. Entomol . 146 : 457 – 464 . Google Scholar CrossRef Search ADS Holway , D. A. , and A. V. Suarez . 2006 . Homogenization of ant communities in mediterranean California: the effects of urbanization and invasion . Biol. Conserv . 127 : 319 – 326 . Google Scholar CrossRef Search ADS Holway , D. A. , A. V. Suarez , and T. J. Case . 2002a . Role of abiotic factors in governing susceptibility to invasion: a test with Argentine ants . Ecology 83 : 1610 – 1619 . Google Scholar CrossRef Search ADS Holway , D. A. , L. Lach , A. V. Suarez , N. D. Tsutsui , and T. J. Case . 2002b . The causes and consequences of ant invasions . Annu. Rev. Ecol. Evol. S . 33 : 181 – 233 . Google Scholar CrossRef Search ADS Kadmon , R. , O. Farber , and A. Danin . 2004 . Effect of roadside bias on the accuracy of predictive maps produced by bioclimatic models . Ecol. Appl . 14 : 401 – 413 . Google Scholar CrossRef Search ADS Kaspari , M . 1993 . Body size and microclimate use in Neotropical granivorous ants . Oecologia 96 : 500 – 507 . Google Scholar CrossRef Search ADS PubMed Korner-Nievergelt , F. , T. Roth , S. von Felten , J. Guélat , B. Almasi , and P. Korner-Nievergelt . 2015 . Bayesian data analysis in ecology using linear models with R, BUGS, and Stan . Academic Press , London, United Kingdom . Lach , L. , and L. M. Hooper-Bui . 2010 . Consequences of ant invasions , pp. 261 – 286 . In L. Lach , C. L. Parr and K. L. Abbott (eds.), Ant ecology . Oxford University Press , New York . Google Scholar CrossRef Search ADS LeBrun , E. G. , R. M. Plowes , and L. E. Gilbert . 2012 . Imported fire ants near the edge of their range: disturbance and moisture determine prevalence and impact of an invasive social insect . J. Anim. Ecol . 81 : 884 – 895 . Google Scholar CrossRef Search ADS PubMed Lenth , R. V . 2016 . Least-squares means: the R package lsmeans . J. Stat. Softw . 69 : 1 – 33 . Google Scholar CrossRef Search ADS Leppänen , J. , K. Vepsäläinen , H. Anthoni , and R. Savolainen . 2013 . Comparative phylogeography of the ants Myrmica ruginodis and Myrmica rubra . J. Biogeog . 40 : 479 – 491 . Google Scholar CrossRef Search ADS Lubin , Y. D . 1984 . Changes in the native fauna of the Galápagos Islands following invasion by the little red fire ant, Wasmannia auropunctata . Biol. J. Linn. Soc . 21 : 229 – 242 . Google Scholar CrossRef Search ADS McPhee , K. , J. Garnas , F. Drummond , and E. Groden . 2012 . Homopterans and an invasive red ant, Myrmica rubra (L.), in Maine . Environ. Entomol . 41 : 59 – 71 . Google Scholar CrossRef Search ADS PubMed Menke , S. B. , and D. A. Holway . 2006 . Abiotic factors control invasion by Argentine ants at the community scale . J. Anim. Ecol . 75 : 368 – 376 . Google Scholar CrossRef Search ADS PubMed Naumann , K. , and R. J. Higgins . 2015 . The European fire ant (Hymenoptera: Formicidae) as an invasive species: impact on local ant species and other epigaeic arthropods . Can. Entomol . 147 : 592 – 601 . Google Scholar CrossRef Search ADS Ness , J. H. , M. A. Morales , E. Kenison , E. Leduc , P. Leipzig‐Scott , E. Rollinson , B. J. Swimm , and D. R. von Allmen . 2013 . Reciprocally beneficial interactions between introduced plants and ants are induced by the presence of a third introduced species . Oikos . 122 : 695 – 704 . Google Scholar CrossRef Search ADS Newbold , T . 2010 . Applications and limitations of museum data for conservation and ecology, with particular attention to species distribution models . Prog. Phys. Geog . 34 : 3 – 22 . Google Scholar CrossRef Search ADS Noordijk , J. , R. Morssinkhof , P. Boer , A. P. Schaffers , T. Heijerman , and K. V. Sýkora . 2008 . How ants find each other; temporal and spatial patterns in nuptial flights . Insect. Soc . 55 : 266 – 273 . Google Scholar CrossRef Search ADS O’Dowd , D. J. , P. T. Green , and P. S. Lake . 2003 . Invasional ‘meltdown’ on an oceanic island . Ecol. Lett . 6 : 812 – 817 . Google Scholar CrossRef Search ADS Ouellette , G. D. , F. A. Drummond , B. Choate , and E. Groden . 2010 . Ant diversity and distribution in Acadia National Park, Maine . Environ. Entomol . 39 : 1447 – 1456 . Google Scholar CrossRef Search ADS PubMed Pearson , B. , and A. R. Child . 1980 . The distribution of an esterase polymorphism in macrogynes and microgynes of Myrmica rubra Latreille . Evolution . 34 : 105 – 109 . Google Scholar CrossRef Search ADS PubMed Peterson , A. T . 2003 . Predicting the geography of species’ invasions via ecological niche modeling . Q. Rev. Biol . 78 : 419 – 433 . Google Scholar CrossRef Search ADS PubMed Peterson , A. T. , and Y. Nakazawa . 2008 . Environmental data sets matter in ecological niche modelling: an example with Solenopsis invicta and Solenopsis richteri . Global Ecol. Biogeogr . 17 : 135 – 144 . Phillips , S. J. , M. Dudík , J. Elith , C. H. Graham , A. Lehmann , J. Leathwick , and S. Ferrier . 2009 . Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data . Ecol. Appl . 19 : 181 – 197 . Google Scholar CrossRef Search ADS PubMed Pitt , J. P. , S. P. Worner , and A. V. Suarez . 2009 . Predicting Argentine ant spread over the heterogeneous landscape using a spatially explicit stochastic model . Ecol. Appl . 19 : 1176 – 1186 . Google Scholar CrossRef Search ADS PubMed Porter , S. D. , and D. A. Savignano . 1990 . Invasion of polygyne fire ants decimates native ants and disrupts arthropod community . Ecology . 71 : 2095 – 2106 . Google Scholar CrossRef Search ADS Potts , L. R. , O. F. Francke , and J. C. Cokendolpher . 1984 . Humidity preferences of four species of fire ants (Hymenoptera: Formicidae: Solenopsis) . Insect. Soc . 31 : 335 – 340 . Google Scholar CrossRef Search ADS Prior , K. M. , J. M. Robinson , S. A. Meadley Dunphy , and M. E. Frederickson . 2015 . Mutualism between co-introduced species facilitates invasion and alters plant community structure . Proc. Biol. Sci . 282 : 20142846 . Google Scholar CrossRef Search ADS PubMed R Core Team . 2016 . R: a language and environment for statistical computing. R version 3.3.2 . R Foundation for Statistical Computing, Vienna . Radchenko , A. , and G. W. Elmes . 2010 . Myrmica ants (Hymenoptera: Formicidae) of the Old World . Natura Optima Dux Foundation , Warrzawa, Poland . Reddy , S. , and L. M. Dávalos . 2003 . Geographical sampling bias and its implications for conservation priorities in Africa . J. Biogeogr . 30 : 1719 – 1727 . Google Scholar CrossRef Search ADS Roura-Pascual , N. , C. Hui , T. Ikeda , G. Leday , D. M. Richardson , S. Carpintero , X. Espadaler , C. Gómez , B. Guénard , S. Hartley , et al. 2011 . Relative roles of climatic suitability and anthropogenic influence in determining the pattern of spread in a global invader . Proc. Natl. Acad. Sci. USA 108 : 220 – 225 . Google Scholar CrossRef Search ADS Seppä , P. , and P. Pamilo . 1995 . Gene flow and population viscosity in Myrmica ants . Heredity . 74 : 200 – 209 . Google Scholar CrossRef Search ADS Sturtevant , A. H . 1931 . Ants collected on Cape Cod, Massachusetts . Psyche . 38 : 73 – 79 . Google Scholar CrossRef Search ADS Suarez , A. V. , D. A. Holway , and T. J. Case . 2001 . Patterns of spread in biological invasions dominated by long-distance jump dispersal: insights from Argentine ants . Proc. Natl. Acad. Sci. USA 98 : 1095 – 1100 . Google Scholar CrossRef Search ADS Talbot , M . 1943 . Response of the ant Prenolepis imparis Say to temperature and humidity changes . Ecology . 24 : 345 – 352 . Google Scholar CrossRef Search ADS Tartally , A. , V. Antonova , X. Espadaler , S. Csősz , and W. Czechowski . 2016 . Collapse of the invasive garden ant, Lasius neglectus, populations in four European countries . Biol. Invasions . 18 : 3127 – 3131 . Google Scholar CrossRef Search ADS Tschinkel , W. R . 1988 . Distribution of the fire ants Solenopsis invicta and S. geminata (Hymenoptera: Formicidae) in northern Florida in relation to habitat and disturbance . Ann. Entomol. Soc. Am . 81 : 76 – 81 . Google Scholar CrossRef Search ADS Tschinkel , W. R . 2006 . The fire ants . Harvard University Press , Cambridge, MA . Tschinkel , W. R. , and J. R. King . 2013 . The role of habitat in the persistence of fire ant populations . Plos One . 8 : e78580 . Google Scholar CrossRef Search ADS PubMed Vepsäläinen , K. , H. Ikonen , and M. J. Koivula . 2008 . The structure of ant assemblages in an urban area of Helsinki, southern Finland . Ann. Zool. Fenn . 45 : 109 – 127 . Google Scholar CrossRef Search ADS Verble-Pearson , R. , and S. Pearson . 2016 . European fire ant presence decreases native arboreal insect abundance in Acadia National Park, Maine, USA . Nat. Areas J . 36 : 162 – 165 . Google Scholar CrossRef Search ADS Walin , L. , P. Seppä , and L. Sundström . 2001 . Reproductive allocation within a polygyne, polydomous colony of the ant Myrmica rubra . Ecol. Entomol . 26 : 537 – 546 . Google Scholar CrossRef Search ADS Ward , P. S . 1987 . Distribution of the introduced Argentine ant (Iridomyrmex humilis) in natural habitats of the lower Sacramento Valley and its effects on the indigenous ant fauna . Hilgardia . 55 : 1 – 16 . Google Scholar CrossRef Search ADS Wetterer , J. K. , and A. G. Radchenko . 2011 . Worldwide spread of the ruby ant, Myrmica rubra (Hymenoptera: Formicidae) . Myrmecol. News 14 : 87 – 96 . Wheeler , W. M . 1906 . New ants from New England . Psyche . 13 : 38 – 41 . Google Scholar CrossRef Search ADS Wheeler , W. M . 1908 . A European ant (Myrmica levinodis) introduced into Massachusetts . J. Econ. Entomol . 1 : 337 – 339 . Google Scholar CrossRef Search ADS Zuur , A. , E. N. Ieno , N. Walker , A. A. Saveliev , and G. M. Smith . 2009 . Mixed effects models and extensions in ecology with R . Springer , New York . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of Entomological Society of America. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Environmental EntomologyOxford University Press

Published: Apr 5, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off