In 1994 we addressed four questions related to the rapid range expansion of the nine-banded armadillo (Dasypus novemcinctus Linnaeus, 1758) through the south-eastern and central United States since the species first crossed the Texas border from Mexico in the mid-1800s (Taulman & Robbins, 1996). Those questions were: (1) What factors may have contributed to preventing the population from entering the United States earlier? (2) How might the natural dispersal of the armadillo across the southern United States have been facilitated? (3) What was the current distribution of the armadillo in 1994? (4) What climatic extremes might be expected to limit the ultimate expansion of the species in the United States? That paper included a comprehensive survey of state wildlife agency biologists along and beyond the previously reported range boundary (Humphrey, 1974). We used the responses of those local wildlife professionals as evidence to support a new estimated distribution as of 1994. We concluded that the previously described annual precipitation level of 38 cm (Humphrey, 1974) was still a good estimate of a limiting factor to population expansion to the west.
However, we noted that armadillos had moved northwards beyond the previously predicted winter temperature limit of 9 annual freeze days in the Midwestern states and we suggested that areas with average daily January temperatures of ≤ −2 °C (about 24 annual freeze days) then appeared to describe the northern limit of the permanent armadillo population persistence in the northern parts of the range. We also documented the unification of the original Texas population with the previously distinct group that had expanded out of Florida after the introduction of a cohort there (Talmage & Buchanan, 1954), and the northward expansion of that new unified population into Alabama, Georgia, and as far as southern South Carolina.
In the intervening 19 years more data on the armadillo's range expansion have appeared in the literature, including several new records in New Mexico as recently as 2007 (Frey & Stuart, 2009), additional observations in Nebraska (Freeman & Genoways, 1998), an impressive collection of new sightings in southern and central Illinois (Hofmann, 2009), and new records beyond the previous population boundary in Tennessee (Eichler & Gaudin, 2011). We believed that it might be useful to conduct a new survey of state and academic biologists and conservation officers in order to document any changes to the armadillo's range and to consider whether previous predictions of climatic extremes that might pose limits to the species' distribution still appeared to be reasonable. This paper presents the results of that survey combined with observations reported in the recent literature, a new analysis of the current established range of the armadillo in the United States, and our predictions of factors that may define ultimate distributional limits of the armadillo population based on the most recent 30-year summary climate data.
We sent out surveys by email and postal mail to state biologists and conservation officers, as well as to academic biologists at colleges and universities in 17 US states along and beyond the range boundary of the armadillo as described in our last study (Taulman & Robbins, 1996). The survey enquired about the date and specific location of any observations of live and dead armadillos, ages of any specimens observed, and whether the local biologist believed that there was a persistent, breeding population in the area of his/her familiarity. We also searched the literature for any new information on armadillo sightings in the United States since our last review in 1994.
We mapped the position of each reported sighting along and beyond the previous 1994 population boundary line (Taulman & Robbins, 1996). We categorized records into three temporal groups: sightings within the past year (since summer 2012), records from 1 to 10 years old (2003–2012), and historical sightings before 2003. Sometimes reports indicated the presence or absence of armadillos in entire counties regularly traversed by a respondent. Using all available records of armadillos gleaned from the literature, and the sightings reported to us by local observers, along with their personal opinions about the status of the armadillo population in their areas of familiarity, we created an estimated boundary of the present contiguous range of an established, breeding population of armadillos in the United States.
In order to create an updated predictive map of climatic factors that might limit the future distribution of armadillos we collected recent summaries of temperature and precipitation data in the form of published online compilations that were not available in 1994. We acquired a database of US annual precipitation because dispersal to the west generally appears to depend on a minimum average rainfall and the soil, vegetation, and the invertebrate community that level of precipitation supports (Humphrey, 1974; Taulman & Robbins, 1996). We also used a database of average minimum daily January temperatures to assess distributional limits at the northern range boundary because the successful survival of the armadillo population in the northern parts of the range seems to be limited by winter temperature extremes (Humphrey, 1974; Taulman & Robbins, 1996). Both databases were created by Oregon State University's PRISM Climate Group, incorporate climate data averaged over the period 1981–2010, and are available from the University's Northwest Alliance for Computational Science & Engineering [2013; Matt Doggett, Senior Meteorologist, Northwest Alliance for Computation Science and Engineering (NACSE), pers. comm.].
We received 153 responses to our postal and email surveys. The current range boundary represents our best estimate of the established breeding range of the armadillo, based on local respondents' observations and knowledge of the wildlife occurring in particular local areas in the 17 states included in our survey, and the sightings and records derived from published literature (for a review of literature on armadillos sightings since 1994, see Loughry & McDonough, 2013) (Fig. 1). Individual records are shown within 50 km inside of the proposed range boundary as well as all records received of sightings and areas of non-occurrence outside of that line.
With regard to the many records of armadillos that occur outside of our estimated current population boundary (Fig. 1), we believe those may be occasional individuals that have dispersed from the established population or have been transported into areas beyond the permanent range. Many respondents and publications suggested that armadillos observed beyond our proposed range boundary may have hitched rides on agricultural vehicles and then dropped off at a location remote from their origin or may have been transported to those sites intentionally and then escaped or been released [E. Corey, North Carolina Parks and Recreation (pers. comm.) in northern North Carolina; R. Munkel, Iowa Department of Natural Resources (pers. comm.) in Iowa; Kennedy & Jones (2006) in Texas; Stuart et al. (2007) in New Mexico; Hofmann (2009) in central and northern Illinois; Platt et al. (2009) in South Dakota; and P. Holahan, University of Wisconsin Zoological Museum (pers. comm.) in Wisconsin]. Whether those remote sightings represent natural dispersal or human-assisted transport, we do not believe that the isolated observations of armadillos shown beyond our proposed established range boundary represent a permanent breeding population in those areas. We believe that the reports by many biologists who have never seen an armadillo in the same counties where we also received isolated incidental observations from others, support our contention that armadillos are not currently established in those areas.
Results of the current survey indicate that the range of the armadillo to the west in Texas and Oklahoma has remained stationary since our last analysis in 1994 (Taulman & Robbins, 1996). The range of the armadillo has expanded to the north over the past 19 years, however. As a result of new sightings in central and southern Illinois since 2008, J. Hofmann now believes that armadillos are established in the southern part of the state (J. Hofmann, Illinois Natural History Survey, pers. comm.). We agree that the density and abundance of armadillo observations in Illinois south of about 39°30′ N latitude probably represents an established population in the southern part of the state as of 2013. Our conclusion that the permanent range of armadillos has expanded into other areas indicated in Fig. 1 is based on observations provided by personal communications from 153 responding biologists, as well as from publications appearing after 1994.
We estimate that the current established population of armadillos extends to the west in Texas and Oklahoma to about the isopleth representing an annual precipitation of 50 cm (Fig. 2). The previous annual precipitation isopleth at 38 cm may represent a level of habitat suitability in which armadillos can survive temporarily. We would suggest considering a zone of precipitation between 40 and 50 cm as a marginal sink habitat (Pulliam, 1988) where pioneers or released individuals may survive temporarily, primarily along watercourses, but where a permanent sustaining population probably will not be able to become established (Fig. 2).
The primary climatic factor limiting the armadillo's range to the north-west appears to transition from annual precipitation in south-western Kansas to winter temperature minima in west-central Kansas. We previously predicted that the armadillo's range expansion northwards in the United States would be limited to areas with average daily January temperatures greater than −2 °C (Taulman & Robbins, 1996). Because the coldest winter temperatures in a region determine the potential survivability for armadillos and their young (McNab, 1980), we now suggest estimating the northern range limit using average minimum daily January temperatures. The current species range appears to extend to about the −8 °C isopleth in Kansas (Fig. 3). However, farther eastwards, the current established range does not reach the −8 °C isopleth. On the eastern seaboard, armadillos have not yet reached even the regions where average January minimum temperatures are −4 °C (Fig. 3).
The diversity of responses from Nebraska, Iowa and North Carolina, with scattered observations over many years by some biologists and other respondents reporting no sightings from the same regions, leads us to conclude that armadillos are not established in these states. These historical and recent records in the Midwestern states, up to a latitude where average January minimum temperatures are about −10 °C, may indicate a zone between −8 and −10 °C which represents another marginal sink region, based on winter temperature minima in this case. Armadillos occurring there, whether pioneering dispersers or transported, may be able to survive temporarily during warm seasons but may not be able to establish a persistent breeding population until climate change shifts the −8 °C isopleth northwards (United States Global Climate Research Program, 2009). Although we have predicted distributional limits based on amount of precipitation and average winter temperatures, observations made by the second author (L.W.R.) suggest that sub-freezing temperatures do not limit foraging if forest leaf litter is available; rather it is persistent snow or ice cover that limits armadillo foraging opportunities and eventual starvation. It is assumed that average winter temperatures are associated with the persistence of frozen precipitation.
The rather slow dispersal of armadillos up the east coast in South Carolina since our last survey suggests that some other factors besides winter temperature extremes are contributing to limit population expansion there. It is interesting to note that the range of the extinct armadillo, Dasypus bellus, was also limited to areas very near our current range boundary estimate in Missouri, Illinois and Indiana, and that range boundary also dropped farther south along the east coast, just extending past the northern border of North Carolina at its farthest extent (Klippel & Parmalee, 1984, as cited in Layne, 2003).