2010. Proceedings of the Academy of Science 119(1):87–94

DISTRIBUTION OF THE PLAINS POCKET ( BURSARIUS) IN THE GRASSLAND PHYSIOGRAPHIC REGIONS OF INDIANA

Vanessa S. Quinn: Department of Biology/Chemistry, Purdue University-North Central, Westville, Indiana 46391 USA Chia-Chun Tsai and Patrick A. Zollner: Department of Forestry and Natural Resources, Purdue University, West Lafayette

ABSTRACT. We conducted surveys for plains pocket (Geomys bursarius) in nine counties in central and northern Indiana. The estimated mean (6 1 SD) population from road surveys was 512 6 440 individuals which is lower than previous surveys. Gophers were not found in Warren and Tippecanoe counties where they were historically present and showed a decline in Benton and northern Jasper counties. The decline in northern Jasper County may have been in part due to the timing and intensity of our survey work. Northern Newton County shows mounding activity presumably indicating healthy populations. More than a dozen clusters of mounds were observed along the eastern edge of White County in an area where mounding activity was not previously detected. Finally, we analyzed the spatial pattern of gopher mounds detected during this and three previous surveys in Indiana to delineate eight putative Indiana subpopulations. The geographic extent of these subpopulations incorporates 404 km of roads within suitable habitat. Keywords: Density, distribution, Geomys bursarius, , program DISTANCE

The plains pocket gophers (Geomys bursar- ways and line-transects on selected public lands ius) is a of special concern in Indiana with suitable habitat; (2) to determine burrow- due to a limited distribution in grassland ing activity using the sign-count inventory physiographic regions of northwest Indiana method (Reid et al. 1966); and (3) to combine (Fig. 1; Heaney & Timm 1983; Thorne 1989). data on observed mound locations from all This species was likely distributed throughout gophers surveys within Indiana (Conaway northwest Indiana prior to 1900 (Evermann & 1947; Tuszynski 1971; Thorne 1989) to define Butler 1894). Three surveys conducted since putative subpopulations based upon geograph- 1947 have shown declining populations. Con- ic proximity of mounds. away (1947) conducted a limited survey show- ing discontinuous and scattered Indiana popu- METHODS lations within the range of pre-settlement Study area.—We searched portions of nine populations. The second, and more thorough, Indiana counties within the Iroquois Till Plains survey found gophers in several northwestern region of northwestern Indiana (Gray 2000; Indiana counties with the majority of popula- Benton, Carroll, Cass, Jasper, Newton, Pulaski, tions located in Jasper and Newton counties Tippecanoe, Warren and White counties; (Tuszynski 1971). The most recent survey Fig. 1) for gopher mounding activity from identified population distribution as declining May–August 2008. The grassland physiograph- relative to previous surveys (Thorne 1989). We ic regions of Indiana are the only portions of report results of a survey conducted in the the state where gophers have been located summer of 2008 in the grassland physiographic (Conaway 1947; Tuszynski 1971; Thorne 1989). regions of northwest Indiana in addition to Within this region we searched areas occurring those sites where gophers were previously within potentially suitable habitat (cover class- identified. Our objectives were (1) to determine es of 21 {open}, 31 {barren}, 52 {shrub}, 71 the current distribution of gophers in Indiana {grassland/herbaceous}, or 81 {pasture/hay}) using surveys of roadside and railroad right-of- from USGS Indiana landcover data (USGS 2002). These designations corresponded to Correspondence: e-mail: [email protected] vegetative conditions in which gophers were

87 88 PROCEEDINGS OF THE INDIANA ACADEMY OF SCIENCE

Figure 1.—Location of Iroquois Till Plains region of northwestern Indiana (inset) is the shaded area. The potential plains pocket gopher habitat is indicated by dark shading within the Iroquois Till Plains region. QUINN ET AL.—PLAINS POCKET GOPHER 89 documented to occur during the 1988 survey survey was completed). The number of mounds (Thorne 1989). observed was recorded, and the perpendicular Road survey.—During May–June 2008 we distance from the foot transects to each mound surveyed roadsides located within suitable was recorded in meters. cover-types in the grassland physiographic Data analyses.—Distance sampling analysis: region of northwestern Indiana (Fig. 1) for Data on mound occurrence from both roadside signs of above-ground activity indicated by surveys and walking transects were geo-refer- mounding (Andersen 1988; Jones et al. 1983, enced. We used the program DISTANCE 5.0 1985). Mounds are visible from a vehicle and (Burnham et al. 1980; Buckland et al. 2000; have a unique shape and texture (Thorne 1989). Thomas et al. 2002) to estimate mound density Fresh mounds can be distinguished by their (accounting for decreasing detection probabil- characteristic triangular, or deltaic, shape with ity as a function of distance from observer) for a round patch of soil located at the apex roadside and foot transect surveys. DIS- (Macdonald 2006). Examination of the USGS TANCE estimated density based on the detec- cover types (USGS 2002) within Iroquois Till tion probability and the effective strip width. Plain physiographic province indicated approx- The effective strip width is the distance from imately 1341 km of targeted survey roads the transect at which the number of within the Iroquios till plain (see Study area detected beyond this point exactly equals the above) indicating suitable habitat. We searched number of animals missed within this distance the right of way and surrounding habitat for (Buckland et al. 2000). We fit data with mounds on all road segments within our study different combinations of four types of key area that were not on primary high speed roads functions and three types of series expansion (greater than 55 mph). Searches were conduct- with no constraints on monotonocity. We first ed by a passenger while the driver maintained a used the complete data set of all mounds (n 5 speed of 30–40 km/hr. Once mounds were 544) for analysis. None of the models had a sighted a TrimbleH GeoXTTM handheld GPS good fit to the distribution as indicated by a unit from the GeoExplorerH 2004 series (ETrim- high P-value (P , 0.00001; Zar 2009). We ble Navigation Limited) was used to georefer- pooled the data by grouping observations on ence the position of the vehicle with a precision the same transect as a cluster (n 5 135). We less than one meter. A laser range finder also truncated our data set by deleting all (Bushnell Yardage Pro Sport 450, in yards) or observations closer than three yards (n 5 3) as a measuring tape (in feet and inches) was used to well as the 10% of observations furthest from measure distance from the center of the road to the transect (n 5 14; suggested by Swann et al. the center of the mounds. We recorded date, 2004). The pooled data were analyzed using distance, name of the road, land use type, and ‘‘cluster’’ option in DISTANCE and gained took photos of mounds to confirm any ques- better fits (P 5 0.79). After selecting the best tionable mounding activity (i.e., mounds made combination of key functions and series expan- by gophers, moles, woodchucks, or coyotes). sions, we compared densities of clusters and Mounds without the characteristic shape (Mac- densities of mounds estimated by different donald 2006) were omitted. types of size-biased regression methods provid- Transect survey.—We implemented an addi- ed by DISTANCE. tional 30 walking transects within polygons of Subpopulation definition: First, we estimated suitable habitat on public lands. Within these gopher maximum dispersal distance as the public lands we selected the 30 largest polygons square root of home range size multiplied by (ranging from 0.7–0.9 km2) of suitable habitat 40 (Bowman et al. 2002). We used the upper from the USGS landcover which indicated bound of spring and summer average home these public lands contained polygons of range size in (66 6 23 m2; Zinnel potentially suitable habitat ranging in size from 1992) and found a maximum dispersal distance 0.2 6 0.1 km2. Within each of the 30 selected of 378 m. polygons transects were located by intersecting Next, we combined all known mound loca- the centroid of the polygon with a random tions from all survey periods into a single point along its perimeter. The total length of coverage (Fig. 2). Using this coverage we the 30 transects was 16 km (0.5 6 0.2 km). We calculated the distance (299 6 539 m) to the walked each transect once in July (after road nearest occurrence of another mound in any 90 PROCEEDINGS OF THE INDIANA ACADEMY OF SCIENCE

Figure 2.—Location of all plains pocket gopher mounds observed within the Iroquois Till Plains region and northern Newton County while conducting road, transect, and historic location surveys during the summer of 2008. other survey for each observed mound. We surveys. Where these buffers overlapped (indi- calculated the mean 6 one standard deviation cating a neighboring mound nearer than (944 6 655 m) distance to the nearest mound 2254 m) and included at least three distinct for those mounds further than estimated mound observations we classified those dispersal distance from their nearest neighbor. mounds as belonging to a putative subpopula- We used the upper 95% confidence interval tion of gophers in Indiana. Finally, we took all (2254 m) of this latter distribution to create a of the mound locations defining putative buffer around each mound location from all subpopulations and generated the minimum QUINN ET AL.—PLAINS POCKET GOPHER 91 convex polygon that encompassed all of those multiplied this density estimate by the area of points. We buffered that polygon by 2 km potential habitat and estimated 512 6 440 (approximate estimated distances between geo- gophers in the surveyed area. graphically distinct clusters of mounds) to We surveyed 31 walking transects rather define our putative subpopulations of gophers than the intended 30 because one transect in in Indiana (Fig. 2). Jasper-Pulaski Fish and Wildlife Areas was divided by a ditch we could not cross. We RESULTS located 67 mounds and removed the top 15% of Road and transect survey.—We drove a total the data. The detection functions did not differ of 2345 km of roads during 30 working days of significantly. The D Akaike’s Information roadside surveys. Considering the effective Criterion (AIC) values with different combina- distance from the road at which we could detect tions of series expansion ranged from 0 to 4.11. mounds we searched a total area of 4988 km2.In The estimated mound density provided by the addition to driving roads throughout the Iro- best model (Hazard-rate hermite polynomial) quois Till Plain we included Willow Slough Fish was 2052 6 889 mounds/km2. This mound and Wildlife Area, Jasper-Pulaski Fish and density acquired from walking transect survey Wildlife Area, , Prophetstown converts to 0.2 6 0.1 individuals/hectare. State Park, and the northern half of Newton Historic location survey.—Among the 193 County. We searched a total of 512 km2 of historic locations (Thorne 1989) 36 of the sites potential habitat for gophers (10% of the contained mounds, 117 had no mounding surveyed area). Within this area we located 611 activity, and 40 sites could not be located due mounds, 544 mounds during road survey, and to errors in the reported projections. In northern 67 from the walking transect survey. Jasper County only 6 of the 68 historic sites we The analysis from DISTANCE showed the surveyed contained mounds. By contrast, 30 of best combinations of key function and series 70 historic locations surveyed in Newton County expansion for the roadside surveys were uni- still contained mounds. form and cosine. The goodness of fit test was Subpopulation definition.—Our analysis de- 0.79 when the cluster estimation method was lineated eight subpopulations (Fig. 2) ranging regressing ln(cluster size) against perpendicular in size from 28 to 1443 km2 (mean 5 252 6 distance of mounds (model fit 5 0.71). The 484 km2). The total area of land types suitable estimated density of mound clusters was 26 6 for gophers within these subpopulations was 21 clusters per km2. The effective strip width of 160 km2. This is smaller than the 512 km2 of detection was 20 inches (0.5 m). The average potentially suitable habitat across the total and estimated cluster sizes were four mounds surveyed area. Within these subpopulations per cluster. Combining these values provides an there was a total of 404 km of roadside habitat estimated density of 94 6 79 mounds per km2 passing through suitable habitat while the in our survey area. larger study area encompassed 1342 km of Sparks and Andersen (1988) measured the roads through similar habitat. average mound production rates in Willow Slough Fish and Wildlife Area. From their DISCUSSION calculations we used a mean rate of 19 mounds Consistent with results of previous surveys observed per gopher during June (four mounds/ (Tuszynski 1971; Thorne 1989) our survey day) to convert our estimate of density of documents a decreasing geographic distribution mound clusters to density of individual go- of gophers in Indiana which presumably phers. Since mound persistence was primarily a corresponds to a decline in the overall abun- function of mowing we estimated mounds dance within Indiana. Our survey shows the would be detectable for 21 days following their historic populations in Tippecanoe and Warren creation (three week mowing cycle; pers. comm. counties are absent. Furthermore, the two Kanakee Sands staff). Thus, to estimate the locations in Benton County mark a dramatic number of gophers present we divided the decrease from the 15 locations where Thorne density of mounds (D) by the number mounds (1989) observed gophers. However, it should be produced in 21 days (92 mounds). We convert- noted that Tuszynski (1971) observed mounds ed the mound density from one mound in at five locations in Benton County. Thus, it is 21 days to 0.01 6 0.01 individual/hectare. We unclear if some of the apparent fluctuations in 92 PROCEEDINGS OF THE INDIANA ACADEMY OF SCIENCE populations in Indiana are attributable to around Kankakee Sands and Willow Slough changes in gopher distribution or variation in are areas with the highest species densities and probability of detection across the various are critical for the gopher status in Indiana. surveys. This uncertainty is reinforced by our The combination of this observation and the observation of numerous clusters of mounds in apparent decline of populations in northern eastern White County near the Carroll County Jasper County suggest a need to consider the line. The occurrence of gophers in this vicinity availability of routes for dispersal and re- was not noted by either Tuszynski (1971) or colonization between the northern Newton Thorne (1989). Given the limited vagility and County populations and the more western dispersal range of this fossorial species it is portions of the northern Jasper County popu- unlikely that these new gopher populations lations where gophers once occurred. resulted from dispersal. Thus, the parsimonious Variation in detection probability has been explanation is that considerable variation in alluded to as a mechanism possibly underlying detection probability exists when using the several patterns we observed. When interpret- presence of mounding activity to assess the ing our results it is important to recall that we occurrence of gophers. A consequence of this sampled using a distance-based technique to variation is that apparent species absences in estimate densities. Distance-based estimates mound-based surveys may reflect a failure to incorporate decaying probability of detection detect the presence of mounding activity at the as a function of the distance between the time of sampling rather than the true absence observer and the mound in estimating densities. of the species from a locality. Thus, our surveys should have been more Our results suggest the status of gophers in explicit about the extent of the area surveyed northern Jasper County may be an area of relative to the historic work; and, in turn, our considerable conservation concern for the estimated densities are much lower than those Indiana population of this species. Only 9% developed by Thorne (1989) and Tuszynski (6/68) of the historic locations we revisited in (1971). Another of our results that merits this northern Jasper County contained mounds consideration is the higher density estimates while 43% (30/70) of the historic locations we obtained from our foot transect data relative to revisited in northern Newton County contained the roadside surveys. It is intuitive that an mounds. Some of this difference is undoubtedly observer walking on the ground will conduct a attributable to methodological differences. The much more thorough inventory than one while historic sites in northern Jasper County were driving at 30 km per hour. Our use of the the last field surveys we conducted in July, and distance-based analysis criteria elucidates the mounding activity is known to decrease as the implications of the increased sampling efficiency summer progresses. Detection probabilities for the resulting density estimates. Thus, future also decline with taller vegetation later in the surveys might consider expanded use of sampling summer. Based on this variation we expect from walking transects in selected strata of the some of the apparent declines in northern sampling area to decrease bias in the estimates. Jasper County may not reflect as dramatic a Previous research has demonstrated the use of decline in gopher status. Furthermore, the distance-based sampling techniques to estimate historic locations in northern Jasper and the density of many biological phenomena Newton counties were north of the Iroquois (Thomas et al. 2002), and this approach holds Till Plain ecotype where we implemented our great promise for estimating mound densities of comprehensive roadside surveys. Nonetheless, numerous species of gophers. the trend that only six of 68 historic locations in In concurrence with surveys conducted northern Jasper County continued to contain roughly 20 (Thorne 1989) and 40 years ago mounds is of concern. The two remaining (Tuszynski 1971), our results suggest that the historic mound locations occurred in roadside geographic distribution and presumably abun- habitat. Based upon the cover types where we dance is declining. However, these conclusions observed mounds throughout our entire survey, are based upon surveys of mounds; and our it is clear that roadside habitats are critical for results also suggest that inference regarding gophers throughout Indiana. abundance based upon mound surveys may be Our surveys suggest northern Newton Coun- problematic. Our use of distance-based analy- ty still contains large populations. The areas ses undoubtedly provided more realistic treat- QUINN ET AL.—PLAINS POCKET GOPHER 93 ment of mound detection probability and thus serve as hypothetically distinct populations and effective area surveyed and ultimately mound those hypotheses should be assessed using density. However, the use of the mound density genetics estimates of the relatedness of individ- as a surrogate for gopher density can be uals from these populations. problematic for a number of reasons (Reid et al. 1966; Bandoli 1987). The frequency of ACKNOWLEDGMENTS mounding activity may have more to do with The assistance of Valerie Clarkston and Ben local soil and vegetative conditions than with Gillum was critical in completing the field actual density so the use of this sampling work. Bob Porch and the land managers at technique for anything beyond presence and Willow Slough, Jasper-Pulaski, and the Na- absence data is likely to be challenging. Thus, ture’s Conservancy at Kankakee Sands provid- future studies actual densities from trapping ed invaluable advice and expert local knowl- surveys should be paired with concurrent edge. This project was funded by the Indiana mound surveys to discern the feasibility of Department of Natural Resources, Division of mound-based indices for insights into the status Fish and Wildlife through the State Wildlife of gophers in Indiana. Such live trapping of Grant Program and public contributions to the gophers could provide genetic samples which Indiana Nongame Wildlife Fund. Finally, the could be used to estimate the connectivity of Department of Forestry and Natural Resources remnant populations. These connectivity values at Purdue University and the Department of could provide insights into the likelihood that Biology and Chemistry at Purdue North human land uses (e.g., agricultural fields or Central both provided invaluable support for major highways) are serving as barriers to the the students and faculty involved in this movements of gophers within Indiana. Finally, research. We thank Dr. John O. Whitaker, our work suggests that roadside strips of one anonymous reviewer, and Dr. James Berry vegetation continue to be a critical source of for helpful and insightful reviews of previous remnant habitat for this species in Indiana. We versions of this manuscript. do not believe that any particular vegetative type (e.g., prairie grass restoration) was critical. LITERATURE CITED Instead the narrow strips of remnant grasslands Andersen, D.C. 1988. Tunnel-construction methods of any type simply represent the majority of the and foraging path of a fossorial herbivore, remaining habitat and thus are invaluable for Geomys bursarius. Journal of Mammalogy 69: that reason. 565–582. Future roadside mound surveys in Indiana Bandoli, J.H. 1987. Activity and plural occupancy should be conducted within boundaries of the of burrows in Botta’s pocket gopher Thomomys putative subpopulations defined in our results bottae. American Midland Naturalist 118:10– 14. rather than across the entire Iroquois Till Bowman, J., A.G. Jochen & L. Fahrig. 2002. Plains eco-region. Such refinements in search Dispersal distance of is proportional effort will dramatically reduce the effort to home range size. Ecology 83:2049–2055. necessary to conduct roadside surveys from Buckland, S.T., I.B.J. Goudie & D.L. Borchers. 1342 km of road driven to 404 km driven. Such 2000. Wildlife population assessment: Past devel- reduction in total search effort will make it opments and future directions. Biometrics feasible to conduct multiple searches during 56:1–12. annual surveys, greatly reducing the chance of Burnham, K.P., D.R. Anderson & J.L. Laake. 1980. failing to detect gophers where they are present. Estimation of density from line transect sampling Furthermore, such intensive sequences of sur- of biological populations. Wildlife Monographs veys within these focal areas where they have No. 72. 202 pp. historically been found in Indiana can be used Conaway, C.H. 1947. The life history and ecology of the pocket gopher in Indiana. B.S. thesis, Purdue to construct a history of when each subpopu- University, West Lafayette, Indiana. 35 pp. lation occurred which could be used to develop Evermann, B.W. & A.W. Butler. 1894. Preliminary estimates of the probability or persistence of list of Indiana mammals. Proceedings of the Indiana’s population. Finally, these putative Indiana Academy of Science 3:124–139. subpopulations, which were determined based Gray, H.H. 2000. Physiographic divisions of In- upon the spatial pattern of observed mounds diana, Special Report 61. Indiana Geological across three surveys and 38 years, can also Survey. 94 PROCEEDINGS OF THE INDIANA ACADEMY OF SCIENCE

Heaney, L.R. & R.M. Timm. 1983. Relationships of Thomas, L., S.T. Buckland, K.P. Burnham, D.R. pocket gophers of the Geomys from the Anderson, J.L. Laake, D.L. Borchers & S. Strind- Central and Northern . University of berg. 2002. Distance sampling. Pp. 544–552, In Museum of Natural History. Lawrence, Encyclopedia of Environmetrics (A.H. El-Shaarawi, Kansas. 59 pp. W.W. Piegorsch & W. Piegorsch, eds.). John Wiley Jones, J.K., Jr, D.M. Armstrong, R.S. Hoffman & C. and Sons, Boston. Jones. 1983. Mammals of the Northern Great Plains. Thorne, D.H. 1989. Plains pocket gopher distribu- University of Press. Lincoln. 422 pp. tion and nutritional status in Indiana. Indiana Jones, J.K., Jr, D.M. Armstong & J.R. Choate. 1985. Department of Natural Resources. 37 pp. Guide to the Mammals of the Plains States. Tuszynski, R.C. 1971. The ecology of the pocket University of Nebraska Press, Lincoln. 371 pp. gopher (Geomys bursarius illinoensis) in Indiana. Macdonald, D.W. 2006. The Encyclopedia of M.S. thesis, Purdue University, West Lafayette, Mammals. Pp. 216–219. Oxford University Press. Indiana. 63 pp. London. U.S. Geological Survey. 2002. Land cover in Reid, V.H., R.M. Hansen & A.L. Ward. 1966. Indiana, derived from the USGS, Version 06-03- Counting mounds and earth plugs to census 99 (United States Geological Survey, 30-Meter mountain pocket gophers. Journal of Wildlife Grid). Publisher: USGS EROS Data Center. Sioux Falls, . Management 30:327–334. th Sparks, D.W. & D.C. Andersen. 1988. The relation- Zar, J.H. 2009. Biostatistical Analysis. 5 Ed. ship between habitat quality and mound building Prentice Hall. New York. 960 pp. by a fossorial , Geomys bursarius. Journal Zinnel, K.C. 1992. Behavior of free-ranging pocket Geomys bursarius of Mammalogy 69:583–587. gophers ( ). Ph.D. thesis. Univer- sity of Minnesota, Minneapolis, Minnesota. Swann, D.E., C.C. Hass, D.C. Dalton & S.A. Wolf. 2004. Infrared-triggered cameras for detecting wildlife: An evaluation and review. Wildlife Manuscript received 11 November 2009, revised 18 Society Bulletin 32:357–365. June 2010.