Are road kills valid indicators of armadillo population structure?
Inferring the population structure of asocial animals can be problematic. Individuals tend to be widely scattered and secretive so that obtaining an accurate picture of population demography is a time-consuming and difficult process. For example, the nine-banded armadillo (Dasypus novemcinctus) is a relatively asocial crepuscular mammal that inhabits richly vegetated areas throughout the southern United States (Newman, 1913; Kalmbach, 1943; Taber, 1945; Fitch et al., 1952; Talmage and Buchanan, 1954; Humphrey, 1974; Galbreath, 1982; McBee and Baker, 1982; Wetzel, 1985; McDonough, 1992). Despite their present ubiquity, little is known about armadillo population structure. This is probably due in part to the difficulty in locating and capturing individuals within their preferred habitats. Such information is of interest for determining the population attributes that have made armadillos such successful invaders (Newman, 1913; Humphrey, 1974; McBee and Baker, 1982; Wetzel, 1985), and because there is almost no detailed information about the population biology of this little known species.
Road-killed armadillos are a conspicuous and abundant sight along roadways in the southern United States. In most cases, the animals are intact enough to determine the age, sex and reproductive condition of the individual. Such specimens represent an easier and quicker means of collecting data on population structure than by catching individuals (see Case, 1978). However, the crucial assumption is that road-killed individuals are representative of the population as a whole. We tested this assumption by comparing data on population structure derived from road-killed armadillos with those obtained from an adjacent live-caught population.
We collected road-killed armadillos during June through August of 1992-1994 at the Tall Timbers Research Station, located near Tallahassee, Florida. The station is bisected by county highway 12. Road kills were collected along this road, between the eastern and western boundaries of the station (approximately 5 km). During the same time, we also captured live armadillos on the Tall Timbers property as part of a long-term project on armadillo behavior and ecology (e.g., Loughry and McDonough, 1994).
Live armadillos were captured using a large dip net attached to a 2-2.5 m pole (McDonough, 1992, 1994; Loughry and McDonough, 1994). Once captured, all animals were sexed and weighed to the nearest 0.1 kg with a spring balance. The reproductive condition of females was categorized as definitely lactating if the nipples were large and protruding, definitely not lactating if the nipples were barely visible, and possibly lactating if the nipple size was intermediate between the first two categories. In addition, all animals were measured (in cm) to obtain data on overall body size: (1) tail base was the circumference of the tail at its juncture with the pelvic shield of the dorsal carapace; (2) tail length was the length of the tail from the base to the tip (animals lacking a portion of the tail were not measured); (3) front carapace was the length of the anterior edge of the scapular shield of the carapace (i.e., at the juncture with the head); (4) front band was the length of the posterior edge of the scapular shield; and (5) back band was the length of the ninth, most posterior, movable band. These last three measurements were taken from the bottom right to bottom left side of the animal. When first captured, animals were classified by weight (McDonough, 1992, 1994) as juveniles (young of the year [less than]2.5 kg), yearlings (2.5-3.5 kg), or adults ([greater than]3.5 kg). Weight reliably separates adults from juveniles, but our classification of yearlings was conservative because some known-age yearlings weighed nearly as much as adults (see below, and McDonough, 1992).
TABLE 1. - Results of simple and multiple regressions between armadillo body size measurements and weight for live animals at Tall Timbers
Measurement r Partial F
Front carapace 0.95 1.13 Front band 0.97 19.48(**) Back band 0.97 3.94(*) Tail base 0.88 0.94 Tail length 0.76 3.51
* P [less than] 0.05; ** P [less than] 0.01; for simple regressions all r-values were highly significant (P [less than] 0.0001, df = 1,209)
Similar data were obtained, where possible, for all road-killed armadillos. The one exception concerned age. Road-killed animals were often much lighter than expected because of the loss of some body parts. Thus, we could not use weight as a reliable indicator of age. However, a regression using the Statview statistical package (Abacus Concepts, 1992), revealed that, among living members of the Tall Timbers population, body size measurements were each highly correlated with weight (Table 1). In addition, each measure was significantly different among age groups (Table 2). Because body size measurements are presumably correlated with one another, we used multiple regression to identify the more important correlates with weight. This analysis was also highly significant (multiple r = 0.98, P [less than] 0.0001, df = 5, 178), and indicated that length of the front band was the most important correlate (Table 1). Thus, for road-killed individuals not previously captured, we used length of the front band to assign them to a particular age category (n = 25/27 that had not been caught while alive). Road kills were assigned to an age category if their front band length was within 1 SD of the mean for that age group. There was some potential for misclassifying yearlings as adults (and vice versa) by this procedure because the distributions of adults and yearlings exhibited some overlap (see Table 2). In practice this was not a problem because none of our road kills fell on the borderline. Because of damage due to impact with automobiles, we were not always able to reliably measure front band length. In these cases, we used whatever body size measurements were available to assign individuals to a [TABULAR DATA FOR TABLE 2 OMITTED] particular age class. This procedure seems justified because of the high correlations of each measure with body weight (Table 1) and the significant differences between age classes in each measure (Table 2).
The validity of our analyses hinges on the assumption that road-killed and live-caught armadillos were sampled with equal effort. Sampling of live armadillos at Tall Timbers involved intensive searches of particular areas on foot and surveys by car along roads on the property for 4-7 h per night (from approximately 1600-2300 h; total field time = 1650 person-hours over 147 days from 1992-1994). Because of the long-term nature of our study, the cumulative record of live captures represents a nearly complete census of all the individuals in the population. We did not spend nearly as much time searching for road kills, but we believe our sample of road kills is a complete record of all animals hit by automobiles during the summers at Tall Timbers. In order to reach the areas where we were attempting to catch live armadillos, we had to pass along the highway that bisects the property, once at ca. 1600 h and again at approximately 2200 h. Thus, each of the 147 days that we sampled for living armadillos, we also sampled for road kills. In addition, we passed back and forth along the road on days we did not go out in the field and station personnel also saved carcasses for us. This extensive sampling makes it unlikely that we missed any road kills or that carcasses were removed by scavengers before we sampled them.
When two populations are compared, it is difficult to know which population is responsible for generating any differences between the two. For example, differences between our road-killed and live-caught populations could be due to some feature of the road-killed group, or it could be that our live-caught population is different in some respect. We compared data from the live-caught Tall Timbers population with data on live-caught armadillos from the Hagerman National Wildlife Refuge, Pottsboro, Texas (collected in 1990-1991, see McDonough and Loughry, 1995) and the Rob and Bessie Welder Wildlife Refuge, Sinton, Texas (obtained from 1987-1991, see McDonough, 1992, 1994). We assumed that if the demographies of the live-caught populations did not differ, then any differences between our live-caught and road-killed populations were probably due to features of the road-killed group.
For statistical analyses we calculated the proportion of each population made up of each age and sex category. These proportions were compared between the road-killed and living populations by contingency table analyses (Fisher's exact and chi-square tests), using the Instat statistical package. Tests of statistical significance employed a P value of 0.05.
TABLE 3. - Demography of live-caught and road-killed populations of nine-banded armadillos in Florida (Tall Timbers) and Texas
Population Males Females Total
Tall Timbers - Live-caught
Juveniles 18 38 56 Yearlings 3 1 4 Adults 66 60 126 Total 87 99 186
Tall Timbers - Road kills
Juveniles 1 1 2 Yearlings 0 2 2 Adults 13 9 23(*) Total 14 12 27
Welder Wildlife Refuge, Texas
Juveniles 36 59 111(*) Yearlings 8 3 11 Adults 116 111 227 Total 160 173 349
Hagerman National Wildlife Refuge, Texas
Juveniles 4 13 17 Yearlings 0 0 0 Adults 18 17 35 Total 22 30 52
* Additional individuals are included because in some cases we were able to determine the age, but not the sex, of an animal
Comparisons between the road-killed and live-caught Tall Timbers populations revealed no significant differences in sex ratios (overall or for each age group separately, Fisher's test, P [greater than] 0.54 in all comparisons, Table 3) or in the proportion of females in each reproductive condition (chi-square, P = 0.754, df = 2, Table 4). The primary difference between road-killed and living populations of armadillos was in age structure. Only two juveniles (7.4%) were collected as road kills, but juveniles were much more common in the live-caught population (chi-square = 7.80, P = 0.022, df = 2, Table 3).
Comparisons of the live-caught Tall Timbers population with two live-caught populations from Texas revealed no differences in age structure (chi-square = 2.22, P = 0.69, df = 4) or sex ratios (adults only, juveniles only, or for all ages combined, chi-square, P [greater than] 0.45, df = 2 in all cases, see Table 3). However, there was a slight trend within all three populations for the sex ratio of juveniles to be female-biased (Fisher's tests: Tall Timbers, P = 0.08; Welder, P = 0.14; Hagerman, P = 0.28, Table 3).
TABLE 4. - Reproductive condition of living and road-killed adult female armadillos at Tall Timbers. Numbers of females in each reproductive condition do not necessarily add up to the total number of females because some individuals were inadvertently not checked. Total number of living females does not equal total number of females in Table 3 because some females were lactating in 1 yr and not lactating in another and thus are included twice here but only once in Table 3
Total Definitely Possibly Not Population females lactating lactating lactating
Live-caught 70 27 11 24 Road-kills 9 3 2 2
Our analyses indicate that road kills may be used for inferring the demography of adult armadillos, but they provide a misleading picture of age structure. We are unaware of any similar studies that have evaluated the utility of road kills for inferring population structure. However, Rolley and Lehman (1992) have shown that population densities of road-killed raccoons (Procyon lotor) were not reliable predictors of harvest densities. Thus, their study supports our own in suggesting that data from road kills be evaluated cautiously. While not concerned with road kills, studies comparing the demography of hunter-killed vs. live-trapped populations also have found significant differences. For example, Kane and Litvaitis (1992) found that a live-trapped population of black bears (Ursus americanus) contained fewer females and was older than a hunter-killed group. It seems intuitively likely that hunter-killed populations would differ from their live-trapped counterparts because of the preference of hunters for particular types of individuals. Road kills, on the other hand, might represent a more random sample of the individuals present. Our data suggest otherwise and indicate that adult armadillos are more at risk from automobiles than are juveniles.
Adult male and female armadillos were equally likely to be killed by automobiles. There are several lines of evidence that make this result unsurprising. First, based on the results from our live-caught population, males and females were equally abundant at Tall Timbers. Thus, both sexes were presumably equally likely to encounter cars. Second, males and females have similar home range sizes (Clark, 1951; Fitch et. al., 1952; Jacobs, 1979; Galbreath, 1982; Breece and Dusi, 1985; McDonough, 1992), so they would be equally likely to overlap with roadways. Finally, all our road kills were collected during the summer, which is the peak of the breeding season (McDonough, 1992). Mating involves males maintaining proximity to females (Gause, 1980; McDonough, 1992). It may be that males and females suffer automotive casualties at equal rates because this mortality occurs while they are paired. However, this last point may be unlikely since we rarely collected two road kills in the same place on the same night.
The fact that three live-caught armadillo populations from different locations had similar demographies suggests that differences between the road-killed and live-caught Tall Timbers populations were due to the make-up of the road-killed group. The only significant difference was the virtual absence of juvenile road kills, probably due to the smaller home ranges of juveniles (Breece and Dusi, 1985; McDonough, 1992). Juvenile armadillos appear to remain close to their natal burrow throughout much of their 1st summer (pers. observ.; see also Loughry and McDonough, 1994), so it is unlikely that they would range far enough to be at risk from automobiles. As an alternative explanation, assuming juveniles encounter cars as frequently as do adults, juveniles might still be less likely to be killed because they are more vigilant and flee from approaching predators sooner than do adults (McDonough and Loughry, 1995).
There are two general features of the demography of our live-caught populations that are worth mention. First, in all populations, the number of yearlings observed was quite small, suggesting that recruitment into the natal population is limited. At present, it is unknown whether low recruitment might be due to high juvenile mortality, long-distance dispersal of juveniles, or some other effect. However, an alternative explanation is that there were more yearlings in the population, but they were misclassified as adults. Most live-caught yearlings were animals that had been caught as juveniles, so we knew their age. When captured as yearlings, some of these individuals were nearly adult-sized. Thus, it is possible that some large animals that we caught for the first time were mistakenly classified as adults instead of as yearlings. While this problem does generate difficulties in evaluating recruitment, it does not affect our population comparisons because all populations were affected by this bias in the same way. Additionally, in both the Welder and Tall Timbers populations, in which sampling was long-term and extensive, few yearlings were ever captured. This suggests that juvenile recruitment is indeed quite low.
A second point is that all three live-trapped populations had a mildly female-biased juvenile sex ratio. At present, there are a number of hypotheses that might explain this pattern (review in Clutton-Brock and lason, 1986). We currently do not have the necessary data to evaluate the different alternatives, but differential mortality of male juveniles is suggested by the fact that the number of male and female litters did not differ significantly from parity in populations where such data were available (e.g., Tall Timbers: 19 female, 12 male litters, Fisher's test, P = 0.61). In any case, the consistency of the pattern suggests that there may be something worth investigating.
An important caveat to our findings is that all our data on road kills come from the summer months. It is possible that with a year-round sampling regime road kills would provide a more accurate picture of armadillo demography. In particular, the underrepresentation of juveniles might be eliminated by sampling road kills through the autumn and winter, when these individuals may disperse from their natal burrows (pers. observ.). However, armadillos may be less likely to be killed by cars at these times of year because they are less active and shift their activity to daylight hours (Layne and Glover, 1978). More data will be needed to evaluate these alternatives.
Acknowledgments. - We thank the staffs of the Rob and Bessie Welder Wildlife Foundation, Hagerman National Wildlife Refuge and the Tall Timbers Research Station for their help and support. We especially thank F. Sanders of Tall Timbers who collected some of the road kills for us. This work was funded by a Welder Wildlife Foundation Graduate Research Fellowship and grants from Sigma Xi and the Theodore Roosevelt Memorial Fund to C.M.M., and faculty research awards from Valdosta State University and a grant from the American Philosophical Society to W.J.L. We thank D. Bogyo for allowing us to use his Instat statistical package and B. J. Bergstrom for help in tracking down some important references. This is contribution number 453 of the Welder Wildlife Foundation.
ABACUS CONCEPTS. 1992. Statview. Abacus Concepts, Inc., Berkeley, California 466 p.
BREECE, G. A. AND J. L. DUSI. 1985. Food habits and home ranges of the common long-nosed armadillo Dasypus novemcinctus in Alabama, p. 419-427. In: G. G. Montgomery (ed.). The evolution and ecology of armadillos, sloths, and vermilinguas. Smithsonian Institution Press, Washington, D.C.
CASE, R. M. 1978. Interstate highway road-killed animals: a data source for biologists. Wild. Soc. Bull., 6:8-13.
CLARK, W. K. 1951. Ecological life history of the armadillo in the eastern plateau region. Am. Midl. Nat., 46:337-358.
CLUTTON-BROCK, T. H. AND E. R. IASON. 1986. Sex ratio variation in mammals. Q. Rev. Biol., 61:339-374.
FITCH, H. S., P. GOODRUM AND C. NEWMAN. 1952. The armadillo in the southeastern U.S. J. Mammal., 33:21-37.
GALBREATH, G. J. 1982. Armadillo, p. 71-79, In: J. A. Chapman and G. A. Feldhamer (eds.). Wild mammals of North America. Johns Hopkins University Press, Baltimore.
GAUSE, G. E. 1980. Physiological and morphometric responses of the nine-banded armadillo (Dasypus novemcinctus) to environmental factors. Ph.D. Dissertation, University of Florida, Gainesville 110 p.
HUMPHREY, S. R. 1974. Zoogeography of the nine-banded armadillo in the United States. BIOSCIENCE, 24:457-462.
JACOBS, J. 1979. Behavior and space usage patterns of the nine-banded armadillo (Dasypus novemcinctus) in southwestern Mississippi. M.S. Thesis, Cornell University, Ithaca, New York. 132 p.
KALMBACH, E. R. 1943. The armadillo: its relation to agriculture and game. Texas Game, Fish and Oyster Commission, Austin. 61 p.
KANE, D. M. AND J. A. LITVAITIS. 1992. Age and sex composition of live-captured and hunter-killed samples of black bears. J. Mammal., 73:215-217.
LAYNE, J. N. AND D. GLOVER. 1978. Activity cycles of the nine-banded armadillo (Dasypus novemcinctus) in southern Florida, p. 140. In: R. Obrtel, G. Folk and J. Pellantova (eds.). Abstracts of papers from the second International Theriological Congress. Institute of Vertebrate Zoology, Brno.
LOUGHRY, W. J. AND C. M. MCDONOUGH. 1994. Scent discrimination by infant nine-banded armadillos. J. Mammal., 75:1033-1039.
MCBEE, K. AND R. J. BAKER. 1982. Dasypus novemcinctus. Mamm. Species, 162:1-9.
MCDONOUGH, C. M. 1992. Behavioral ecology of nine-banded armadillos in a riverine habitat. Ph.D. Dissertation, University of California, Davis. 101 p.
-----. 1994. Determinants of aggression in nine-banded armadillos. J. Mammal., 75:189-198.
----- W. J. LOUGHRY. 1995. Influences on vigilance in nine-banded armadillos. Ethology, 100: 50-60.
NEWMAN, H. H. 1913. The natural history of the nine-banded armadillo of Texas. Am. Nat., 47:513-539.
ROLLEY, R. E. AND L. E. LEHMAN. 1992. Relationships among raccoon road-kill surveys, harvests, and traffic. Wild., Soc. Bull., 20:313-318.
TABER, F. W. 1945. Contributions on the life history and ecology of the nine-banded armadillo. J. Mammal., 26:211-226.
TALMAGE, R. V. AND G. D. BUCHANAN. 1954. The armadillo (Dasypus novemcinctus). A review of its natural history, ecology, anatomy and reproductive physiology. Rice Inst. Pamphlet. Mono. Biol., 4:1-135.
WETZEL, R. M. 1985. Taxonomy and distribution of armadillos, Dasypodidae, p. 23-46. In: G. G. Montgomery (ed.). The evolution and ecology of armadillos, sloths, and vermilinguas. Smithsonian Institution Press, Washington, D.C.
|Printer friendly Cite/link Email Feedback|
|Author:||Loughry, W.J.; McDonough, Colleen M.|
|Publication:||The American Midland Naturalist|
|Date:||Jan 1, 1996|
|Previous Article:||Growth and physical development of captive-raised black-footed ferrets (Mustela nigripes).|
|Next Article:||Movements and habitat use of subadult alligator snapping turtles (Macroclemys temminckii) in Louisiana.|