Comparisons Between Nine-Banded Armadillo (Dasypus Novemcinctus) Populations in Brazil and the United States

Rev. BioI. Trop. 46(4): 1173-1183,1998 www.ucr.ac.cr www.ots.ac.cr www.ots.duke.edu

Comparisons between nine-banded armadillo (Dasypus novemcinctus) populations in Brazil and the United States

W. J. Loughry and Colleen M. McDonough

Department of Biology, Valdosta State University, Valdosta, GA 31698-0015, U.S.A. Fax: 912-333-7389; email: [email protected]

Received 21-1V-1998. Corrected 1O-Vl11-1998. Aceptado 20-Vl11-1998.

Abstract: We compared characteristics of a population of nine-banded armadillos (Dasypus novemcinctus) studied in the southern United States with a population found in the Atlantic coastal rainforest of Brazil. Adult armadillos in Brazil weighed less than those in the U.S., but when weight was accounted for, did not differ in other measures of body size. However, juveniles in the U.S. were proportionately bigger than those in Brazil. Armadillos in Brazil were less abundant (numbers sighted per h of observation) and were active later at night than those in the U.S. Adult sex-ratios were male-biased in both populations: Finally, there was no significant difference in the incidence of litterrnate associations observed in the two populations, but groups of juveniles (which included non-litterrnates) were observed more frequently in the U.S. Many of these differences may be due to the fact that armadillos are hunted extensively in Brazil but not in the United States.

Key words: Dasypus novemcinctus, armadillos, Brazil, United States, population differences.

Many charactersitics of animal Breece and Dusi 1985, Redford 1985, Sikes et populations vary intraspe~ifically, presumably at. 1990, Lippert 1994). Females give birth to as the result of adaptation to conditions that litters of genetically-identical quadruplets in vary geographically (Lott 1991, Foster and the spring (Newman and Patterson 1910, Endler in press). Comparisons of populations Patterson 1913, Storrs and Williams 1968, inhabiting different areas can yield insights Prodohl et at. 1996), with litters first emerging into the evolutionary forces generating from their natal burrows from late spring differences between populations. In addition, through the summer (Loughry and McDonough the extent of intraspecific variation provides 1994). Juveniles (young of the year) remain in information on how generalizable data from close proximity, foraging together and sharing one population are to populations in other parts the same burrows through some of their first of a species' range. summer (Taber 1945, Galbreath 1982, Nine-banded armadillos (Dasypus McDonough and Loughry 1995, Prodohl et at. novemcinctus) are found from northern 1996), but litters appear to break up (due to Argentina to the southern United States dispersal or mortality) by the fall (McDonough (Humphrey 1974, Wetzel1982, 1985,Taulman and Loughry unpublished data). and Robbins 1996). They are relatively asocial, The above account. is potentially burrowing mammals (Newman 1913, problematic because almost all of this Galbreath 1982, McBee and Baker 1982), that information is derived from studies performed are mostly active at night (McDonough and in the northern-most part of the species' range Loughry 1997a), and feed primarily on insects (i.e., in the United States; reviews in Kalmbach (Kalmbach 1943,Clark 1951,Fitch et at. 1952, 1943, Taber 1945, Talmage and Buchanan 1174 REVISTA DE BIOLOGIA TROPICAL

1954, Galbreath 1982, McBee and Baker the analyses reported here we pooled data over 1982). Nine-banded armadillos have only the entire four year period for comparison with recently colonized the U.S. (Humphrey 1974, those collected in Brazil. While this makes the Taulman and Robbins 1996), so data from U.S. data set large in comparison with that from these studies may not be representative of Brazil, it seems appropriate because we are populations in more ancient parts of the interested in average differences between the species' range, living in the kinds of two populations, rather than transient conditions under which the species differencesdue to yearly variation. However,to presumably evolved. Comparisons with non- insure that pooling was appropriate, we North American populations of D. performed separate comparisons of each year's novemcinctus are required to determine if this data from Tall Timbers with the Brazilian data is a legitimate concern. In this study, we and obtained exactly the same pattern of results. addressed this issue by collecting data on a Data collection followed procedures population of nine-banded armadillos located developed during the Florida study in northern Florida (Loughry and (McDonough and Loughry 1997a, Loughry McDonough 1996, McDonough and Loughry and McDonough 1998). In both populations, 1997a) and, using the same methods, a we censused the property during both daylight population located in the Atlantic coastal and evening hours (200 days for a total of 2273 rainforest of Brazil. These two data sets person-hours of field time in the U.S., in provide an opportunity to examine the Brazil, 108 days and 958 person-hours). A question of how similar armadillos are in daily census typically lasted 4-6 h and was these two widely separated locales. While conducted by walking or driving along trails or our data are not an exhaustive account of the roads on each property. Spotlights and miner's population attributes of armadillos in either lamps were used to observe animals after dark. location, they allow at least a preliminary The total linear distance censused was assessment of the extent of intraspecific approximately 25 km at each site. A rotating variation that can occur in D. novemcinctus. schedule of observations was used, such that a portion of the entire sampling area was MATERIALSAND METHODS censused each night, followed by another portion the next night, and so on until the Data on Brazilian armadillos were entire area had been sampled. This schedule collected from 20 January-l June, 1996, at the was then repeated for the duration of the field Po~o das Antas Biological Reserve (5500 ha), season. Both sites were comprised of a variety located approximately 70 km north of Rio de of habitat types (in Brazil: rainforest, swamp, Janiero. The reserve contains one of the largest grassland and disturbed habitats, see Dietz et remaining fragments of Atlantic coastal al. 1997, Loughry and McDonough 1997, in rainforest and is the site of conservation efforts the U. S.: hardwood hammocks, wetlands, to preserve the golden-lion tamarin fields and upland pine areas, Brennan et al. (Leontopithecus rosalia, Kleiman et al. 1986, 1998). Visibility to the sides of roads and trails Dietz et al. 1997). Data for the u.S. come from varied considerably depending on habitat type. a four year (1992-1995) study of armadillos While we did not attempt to measure located on the Tall Timbers Research Station visibilities systematically in the two locales, it (1600 ha), near Tallahassee, Florida (see was our impression that they were relatively Loughry and McDonough 1996, McDonough similar overall. Thus, the total area sampled at and Loughry 1997a). There was some year-to- each site was roughly the same. year variation in the characteristics of the We attempted to capture armadillos Florida population (Loughry and McDonough observed during censuses with large dip nets 1996,McDonough and Loughry 1997a), but in attached to a 1.5 m pole. In Brazil, we also LOUGHRY & McDONOUGH: Comparisons between annadillo populations 1175 used live-traps placed in the mouths of active juvenile D. novemcinctusappear very similar to burrows. Once caught, animals were sexed, adult D. septemcinctus. Thus, our data on weighed, and marked for long-range juvenile abundances and the timing of activity identification by gluing unique colors and may be slightly biased by the inclusion of some shapes of reflective tape on various parts of the data from D. septemcinctus. This is unlikely to carapace. A small tissue sample for genetic be a large problem because D. septemcinctus studies was taken from one or both ears using was relatively rare on the reserve and typically an ear-notcher. These notches were taken from found in differenthabitatsfromD. novemcinctus different locations on the ears of different (Loughry and McDonough 1997). In addition, individuals, further contributing to most observations of juvenile D. novemcinctus identification of particular animals. Animals were from animals that had been captured, were markedfor permanent identification using marked and subsequentlyresighted. ear tags and, in the U.S., with passive Using the data collected by these methods, transponder (PIT) tags. Five body size we compared five characteristics of our two measurements (all in cm) were obtained from populations. (1) We compared morphology by all captured animals: (1) tail base = the analyzing differences in body weight and body circumference of the tail at itsjuncture with the size measurements with t-tests. Body size body near the pelvic shield of the dorsal measurements are correlated with body weight carapace; (2) tail length =the length of the tail (Loughry and McDonough 1996), so we from the base to the tip (animals who were performed a second analysis of body size missing a portion of the tail were not differences using an ANCOVA in which body measured); (3) front carapace = the length of weight was the covariate. For both sets of the anterior edge of the scapular shield of the analyses, we averaged data for individuals carapace (i.e., at thejuncture with the head); (4) with multiple measurements (except where the front band =the length of the posterior edge of animal changed age groups, e.g., data for an the scapular shield; and (5) back band = the individual first caught as a juvenile then length of the most posterior movable band. subsequently as an adult were not averaged but Animals were assigned to age categories based treated as independent points). Although on weight (McDonough 1992, 1994, Loughry armadillos may exhibit some sexual and McDonough 1996) as juveniles (young of dimorphism (McDonough 1992), we pooled the year, < 2.5 kg), yearlings (2.5-3.5 kg), or data for males and females in these analyses adults ( > 3.5 kg). Because of the smaller size because we found little evidence of sexual of Brazilianarmadillos (see below), individuals dimorphism in our populations and because there were only classified as adults (2:2.5 kg) or the small number of females caught in Brazil juveniles « 2.5 kg). Consequently, we omitted (see below) did not permit statistical the yearling age class of U.S. animals from all examination of body size for each sex of the analyses reported here. Finally, the time separately. (2) We compared abundances by of day (in local DST) at which the animal was calculating the number of individuals (adults, first seen was recorded. juveniles, and total animals) observed per h of During our censuses we observed more observation in each locale. These data were armadillosthan we were able to catch. For these further subdivided into the number of animals individuals we noted the time at which the observed during daylight hours (8:00-16:00 animalwas sighted and its age (aseitherjuvenile local DST) and at night (16:00-24:00 local or adult on the basis of body size). If the DST; cf. McDonough. and Loughry 1997a). genitalia were visible we were able to classify Comparisons between the two populations the individualby sex as well. The seven-banded were made using t-tests. (3) We further armadillo,D. septemcinctus,also occurs at P~o examined activity by using t-tests to compare das Antas (Loughry and McDonough 1997)and the times of day when individuals were fITst 1176 REVISTA DE BIOLOGIA TROPICAL sighted in the two populations (cf. McDonough littermate associations. In contrast, we often and Loughry 19~7a). These data included observed such groups in the U.S. To evaluate sightings of animals that were subsequently this difference we determined, for each captured as well as animals that were seen but population, the number of litters in which a not caught.All times were in local DST. There juvenile was observed with at least one other is a potential problem with this analysis in that littermate. Multiple observations of the same we observed armadillos in Brazil during the litter were ignored so that if a litter was austral summer and on through the fall, when observed as an association at least once, it was hours of daylight were diminishing (all data for classified as exhibiting association. Only litters the U.S. come from June-August). Thus, it is in which a juvenile was never observed with a possible we might find no difference between sibling were classified as singletons. We then U.S. and Brazilian armadillos in the time of day compared the proportion of singletons between at which they were observed, but Brazilian the two populations using Fisher's exact test. armadillos would have been active after dark This analysis used data from litters identified whereas U.S. animals would have been active with microsatellite DNA markers (Prod6hl et during daylight. In an attempt to eliminate this at. 1996). We also performed a second analysis problem, we performed a second analysis in in which we examined the number of times which we subtractedthe time of sunset from the juveniles in each population were observed time at which an animal was observed. Thus, alone versus in proximity to another juvenile animals with positive values were observed (regardless of whether this otherjuvenile was a after sunset and those with negative values littermate or not). Again, we pooled data from before.sunset. Data on the timing of sunset at the same individuals so that if a juvenile was each site were obtained from the Astronomical observed associating with another juvenile at Almanac (U.S. Naval Observatory 1995). (4) least once, it was classified as exhibiting Sex-ratios of adults, juveniles and the entire association. The incidence of singletons population were compared between the two observed in each population was then sites using data from captured animals. compared with Fisher's exact test. Because litters consist of identical quadruplets A final note concerns comparisons (Newman and Patterson 1910, Newman 1913, between juveniles at the two sites. Such Storrs and Williams 1968, Prod6hl et at. 1996), comparisons are potentially problematic if it may be more appropriate to examine sex- juveniles are not of similar age. Forexample, ratios of litters rather than of individual larger body size or the absence of littermate juveniles. In our analyses we compared associations could be due to juveniles. at one juvenile sex-ratios using both numbers of male site being older than juveniles at the other site. and female litters and numbers of male and There are several lines of evidence that sugge!;t female individuals. Littermates were identified this was not the case. First, field assistants using microsatellite DNA markers (Prod6hl et involved in the golden-lion tamarin project at at. 1996). Sex-ratios within populations were Po<;o das Antas reported first observing compared using Chi-square tests and between juvenile armadillos at the beginning of the populations with Fisher's exact tests. (5) austral summer (late November to early Juvenile armadillos tend to .associate (e.g., December), which is similar to the timing of forage together) with their littermates for some emergence from natal burrows at Tall Timbers time after they first emerge from their natal (i.e., at the beginning of summer in Mayor burrows (Loughry and McDonough 1994, early June; Loughry and McDonough 1994). McDonough and Loughry 1995, Prod6hl et at. These assistants are in the reserve from dawn 1996, Loughry and McDonough 1998). to dusk six days a week each week of the year. Although we caught a number of juveniles in As such, they are the individuals most likely to Brazil (see below), we rarely observed encounter juvenile armadillos and thus, their LOUGHRY & McDONOUGH: Comparisons between armadillo populations 1177 observations can be considered reliable. In covariance with body weight as the.covariate addition, the timing of juvenile emergence at showed that, for a given body weight, adult Po~o das Antas is consistent with data armadillos in the u.S. and Brazil did.not differ documenting the timing of births for a in any body size measurement (all p > 0.05). population in Paraguay. In Paraguay, most However, juveniles in the U.S.. had births occurred in September or early October proportionately larger front band (p =0.02) and (K. Hill pers. comm.), which would lead to front carapace lengths (p = 0.045) than did juvenile emergence in November or December. Brazilian juveniles, even when variation in Thus, we conclude that the majority ofjuveniles body weight was accounted for. in our two populations were of similar age. Abundance: The total number of RESULTS armadillos observed per h of observation was significantly greater in the U.S. than in Brazil Body size: Both adult and juvenile nine- whether one compared the number of animals banded armadillosin the U.S. were significantly sighted over the entire daily sampling period,

TABLE 1

Mean (:!:SE) body size measurements of nine-banded armadillos in Brazil and the United States

Poo das Antas Tall Timbers

Measure Juveniles Adults Juveniles Adults Weight (kg) 1.16 (0.14) 3.13 (0.08) 1.30 (0.05) 4.11 (0.03) Front carapace (em) 14.26 (0.58) 19.97 (0.22) 15.78 (0.20) 21.64 (0.07) Front band (em) 21.31 (1.02) 31.00 (0.35) 22.90 (0.32) 34.64 (0.11) Back band (em) 21.85 (1.16) 33.02 (0.30) 23.16 (0.37) 36.50 (0.13) Tail base (em) 9.39 (0.50) 13.66 (0.16) 10.70 (0.2)) 15.48 (0.05) Tail length (em) 23.18 (1.10) 27.96 (0.59) 27.36 (0.39) 32.91 (0.16) n 13 21 104 201 larger than their Brazilian counterparts (Table only during the daytime, or only at night I). Adults were significantly different in every (Table 2). This same pattern was found when measurement (ANOYA,all p ~ 0.0001); while data for adults were analyzed separately (Table juveniles differed in front carapace length (p = 2). Juveniles were also more abundant in the 0.04) and tail length (p =0.0012). Analysis of U.S. than in Brazil when the data included the TABLE2

Average number of armadillos observed per h of observation (:!:SE) in Brazil and the United States. Sample sizes are the number of days of observation. Statistical comparisonsare within each row with *p < 0.05, **p < 0.01, ***p < 0.001 Poo das Antas Tall Timbers All hours of observation All animals 0.41 (0.05) 1.05 (0.06) 7.33*** Juveniles 0.10 (0.02) 0.23 (0.02) 3.86*** Adults 0.29(0.05) 0.82 (0.05) 6.47*** n 108 200 Day only (8:00-16:00) All animals 0.16 (0.11) 0.71 (0.12) 2.75** Juveniles 0.15(0.11) 0.39 (0.09) 1.45 Adults 0.01 (0.01) 0.33 (0.08) 2.36* n 33 85 Night only (16:00-24:00) All animals 0.48 (0.06) 1.16 (0.06) 7.41*** Juveniles 0.13 (0.03) 0.22 (0.02) 2.50* Adults 0.33 (0.05) 0.94 (0.06) 7.17*** n 100 190 1178 REVISTA DE BIOLOGIA TROPICAL entire day or nighttime only (Table 2). not be determined, they are excluded from However, juvenile abundances did not differ Table 4). In the U.S., the overall sex-ratio was during daytime (Table 2). not significantly biased (Chi-square =0.46, p> Activity: The time of day when 0.25), but was male-biased for adults (Chi- armadillos were first observed was square = 5.21, p < 0.025) and female-biased significantly later in Brazil than in the U.S. in for juveniles (Chi-square = 4.20, P < 0.05; TABLE3

Average time of first sighting (:t SE) for nine-banded armadillos in Brazil and the United States. Sample sizes for adults and juveniles in each population do not add up to the sample size for all animals because in some cases we observed individuals briefly and could not reliably age them. These animals are included in the all animals category. but excluded from the age group data. Statistical comparisons are within each row with * p < 0.05. ** p < 0.01. *** P < 0.001

Poo das Antas Tall Timbers All animals Time of day 20.78 (0.18) 19.31 (0.07) 6.99*** Hours from sunset 1.79 (0.19) -0.71 (0.07) 11.73*** n 154 1155 Juveniles only Time of day 19.23 (0.39) 17.61 (0.19) 3.04** Hours from sunset -0.22 (0.45) -2.40 (0.19) 3.99*** n 35 265 Adults only Time of day' 20.99 (0.16) 19.,82 (0.07) 5.60*** Hours from sunset 2.21 (0.17) -0.21 (0.07) 11.46*** n 101 883 comparisons using data from all animals, and Table 4). The sex-ratio bias for juveniles in when adults and juveniles were analyzed the U.S. disappeared when the number of male separately (Table 3). Analyzing the timing of and female litters were compared rather than activity relative to sunset showed again that, in the numberof individuals(Chi-square=0.71, all cases, Brazilian armadillos w~re active later P > 0.25). The overall sex-ratio differed than those in the U.S. (Table 3). between the two populations (Fisher's test, p = Sex-ratios: There was evidence of sex- 0.03), but not when adults and juveniles were ratio biases within each population (Table 4). examined separately (for juveniles this In Brazil, the overall population sex-ratio was included comparisons of numbers of male and female litters as well as numbers of TABLE4 individuals, all p > 0.07).

Sex-ratios of adult and juvenile armadillos in Brazil Associations among juveniles: In Brazil and the United States. For juveniles, the number we observed a littermate association in only of litters are indicated parenthetically one of 15 litters (6.7%). Littermate associations occurred in 18 of 69 litters Po~o das Antas Tall Timbers (26.1%) in the U.S., but the difference in the Group Males Females Males Females All animals 25 10 163 151 proportion of litters observed as singletons in Adults only 16 6 121 88 the two populations was not significant Juveniles only 9 (9) 4 (4) 42 (31) 63 (38) (Fisher's test, p = 0.17). However, juveniles did associate with one another more frequently significantly male-biased (Chi-square =6.43, P in the U.S. when associations with non- < 0.025)as was the sex-ratiofor adults(Chi- littermates were included. Groups ofjuveniles square =4.55, P < 0.05; Table 4), but not for were observed in 27 of 56 cases (48%) in the juveniles (Chi-square = 1.92, P > 0.10; two U.S., but just once in 15 instances in Brazil. juveniles were found dead and their sex could This difference in the proportion of juveniles LOUGHRY & McDONOUGH: Comparisons between armadillo populations 1179 observed as singletons was significant between and the U. S. may be no more extensive than the two populations (Fisher's test, p = 0.003). the differences between various U. S. populations. At present, we cannot completely DISCUSSION address this issue because, with the exception of body size, there are no published data on the Our results demonstrate that substantial variables we measured for any other D. differences do exist between at least one novemcinctus population. Adult body weights populationof D. novemcinctusin Brazil and one of D. novemcinctus range from 3-7 kg (McBee in the United States. Armadillos at Po~o das and Baker 1982). However, consistent with the Antas were less abundant, smaller, and active findings of the present study,populations found later at night than armadillos at TallTimbers. In north of the Amazon appear to weigh more than addition, the Brazilian population exhibited a those found south (Wetzel and Mondolfi 1979, strongly male-biased sex-ratio and a lack of Wetzel 1982,Schaller 1983, Redford 1994,but association among juveniles. Two valid see Hill et at. 1997). Such differences occur in criticisms of this study concern the relatively other aspects of morphology as well. For small sample size from Brazil, particularly for example, South American populations of D. comparisons of body sizes and sex-ratios, and novemcinctus often have only eight movable the fact that all the Brazilian data were derived bands whereas populations in the U.S. from a single field season. With regard to normally have nine or more (Wetzel and sample size, we would point out that our data Mondolfi 1979,Wetzel 1982, 1985). represent the largest reported sample from a Setting aside the issue of how Brazilian population of D. novemcinctus (cf. representative our two populations are of Wetzeland Mondolfi 1979,Wetzel 1982,Carter North and South American populations in and Encarna~ao 1983, Schaller 1983, Carter general, it still remains to explain why our two 1985,Redford 1994). In addition, although our populations were so different from one Brazilian data come from a single field season, another. One possible explanation concerns this field season lasted twice as long as one of the fact that humans hunt armadillos our field seasons in the U. S. (> 100days versus extensively (albeit, illegally) at Po~odas Antas ca. 50 days), so our sampling effort in Brazil (C. Ruiz-Miranda pers. comm.) but not at Tall representsthe equivalent of over 2 years of data Timbers. Although we had no problems in collection in the U. S. Finally, as mentioned detecting armadillos at a distance in Brazil, it earlier, when we compared our Brazilian data was much more difficult to approach active separately with each year of data from Tall animals in order to catch them. Almost Timbers, the same pattern of differences invariably, our approach was detected before emerged (note that these analyses reversed the we were close enough to use our nets and, problem such that the Brazilian data set was upon detection, the animals immediately ran now twice as large as that from the U. S.). Thus, into the forest. At TallTimbers, mortality from we believe our data provide an adequate basis human hunters is almost non-existent (pers. for evaluating potential variation between the observ.) and we rarely have problems with two populations. animals fleeing from us before we can attempt Another important issue concerns how to capture them. While we did not attempt to representative these two populations are of do so, it would be interesting to compare levels North and South American populations in of vigilance in the two populations (e.g., general. North American populations of D. McDonough and Loughry 1995). We predict novemcinctus can vary dramatically, e.g., in that Brazilian armadillos will be more vigilant sizes of home ranges and in the incidence of than those in the U.S., thus enabling them to aggression (review in McDonough 1992). detect potential predators sooner. The Thus, the differences we report between Brazil influence of human hunting could account for 1180 REVISTA DE BIOLOGIA TROPICAL

many of the differences between the two All other xenarthra at Po~odas Antas were rare populations. For example, hunting could (Loughry and McDonough 1997), so it seems generate the lower abundances (see also Hill et improbable that competition among these al. 1997) and later timing of activity exhibited species is severe. D. novemcinctus is among by Brazilianarmadillos, and might contributeto the largest of the armadillos present (Loughry the lack of associations among juveniles and McDonough 1997) and thus it seems observed in Brazil either directly by removing likely that the other species would show more young or indirectly by promoting the break-up negative effects of competition with nine- of litters to avoid detection by human banded armadillos than vice-versa. Data on "predators". However,the impact of huntingon habitat preferences suggest that each species Brazilianarmadillo populations will be difficult utilizes different parts of the reserve, to evaluate because, as hunting is illegal, minimizing opportunities for competitive hunters are unwilling to supply data on hunting interactions (Loughry and McDonough 1997), successor the demographyof killed individuals. and D. novemcinctus occupies what are Hunting by humans is the most presumably the most 'favorable sites (i.e., in conspicuous difference in sources of mortality primary forest, near the rich feeding grounds between our two populations, but it is possible of the swamps; Loughry and McDonough that differences in the types of non-human 1997), so they would seem least likely to be predators present at each site also contribute to impacted by the presence of other xenarthra. the population differences we report. However, Extensive hunting of armadillos in Brazil this seems unlikely to be a major influence. may explain some of the observed differences There are few large predators present at either between U.S. and Brazil populations, but it site capable of killing an adult armadillo (pers. seems unlikely to explain differences in body observ.). Juvenile mortality due to predation size and sex-ratiobiases. Hunting could explain can be quite high in the U. S. (McDonough and body size differences if hunters preferentially Loughry 1997b) and we have no evidence to took only large animals (but see Abrams and suggest that it is significantly higher in Brazil. Rowe 1996), but it appears that most hunting Nonetheless, additional data will be required to involves setting traps in burrows or across detennine the potential importance of non- heavily used trails (c. Ruiz-Miranda pers. human predators in generating the differences comm.), so large individuals would be unlikely between our two populations. to be selectively taken. Bergmann's rule (e.g., One alternative to hunting as an James 1970) is also an unlikely explanation. explanation for the differences between Tall Timbers is further north of the equator Brazilian and U.S. armadillos is that, in the (latitude= 30° 40') than Po~odasAntas is south U.S., D. novemcinctus is the only xenarthran (Iatittude = 22°31') so a latitudinalgradientin present, whereas in Brazil there are more body size might seem plausible. However, species of armadillos (Cabassous unicinctus Wetzeland Mondolfi (1979) report data fromD. and Dasypus septemcinctus; Euphractus novemcinctus populations in Venezuela which sexcinctus used to occur but may now be are as large as the animals in Florida. Resultsof locally extinct; Loughry and McDonough analysis of covariance showed that U.S. 1997) as well as the southern collared anteater juveniles were proportionately bigger than their (Tamandua tetradactylus; Loughry and Brazilian counterparts when variation in body McDonough 1997). Thus, competitive weight was accounted for, suggestingthat larger interactions with other xenarthra in Brazil adult size is accomplished by increased growth might lead to lower abundances as well as as juveniles. Whether this represents a genetic alterations in the timing of activity so as to change in developmental growth patterns or avoid overlap with these other species. This merely a response to increased resource explanation seems unlikely for several reasons. availability is unknown, LOUGHRY & McDONOUGH: Comparisons between armadillo populations 1181

Sex-ratios of adults in both Brazil and the site confounds any comparisons based on other U.S. were male-biased. The skew found in differences. An appropriate comparison would Brazil could be due to small sample size, but require a non-hunted South American the fact that the same pattern was found in the population. Such a population may be difficult U.S. suggests that this could be an intrinisic to find as armadillos are apparently a favored feature of nine-banded armadillo populations. food item in many countries (Eisenberg 1989, However, other U.S. populations do not show Emmons 1990, Redford and Eisenberg 1992, such a bias, nor was one detected at Tall Taulman and Robbins 1996, Hill et at. 1997). Timbers during the first three years (1992- Nonetheless, our data do suggest caution in 1994) of our study (Loughry and McDonough making generalizations about nine-banded 1996). Assuming the bias we have detected is armadillos based on data from just the United real, it is unclear why it exists. In the U.S., States, and point out the need for more data on juvenile males are actually less abundant than more populations of D. novemcinctus in other juvenile females (though not in terms of the parts of the species' range. numbers oflitters produced, see Table 4). This suggests that there may be either increased ACKNOWLEDGMENTS mortality or a higher probability of dispersal out of the population for females as they get We are extremely grateful to C. Ruiz- older. Alternatively, adult females might be Miranda of the Smithsonian Institution less active or less detectable than are adult (National Zoo) for encouraging us to pursue males, but we know of no data to support such study of Brazilian armadillos. Our Brazilian a hypothesis. Juvenile mortality can be high work would not have been possible without the (McDonough and Loughry 1997b), but we exceptional armadillo-catching abilities of M. currently do not have data on sex differences in Coelho, to whom we are deeply indebted. mortality. Recruitment into our Florida Work in Brazil was supported by an population is low for both sexes (Loughry and Organization of American States Fellowship McDonough 1996, 1998), but there is some (to W. J. L), a Valdosta State University evidence that females may be less likely to faculty research award (to C. M. M.), the remain in their natal population than males Smithsonian Institution, TransBrasil, and the (Loughry and McDonough 1998). However, Associa~~io Mico-Leiio-Dourado (AMLD). any explanations of sex-ratio biases will Thanks also to the director of AMLD, D. probably require additional data to confirm Rambaldi, and the director of Po~o das Antas, that such a bias is real. D. Pessamilio, for facilitating our research. In Most earlier studies of D. novemcinctus the U.S., we are grateful to T. Engstrom and L have been performed in the U.S., with Brennan of Tall Timbers Research Station for populations that have presumably been their continued support of our research~ Our established for at most 100 to 200 years armadillo studies in the U.S. have been funded (Humphrey 1974, McBee and Baker 1982, by the American Philosophical Society and. Montgomery 1985,Wetzel 1985, Taulman and Valdosta State University faculty research Robbins 1996). We have shown here that at awards to W. J. L, and by Sigma Xi, the least one U.S. population differs substantially Theodore Roosevelt Memorial Fund, and from a population located in a more Valdosta State University faculty research historically ancient part of the species' range awards to C. M. M. (i.e., Brazil). Thus, the question arises of how representative data on U.S. populations are for REFERENCES the species as a whole. At present we cannot Abrams, P.A. & L. Rowe. 1996. The effects ofpredation answer this question. The extensive hunting of on the age and size of maturity of prey. Evolution armadillos that occurs at our Brazilian study 50: 1052-1061. 1182 REVISTA DE BIOLOGIA TROPICAL

Breece, G. A. & J. L. Dusi. 1985. Food habits and home James, F. C. 1970. Geographic size variation in birds and ranges of the common long-nosed armadillo its relationship to climate. Ecology 51: 365-390. Dasypus novemcinctus in Alabama, p. 419-427. In G. G. Montgomery (ed.). The evolution and ecology Kalmbach, E. R. 1943. The armadillo: its relation to of armadillos, sloths, and vermilinguas. Smithsonian agriculture and game. Texas Game, Fish and Oyster Instil., Washington, D.C. 451 p. Commission, Austin. 61 p.

Brennan, L. A., R. T. Engstrom, S. M. Hermann, S. T. Kleiman, D. G., B. B. Beck, 1. M. Dietz, L. A. Dietz, 1. D. Lindeman, W. K. Moser, K. McGorty, J. Noble & w. E. Ballou & A. F. Coimbra-Filho. 1986. Conservation Palmer. 1998. A five-year land management plan for program for the golden lion tamarin: captive research Tall Tnnbers Research Station property: 1998-2002. and management, ecological studies, education Tall TImbers Research Station, Tallahassee. 106 p. strategies and reintroduction, p. 959-979. In K. Benirschke (ed.). Primates: the road to self-sustaining Carter, T. S. 1985. Armadillos of Brazil. Nat. Geog. Res. populations. Springer-Verlag, New York. 1044 p. Rep. 20: 101-107. Lippert, K. J. 1994. Food habits, distribution and impact Carter, T. S. & C. Encarna~ao. 1983. Characteristics and of the nine-banded armadillo in Missouri. M.S. use of burrows by four species of armadillos in thesis, Southwest Missouri State University, Brazil. J. Mammal. 64: 103-108. Springfield. 88 p.

Clark, W. K. 1951. Ecological life history of the Lott, D. F. 1991. Intraspecific variation in the social armadillo in the eastern Edwards Plateau region. systems of wild vertebrates. Cambridge Univ., Amer. MidI. Nat. 46: 337-358. Cambridge. 238 p.

Dietz, 1. M., C. A. Peres & L. Pinder. 1997. Foraging Loughry, W. J. & C. M. McDonough. 1994. Scent ecology and use of space in wild golden lion discrimination by infant nine-banded armadillos. J. tamarins (Leontopithecus rosalia). Amer. J. Primat. Mammal. 75: 1033-1039. 41: 289-305. Loughry, W. J. & C. M. McDonough. 1996. Are road-kills Eisenberg, J. F. 1989. Mammals of the neotropics. Vol. valid indicators of armadillo population structure? 1. The northern tropics. Univ. Chicago, Chicago. Amer. MidI. Nat. 135: 53-59. 449 p. Loughry, W. J. & C. M. McDonough. 1997. Survey of the Emmons, L. H. 1990. Neotropical rainforest mammals: a Xenarthrans at Po~o das Antas Biological Reserve. field guide. Univ. Chicago, Chicago. 281 p. Edentata 3: 5-7.

Fitch, H. S., P. Goodrum & C. Newman. 1952. The Loughry, W. J. & C. M. McDonough. 1998. Spatial patterns armadillo in the southeastern U. S. 1. Mammal. 33: in a population of nine-banded armadillos (Dasypus 21-37. novemcinctus). Amer. MidI. Nat. 140: 161-169.

Foster, S. A. & 1. A. Endler. (eds.). in press. Geographic McBee, K. & R. J. Baker. 1982. Dasypus novemcinctus. variation in behavior: perspectives on evolutionary Mammal. Spec. 162: 1-9. mechanisms. Oxford Univ., Oxford. McDonough, C. M. 1992. The behavior and ecology of Galbreath, G. J. 1982. Armadillo, p. 71-79. In 1. A. nine-banded armadillos (Dasypus novemcinctus) in Chapman & G. A. Feldhamer (eds.). Wild mammals south Texas. Ph. D. Thesis, University of California, of North America. Johns Hopkins Univ., Baltimore. Davis. 101 p. 1147 p. McDonough, C. M. 1994. Determinants of aggression in Hill, K., 1. Padwe, C. Bejyvagi, A. Bepurangi, F. Jakugi, R. nine-banded armadillos. J. Mammal. 75: 189-198. Tykuarangi & T. Tykuarangi. 1997. Monitoring hunting impact on large vertebrates in the McDonough, C. M. & w. J. Loughry. 1995. Influences on Mbaracayu reserve, Paraguay, using native research vigilance in nine-banded armadillos. Ethology 100: assistants. Cons. BioI. 11: 1339-1353. 50-60.

Humphrey, S. R. 1974. Zoogeography of the nine-banded McDonough, C. M. & w. J. Loughry. 1997a. Influences armadillo in the United States. Bioscience 24: 457- on activity patterns in a population of nine-banded 462. aramdillos. 1. Mammal. 78: 932-941. , .

LOUGHRY & McDONOUGH: Comparisons between annadillo populations 1183

McDonough, C. M. & w. J. Loughry. 1997b. Patterns of novemcinctus) in a northern part of its range. Amer. mortality in a population of nine-banded MidI. Nat. 123: 383-389. annadillos, Dasypus novemcinctus. Amer. MidI. Nat. 138: 299-305. Storrs, E.E., & R. J. Williams. 1968. A study of monozygotic quadruplet annadillos in relation to Montgomery, G. G. (ed.) 1985. The evolution and mammalian inheritance. Proc. Nat. Acad. Sci. ecology of annadillos, sloths, and vermilinguas. (USA) 60: 910-914. Smithsonian Instit., Washington, D.c. 451 p. Taber, F. W. 1945. Contribution on the life history and Newman, H. H. 1913. The natural history of the nine- ecology of the nine-banded annadillo. J. Mammal. banded annadillo of Texas. Amer. Nat. 47: 513-539. 26: 211-226.

Newman, H. H. & 1. T. Patterson. 1910. The development Talmage, R. V. & G. D. Buchanan. 1954. The armadillo of the nine-banded annadillo from primitive streak to (Dasypus novemcinctus). A review of its natural birth, with special reference to the question of history, ecology, anatomy and reproductive polyembryony. J. Morph. 21: 359-423. physiology. Rice Inst. Pamph. Mono. BioI. 4: 1- 135. Patterson, J. T. 1913. Polyembryonic development in Tatusia novemcinctus. J. Morph. 24: 559-684. Taulman, 1. F. & L W. Robbins. 1996. Recent range expansion and distributional limits of the nine- Prodohl, P. A., W. J. Loughry, C. M. McDonough, W. S. banded annadillo (Dasypus novemcinctus) in the Nelson. & J. C. Avise. 1996. Molecular United States. 1. Biogeog. 23: 635-648. documentation of polyembryony and the micro- spatial dispersion of clonal sibships in the nine- U.S. Naval Observatory. 1995. The astronomical banded annadillo, Dasypus novemcinctus. Proc. R. almanac. U. S. Government Printing Office, Soc. Lond. Ser. B 263: 1643-1649. Washington, D.c. 548 p.

Redford, K.H. 1985. Food habits of armadillos Wetzel, R. M. 1982. Systematics, distribution, ecology, (Xenarthra: Dasypodidae), p. 429-437. In G. G. and conservation of South American edentates, p. Montgomery (ed.). The evolution and ecology of 345-375. In M. A. Mares & H. H. Genoways annadillos, sloths, and vennilinguas. Smithsonian (eds.). Mammalian biology in South America. Instit., Washington, D.C. 451 p. Spec. Publ. Ser., Pymatuning Lab. Ecol., Pittsburgh. 539 p. Redford, K.H. 1994. The edentates of the cerrado. EdentataI: 4-10. Wetzel, R. M. 1985. Taxonomy and distribution of annadillos, Dasypodidae, p. 23-46. In G. G. Redford, K.H. & J. F. Eisenberg. 1992. Mammals of the Montgomery (ed.). The evolution and ecology of neotropics. Vol. 2. The southern cone. Univ. annadillos, sloths, and vennilinguas. Smithsonian Chicago, Chicago. 430 p. Instit., Washington, D.C. 451 p.

Schaller, G. B. 1983. Manunals and their biomass on a Wetzel, R. M. & E. Mondolfi. 1979. The subgenera and Brazilianranch. Arq.ZooI.(Univ.SaoPaulo)31: 1-36. species of long-nosed annadillos, genus Dasypus L, p.43-63. In J. F. Eisenberg (ed.). Vertebrate ecology Sikes, R. S., G. A. Heidt. & D. A. Elrod. 1990. Seasonal in the northern neotropics. Smithsonian Instit., diets of the nine-banded armadillo (Dasypus Washington, D.C. 271 p.