SCENT DISCRIMINATION BY INFANT

NINE-BANDED ARMADILLOS

W. J. LOUGHRY AND COLLEEN M. MCDONOUGH

Biology Department, Valdosta State University,Valdosta, GA 31698-0015

We tested the scent-discrimination abilities of infant (i.e., young-of-the-year) nine-bandedarmadillos (Dasypus novemcinctus) in two-choice tests. The amount of time spent near andthe number of touches (with the snout) directed at pads containing various odors wererecorded. Infants spent more time near and investigated more often a pad containing theirown odor over a pad with no odor (indicating infants could detect the test odors), a padcontaining their own odor over one containing odors from a strange infant (i.e., a nonsib-ling), a pad containing odors from a sibling over one containing odors from a strangeinfant, and a pad containing their sibling's scent over a pad containing their own scent.These results indicate the potential for discrimination of kin in this species and furthersuggest that the odors used in discrimination may be individually distinct. This latter resultis surprising because nine-banded armadillos are born in litters of genetically identicalquadruplets.

Key words:fants

Dasypus novemcinctus, armadillos, kin recognition, scent discrimination, in-

One foundation of behavioral ecology isthe recognition that genetic relatedness be-tween individuals influences social interac-

tions and population structure (Dawkins,1976; Hamilton, 1964; Wilson, 1975). Acorollary of this principle is that, all otherthings being equal, individuals should varytheir behavior according to their degree ofrelatedness to other individuals. While there

may be limits to the extent that degrees ofrelatedness are discriminated, animals are

expected to show at least some ability todiscriminate between kin and nonkin, andpossibly between different types of kin (re-views by Blaustein and Porter, 1990;Fletcher and Michener, 1987; Gamboa etaI., 1991; Hepper, 1986; Holmes, 1988;Sherman and Holmes, 1985; Waldman,1988). An intriguing corollary to this ar-gument is that individuals who are geneti-cally identical may not be able to discrim-inate among one another (if thediscrimination cue is genetically based).

Nine-banded armadillos (Dasypus no-vemcinctus) give birth to litters of geneti-

Journal of Mammalogy, 75(4);1033-1039, 1994

cally identical quadruplets (Newman, 1913;Newman and Patterson, 1910). Breedingoccurs in the summer or early autumn, im-plantation of embryos is delayed, but usu-ally is accomplished by November-Decem-ber, and infants are born the followingspring (Enders, 1966; Patterson, 1913). Lit-ters first emerge aboveground in May-July.Littermates share a common burrow andforage near one another for a period rangingfrom a few weeks to several months (Kalm-bach, 1943; McBee and Baker, 1982;McDonough, 1992). Although relativelyasocial, there are reasons to suspect that dis-crimination of kin may be beneficial to ar-madillos. First, aggression can be quiteprevalent in some populations (Denson,1979; McDonough, 1994) and one mightexpect aggression to be directed primarilyat nonkin. Second, adult armadillos tend toexhibit overlapping home ranges (Clark,1951; Fitch et aI., 1952; Herbst and Red-ford, 1991; Jacobs, 1979; Layne and Glov-er, 1977; McDonough, 1992). Again, onemight expect individuals to tolerate more

1033

1034 JOURNAL OF MAMMALOGY Vol. 75, No.4

overlap with kin than with nonkin. Thesebenefits apply mostly to adult animals, butthey currently are speculative because wehave no information on dispersal of litter-mates. Thus, it is unclear whether adult in-dividuals interact with kin and, consequent-ly, whether selection would favorkin-discrimination abilities (cf., Beecher,1991; Blaustein et aI., 1993; Holmes,1984). Infant (i.e., young-of-the-year) ar-madillos may be better candidates for in-vestigating kin discrimination because theydo appear to encounter both kin and nonkin.For example, litters often exhibit home-range overlap (pers. obser.) and there arereports of individuals from two or more lit-ters occupying the same burrow (Kalm-bach, 1943). As in other mammals (Halpin,1986, 1991), kin discrimination may bebased primarily on olfaction, especially be-cause armadillos have poor eyesight and arerelatively nonvocal, but have large, pairedanal glands (Brown, 1985; Haynes and En-ders, 1961). In this study, we examine theability of infant nine-banded armadillos todiscriminate between scents from other in-fants, including between kin and nonkin.We also test the prediction that geneticallyidentical siblings may be unable to distin-guish between one another.

MATERIALS AND METHODS

Experiments were conducted at two locationsover a period of 3 years. From 31 May through7 August 1991, infants were tested on the Roband Bessie Welder Wildlife Refuge, located nearSinton, Texas. In 1992 and 1993, infants were

tested from mid-June to early August at the TallTimbers Research Station, located near Talla-

hassee, Florida. We found no across-year or be-tween-population differences in scent-discrimi-nation ability so data were pooled in allanalyses.

Study areas were intensively searched for in-fant armadillos each day during the late morning(1000-1400 h) and early evening (1600-2300h). We attempted to capture all infants sighted(by hand or with a dip net attached to a pole;see McDonough, 1994), regardless of whetherthere were other littermates present. Each cap-

tured infant was placed in a separate, clean plas-tic tub, where it was kept for the duration of theexperiments. Infants were allowed occasionalaccess to water during captivity, but they werenot fed. No infant was held >48 h, and all were

released at the same time of day and in the samelocation as where they were originally caught.Siblings were defined as same-sex animals of

similar body size, caught in similar locations.While this was our minimal definition of sib-

lings, most siblings were caught together on thesame day and clearly belonged to the same litter(i.e., they frequented the same burrows, inter-acted with one another, had nearly identicalbody-size measurements and facial patterns,etc.). Although we cannot rule out the possibil-ity, it is unlikely that nonsiblings were classifiedas siblings by these criteria. We never foundmore than four same-sized, same-sex individualsin the same area at the same time. Where litters

overlapped, litters always differed in sex or bodysize so that infants could be reliably assigned toa particular litter. However, if we were in doubtabout whether two infants were siblings, theywere not used in the kin-discrimination experi-ments.

Scent acquisition.-Odors were obtained

from armadillos by placing one or more KotexSuper Maxi pads in the plastic container with anindividual for at least 12 h. The pads were thenplaced in airtight plastic freezer bags until used.In a few cases, pads were kept in the freezerbags for up to 72 h so that they could be usedin testing an animal captured after the release ofthe scent donor. These pads remained pungentto us, and there was no evidence that they losttheir potency for the armadillos either (i.e., in-fants continued to investigate and discriminateamong them in the same manner as infants pre-sented with fresher pads). Prior to testing, padswere cut in half to increase the number of padsobtained from each animal.

Test apparatus.-All experiments were con-ducted in a modified, circular, liquid cattle feed-er. The feeder used in Texas was made of white

plastic and measured 150 em in diameter at thebase and 170 em at the top. The feeder used in

Florida was slightly smaller and made of yellowplastic. The floor of both feeders was coveredwith sand over which a sheet of white linoleum

was placed. The linoleum was attached to thesides of the feeder with white duct tape. Blackelectrical tape was used to demarcate four equal-

f'

.

November 1994 LOUGHRY AND MCDONOUGH-SCENT DISCRIMINATION BY ARMADILLOS 1035

sized quadrats inside the feeder. Both feedershad plastic lids with two (in Florida) or four (inTexas) clear windows that provided a means ofobserving the test animals.

Testing procedure.-In all tests, two padswere used, each placed in opposing quadrats (thefeeder was not rotated between tests, so the

same two quadrats were used in all tests). Theadhesive on the back of the pads was used toattach them to the wall of the feeder, just abovethe line of duct tape that attached the linoleum

floor to the sides of the feeder. The pads were,thus, at about the eye level of infants, whichmade it easy to identify when an animal wasinvestigating the pads because it had to raise itssnout to do so. An observer sat unobtrusivelynext to the feeder and observed the test animal

through one of the clear plastic windows in thefeeder cover. The observer always sat in thesame position, and all data were collected by thesame observer. All tests were conducted double-blind so that the observer did not know which

experiment was being performed or which padwas in either quadrat. The positioning of pads ina given test was random and determined by cointoss. After each test, the floor and walls of the

feeder were washed with water and again withisopropyl alcohol to remove any residual odors.

Upon being placed in the apparatus, animalswere given an acclimation period of 2:5 min.Data collection did not begin after this perioduntil the animal showed consistent activity bycrossing through at least two quadrats. Datawere collected for 10 min using a laptop com-puter and consisted of recording the amount oftime spent in each quadrat and the number of

times an infant investigated each pad by touch-ing it with its snout. An infant was defined asbeing in a quadrat if more than one-half its bodywas located within the quadrat. Each infant wastested in as many as four different experiments.The order of experiments was counterbalancedacross subjects, and tests on a given infant wereseparated by a minimum of I h. In addition, noinfant was exposed to the odor of another indi-vidual more than once, and we attempted tominimize the use of the same odor source in the

same type of experiment more than once (i.e., ifinfant Al was tested in the sibling versus non-sibling experiment with odor from its littermateA2 and stranger B I, when A2 was tested in thesame experiment, we did not use B I as the odorsource for the stranger again). Because of limi-

tations on which infants were available at a par-ticular time, we could not control for sex of odordonors in all cases. Thus, in some discrimina-

tions using odors from nonsiblings, the non sib-ling was not the same sex as the test subject. Intotal, 45 infants were tested (21 in Texas, 24 inFlorida). All but one of these infants were tested

in more than one experiment, although not allinfants were tested in all experiments.

Experiments.-Experiment I was designed todetermine if infants could detect the test odors.

In this experiment, an infant was given a choicebetween a clean pad and a pad containing itsown scent. We predicted infants should preferthe pad containing their own scent.

Experiment 2 examined whether infants coulddistinguish between odors. In this case, infants

were given a choice between a pad containingodor from another infant (a nonsibling) and apad containing its own odor. There is no clearprediction about the outcome of this test. Infantsmight prefer their own odor over that of someunfamiliar animal, or they might exhibit height-ened curiosity over this novel odor and, thus,spend more time investigating it. Either out-come, however, provides evidence that infantscan discriminate between the two scents. Indi-

viduals tested in this experiment always camefrom litters that were spatially remote from oneanother (usually> I km), so they were unlikelyto be closely related or to have encountered eachother's scent in the wild.

Experiment 3 tested whether infant odorswere individually distinctive. Infants were givena choice between a pad containing their ownodor and one containing the scent of a sibling.If infant odors are strictly genetically deter-mined, then infants should be incapable of dis-

tinguishing between these two scents. Perhaps astronger test of this hypothesis would be to pres-ent an infant with pads containing scents fromeach of two siblings (i.e., sibling versus sibling,rather than self versus sibling), but this provedlogistically impractical in the present study.

Kin discrimination was tested in experiment4. Infants were given a choice between a pad

containing the scent of their sibling and one con-taining scent from an unfamiliar infant. We hy-pothesized that infants would prefer the scent oftheir sibling over that of a nonsibling. As in ex-

periment 2, individuals tested in this experimentcame from litters that were spatially remote fromone another.

1036 JOURNAL OF MAMMALOGY Vol. 75, No.4

TABLE I.-Results of scent-discriminationtests for infant nine-banded armadillos. Valuesare means (:!: 1 SE) for time spent with morethan one-half the body in the quadrat containingthat odor pad and the number of times an infantinvestigated a pad by touching it with its snout.

Experiment" Time (s) Touches

1. Own odor versus a blank pad (n = 32)

Own 245.96 (32.18) 4.84 (0.88)Blank 121.84 (21.11) 1.41 (0.31)P 0.008 0.0002

2. Own odor versus a stranger's (n = 28)

Own 263.43 (37.41) 3.86 (0.92)Stranger 161.96 (32.06) 2.11 (0.51)P 0.096 0.029

3. Own odor versus a sibling's (n = 22)

Own 129.13 (14.90) 2.45 (0.53)Sibling 191.25 (24.83) 3.86 (0.78)P 0.159 0.042

4. Sibling's odor versus a stranger's (n = 28)

Sibling 220.69 (28.83) 4.50 (0.77)Stranger 123.40 (12.56) 3.29 (0.64)P 0.050 0.241

" P values are derived from Wilcoxon signed-rank tests.

Statistical tests.-Comparisons were madeusing Wilcoxon signed-rank tests to examine in-fant preferences for different scents. Although inmost cases we had specific predictions about theoutcome of particular comparisons, to be con-servative all tests were two-tailed.

RESULTS

Experiment 1 showed clearly that infantscould detect the odors used in the tests (Ta-ble 1). Infants apparently were attracted totheir own odor because they spent overtwice as much time in the quadrat contain-ing this pad as in the quadrat containing theblank pad, and they also investigated theirown pad far more frequently.

The second experiment showed that in-fants discriminated between their own odorand that of a stranger by investigating sig-nificantly more often the pad containingtheir own scent (Table 1). There also was anonsignificant trend for infants to spendmore time in the quadrat containing thispad.

Experiment 3 suggested that infant odorsare individually distinctive. Surprisingly,infants appeared to prefer the scent of theirsiblings to their own (Table 1). Infants in-vestigated significantly more often the padcontaining their sibling's scent, and therewas a nonsignificant trend for them tospend more time in the quadrat containingthis pad.

Experiment 4 provided evidence that in-fant armadillos are capable of kin discrim-ination. Infants spent significantly moretime near and tended to investigate moreoften the pad containing scent from theirsibling than that containing the scent of anonsibling (Table 1).

I

DISCUSSION

These results provide evidence for thescent-discrimination abilities of infant nine-banded armadillos. Infants appear capableof detecting odors produced by conspecificsand of discriminating between the odors ofdifferent individuals. Such discriminationincludes a capacity to distinguish self fromstranger, sibling from stranger, and selffrom sibling.

The implications of experiment 1 are rel-atively straightforward and confirm that, asin most mammals (Halpin, 1986, 1991), ol-faction is an important source of informa-tion for infant armadillos. Infants, likeadults, have poor eyesight and are relativelynonvocal, so their primary sense appears tobe olfaction. Thus, they should have easilydetected the test odors, and experiment 1shows they did. This experiment leavesopen the question of the source of theseodors. Our procedure allowed an amalgamof odors derived from urine, feces, andbody glands to permeate the test pads. In-fants were not observed during the periodwhen they were depositing these odors, sowe do not know the relative contribution ofeach odor source to the pads. Thus, it ispresently unclear which of these sourcesmay be most important in promoting dis-crimination. A likely candidate might be thesecretions from the paired anal glands be-

')

November 1994 LOUGHRY AND MCDONOUGH-SCENT DISCRIMINATION BY ARMADILLOS 1037

.

cause we have observed individuals inves-tigating one another by sniffing in the re-gion of these glands (pers. obser.; see alsoMcDonough, 1994). However, feces alsomay be important, as we have observedadult armadillos burying their feces in away that suggests scent-marking of thehome range (c. M. McDonough, pers. ob-ser.).

The results of experiment 2 also are notsurprising. Infants were able to distinguishbetween their own scent and that of a non-sibling. Similar results have been obtainedfor a number of other species (reviews inBrown, 1979; Halpin, 1986). However,there are two caveats that need to be madeabout the present findings. First, our resultsdo not allow identification of the type ofdiscrimination that may be occurring. In-fants may be distinguishing between famil-iar and unfamiliar odors, they may be dis-tinguishing between self and other (whichwould allow discrimination between two fa-miliar odors), or they may be making someother classification of the scents. Furtherexperiments are needed to identify whichdistinction is actually the relevant one forinfants. A second, related point is that dis-crimination might have been based on thestrength of the odors contained in the padsrather than infant identity (this concern alsoapplies to experiments 3 and 4). This seemsunlikely because all pads appeared heavilyconcentrated with scent to us, but only ex-periments in which the amount of odor ap-plied to the pads is controlled can resolvethis issue.

If one assumes that infant odors are ge-netically determined within an individual,the results of experiment 2 and the fact thatarmadillo siblings are genetically identicalto one another lead to the logical inferencethat infants should be able to discriminatebetween the odor of a sibling and that of anon sibling (indeed, it would be the sameexperiment as experiment 2). The results ofexperiment 3 call this assumption into ques-tion and suggest that infant odors are notlitter-specific, but are individually distinc-

tive. Such distinctiveness might arise ifsiblings feed on slightly different types offood (Leon, 1975; Porter et aI., 1989) orbecause of maternal effects that make these

genetically identical siblings phenotypicallydifferent (Storrs and Williams, 1968).

The results of experiment 4 show that, inspite of intra-litter variability in scent sig-natures, infants are still capable of discrim-inating between kin and nonkin. This studyis, thus, consistent with studies on a varietyof other mammals that also have demon-

strated scent-based kin discrimination (re-view in Halpin, 1991; see also Koprowski,1993). However, the present study leavesopen the question of the underlying mech-anism(s) promoting kin discrimination.Based on our experimental design, infantscould be discriminating kin from nonkin onthe basis of past association (i.e., familiar-ity-Holmes and Sherman, 1983) or phe-notype matching (Lacy and Sherman,1983). Kin discrimination also might be anepiphenomenon of other distinctions infantsmake between individuals (Grafen, 1990;Hare, 1992, 1994). Thus, while we haveshown that infants can discriminate kin

from nonkin, we do not know how they ac-complish this.

Assuming infants do discriminate be-tween kin and nonkin, why should they? Atpresent, we can only speculate on potentialbenefits. Siblings appear to maintain con-tact while foraging and share a commonburrow for a variable length of time duringtheir first summer (pers. obser.). Such be-havior might provide anti predator and ther-moregulatory benefits. Of course, infantscould reap these same benefits by associ-ating with nonkin as well. Perhaps it is notthe increased benefits of associating withkin that are important, but the reducedcosts. If associating with other armadillos iscostly (in terms of increased competitionfor food, etc.), perhaps it is better to incurthese costs among kin than among nonkin.Clearly, more data are needed before wecan begin to understand the costs and ben-efits of social behavior in armadillos.

1038 JOURNAL OF MAMMALOGY Vol. 75, No.4

While these results indicate the potentialfor scent discrimination by infant nine-banded armadillos, they leave at least twoimportant questions unanswered. First,while we have demonstrated the potentialfor kin discrimination by infants, it remainsto be shown that such discrimination actu-

ally is manifested in the field and has func-tional consequences (Blaustein et a!., 1991;Gamboa et a!., 1991; Holmes, 1988). Thatis, we have shown that infants can obtaininformation about their genetic relatedness(or familiarity) to another infant from itsscent. We do not know if infants actuallyuse this information. To begin addressingthis issue, we plan tests in which two in-fants are placed in an arena to see if siblingsand non siblings behave differently towardone another. The second question concernsthe maintenance of kin-discrimination abil-

ities. Are older animals also capable of sim-ilar scent discriminations, or are such abil-

ities unnecessary because of infrequentcontact between kin? We tested infants that

varied widely in age (based on bodyweight-McDonough, 1992). As infants getolder and heavier, litters appear to break up,so one might expect kin discrimination tobe less important to older infants. Whilethis may be so, it apparently did not hinderability of infants to discriminate amongodors in the present study. We divided in-fants into two groups, light (lower thanmean infant body weight) and heavy (great-er than mean infant body weight) and couldfind no difference in discrimination ability.However, to fully understand the potentialfor kin-biased behavior in older armadillos,we will need to know the fate of littermates

as they age, their potential to interact withkin as adults, and the kin-discriminationabilities of yearling and adult armadillos. Atpresent, there is no information availablerelevant to any of these issues.

ACKNOWLEDGMENTS

We thank the staff of the Rob and BessieWelder Wildlife Foundation and the Tall Tim-bers Research Station for all their help and sup-

port. This work was funded by a Welder WildlifeFoundation Graduate Research Fellowship andgrants from Sigma Xi and the Theodore Roo-sevelt Memorial Fund to C. M. McDonough,and a faculty research award from Valdosta StateUniversity and a grant from the American Philo-sophical Society to W. J. Loughry. We especiallythank J. Rasmussen and T. Engstrom for theirhelp with this project. B. J. Bergstrom, M. E.Smith, and two anonymous reviewers providedvaluable comments on earlier drafts of this pa-per. This is contribution no. 4 I9 of the WelderWildlife Foundation.

('

...,

LITERATURE CITED

BEECHER, M. D. 1991. Successes and failures of par-ent-offspring recognition in animals. Pp. 94-124, inKin recognition (P. G. Hepper, ed.). Cambridge Uni-versity Press, Cambridge, United Kingdom, 457 pp.

BLAUSTEIN, A. R., AND R. H. PORTER. 1990. The ubiq-uitous concept of recognition with special referenceto kin. Pp. 123-148, in Interpretation and explana-tion in the study of animal behavior. Interpretation,intentionality, and communication (M. Bekoff andD. Jamieson, eds.). Westview Press, Boulder, Colo-rado, I: 1-505.

BLAUSTEIN, A. R., M. BEKOFF, J. A. BYERS, AND T. J.DANIELS. 1991. Kin recognition in vertebrates: whatdo we really know about adaptive value? AnimalBehaviour, 41:1079-1083.

BLAUSTEIN, A. R., T. YOSHIKAWA, K. ASOH, AND S. C.WALLS. 1993. Ontogenetic shifts in tadpole kin rec-ognition: loss of signal and perception. Animal Be-haviour, 46:525-538.

BROWN, R. E. 1979. Mammalian social odors: a crit-ical review. Advances in the Study of Behavior, 10:103-162.

-. 1985. Armadillos, sloths, anteaters, and pan-

golins: orders Edentata and Pholidota. Pp. 732-738,in Social odours in mammals (R. E. Brown and D.W. Macdonald, eds.). Clarendon Press, Oxford,

United Kingdom, 2:1-882.CLARK, W. K. 1951. Ecological life history of the ar-

madillo in the eastern plateau region. The AmericanMidland Naturalist, 46:337-358.

DAWKINS, R. D. 1976. The selfish gene. Oxford Uni-versity Press, Oxford, United Kingdom, 224 pp.

DENSON, R. D. 1979. Aggression and tumbling amongarmadillo. The Southwestern Naturalist, 24:697-698.

ENDERS, A. C. 1966. The reproductive cycle of thenine-banded armadillo (Dasypus novemcinctus). Pp.205-310, in Comparative biology of reproduction inmammals (1. W. Rowlands, ed.). Academic Press,New York, 559 pp.

FITCH, H. S., P. GOODRUM, ANDC. NEWMAN. 1952. Thearmadillo in the southeastern U.S. Journal of Mam-

malogy, 33:21-37.FLETCHER, D. J. C., AND C. D. MICHENER (EDS.). 1987.

Kin recognition in animals. John Wiley & Sons,New York, 465 pp.

~

November 1994 LOUGHRY AND MCDONOUGH-SCENT DISCRIMINATION BY ARMADILLOS 1039

GAMBOA, G. J., H. K. REEVE, AND W. G. HOLMES.

1991. Conceptual issues and methodology in kin-recognition research: a critical discussion. Ethology,88:109-127.

GRAFEN, A. 1990. Do animals really recognize kin?Animal Behaviour, 39:42-54.

HALPIN, Z. T 1986. Individual odors among mam-mals: origins and functions. Advances in the Studyof Behaviour, 16:40-70.

-. 1991. Kin recognition cues of vertebrates.Pp. 220-258, in Kin recognition (P. G. Hepper, ed.).Cambridge University Press, Cambridge, UnitedKingdom, 457 pp.

HAMILTON, W. D. 1964. The genetical evolution ofsocial behaviour (I and II). Journal of TheoreticalBiology, 7: I-52.

HARE, J. E 1992. Colony member discrimination byjuvenile Columbian ground squirrels (Spermophi/usco/umbianus). Ethology, 92:301-315.

-. 1994. Group member discrimination via fa-miliarity with substrate-borne chemical cues by Co-lumbian ground squirrels. Animal Behaviour, 47:803-813.

HAYNES, J. E, AND A. C ENDERS. 1961. The compo-sition of the anal glands of Dasypus novemcinctus.The American Journal of Anatomy, 108:295-301.

HEPPER, P. G. 1986. Kin recognition: functions andmechanisms. Biological Reviews, 61 :63-93.

HERBST, L., AND K. REDFORD. 1991. Home-range sizeand social spacing among female common long-nosed armadillos (Dasypus novemcinctus). NationalGeographic Research and Exploration, 7:236-237.

HOLMES, W. G. 1984. Sibling recognition in thirteen-lined ground squirrels: effects of genetic relatedness,rearing association, and olfaction. Behavioral Ecol-ogy and Sociobiology, 14:225-233.

-. 1988. Kinship and the development of socialpreferences. Pp. 389-413, in Handbook of behav-ioral neurobiology. Developmental psychobiologyand behavioral ecology (E. M. Blass, ed.). PlenumPress, New York, 9:1-453.

HOLMES, W. G., AND P. W. SHERMAN. 1983. Kin rec-

ognition in animals. American Scientist, 71:46-55.JACOBS, J. 1979. Behavior and space usage patterns

of the nine-banded armadillo (Dasypus novemcinc-tus) in southwestern Mississippi. M.S. thesis, Cor-nell University, Ithaca, New York, 132 pp.

KALMBACH, E. R. 1943. The armadillo: its relation toagriculture and game. Texas Game, Fish and OysterCommission, Austin, Texas, 61 pp.

KOPROWSKI, J. L. 1993. The role of kinship in fieldinteractions among juvenile gray squirrels (Sciurus

"

il

caro/inensis). Canadian Journal of Zoology, 71 :224-226.

LACY, R. C, AND P. W. SHERMAN. 1983. Kin recog-nition by phenotype matching. The American Nat-uralist, 121:489-512.

LAYNE, J. N., AND D. GLOVER. 1977. Home range of- the armadillo in Florida. Journal of Mammalogy, 58:

411-413.

LEON, M. 1975. Dietary control of the maternal pher-omone in the lactating rat. Physiology and Behavior,14:311-319.

McBEE, K., AND R. J. BAKER. 1982. Dasypus no v-emcinctus. Mammalian Species, 162:1-9.

McDoNOUGH, C M. 1992. Behavioral ecology ofnine-banded armadillos in a riverine habitat. Ph.D.

dissert., University of California, Davis, 101 pp.-. 1994. Determinants of aggression in nine-

banded armadillos. Journal of Mammalogy, 75:189-198.

NEWMAN, H. H. 1913. The natural history of the nine-banded armadillo of Texas. The American Natural-ist,47:513-539.

NEWMAN, H. H., AND J. T. PATTERSON. 1910. The de-

velopment of the nine-banded armadillo from prim-itive streak to birth, with special reference to thequestion of polyembryony. Journal of Morphology,21 :359-423.

PATTERSON, J. T. 1913. Polyembryonic developmentin Tatusia novemcinctus. Journal of Morphology,24:559-684.

PORTER, R. H., S. A. McFAYDEN-KETCHUM, AND G. A.

KING. 1989. Underlying bases of recognition sig-natures in spiny mice (Acomys cahirinus). AnimalBehaviour, 37:638-644.

SHERMAN, P. W., AND W. G. HOLMES. 1985. Kin rec-

ognition: issues and evidence. Pp. 437-460, in Ex-perimental behavioral ecology and sociobiology (B.Holldobler and M. Lindauer, eds.). Sinauer Associ-ates, Inc., Publishers, Sunderland, Massachusetts,488 pp.

STORRS, E. E., AND R. J. WILLIAMS. 1968. A study ofmonozygotic quadruplet armadillos in relation tomammalian inheritance. Proceedings of the NationalAcademy of Sciences, 60:910-914.

WALDMAN, B. 1988. The ecology of kin recognition.Annual Review of Ecology and Systematics, 19:543-571.

WILSON, E. O. 1975. Sociobiology, the new synthesis.Belknap Press, Cambridge, Massachusetts, 697 pp.

Submitted 8 October 1993. Accepted 24 January 1994.

Associate Editor was Stephen H. Vessey.