Richard B. Loomis told me that larval chiggers are able to survive under water for several days but that warm water will hasten their demise.

Infestations of larval bot flies (_Sarcophaga cistudinis_) were noted in several turtles at the Damm Farm and, upon closer scrutiny, were found to be common in preserved specimens from other areas. Larvae were always found in flask-shaped pockets (Pl. 27, Fig. 2) beneath the skin; the pockets opened to the outside by a small hole, the edges of which were dried and discolored. One larva sometimes protruded from the opening. The inside of the pocket is lined with smooth, skinlike tissue. Heavily infested box turtles may have four or five such pockets, each containing one to many larvae. The most frequent sites of the pockets are the skin of the axillary and inguinal regions, and the skin of the limbs and neck, especially near the bases of these members. Subadults were more heavily infested than older adults; no infestations of hatchlings or small juveniles were noted.

An adult female, infested with bot fly larvae when she was removed from her hibernaculum in late October, 1955, bore no trace of larvae or of the pocket that had contained them when she was recaptured the following June. According to Rokosky (1948), the larvae eventually fall to earth and pupate. The individuals of _T. carolina_ studied by him were not re-infested by adult bot flies; one turtle ate some of the larvae that dropped from its body.

The manner in which box turtles are infested by bot fly larvae is uncertain. Possibly the eggs are picked up accidentally or laid on the skin while box turtles are foraging in dung. Belding (1952:841) cla.s.sifies the genus _Scarophaga_ as semi-host-specific, depositing eggs in open wounds.

McMullen (1940), Rodeck (1949), and Rainey (1953), described individuals of _T. ornata_ parasitized by _S. cistudinis_. Rokosky (1948) and Peters (1948:473) reported infestations in _T. carolina_.

Infestations were the cause of death in the instances noted by Rainey and Rokosky.

PREDATORS

Few first-hand observations on predators of _T. ornata_ are available and I have found little direct evidence of predation in the course of this study. In general, adults of the species seem to have few natural enemies other than man. Several of my colleagues at the University of Kansas have observed dogs carrying box turtles in their mouths or chewing on them. Frank B. Cross told me his dog caught and ate young _T. ornata_ in Payne County, Oklahoma, and A. B. Leonard once saw a badger carrying one in Dewy County, Oklahoma. At the Reservation, a freshly killed juvenile was found beneath the nest of a crow (_Corvus brachyrhynchos_) and remains of a hatchling were found in a scat of a copperhead (_Agkistrodon contortrix_).

Dr. Fred H. Dale, Director of the Patuxent Research Refuge, Laurel, Maryland, kindly furnished photostatic copies of cards, from the Division of Food Habits Research of the U. S. Fish and Wildlife Service, recording the instances in which _Terrapene ornata_ was listed as a food-item. In one instance the stomach of each of two nestlings, in the same nest, of the White-necked Raven (_Corvus cryptoleucus_) in Terry County, Texas, contained remains of recently hatched ornate box turtles; the remains of one turtle made up 64 per cent of the contents of one stomach, and parts of three turtles made up 80 per cent of the contents of the other stomach. Each of two stomachs of the coyote (_Canis latrans_) from Quay County, New Mexico, contained a "trace" of ornate box turtle.

Wild carnivores known to occur on the Damm Farm were racc.o.o.ns (_Procyon lotor_), striped skunks (_Mephitis mephitis_), badgers (_Taxidea taxus_), and coyotes (_Canis latrans_); all were suspect as predators of ornate box turtles.

On December 10, 1953, ten dead box turtles (eight adults and two juveniles) were discovered at the top of a cut bank on the Damm Farm, within a few feet of a burrow that was used at least part of the time by a striped skunk. The condition of the turtles suggested that they had lain in the open for several weeks. The heads and legs were missing from most of the turtles and tooth marks were discernible on several of the sh.e.l.ls. A logical explanation of this occurrence is that the turtles, using the burrow as a hibernaculum, were ousted by a predator that also inhabited the burrow. Turtles moving about sporadically in late autumn may be quickly chilled by a sudden drop in temperature and therefore be more susceptible to predation than at other times of the year. Two of my colleagues at the Museum of Natural History informed me that they had observed similar concentrations of dead _T. ornata_ in winter.

In July, 1952, H. B. Tordoff collected eight sh.e.l.ls of juvenile _T.

ornata_ in a dry creek bed near Sharon, Barber County, Kansas. Some of the sh.e.l.ls had small tooth-punctures. The stream bed habitat and the appearance of the tooth punctures tended to incriminate racc.o.o.ns as predators. Racc.o.o.ns, more than any other carnivore mentioned above, possess the manual dexterity necessary to pry open the sh.e.l.l of a box turtle and bite away the soft parts. Badgers and possibly coyotes are probably the only local carnivores (excluding large dogs) that could crack open the sh.e.l.l of an adult turtle by sheer force.

Adults of _T. ornata_, since they occasionally molest small juveniles, must be considered in the category of predators. When captive adults and juveniles were fed from the same container in the laboratory, the turtles occasionally bit one another accidently. Serious injury to the young was prevented by watching the adults closely and moving them away when they caught a smaller turtle by the leg or head. Similar accidents presumably occur in nature; juveniles and adults were sometimes found feeding side by side. William R. Brecheisen told me that adults kept in a stock tank at his farm in the summer of 1955 regularly and purposefully chased and bit small juveniles in the same tank. Brecheisen gave me a juvenile that had been so bitten; the right side of its head was badly damaged (the eye gone and a portion of the bony orbit broken) but was partly healed. Ralph J. Donahue told me that he saw an adult _T. ornata_ attack a juvenal _T. carolina_, and provided a photograph of the incident. The juvenile was not injured.

Although small box turtles may occasionally be caught and killed by adults in nature, this seems not to const.i.tute a major source of predation on the young.

Other animals that may prey upon young box turtles occasionally (and that were known to occur at the Damm Farm) are bullsnakes (_Pituophis catenifer_), red-tailed hawks (_Buteo jamaicensis_), marsh hawks (_Circus cyaneus_), crows (_Corvus brachyrhynchos_), and opossums (_Didelphis marsupialis_), and domestic cats.

Nest predators probably have greater effect on populations of _T.

ornata_ than do predators of hatchlings, juveniles, and adults. Four robbed nests were found at the Damm Farm; in each instance, striped skunks were thought to be the predators. E. H. Taylor told me that he once saw a bullsnake swallow an entire clutch of newly laid eggs before the female turtle could cover the nest.

DEFENSE

Box turtles rely for protection on the closable sh.e.l.l and on inconspicuousness; defense reactions, except in the rare instances that biting is provoked, are purely pa.s.sive.

Box turtles handled in the course of field work varied widely in their reactions. Many struggled violently when being measured or marked whereas others were completely pa.s.sive, closing the sh.e.l.l tightly and making it difficult for me to examine the soft parts of the body.

These differences in behavior did not seem to be correlated either with s.e.x or with age; generally lessened activity was a.s.sociated with suboptimum body temperatures. All box turtles found in the field were extremely wary. As soon as one sighted me (sometimes at a distance of 200 feet or more), it became motionless with sh.e.l.l raised from the ground and neck extended (Pl. 28, Fig. 5). Some turtles remained in this motionless stance for half an hour or more, finally moving slowly away if I remained motionless. Turtles made no attempt to escape until I approached them closely or until they were in danger of being trampled by my horse; they would then move away with remarkable rapidity. Box turtles seemed unaware of an intruder until he could be seen or until he touched the turtle. When a turtle was approached from the rear, whistling, finger snapping, and normal footfalls did not attract its attention. Latham (1917:16) observed corresponding behavior in _T. carolina_. Wever and Vernon (1956) found the ear of _T. carolina_ to be keenly sensitive to sounds in the range of 100-600 cycles per second but progressively less sensitive to sounds of higher and lower frequencies. Surely a predator as stealthy as a coyote could approach a box turtle unseen and could quickly bite off at least one of the turtle's legs. Many of the mutilated box turtles that I observed may have survived such encounters with carnivores. The tendency of some individuals, when handled, to over-extend the limbs and neck (rather than closing the sh.e.l.l) in an attempt to escape, would make them easy victims for any predator.

Ornate box turtles were kept in my home, along with several cats.

Initial behavior was characterized by mutual wariness; subsequently the cats would follow a turtle about the house for a time, occasionally pawing at an exposed limb. The turtles withdrew only when touched or when approached from the front. After a day or two the cats and turtles ignored each other, often eating and drinking from the same dishes without incident. Under these circ.u.mstances the cats, I believe, could easily have killed or injured the turtles. A turtle would occasionally gain the respect of a cat by biting it.

The strong odor sometimes given off by box turtles is produced by the secretions of four musk glands, two situated anteriorly on each side and opening by small, nearly invisible apertures beneath the fourth marginal scute. According to Hoffman (1890:9), two other musk glands, opening beneath the eighth marginal scute on each side, are also present in _Terrapene_; these posterior glands were not found in the several specimens of _T. ornata_ that I dissected.

Strong odors were produced by nearly all small juveniles until they became accustomed to being handled. Older juveniles and adults produced strong odors only in response to pain or injury, as, for example, when they were killed in the laboratory prior to preservation or when they were being marked in the field. Young box turtles were capable of producing strong odors as soon as they hatched.

Norris and Zweifel (1950:3) considered the odor produced by _T. o.

luteola_ to issue from the "... concentrated, highly pungent urine...." voided by individuals when they were disturbed, and thought the production of odor to be a defense mechanism. Neill (1948b:130) reported that hatchlings of _T. carolina_ with unhealed umbilical scars emitted a musky odor comparable to that of the stinkpot, _Sternotherus odoratus_; he thought the capacity to produce this odor was lost at about the time that the plastral hinge became functional.

The function of musk glands in _Terrapene_ and, in all other turtles, is unknown. Since biting and nuzzling of the edges of the sh.e.l.l is an integral part of the courtship of many turtles, odor produced by the musk glands may well be a means of social recognition or of s.e.xual stimulation. Repellant odor may have a protective value in young box turtles but it is unlikely that larger predators would be frightened away or even discouraged by odor alone. In this respect Neill (_loc.

cit._) and I concur.

DISCUSSION OF ADAPTATIONS

Most of the morphological characteristics distinguishing box turtles from other North American emyid turtles, the most notable of which is the movable plastron, are modifications that have evolved as a result of selectional pressures favoring adaptation to more or less terrestrial existence. Similar adaptations have arisen independently in several branches of the emyid stock (see introduction). The genus _Terrapene_ seems to have departed farther from a generalized emyid form than have other kinds of box-turtle-like chelonians. In a morphological sense, _Terrapene ornata_ is clearly the most specialized member of its genus now occurring in the United States (my own studies have revealed that populations in western Mexico now referred to as _T. klauberi_ and _T. nelsoni_ are as specialized as _T. ornata_ in some respects but more generalized in others). The present ecological study has demonstrated that _T. ornata_ is specialized in habits as well as in structure. It is concluded that these specializations (of more generalized and perhaps more primitive conditions as, for example in _T. carolina_) const.i.tute adaptation for terrestrial existence in open, semiarid habitats. These adaptations in _T. ornata_ have resulted, in a few instances, in unique habits and structures; however, in most instances the adaptations have produced slight but recognizable changes that are definable only by degree of difference from other species of box turtles.

The closable sh.e.l.l of box turtles is of obvious survival value in providing protection for the soft parts of the body. In most of the species of _Terrapene_, the lobes of the plastron completely close the openings of the sh.e.l.l; closure is so tightly effected in some individuals that it is difficult to insert the blade of a knife between the adpressed margins of carapace and plastron. In _T. ornata_ nevertheless, both lobes of the plastron are deficient on their lateral margins; four narrow openings remain when the lobes are drawn shut. Emargination of the plastron has occurred at the places where the limbs rub against it during locomotion. This reduction of the plastron permits the body to be held off the ground during forward locomotion and seemingly permits a generally freer range of movement for the limbs. The possible disadvantages of an imperfectly closable sh.e.l.l seem to be compensated for by increased mobility. Reduction of the plastron is correlated with a general lightening of the sh.e.l.l, probably a.s.sociated with the increased vagility of this species.

Lightening of the sh.e.l.l is evident also in the relatively thin, loosely articulated bony elements. Sh.e.l.ls of adult _T. ornata_ that are old and weathered, or macerated (unless they are partly co-ossified because of injury), can nearly always be disarticulated with ease, whereas the bony elements in the sh.e.l.ls of adult _T.

carolina_ (all races) are nearly always co-ossified or separable only after prolonged maceration.

The relatively low, flattened sh.e.l.l of _T. ornata_ is an adaptation a.s.sociated with the tendency to seek shelter in the limited s.p.a.ce of earthen forms, burrows, or small natural cavities in the course of the warm season and to burrow more deeply into the ground in winter.

_Terrapene ornata_ is, in fact, the only species of the genus that may be considered an habitual burrower. Individuals of _T. carolina_ tend to seek shelter in the warm season by making forms in dense vegetation or by digging into yielding substrata such as mud or humus, although they may burrow deeply into the earth in winter. Extreme weakness or absence of the middorsal keel of _T. ornata_ seems to be a modification a.s.sociated with burrowing habits and general adaptation to terrestrial life; the keel is similarly reduced in testudinids.

Retention of epidermal laminae (as opposed to regular exfoliation of the older parts of scutes) occurs in all box turtles, in several other groups of terrestrial emyids, and in testudinids. The phenomenon is here considered to be a specialization of scute shedding--developed in terrestrial and semiterrestrial chelonians--that provides additional protection to the sh.e.l.l against wear and minor injuries.

General shortening of digits--the result of reduction in number of phalanges as well as in their length, and to a lesser degree the shortening of metapodial elements--has occurred in several groups of chelonians with terrestrial tendencies (the opposite--lengthening of phalanges and metapodials, and hyperphalangy--has occurred in certain groups that are highly aquatic). The pes of box turtles has remained relatively unchanged in this respect; a few phalanges on the lateral digit have been lost (especially in three-toed forms), but little reduction in length has occurred. The chief modification of the pes is a general narrowing brought about by the tendency of the digits to be crowded together, one on top of the other, rather than spread in a horizontal plane. Considerably more modification is seen in the ma.n.u.s of _Terrapene_. Phalangeal formulae (expressing the number of phalanges from the first digit outward) range from 2-3-3-3-2 (primitive in _Terrapene_) to 2-3-3-2-2 in the races of _carolina_ and have the same range in the species of eastern Mexico. Extreme reduction in number (2-2-2-2-2) as well as general shortening of phalanges occurs in _T. ornata_. The formula is the same in the one specimen of _T. klauberi_ that has been skeletonized. This modification of the forelimb in _T. ornata_ has produced a more rigid, stronger ma.n.u.s that is well adapted to the requirements of burrowing and to locomotion over unyielding substrata. Shortening of the ma.n.u.s (and, to a lesser extent, the pes) has been accompanied by reduction and loss of interdigital webbing. It is noteworthy that _T. ornata_ has achieved the same reduction in number of phalanges as _Gopherus_, which displays the extreme of specialization in this respect among North American turtles. The ma.n.u.s in _T. ornata_ is not shortened so much as in _Gopherus_.

The first toe in males of _T. ornata_ is uniquely widened, thickened, and inturned. Males of some other species of _Terrapene_ have greatly enlarged rear claws, some of which turn slightly inward, but none has the flexed first toe hooklike as it is in _ornata_ (a modified first toe, resembling that described for _T. ornata_, has been observed in a live male of _T. klauberi_ [now KU 51430] since the preparation of this ma.n.u.script). In males of _T. ornata_ the penultimate phalanx of the first toe has a normal, vertical articular surface on its proximal end. However, the distal articular surface (when viewed from the distal end of the phalanx) has its axis rotated away from the vertical plane approximately 45 degrees in a counterclockwise direction. As the foot is p.r.o.nated and extended, and as the digits are flexed, there is a concomitant inward rotation of the first metatarsal at its proximal joint; this rotation, combined with the divergent planes of the articulating surfaces on the penultimate phalanx, cause the ungual phalanx to be flexed at right angles to the inner side of foot, in a plane perpendicular to that of the other toes (Fig. 21).

The precise function of the modified first toe of males is unknown, although it is reasonably safe to a.s.sume that the modification is closely a.s.sociated with clasping during coition. In the matings that I observed, the inturned first claw of the male secured a hold on the female's rump or just beneath her legs, whereas the remaining three toes gripped the edge of her plastron. The combined hold, on sh.e.l.l and skin, clearly affords the male a more secure position during coitus (whether the female clasps his legs with hers or not) than would a hold on skin or sh.e.l.l alone. Possibly intromission can be maintained in this position even when the female is attempting to escape. In males the plastron is less concave in _T. ornata_ than in _T.

carolina_. Furthermore, males of _T. ornata_ are, on the average, smaller than females, whereas the reverse is true in _T. carolina_.

Possibly the ability of the male to secure an especially firm grip on the female enhances the probability of small males mounting and inseminating larger females, whereas successful matings might otherwise be limited to pairs in which the male was the larger member.

It is worthy of note that turtles of the genus _Terrapene_ are seemingly the only North American emyids that carry out the entire process of mating on land; other, semiterrestrial emyids (for example, _Clemmys insculpta_ and _Emydoidea blandingi_) return to water for actual coition, although the precoital behavior sometimes occurs on land.

Nearly all gradations from a fully developed zygomatic arch to a greatly reduced arch can be observed in skulls of the various species of _Terrapene_ (Fig. 2) (Taylor, 1895:586, Figs. 2-7). The highest degree of reduction is achieved in _T. ornata_ and _T. klauberi_, both of which lack the quadratojugal bone and have no zygomatic arch whatever (except for an occasional, poorly defined anterior vestige formed by the postfrontal, the jugal, or both). Reduction of the zygoma clearly represents modification of a more generalized, complete arch. As yet there is no clear evidence that reduction of the zygomatic arch is of adaptive value. It is noteworthy, however, that similar reduction of the arch has occurred independently in a number of emyid and testudinid groups, nearly all of which have terrestrial or semiterrestrial habits. Although discussion of phyletic lines in _Terrapene_ is beyond the scope of this report, I tentatively suggest that reduced zygomatic arches have arisen independently in more than one group of _Terrapene_ and that similar reduction of the arch in two species of the genus does not necessarily indicate an especially close relationship of such species.

In a recent survey of cloacal bursae in chelonians, Smith and James (1958:88) reported _T. ornata_ and _T. mexicana_ to be among the few emyids that lacked these structures; in the opinion of the authors (_op. cit._:94) cloacal bursae evolved in chelonians that required an accessory respiratory organ for long periods of quiescence (hibernation or aestivation) under water, and were secondarily lost in terrestrial forms that hibernated on land. The a.s.sumption is a reasonable one, at least in regard to emyids and testudinids. Lack of cloacal bursae in _T. ornata_ and in all testudinids, can be correlated with the completely terrestrial habits of those turtles.

Cloacal bursae seem to be vestigial in the species of _Terrapene_ possessing them and to be of little or no use as respiratory structures (except perhaps in _T. coahuila_).

In most of the species of _Terrapene_ the carapace has a pattern of pale markings on a darker background; however, unicolored individuals are the rule in certain populations (for example, at the western edge of the range of _T. carolina_ and in _T. ornata luteola_) and occur as occasional variations in other populations (in _T. yucatana_, _T.

mexicana_, and, throughout the range of _T. carolina_, albeit more commonly in the southeastern part of the range). Personal observation of interspecific and ontogenetic variation of color patterns of box turtles has convinced me that a basic pattern of more or less linear radiations is the one from which all other patterns (including spots, blotches, rosettes, and the unicolored condition) can be derived, and, that the radial pattern is generalized and primitive for _Terrapene_ (possibly for all emyids and testudinids as well). In the light of this conclusion, the radial pattern of _T. ornata_ may be considered generalized. I suspect, however, that the pattern of a living species most closely approaching that of the primitive ancestral stock of _Terrapene_ is the pattern of fine, wavy, dark radiations (on a paler background) present in young examples of _T. coahuila_.

Box turtles in general have lower reproductive potentials (as indicated by fewer eggs and longer prep.u.b.eral period) than do most aquatic emyids. This low potential seems to be compensated for by a lower rate of postnatal mortality (especially in adults) due to the protection afforded by the closable sh.e.l.l and the ability to recover from serious injury. _Terrapene o. ornata_ and _T. c. carolina_ are the only box turtles the life histories of which are known well enough to permit significant comparison. The reproductive potentials of _T.

o. ornata_ and _T. c. carolina_ seem to be much the same.