University of Kansas Publications
Museum of Natural History

Volume 11, No. 10, pp. 527-669, 16 pls., 29 figs.

March 7, 1960

Natural History of the Ornate Box Turtle, Terrapene ornata ornata Agassiz

BY
JOHN M. LEGLER

University of Kansas
Lawrence
1960

University of Kansas Publications, Museum of Natural History
Editors: E. Raymond Hall, Chairman, Henry S. Fitch, Robert W. Wilson
Volume 11, No. 10, pp. 527-669, 16 pls., 29 figs.
Published March 7, 1960

University of Kansas

PRINTED IN
THE STATE PRINTING PLANT
TOPEKA, KANSAS
1960

Natural History of the Ornate Box Turtle, Terrapene ornata ornata Agassiz

BY

John M. Legler

CONTENTS

INTRODUCTION

The ornate box turtle, Terrapene o. ornata Agassiz, was studied more or less continuously from September, 1953, until July, 1957. Intensive field studies were made of free-living, marked populations in two small areas of Douglas County, Kansas, in the period 1954 to 1956. Laboratory studies were made, whenever possible, of phenomena difficult to observe in the field, or to clarify or substantiate field observations. Certain phases of the work (for example, studies of populations and movements) were based almost entirely on field observation whereas other phases (for example, growth and gametogenic cycles) were carried out almost entirely within the laboratory on specimens obtained from eastern Kansas and other localities.

A taxonomic revision of the genus Terrapene was begun in 1956 as an outgrowth of the present study. The systematic status of T. ornata and other species is here discussed only briefly.

Objectives of the study here reported on were: 1) to learn as much as possible concerning the habits, adaptations, and life history of T. o. ornata; 2) to compare the information thus acquired with corresponding information on other emyid and testudinid chelonians, and especially with that on other species and subspecies of Terrapene; 3) to determine what factors limit the geographic distribution of ornate box turtles; and, 4) to determine the role of ornate box turtles in an ecological community.

Acknowledgments

The aid given by a number of persons has contributed substantially to the present study. I am grateful to my wife, Avis J. Legler, who, more than any single person, has unselfishly contributed her time to this project; in addition to making all the histological preparations and typing the entire manuscript, she has assisted and encouraged me in every phase of the study. Dr. Henry S. Fitch has been most helpful in offering counsel and encouragement. Thanks are due Professor E. Raymond Hall for critically reading the manuscript.

Special thanks are due also to the following persons: Professor A. B. Leonard for helpful suggestions dealing with photography and for advice on several parts of the manuscript; Professor William C. Young for the use of facilities at the Endocrine Laboratory, University of Kansas; Professor Edward H. Taylor for permission to study specimens in his care; Dr. Richard B. Loomis for identifying chigger mites and offering helpful suggestions on the discussion of ectoparasites; Mr. Irwin Ungar for identification of plants; and, Mr. William R. Brecheisen for allowing me to examine his field notes and for assistance with field work. Identifications of animal remains in stomachs were made by Professor A. B. Leonard (mollusks, crustaceans), Dr. George W. Byers (arthropods), and Dr. Sydney Anderson (mammals).

Miss Sophia Damm generously permitted the use of her property as a study area and Mr. Walter W. Wulfkuhle made available two saddle horses that greatly facilitated field work. The drawings (with the exception of [Fig. 21]) are by Miss Lucy Jean Remple. All photographs are by the author.

I am grateful also to the Kansas Academy of Science for three research grants (totaling $175.00) that supported part of the work. The brief discussion of taxonomic relationships and distribution results partly from studies made by means of two research grants (totaling $150.00), from the Graduate School, University of Kansas, for which I thank Dean John H. Nelson.

Systematic Relationships and Distribution

Turtles of the genus Terrapene belong to the Emyidae, a family comprising chiefly aquatic and semiaquatic species. Terrapene, nevertheless, is adapted for terrestrial existence and differs from all other North American emyids in having a hinged and movable plastron and a down-turned (although often notched) maxillary beak. Emydoidea blandingi, the only other North American emyid with a hinged plastron, lacks a down-turned beak. The adaptations of box turtles to terrestrial existence (reduction of webbing between toes, reduction in number of phalanges, reduction of zygomatic arch, and heightening of shell) occur in far greater degree in true land tortoises of the family Testudinidae. Four genera of emyid turtles in the eastern hemisphere (Cuora, Cyclemys, Emys, and Notochelys) possess terrestrial adaptations paralleling those of Terrapene but (with the possible exception of Cuora) the adaptations are less pronounced than in Terrapene. A movable plastron has occurred independently in two groups of emyids in the New World and in at least three groups in the Old World.

The genus Terrapene, in my view, contains seven species, comprising 11 named kinds. Of these species, five are poorly known and occur only in Mexico. Terrapene mexicana (northeastern Mexico) and T. yucatana (Yucatan peninsula) although closely related, differ from each other in a number of characters. Similarly, Terrapene klauberi (southern Sonora) and T. nelsoni (Tepic, Nayarit—known from a single adult male) are closely related but are considered distinct because of their morphological differences and widely separated known ranges. Terrapene coahuila, so far found only in the basin of Cuatro Ciénegas in central Coahuila, is the most primitive Terrapene known; it differs from other box turtles in a number of morphological characters and is the only member of the genus that is chiefly aquatic.

Two species of Terrapene occur in the United States. Terrapene carolina, having four recognized subspecies, has a nearly continuous distribution from southern Maine, southern Michigan, and southern Wisconsin, southward to Florida and the Gulf coast and westward to southeastern Kansas, eastern Oklahoma and eastern Texas, and characteristically inhabits wooded areas.

Terrapene ornata is a characteristic inhabitant of the western prairies of the United States, and ranges from western and southern Illinois, Missouri, Oklahoma, and all but the extreme eastern part of Texas, westward to southeastern Wyoming, eastern Colorado, eastern and southern New Mexico, and southern Arizona, and, from southern South Dakota and southern Wisconsin, southward to northern Mexico ([Fig. 1]). It is the only species of the genus that occurs in both Mexico and the United States. The northeasternmost populations of T. ornata, occurring in small areas of prairie in Indiana and Illinois, seem to be isolated from the main range of the species. The ranges of T. ornata and T. carolina overlap in the broad belt of prairie-forest ecotone in the central United States. Interspecific matings under laboratory conditions are not uncommon and several verbal reports of such matings under natural conditions have reached me. Nevertheless, after examining many specimens of both species and all alleged "hybrids" recorded in the literature, I find no convincing evidence that hybridization occurs under natural conditions.

Terrapene ornata differs from T. carolina in having a low, flattened carapace lacking a middorsal keel (carapace highly arched and distinctly keeled in carolina), and in having four claws on the hind foot (three or four in carolina), the claw of the first toe of males being widened, thickened, and turned in (first toe not thus modified in carolina). Terrapene ornata is here considered to be the most specialized member of the genus by virtue of its reduced phalangeal formula, lightened, relatively loosely articulated shell, reduced plastron, and lightly built skull, which completely lacks quadratojugal bones ([Fig. 2]); most of these specializations seem to be associated with adaptation for terrestrial existence in open habitats.

Fig. 1. Geographic distribution of Terrapene ornata. Solid symbols indicate the known range of T. o. ornata and hollow symbols the known range of T. o. luteola. Half-circles show the approximate range of intergradation between the two subspecies. Triangles indicate localities recorded in literature; specimens were examined from all other localities shown. Only peripheral localities are shown on the map.

Two subspecies of T. ornata an recognized. Terrapene o. luteola, Smith and Ramsey (1952), ranges from northern Sonora (Guaymas) and southern Arizona (southern Pima County) eastward to southeastern New Mexico and Trans-Pecos, Texas, where it intergrades with T. o. ornata; the latter subspecies is not yet known from Mexico but almost surely occurs in the northeastern part of that country. The subspecies luteola differs from ornata in being slightly larger and in having more pale radiations on the shell (11 to 14 radiations on the second lateral lamina in luteola, five to eight in ornata). In individuals of luteola the markings of the shell become less distinct with advancing age and eventually are lost; shells of most old individuals are uniform straw color or pale greenish-brown; this change in coloration does not occur in T. o. ornata.

Fig. 2. Dorsal and lateral views of skull of T. o. ornata (a and b) (KU 1172, male, from 6 ml. S. Garnett, Anderson Co., Kansas) and of T. carolina (c and d) (KU 39742, from northern Florida). Note the relatively higher brain-case and the incomplete zygomatic arch in T. o. ornata. All figures natural size.

Fossils

Of the several species of fossil Terrapene described (Hay, 1908b:359-367, Auffenberg, 1958), most are clearly allied to Recent T. carolina. One species, Terrapene longinsulae Hay, (1908a:166-168, Pl. 26) from "… the Upper Miocene or Lower Pliocene…." of Phillips County, Kansas, however, is closely related to T. ornata (if not identical). I have examined the type specimen of T. longinsulae. Stock and Bode (1936:234, Pl. 8) reported T. ornata from sub-Recent deposits near Clovis, Curry County, New Mexico.

Economic Importance

Ornate box turtles, referred to as "land terrapins" or "land tortoises" over most of the range of the species, are regarded by most persons whom I have queried as innocuous. These turtles occasionally damage garden crops and have been known to eat the eggs of upland game birds. Terrapene ornata is seldom used for food. A. B. Leonard told me the species was eaten occasionally by Arapaho Indians in Dewey County, Oklahoma. Several specimens in the University of Kansas Archeological Collections were found in Indian middens in Rice County, Kansas, from a culture dated approximately 1500 to 1600 A. D. The flesh of T. ornata occasionally may be toxic if the turtle has eaten toxic fungi as has been recorded for T. carolina (Carr, 1952:147).

Study Areas

Preliminary studies and collections of specimens were made at a number of localities in northeastern Kansas in 1953 and 1954. Two small areas were finally selected for more intensive study. One of these areas, the University of Kansas Natural History Reservation, five and one-half miles north-northeast of Lawrence in the northeasternmost section of Douglas County, Kansas, is a tract of 590 acres maintained as a natural area for biological investigations. Slightly less than two thirds (338 acres) of the Reservation is wooded; the remainder consists of open areas having vegetation ranging from undisturbed prairie grassland to weedy, partly brushy fields (Fitch, 1952). Although ornate box turtles were not numerous at the Reservation, the area was selected for study because: 1) there was a minimum of interference there from man and none from domestic animals; 2) the vegetation of the Reservation is typical of areas where T. ornata and T. carolina occur sympatrically (actually only one specimen of T. carolina has been seen at the Reservation); and, 3) availability of biological and climatological data there greatly facilitated the present study. Actual field work at the Reservation consisted of studies of hibernation and long-term observations on movements of a few box turtles.

A much larger number of individuals was intensively studied on a tract of land, owned by Sophia Damm, situated 12 miles west and one and one-half miles north of Lawrence in the northwestern quarter of Douglas County, Kansas. The Damm Farm lies on the southern slope of a prominence—extending northwestward from Lawrence to Topeka—that separates the Kansas River Valley from the watershed of the Wakarusa River to the south. The prominence has an elevation of approximately 1100 feet and is dissected on both sides by small valleys draining into the two larger river valleys.

The Damm Farm (see [Pl. 15]) has a total area of approximately 220 acres. The crest of a hill extends diagonally from the middle of the northern edge approximately two thirds of the distance to the southwestern corner. Another hill is in the extreme northwestern corner of the study area.

The northeastern 22 acres were wooded and had small patches of overgrazed pasture. Trees in the wooded area were Black Walnut (Juglans nigra), Elms (Ulmus americana, U. rubra), Cottonwood (Populus deltoides), and Northern Prickly Ash (Xanthoxylum americanum). The areas used as pasture had thick growths of Buckbush (Symphoricarpos orbiculatus) mixed with short grasses (Bromus japonicus, Muhlenbergia Schreberi, and Poa pratensis). Farm buildings were situated in the wooded area at the end of an entry road. The southeastern 74 acres were cultivated; corn, wheat, and milo were grown here and fallow fields had a sparse growth of weeds.

Most of the western two thirds of the study area, comprising 124 acres, was open rolling prairie (hereafter referred to as "pasture") upon which beef-cattle were grazed ([Pl. 16, Fig. 1]; [Pl. 17, Fig. 1]; [Pl. 18, Fig. 2]). Rock fences ([Pl. 17, Fig. 2]) two to four feet high bordered the northern edge, southern edge, and one half of western edge of the pasture. A wagon track lead from a gate on the entry road, along the crest of the hill, to a gate in the southern fence. Except for the latter gate and for ocassional under-cut places in low areas, there were no openings in the rock fences through which box turtles could pass. A few trees—American Elm, Hackberry (Celtis occidentalis), Red Mulberry (Morus rubra), Osage Orange (Maclura pomifera), Black Cherry (Prunus serotina), Box-Elder (Acer Negundo), and Dogwood (Cornus Drummondi)—were scattered along fences at the borders of the pasture and in ravines. Larger trees in a small wooded creek-bed at the southwestern edge of the pasture were chiefly Cottonwood, American Elm, Red Mulberry, and Black Willow (Salix nigra). The only trees growing on the pasture itself were a few small Osage Orange, none of which bore fruit.

Paths were worn along fences by cattle and in several places near the fence, usually beneath shade trees, there were large bare places where cattle congregated. Vegetation near paths and bare places was weedy and in some places there were tall stands of Smooth Sumac (Rhus glabra).

Rich stands of prairie grasses occurred along the top of the hill in the pasture; bluestems (Andropogon gerardi, A. scoparius) were the dominant species and Switchgrass (Panicum virgatum) and Indian grass (Sorghastrum nutans) were scattered throughout. A number of small areas on top of the hill were moderately overgrazed, as indicated by mixture of native grasses with an association of shorter plants consisting chiefly of Ragweed (Ambrosia artemisiifolia var. elatior), Mugwort (Artemisia ludoviciana), Japanese Chess (Bromus japonicus), and Asters (Aster sp.).

The upper parts of the hillsides were overgrazed moderately to heavily. Limestone rocks of various sizes were partly embedded in soil or lay loose at the surface. Depressions beneath rocks provided shelter for box turtles as well as for other small vertebrates. Native grasses were sparse in this area and gave way to Sideoats Grama (Bouteloua curtipendula), extensive patches of Smooth Sumac, and scattered colonies of Buckbrush.

Tall grasses were dominant on the lower hillsides and small patches of Slough grass (Spartina pectinata) grew in moist areas. Ravines originated at small intermittent springs on the sides of the hill. The banks of ravines were high and steep and more or less bare of vegetation. High, dense stands of Slough grass grew at intermittent springs and along the courses of ravines; sedges (Carex, sp.) grew where small pools of water formed and created marshy conditions. Prairie grasses along the tops of ravine embankments formed a narrow overhanging canopy of vegetation that was accentuated in many places where the sod was under-cut by erosion or by the activities of burrowing animals ([Pl. 18, Fig. 1]). Box turtles frequently sought shelter beneath this vegetational canopy or burrowed beneath the sod.

On the highest part of the pasture near the entry road several small areas were nearly bare, presumably because of heavy overgrazing; grasses (except for scattered clumps of Bouteloua curtipendula and Setaria lutescens) were absent and dominant vegetation consisted of Buffalo-bur (Solanum rostratum), Blue Vervain (Verbena hastata), Mullein (Verbascum Thapsus), Ragweed, Asters, and a few Prickly Pear (Opuntia humifusa). Two small areas on the pasture completely lacked vegetation; these may have been wallows or the sites of old salt-licks.

Three shallow stock ponds, behind earthen dikes in ravines, were present on the pasture. The pond near the farm buildings ("House Pond") and that in the southwestern part of the pasture ("Far Pond") were present when studies of box turtles were begun. The largest pond, in a deep ravine in the northern part of the pasture, was constructed in June, 1956, and became filled in approximately one month (Pls. [16] and [18]). Pond embankments were chiefly bare of vegetation because of trampling by cattle; in a few places at the edge of the water, or in places too steep for cattle to walk, there were small patches of weeds, sedges, and Slough Grass. The ponds contained some water at all times of the year. The only vertebrates permanently inhabiting the ponds in the course of my studies were Bullfrogs (Rana catesbeiana) and Leopard frogs (Rana pipiens).

The three parts of the pasture in which studies were concentrated were designated as separate subdivisions. The northwest corner area (28 acres) was triangular and bounded on two sides by rock fences and on its third side by a deep ravine. The southern ravine area (17 acres) constituted the part of the lower southern hillside drained by a series of ravines. The house pond area (seven acres) surrounded "House Pond." Habitat in these three subdivisions of the pasture was especially favorable for box turtles.

Materials and Methods

Observations were made at the Damm Farm on 102 days in the two-year period beginning in Autumn, 1954; observations were concentrated in the period from May to October although some observations were made in every month, January and February excepted. Field work was done chiefly in daylight hours but a few trips were made to the study area at night.

Routine handling of each turtle captured at the Damm Farm consisted of: marking, weighing and measuring turtle; recording the exact place of capture, body temperature and environmental temperature; and, recording miscellaneous items such as the presence of ectoparasites, injuries, distinctive markings, and in some instances, the approximate age of the turtle.

Excursions on the Damm Farm were made on foot in 1954 and 1955, and, in 1956, on horseback. By using a horse, more ground could be covered per unit of time, a better view could be obtained of immediate surroundings, and, cattle on the area, being accustomed to horses, did not become agitated as they would when unmounted persons were nearby.

The entire study area could not be inspected thoroughly in a single day. It was usually more profitable to find and mark turtles along fences, in ravines, or in other open areas, and subsequently to follow their movements away from these areas by means of trailing threads. Turtles could be observed from a distance through binoculars. Cultivated areas were regularly scanned with binoculars but turtles were seldom seen there. Behavior was observed by sitting motionless on rock fences or in a blind on top of a stepladder.

No box turtles were removed from the study area. Specimens obtained in other areas were used for studies of growth, reproduction, and food habits. Measurements, weights, and data concerning temperature and ectoparasites were obtained from specimens collected elsewhere as well as from individuals on study areas.

Turtles were obtained by hand-collecting and in unbaited traps; the number captured in a single day ranged from 12 to none. Traps, like those used by Packard (1956:9) for tree squirrels, were set in the mouths of burrows and dens, or—with leads to channel animals into the trap—along ravines and rock fences. Traps set in the open were covered to prevent death of turtles from overheating in direct sunlight. Live-trapping provided much valuable data, although quail, rabbits, opossums, and box turtles were caught with about equal frequency in the traps.

Turtles were marked by notching the marginal scutes of the carapace by means of a hacksaw blade, following the code system described by Cagle (1939). Notches, one eighth to one quarter of an inch deep and wide could be cut more quickly than filed and were more evident than drilled holes which often became plugged with soil and obscured. Hatchlings and juveniles were notched with a sharp knife.

Movements of individual turtles were studied by means of a turtle-trailing device—similar to the kind first described by Breder (1927) and later modified by Stickel (1950:355-356)—a tin can, cut to fit the shell of a turtle, with an axle that bore a spool of thread ([Pl. 27, Fig. 1]). The device was taped to the turtle; the free end of the thread was tied to a stationary object. Thread payed out from the spool through a guide-loop and marked the course of the turtle as it moved away from the starting point. Because of its great strength and elasticity (as compared to cotton), nylon sewing thread was used in trailers. Ordinarily, turtles were unable to break the thread if it became snarled or was expended. Cattle frequently tangled the thread and displaced it but did not often break it. Ordinary spools were cut down on a lathe so they would hold 600 to 800 yards of thread. Turtle-trailing provided an accurate record of where and how far a turtle had traveled, and to a lesser extent, the sort of activity in which the turtle had been engaged (evidence of feeding, forms, or trial nest holes). Trailers seemed not to alter the normal activity of turtles.

Prominent landmarks were rare or wanting in most places on the pasture. Locations of captures (or reference points in the movements of trailer-turtles) were determined by triangulation with a Brunton compass, using trees along fences as known points of reference. Rough maps were made in the field and used later, along with compass readings and measurements, to make a more precise record of movements and captures on a large map (scale, 100 feet to one inch) of the study area. Mapped points of capture in grassy areas were accurate within ten to twenty feet; points of capture in areas where landmarks were nearby were nearly exact. Areas were measured with a planimeter; distances traveled by individuals were measured with a cartometer.

Turtles were measured in the field to the nearest millimeter with large wooden calipers (of the type used by shoe salesmen) and a clear plastic ruler. Measurements in the laboratory, especially in studies of growth, were made, to the nearest tenth of a millimeter with dial calipers. Measurements made on each specimen examined in the field were: length of carapace, width of carapace, length of plastron (sum of lengths of forelobe and hind lobe), width of plastron (at hinge), and height. All measurements were made in a straight line. A spring scale of 500 gram capacity, used in the field, gave weights accurately within three grams. A triple-beam balance was used in the laboratory. Unless otherwise noted, measurements are expressed in millimeters and weights are expressed in grams.

Body temperatures were taken by means of a quick-reading Schultheis thermometer inserted into the distal portion of the large intestine with the bulb directed ventrally to avoid puncturing the bladder. Body temperature of turtles were altered little or not at all in the few seconds the turtles were held and no attempt was made (except for small juveniles) to insulate them from the warmth of my hands. Data recorded with body temperature were: air temperature (in shade, approximately one inch from turtle); ground temperature (or water temperature); behavior of turtle; weather conditions; nature of vegetation or other cover; and, time of day. Unless otherwise noted, temperatures are expressed in degrees Centigrade.

A maximum-minimum thermometer was installed near the buildings at the Damm Farm. Notes on general weather conditions were made on each visit to the study area. Additional climatological data were obtained from the U. S. Weather Stations in Topeka and Lawrence, from records at the Reservation, and from official bulletins of the U. S. Weather Bureau.

Stomachs and gonads were removed and preserved by standard techniques soon after specimens were killed. The dates given to gonads were, in all instances, the dates when the specimens were killed. Eggs were prepared for incubation in the manner described by Legler (1956). Females laying or containing eggs used in studies of incubation were preserved for further studies and comparison with young hatched from the eggs. Histological preparations were fixed in ten per cent formalin or Bouin's fluid, embedded in paraffin, and stained with hematoxalin and eosin.

Terminology

Names used for the epidermal and bony parts of the shell follow the classification proposed by Carr (1952:35-39). The terms "scute," "lamina," and "scale" are used here more or less interchangeably for the epidermal parts as are the terms "plate," "bone," and "element" for the bony parts of the shell.

The term "form" is used here in the same sense that Stickel (1950:358) used it in her study of T. carolina—to indicate a depression or cavity made by a turtle in vegetation or soil. Forms correspond closely in shape and size to shape and size of the turtle. Forms of T. ornata differ from those of T. carolina chiefly in being made most often in soil, over which there is a minimum of vegetational cover. The term "den" refers to natural cavities (or cavities of unknown origin) beneath rocks, in rock fences, or in cut banks. The term "burrow," unless otherwise noted, refers to burrows made by animals other than box turtles.

HABITAT AND LIMITING FACTORS

The known range of T. ornata includes the southern half of the Grassland Biome, part of the Desert Biome, and that part of the Temperate Deciduous Forest Biome known as the Prairie-Forest Ecotone. The species is found in microhabitats that differ widely in food supply, temperature, moisture, and kind of soil. In spite of its relatively high degree of morphological specialization, T. ornata is remarkably versatile in regard to habitat requirements.

Ornate box turtles are relatively inconspicuous in natural surroundings and collectors seldom seek out and obtain specimens under completely natural conditions as may be done with certain other reptiles and amphibians by turning rocks, tearing apart logs, or setting traps. Most series of specimens are obtained by hunting after rains on roads or other natural breaks in vegetational cover. Detailed information on habitat preferences is lacking.

Low temperature seems to be an important factor limiting the distribution of T. ornata in the northern part of its range. Box turtles, like nearly all other reptiles occurring at these latitudes, spend the winter in underground hibernacula. The depth to which the ground freezes in the coldest part of the winter is therefore a critical factor. The ground freezes to an average depth of 30 inches or less over most of the range of the species; only in the extreme northern part of the range (southern South Dakota, southeastern Wyoming) does the ground freeze to an average depth of as much as 35 inches. Average depth of freezing is, in fact, less than 15 inches over more than one half the range of the species. The average number of frost-free days per year ranges from 130 to 140 days in the northern part of the range to more than 250 days in the southwestern part of the range.

Terrapene ornata occurs from near sea level to elevations of more than 5000 feet. Both subspecies are found at both high and low elevations but luteola is more consistently taken at high elevations than ornata. The latter subspecies commonly occurs at elevations above 4000 feet on the high plains in extreme western Kansas and eastern Colorado; the highest elevation from which I have examined specimens of T. o. ornata is between 4600 and 4700 feet near Akron, Washington County, Colorado. The greater part of the known range of T. o. luteola lies above 3000 feet.

Norris and Zweifel (1950:1) observed T. o. luteola on the Jornada del Muerto, an elongate plain approximately 4500 feet high, in southeastern Socorro County, New Mexico; box turtles were abundant on the level part of the plain and on the bordering foothills but not at higher elevations where the substratum was rocky. The authors otherwise noted no preference for any kind of soil. The principal elements of the plant associations in which the turtles were found were creosote bush, yucca, mesquite, juniper, tarbush, and grasses. Lewis (1950:3) reported that T. ornata luteola inhabited the yucca-grassland zone in Dona Ana County, New Mexico; he stated (op. cit.: 10) that individuals were commonly found on roads after rains and in cloudy weather. No specimens were taken at altitudes higher than 4300 feet.

I have examined specimens of luteola from elevations of approximately 5500 feet in Cochise County, Arizona, and Lincoln County, New Mexico. These localities are probably at or near the maximum elevation at which the species occurs. The texture of the substrate is the most important factor limiting vertical distribution. Ornate box turtles, like nearly all other turtles, excavate nests; T. ornata is a burrower, at least for purposes of hibernation. Populations of the species, therefore, could not survive in areas of hard unyielding substrata. Such substrata seem to be the most important factor limiting altitudinal distribution.

Most of the area in which T. ornata occurs is semiarid or arid. Average precipitation in the warm season (April through September) varies from approximately 25 inches in the northeast to less than ten inches in the southwest. In drier parts of the range, precipitation is unevenly distributed over the warm season. Long, hot, dry periods are unfavorable for reptilian activity. T. ornata, like many other reptiles inhabiting dry regions, survives long periods without water by seeking shelter (usually underground) and remaining quiescent. Populations of the subspecies luteola live under far more rigorous conditions in this respect than do the more northern populations. Specimens of luteola from Arizona that were kept for several years in the laboratory under dry conditions and fed adequately, but at infrequent intervals, were able to remain healthy and even to grow whereas examples of ornata kept under the same conditions soon languished and died; luteola seems to be physiologically adapted for existence under arid conditions, where normal activity is sometimes possible for only a few weeks in the year.

The prairies of Nebraska, Kansas, Oklahoma, and northern Texas seem to provide the most nearly optimum habitat for the species; in these regions box turtles are active on a large majority of the days from April to October in years having average or better than average precipitation and population density seems to be greater than in the more arid parts of the range.

Activities of man have probably affected the density of populations of the ornate box turtle in many parts of its range but appear not to have acted as limiting factors except in certain areas along the northern edge of the range (Blanchard, 1923:19-20, 24) where disruption of grassland through intensive cultivation probably has excluded the species. Unlike certain other reptiles of the Great Plains (Fitch, 1955:64), T. ornata seems not to have been affected—either by direct decimation of populations or by disruption of habitat—by intensive zoological collecting in restricted areas. Environmental changes such as those resulting from overgrazing and erosion, or from protection of the habitat from grazing could be expected to cause long-term changes in populations of ornate box turtles.

Terrapene o. ornata is an omnivorous, opportunistic feeder, primarily insectivorous but able to subsist on nearly any sort of animal or vegetable food. The general food habits of luteola are poorly known but probably resemble those of ornata. Although kind of food available probably does not limit the distribution of T. ornata there are indications that it influences population density. In Kansas, for example, dung insects are an important staple in the diet and box turtles were found always to be more numerous in areas where domestic cattle provided an abundant supply of dung than elsewhere. A similar relationship probably existed in former times between box turtles and native ungulates. Near extinction of buffalo in the Great Plains possibly caused a decrease in populations of box turtles. Henry S. Fitch told me that the number of T. ornata at the Reservation gradually declined after cattle were removed from the area in 1948.

In summary, the distribution of T. ornata seems to be limited by: 1) Presence of a substrate too hard to permit digging of nests and forms (southwestern and western edges of range); 2) temperatures causing the ground to freeze deep enough (approximately 30 inches) to kill turtles in hibernacula (northern edge of range); and, 3) the lack of one or more relatively wet periods in the course of the warm season, preventing at least temporary emergence from quiescence (southwestern edge of range).

HABITAT IN KANSAS

Clarke (1958:40-45) reported T. o. ornata in all terrestrial communities studied in Osage County; he considered the subspecies to be characteristic of the "… cultivated-field community …" and to be of frequent occurrence in (but not characteristic of) the "… Oak-Walnut Hillside Forest …, Buckbrush-Sumac …, and Prairie communities …". Brennan (1937:345) found T. o. ornata to be equally abundant in mixed prairie and prairie-streamside habitats in Ellis County; the subspecies was much rarer on rocky hillsides and in the habitat surrounding prairie ponds. Carpenter (1940:641) listed T. o. ornata as an inhabitant of "… tall and mixed-grass prairies …" (also in Oklahoma and Nebraska). Fitch (1958:99) found the order of preference for habitats at the Natural History Reservation to be grazed pasture land, woodland, open fields with undisturbed prairie vegetation, and fallow fields with a rank growth of weeds.

At the Damm Farm the greatest number of box turtles was collected on the pasture, especially in three areas designated in [Plate 15] as the "northwest corner," "southern ravine," and "house pond" areas. These three areas had several features in common. All contained ravines and rocky slopes that provided many places of concealment (dens, burrows of larger animals, and suitable substrate for the excavation of earthen forms). All contained water (in ponds and intermittent streams) for most of the year; and, all were frequented daily by cattle that left an abundant supply of dung in which box turtles foraged. In addition, each of the three areas contained at least one mulberry tree, under which fruit was abundant in the months of June and July.

The relative numbers of box turtles found in different areas on the Damm Farm were, of course, governed to some extent by my activity in these areas and by the relative ease with which box turtles were seen in different types of vegetational cover. Turtles were more easily seen in the pasture (especially in sparsely vegetated or denuded areas) where much of my field work was done on horseback, than in the wooded areas, where excursions were usually made on foot. It was evident, however, after mapping known ranges and studying patterns of movement in marked turtles, that concentrations in the three above-mentioned areas of pasture were an indication of actual preference by turtles for the more favorable habitat in these areas rather than the result of incomplete sampling.

REPRODUCTION

Mating

Mating takes place throughout the season of activity but is most common in spring—soon after emergence from hibernation—and in autumn. Turtles frequently copulated in the laboratory in spring and autumn. Copulation was observed under natural conditions on several occasions but only once at the Damm Farm.

Norris and Zwiefel (1950:4) saw two captive individuals of T. o. luteola copulating on 12 August; copulation lasted two hours. Brumwell (1940:391-2) gave the following description of mating in T. o. ornata. A male pursued a female for nearly half an hour, first nudging the margins of her shell and later approaching her rapidly from the rear and hurling himself on her back in an attempt to mount, at the same time emitting a stream of liquid from each nostril. The liquid was presumably water; both sexes had imbibed water in a pond just before courtship began. Brumwell suggested that pressure on the plastron of the male had forced the water out his nostrils. The pair remained in the coital position for 30 minutes after the male had achieved intromission. In another instance, Brumwell (loc. cit.) saw four males pursuing a single female, the males exhibiting the same behavior (nudging and lunging) outlined above. Males that attempted to mount other males were repelled by defensive snapping of the approached male. The female also snapped at some of the males that tried to mount her. One male was finally successful in mounting and was henceforth unmolested by the other males. Brumwell suggested that shell biting and tapping may be methods of sex-recognition.

In the several instances of mating that I observed, the male, after mounting the shell of the female ([Pl. 28]), gripped her, with the first claws of his hind feet, just beneath her legs or on the skin of the gluteal region and, with the remaining three claws, gripped the posterior edges of her plastron. In most instances the female secured the male's legs by hooking her own legs around them. The coital position of T. ornata seems to differ from that of T. carolina, at least in regard to the position of the male's legs. The coital positions of T. carolina illustrated by Cahn (1937:94, Fig. 13) are physically impossible for T. ornata.

In T. ornata the pressure exerted on the male's legs by the female probably impairs circulation and probably is painful to the male, especially after coitus, when the male falls backward but is still held by the female. The heavily developed musculature of the legs of males may be an adaptation to strengthen the legs for this temporary period of stress. Evans (1953:191) and Cahn and Conder (1932:87-88) observed the hind legs of males of T. carolina to be noticeably weakened after copulation, causing the males to remain inactive for several hours.

Evans (op. cit.) observed 72 matings of T. carolina and divided the process into three phases as follows: 1) circling, pushing and biting by the male; 2) mounting (female with shell closed); and, 3) coition (female with shell open). Penn and Pottharst (1940:26) reported that captive T. carolina in New Orleans mated chiefly under conditions of optimum temperature (21 to 27° C.) and high humidity; some matings took place in a pool of water. Males pushed females about after mating, often rolling them over several times.

Because ornate box turtles observed by me were able easily to right themselves from an inverted position on substrata of all kinds, males left lying on their backs after copulation are probably in no danger of perishing in this position, as was suggested by Allard (1939) for T. carolina.

Insemination

Oviducts of several females were flushed by means of a pipette to determine whether they contained sperm. Approximately half of the females captured in May, 1956, had sperm in their oviducts, but females captured in June and July did not. Sperm flushed from the oviducts were in clumps of several hundred and showed no sign of motility a few minutes after the female was anesthetized with chloroform. No sperm were found in the oviducts of immature females but one female of nearly adult size was observed in copulation with a mature male.

Thorough examination of microscopic sections of oviduct (taken at various times in the season of activity) usually revealed a few sperm lodged in the folds ([Pl. 19, Fig. 8]) of the cephalic as well as the caudal portion of the tube, but no specialized seminal receptacles such as occur in snakes (Fox, 1956) were present. Fertilization without reinsemination probably occurs in T. ornata. Ewing (1943) and Finneran (1948:126) reported that females of T. carolina produced fertile eggs for periods of four and two years, respectively, after being removed from all contact with males.

Sexual Cycle of Males

Testes were preserved in each month from April to October. The following description of spermatogenesis is based chiefly on material collected in 1955, although testes were preserved also in 1954. Comparison of material obtained in 1954 and 1955 revealed that spermatogenesis began earlier and was more advanced on any given date in 1955 than in 1954.

Testes of mature individuals are pale yellow and slightly oblong. The epididymis is ordinarily dark brown or black and contrasts sharply with the color of the testes. Size of testes was expressed as the average length (greatest diameter) of both testes. Testes are smallest in April, immediately after emergence from hibernation, and largest in early September ([Pl. 20, Figs. 3-4]). They are nearly spherical when of maximum size; increase in bulk, therefore, is relatively greater than the increase in size shown in [Figure 3]. They increase in size from April until early June, recede during most of June, and again increase in size in July and August. They remain large from early September until hibernation is begun, becoming only slightly smaller in late September and October.

Increase in size following emergence from hibernation may be due in part to proliferation of the sustentacular cytoplasm. Decrease in size in early June is correlated with the end of the period of most active mating; maximal size is coincident with the peak of the spermatogenic cycle in early September.

Fig. 3. Seasonal fluctuations in size (average greatest diameter) of testes in T. o. ornata as determined by examination of 40 specimens from eastern Kansas.

Spermatogenesis (refer to [Pl. 19, Figs. 1-5]) begins in early May when a few spermatogonia appear in the seminiferous tubules. The histological appearance of testes preserved in April and May is much the same. Nuclei of Sertoli cells, which outnumber the spermatogonia, are evident at the periphery of the tubules and the clear cytoplasm of the cells extends into and nearly fills the lumina. The few darkly stained spermatids that are present in April are cells that probably were produced in the previous summer. Sperm are present in small groups within the sustentacular cytoplasm, but ordinarily are absent in the lumina.

Primary spermatocytes appear in the tubules from mid-May to early June. By mid-May there are practically no sperm at any place in the tubules. The sustentacular cytoplasm has a less compact arrangement in late May than in April.

Spermatogenesis is well under way by mid-June; at this time, two or three distinct layers of primary and secondary spermatocytes are present and these cells outnumber the Sertoli cells. The lumina are filled with cellular detritus and are no longer bordered by a clear ring of sustentacular cytoplasm. No sperm are present.

Spermatids appear in late June and a few of them undergo metamorphosis in early July; by mid-July, spermatids and secondary spermatocytes are the dominant cells in the seminiferous tubules, although spermatogonia are still active.

By late August, clusters of sperm and metamorphosing spermatids surround the Sertoli cells; large numbers of sperm as well as sloughed cells representing various spermatogenic stages are present in the lumina. Secondary spermatocytes are still evident near the periphery of the tubules but they are much less numerous than spermatids. The germinal epithelium is still semiactive and small groups of primary spermatocytes are present in nearly all of the tubules.

The spermatogenic cycle is completed in the latter half of October when most of the spermatozoa pass into the epididymides. A few spermatozoa and spermatids remain in the seminiferous tubules during hibernation. Although no testicular material was obtained from hibernating turtles, comparisons of sections made in October and April show that the germinal epithelium remains inactive from autumn until spring. Possibly some spermiogenesis takes place in the early phases of hibernation or in the period in late autumn when turtles are intermittently active. It is uncertain whether the reorganization of the sustentacular cytoplasm occurs in autumn, in spring, or in the course of hibernation.

The seminiferous tubules of immature males are small, lack lumina, and contain a few large but inactive spermatogonia ([Pl. 19, Fig. 6]). The testes of specimens that were nearly mature contained primary and secondary spermatocytes but lacked lumina; it was thought that such individuals would have matured in the following summer and bred in the following autumn.

Mature sperm were found in epididymides at all times of the year but were most numerous in spring and autumn, the period between spermatogenic cycles ([Pl. 19, Fig. 7]). Sperm expelled from the epididymides in autumn matings are seemingly replaced by others from the seminiferous tubules; the epididymides become much smaller when their supply of sperm is nearly exhausted after spring mating.

Risley (1938:304) found the testes of the common musk turtle, Sternotherus odoratus, to be largest in August and smallest in early May. Recession of testes in spring was coincident with the period of active breeding; increase in size, later in the season, corresponded to increasing spermatogenic activity and enlargement of seminiferous tubules. Altland (1951:600-603) found the spermatogenic cycle of Terrapene carolina to be nearly like that of Sternotherus odoratus. Fox (1952) found that testes of garter snakes (Thamnophis sirtalis and T. elegans) in California reached a peak of spermatogenic activity in midsummer, regressed in the latter half of the summer, and were inactive in winter.

The spermatogenic cycle of T. ornata as here reported, differs in no important respect from those of Thamnophis, Sternotherus odoratus, or Terrapene carolina, except that in T. ornata the cycle begins and ends somewhat later in the season of activity. In most of the lizards that have been studied (Fox, 1952:492-3), spermatogenesis reaches a peak in spring (more or less coincident with the mating period and with ovulation) and the germinal epithelium remains active in winter. Sternotherus, Terrapene, and Thamnophis are alike in completing spermatogenesis late in the season and storing spermatozoa, in the seminiferous tubules or in the epididymides, during hibernation.

It is noteworthy that, in the turtles and snakes mentioned above, sperm produced in autumn are used to fertilize eggs laid in the following year, and mating [with the exception of Thamnophis elegans, (Fox, 1956)] occurs in both spring and autumn. It is not definitely known in any of these instances, whether sperm resulting from autumn or spring inseminations (or both) fertilize the eggs. Risley (1933:693) found motile sperm in the oviducts of female Sternotherus odoratus that had recently emerged from hibernation; he believed that spring mating, although it commonly occurred, was not necessary to fertilize eggs. Disadvantages, if any, of completing spermatogenesis well in advance of ovulation seem to be at least partly counteracted by two annual mating periods or by mating throughout the season of activity.

Sexual Cycle of Females

The following account of oögenesis is based on examination of preserved ovaries from 68 mature specimens. The ages of most specimens were known, inasmuch as the specimens were used in studies of growth as well as gametogenesis. Other data were obtained from adult females that were dissected but not preserved, and from immature females.

Fig. 4. Seasonal fluctuations in ovarian weight in T. o. ornata, as determined by examination of 60 specimens from eastern Kansas.

Size of ovarian follicles was determined by means of a clear plastic gauge containing notches 5, 10, 15, 20, and 25 millimeters wide. The number of follicles within a given size range could be quickly determined by finding the smallest notch into which the follicles fit. It was necessary to weigh all ovaries after preservation since some of them had not been weighed when fresh. Since all ovarian samples were preserved in the same manner, weights remained relatively the same. Preserved material was lighter than fresh by an average of 13 per cent. Follicles less than one millimeter in diameter were not counted. Corpora lutea and corpora albicantia were studied under a binocular dissecting microscope. No histological studies were made of the female reproductive system.

Ovarian follicles and oviducal eggs were recorded separately for the right and left sides. Each ovary was always kept associated with the oviduct of the same side, but in some instances it was not recorded whether the organs were left or right.

Ovaries ordinarily weighed most in October, March, and April, when most females contained enlarged follicles, and least in August and September when the supply of enlarged follicles was usually exhausted (Figs. [4] and [5]).

Fig. 5. The seasonal occurrence of enlarged ovarian follicles in females of T. o. ornata, expressed, for each month, as the percentage of total females that contained two or more follicles having diameters greater than 15 mm. Total number of females in each of the samples is shown in parentheses at the top of each bar.

The ovarian cycle begins in July or August, after ovulation has occurred. At that time many minute follicles form on the germinal ridges of the ovaries. On the basis of the material that I examined, it seems that ovarian follicles either grow to nearly mature size in the season preceding ovulation and remain quiescent over winter or grow rapidly in the period of approximately six weeks between spring emergence and ovulation. Altland (1951:603-5) reported that the former condition was the usual one in T. carolina; he suggested that possibly some of the enlarged follicles were absorbed during hibernation.

Examination of yolks of oviducal eggs revealed that follicles mature when they reach a diameter of 16 to 20 millimeters and a weight of two to two and one-half grams ([Pl. 20, Fig. 1]).

The enlarged follicles remaining on the ovaries after ovulation (excluding those smaller than six mm.) can be grouped according to diameter as: large (greater than 15 mm.), medium (11 to 15 mm.), and small (six to 10 mm.). Ten females collected in the period from June 2 to 8, after they had ovulated, all had follicles falling in at least one of these size groups, and eight had follicles falling in two or more of the groups. In females having enlarged follicles of more than one of the size groups, there were several follicles in each of two groups and no follicles, or only one follicle, in the remaining group. Enlarged follicles represent future clutches but whether the enlarged follicles will be ovulated in the same season or in a later season is questionable.

Evidence found in the present study suggested that at least a few females lay more than one clutch of eggs per year. Among 34 specimens obtained in June and July, eight (24 per cent) had corpora lutea (or easily discernible corpora albicantia) and at least two follicles more than 15 millimeters in diameter; in three specimens (9 per cent) the ovaries bore fresh corpora lutea (representing recent ovulations) and a set of older corpora lutea (representing ovulations that had occurred several weeks previously). It was thought that each of these eleven females (33 per cent of sample) had produced or would have produced two clutches of eggs in the season of its capture. The number of large follicles present after the first set of ovulations (mean, 3.5) was fewer in most instances than the average clutch-size (see below), indicating that second clutches are smaller than first clutches. Smaller second clutches were found also in T. carolina (Legler, 1958).

Further evidence for multiple clutches was the absence of enlarged ovarian follicles in some females obtained in September. Atretic follicles, ordinarily orange, brown, or purplish, were observed on the ovaries of many of the females examined; in most instances, not more than two follicles of the small or medium size groups were atretic. Atresia was in no instance great enough to account for the complete loss of enlarged follicles.

Further study probably will show that many of the females laying in May and early June lay again before the end of July, and that eggs in the oviducts of females captured in the latter month frequently represent second clutches. Under favorable conditions, eggs laid by the end of July would have a good chance of hatching before the advent of cold weather in autumn; turtles hatching too late to escape from the nest could burrow into its sides and probably escape freezing temperatures.

Cagle's findings concerning Pseudemys scripta (1950:38) and Chrysemys picta (1954:228-9) suggest that these species lay more than one clutch per season, at least in the southern parts of their ranges. Carr (1952) indicated that multiple layings were known in most species of marine turtles (families Dermochelydae and Chelonidae) and strongly suspected in other species. Other turtles recorded to have produced multiple clutches in a single season (based chiefly on captive specimens or cultured populations) include: the starred tortoise, Geochelone elegans (Deraniyagala, 1939:287); the Asiatic trionychid, Lissemys punctata (op. cit.:304); the diamond-backed terrapin, Malaclemys terrapin (Hildebrand and Prytherch, 1947:2); and the Japanese soft-shelled turtle, Trionyx japonicus (Mitsukuri, 1895, cited by Cagle, 1950:38).

There is a marked alternation of ovarian activity in T. ornata, one ovary being more active than its partner in a given season. The less active ovary is more active than its partner in the following season. For example, a specimen killed in July had four corpora lutea on the right ovary and two on the left and there were five enlarged follicles (of the medium size group), representing the next set of eggs to be ovulated, four on the left ovary and one on the right. Similar alternation of ovarian activity was observed, to a greater or lesser extent, in nearly all of the females examined. Many subadult females that were approaching their first breeding season (as evidenced by the presence of large ovarian follicles but no indication of former ovulation) had but one active ovary. This may account in part for the tendency of small, young females to lay clutches smaller than average. One ovary may become senile in old females before its partner does; this may explain the occasional absence or atrophy of one ovary in large females that I have examined.

In all the specimens examined, it was evident that ovulation had occurred or would occur in two successive seasons. Senile or young females might, however, be expected to skip a laying season if only one ovary was functioning.

After ovulation, the collapsed follicle assumes a cuplike shape and becomes a glandular corpus luteum ([Pl. 20, Fig. 2]). Corpora lutea are approximately eight millimeters in diameter and are easily discernible at least until the eggs are laid; they are somewhat less distinct after preservation. Corpora lutea undergo rapid involution following oviposition and, after two to three weeks, are little more than small puckerings on the ovarian epithelium. At this stage they are properly referred to as corpora albicantia and are discernible only after careful examination of the ovary under low magnification. Corpora albicantia remain on the ovary until April of the year following ovulation but disappear in May and are never present after the new set of eggs is ovulated. Ovaries of some sub-adults (that would have laid first in the season following capture) contained enlarged follicles and, but for their lack of corpora lutea and corpora albicantia, were indistinguishable from those of older, fully mature females.

Altland (1951:605-610) gave a histological description of the corpus luteum of Terrapene carolina. Corpora lutea were glandular and filled with lipoidal material until the eggs were laid. Atresia of corpora lutea began when eggs were laid, was completed by mid-August, and was coincident with atresia of large follicles that did not undergo ovulation. Altland did not describe the gross external appearance of the corpus albicans.

The corpus luteum of oviparous reptiles seems to be closely associated with the intrauterine life of the eggs and, in viviparous reptiles, it may be an important factor in maintaining optimum gestational environment; however, its functions in all reptiles are poorly understood (Miller, 1948:200-201).

Information gleaned from records of gravid females and known dates of nesting suggests that eggs are retained in the oviducts two to three weeks before laying. Once they are ovulated, the eggs are exposed to but few hazards until laid; counts of corpora lutea are an accurate indication of the number of eggs laid. In the gravid females examined by me, number of corpora lutea on the ovaries was equal, in all but one instance, to the number of oviducal eggs. In the single instance in which an extra corpus luteum was found, one egg had probably been laid before the specimen was captured. The high incidence of correspondence between counts of corpora lutea and counts of oviducal eggs indicates also that T. ornata deposits the entire complement of oviducal eggs at one time, not singly or in smaller groups.

Extrauterine migration of ova, whereby eggs from one ovary pass into the oviduct of the opposite side, is of common occurrence in T. ornata and is known to occur also in T. carolina, Chrysemys picta, Emydoidea blandingi, Pseudemys scripta, Cnemidophorus sexlineatus, and in several mammals (Legler, 1958). This ovular migration may serve to redistribute eggs to the oviducts when the ovaries are functioning at unequal rates.

The eggs acquire shells soon after they enter the oviducts. No shell-less eggs were found in oviducts but several specimens of T. ornata had oviducal eggs, the thin, parchmentlike shells of which lacked the outer calceous layer; in these specimens the corpora lutea were fresh, probably not more than two days old. Eggs that had remained in the oviducts longer had a calceous layer on the outside of the shell. Eggs having incompletely developed shells were successfully incubated in the laboratory. Cagle (1950:38) found shelled but yolkless eggs in the oviducts of several Pseudemys scripta but found no yolkless eggs in nests. No yolkless eggs were found in specimens of T. ornata in the course of the present study.

The uterine portion of the oviducts becomes darkened (pale gray to intense black) in the breeding season. Darkening of oviducts seemed to coincide with the period when eggs were in the oviducts and it persisted for a variable length of time after the eggs were laid. Oviducts of immature females were ordinarily pale.

Nesting

Ornate box turtles nest chiefly in June. Some females nest as early as the first week of May or as late as mid-July but the nesting season reaches its peak in mid-June. Eggs nearly ready to be laid were in oviducts (determined by bimanual palpation in the field or by dissection in the laboratory) of many females captured in June; nearly half of the records so obtained were in the second week of that month. Early records of shelled oviducal eggs were April 25 (specimen from Ottawa County, Oklahoma), May 5, and May 22. The two latest records are for females retaining oviducal eggs on July 2 and 11. Known dates for nesting of free-living females were distributed rather evenly through the month of June. It is worthy of note that all (four) of the nestings known to occur in July were by captive females. Females of T. ornata, like those of some other turtles (Cagle and Tihen, 1948; Risley, 1933:694), seem to retain their eggs until conditions are suitable for nesting. Most of the reports in the literature of nesting after mid-July represent records for captive females.

Nests of T. o. ornata were so well-concealed that they were difficult to find even when a gravid female had been followed to the approximate location by means of a trailing thread. Females spend one to several days seeking a site for the nest, usually traveling a circuitous route within a restricted area. Movements of nest-seeking females were more extensive than those of males and non-gravid females observed in the same periods.

Activities of one gravid female, typical in most respects of the activities of several other gravid females observed (for periods of one to 23 days) at the Damm Farm, illustrate pre-nesting behavior ([Fig. 29]). A trailer was attached to the female on the morning of June 7. She was recovered early on the following afternoon; her movements in the elapsed period had been restricted to a small, deep, ravine 150 feet long and 20 to 30 feet wide. She had traversed each edge of the ravine at least once and had crossed it six or seven times, keeping mostly to areas on the upper parts of south—or west—facing slopes where vegetation was sparse or lacking. In six places she had dug into the ground, probably to test the suitability of the soil for nesting. In three places she dug beneath rocks that jutted out from the bank, and in two places merely scratched away the upper crust of soil. Her most recent attempt at digging (probably late the previous evening or in early morning on the day of her capture) consisted of a flask-shaped cavity that, but for the lack of eggs and a covering of earth, was like a completed nest ([Pl. 21, Fig. 1]). The cavity was 55 millimeters deep, 80 millimeters wide at the bottom, and 60 millimeters wide at the opening. For several inches about the opening the earth was slightly damp. That piled on the rim of the opening was of the consistency of thick mud, indicating that the female had voided fluid first on the surface of the earth and again inside the cavity to soften the soil. Subsequently during eight days her activities were similar but not so extensive as on the day described above. It was determined by daily palpation that she laid her eggs somewhere in the general area of the ravine on June 15 but the nest could not be found.

No completed nests containing eggs were discovered at the Damm Farm but the locations of several robbed nests and partly completed nests provided some information on preferred sites. The nests found were on bare, well-drained, sloping areas and were protected from erosion by upslope clumps of sod or rocks. The nest cavity illustrated in [Plate 21] was at the edge of the sod-line on the upper lip of the west-facing bank of a ravine. One nest had been excavated in a shallow den beneath an overhanging limestone rock. Three nests were on west- or south-facing slopes and one was on the north-facing bank of a ravine. Box turtles presumably select bare areas for nesting because of the greater ease of digging. One female at the Damm Farm was thought to have laid her eggs in a cultivated field and William R. Brecheisen told me he discovered two nests in a wheat field being plowed in July, 1955.

The repeated excavation of trial nest cavities presumably exhausts the supply of liquid in the female's bladder. Frequent imbibing of water is probably necessary if the search for a nesting site is continued for more than a day or two. Standing water was usually available in ponds, ravines, ditches, and other low areas at the Damm Farm in June. Nesting in June, therefore, is advantageous not only because of the greater length of time provided for incubation and hatching but also because of the amount of water available for drinking. Females can probably be more selective in the choice of a nesting site if their explorations are not limited by lack of water.

Females of T. ornata, in all instances known to me, began excavation of their nests in early evening and laid their eggs after dark; Allard (1935:328) reported the same behavior for T. carolina.

William R. Brecheisen, on July 22, 1955, at his farm, two miles south and one mile west of Welda, Anderson County, Kansas, observed that a large female began digging a nest in an earth-filled stock tank at 6:00 P. M. At first she moved her body about on the surface of the earth, loosening it and pushing it aside with all four legs, making a depression approximately two inches deep and large enough to accommodate her body. At 7:30 P. M. she began digging alternately with her hind feet at the bottom of the depression. Digging continued until 10:00 P. M., at which time the nest cavity was three inches deep, and three inches in diameter, with a smaller opening at the top. Six eggs were laid in the next half-hour. Covering of the nest probably took more than one hour but observations were terminated after the final egg was laid. By the following morning the nest-site had been completely covered and was no different in appearance from the rest of the earthen floor of the tank. (Brecheisen observed more of the nesting than anyone else has recorded and I am obliged to him for permission to abstract, as per the above paragraph, the notes that he wrote on the matter.)

A nest made by a captive female at the Reservation was of normal proportions except for an accessory cavity that opened from the neck of the nest, immediately below the surface of the ground. This smaller cavity contained a single egg. This peculiar nest may have resulted from the efforts of two different females since several were kept in the same outdoor pen.

Ten adult females were kept in an outdoor cage in the summer of 1955. The cage was raised off the ground on stilts and its floor was covered with 12 inches of black, loamy soil. A small pan of water was always available in the cage and the turtles were fed greens, fruit, and table scraps each evening. Nesting activity was first noted on June 21, when one of the females was digging a hole in a corner of the enclosure. She dug with alternate strokes of her fully-extended hind legs in the manner described (Legler, 1954:141) for painted turtles (Chrysemys picta bellii). Nevertheless, digging was much less efficient than in Chrysemys, because of the narrow hind foot of the female T. ornata; approximately half of the earth removed by any one stroke rolled back into the nest or was pulled back when she reinserted her leg. The female stopped digging when I made sudden movements or held my hand in front of her. Digging continued for approximately 45 minutes; then the female moved away and burrowed elsewhere in the cage. The nest cavity that she left was little more than a shallow depression. Three other females were digging nests early in the evening on July 3, 5, and 8; in each of these instances the female stopped digging to eat when food was placed in the cage and completed the nesting process, unobserved, later in the evening. In each instance where nest-digging by captive females was observed, the hind quarters of the female rested in a preliminary, shallow depression, and the anterior end of the body was tilted upward at an angle of 20 to 30 degrees. In late June and early July several eggs were found, unburied, on the floor of the cage and in the pan of water.

The excavation of a preliminary cavity by captive females may not represent a natural phenomenon. Allard (1935) made no mention of it in his meticulous description of the nesting process in T. carolina. It is worthy of mention, however, that Booth (1958:261) reported the digging of a preliminary cavity by a captive individual of Gopherus agassizi.

Eggs

The number of eggs in 23 clutches ranged from two to eight (mean, 4.7 ± 1.37 σs]); clutches of four, five, and six eggs were most common, occurring in 18 (78 per cent) instances. The tendency for large females to lay more eggs than small females ([Fig. 6]) was not so pronounced as that reported by Cagle (1950:38) for Pseudemys scripta. The small size of T. ornata, in comparison with other emyid turtles, seemingly limits the number of eggs that can be accommodated internally. The number of eggs per clutch in T. carolina [2 to 7, average 4.2, Allard (1935:331)], is nearly the same as that of T. ornata.

Fig. 6. The relation of plastral length to number of eggs laid by 21 females of T. o. ornata from eastern Kansas.

Shells of the eggs are translucent and pinkish or yellowish when the eggs are in the oviducts. After several days outside the oviducts the shells become chalky-white and nearly opaque. Eggs incubated in the laboratory retained the pinkish color somewhat longer than elsewhere on their under-surfaces, which were in contact with moist cotton, but eventually even this part of the shell became white. Infertile eggs remained translucent and eventually became dark yellow, never becoming white; they could be distinguished from fertile eggs on the basis of color alone. Shells of infertile eggs became brittle and slimy after several weeks.

The outer layer of the shell of a freshly laid egg is brittle and cracks when the egg is dented. After a few days, when the eggs begin to expand, the shell becomes flexible and has a leathery texture. The shell is finely granulated but appears smooth to the unaided eye. The granulations are approximately the same as those illustrated by Agassiz (1857:Pl. 7, Fig. 18) for T. carolina.

Eggs are ellipsoidal. Data concerning size and weight (consisting of mean, one standard deviation, and extremes, respectively) taken from 42 eggs (representing 9 clutches) within 24 hours after they were laid, or dissected from oviducts, are as follows: length, 36.06 ± 2.77 (31.3-40.9); width, 21.72 ± 1.04 (20.0-26.3); and weight, 10.09 ± 1.31 (8.0-14.3). There was a general tendency for smaller clutches to have larger eggs; the largest and heaviest were in the smallest clutch (two eggs) and the smallest were in the largest clutch (eight eggs). Risley (1933:697) reported such a correlation in Sternotherus odoratus, as did Allard (1935:331) in T. carolina. Measurements in the literature of the size of eggs of T. ornata suggest a width greater than that stated above, probably because some eggs already had begun to expand when measured.

Eggs of T. ornata expand in the course of incubation, as do other reptilian eggs with flexible shells, owing to absorption of water. In the laboratory, 48 eggs increased by an average of approximately three grams in weight and three millimeters in width over the entire period of incubation; increase in width coincided with decrease in length. Cotton in incubation dishes was kept moist enough so that some water could be squeezed from it. When the cotton was constantly moist, eggs showed a fairly steady expansion from the first week of incubation until hatching. The process could be reversed by allowing the cotton to dry. Eggs that were allowed to dry for a day or more became grossly dented or collapsed. Eggs at the periphery of the incubation dish were ordinarily more seriously affected by drying than were those at the center or in the bottom of the dish. A generous re-wetting of desiccated eggs and cotton caused the eggs to swell to their original proportions within 24 hours. Recessions occurred, however, even in the clutches that received the most nearly even amount of moisture. Increases in weight and size seemed to reach a peak in the middle of the incubation period and again immediately before hatching. Infertile eggs expanded in the same manner as fertile eggs in the first week or two of incubation, but thereafter gradually regressed in bulk or failed to re-expand after temporary periods of dryness. Fertile eggs that were in good condition had a characteristically turgid, springy feel and could be bounced off a hard surface.

Temporary lack of moisture usually did not kill embryos; prolonged dryness, combined with high temperatures, probably could not be tolerated. Lynn and Ullrich (1950), by desiccating the eggs of Chrysemys picta and Chelydra serpentina, produced abnormalities in the young ranging from slight irregularities of the shell to eyeless monstrosities; eggs desiccated in the latter half of incubation produced a higher percentage of abnormal young than eggs that were desiccated earlier.

In 1956, three fertile eggs, from clutches that were at different stages of incubation, were immersed in water for 48 hours. The eggs rested on the bottom of the bowl in the same position in which they had been placed in the incubation dishes; when turned, they returned invariably to the original position. The embryos in two of the eggs (one and 27 days old at the time of immersion) were still living ten days after the eggs were removed from the water; the embryo in the remaining egg (21 days old at the time of immersion) was dead. Eggs immersed in water increased in size and weight at the same rate as eggs in incubation dishes, indicating that absorption of water probably operates on a threshold principle, the amount absorbed being no more than normal even under wet conditions.

Natural nests usually are in well-drained areas, but water probably stands in some nests for short periods after heavy rains. Provided the nest cavity itself is not damaged, water in the nest is probably more beneficial than harmful to the eggs; however, nests that are inundated during floods probably have little chance of survival.

Embryonic Development

Eggs were examined by transmitted light in the course of incubation. At the time of laying (or removal from oviducts) no embryonic structures were discernible even in eggs that had been retained in the oviducts of captive females some weeks past the normal time of laying; a colorless blastodisc could be seen if eggs were opened. Embryonic structures first became visible at eight to ten days of incubation; at this time vascularization of the blastodisc was evident and the eyes appeared as dark spots. Heart beats were observed in most embryos by the fifteenth day but were evident in a few as early as the tenth day. The pulse of a fifteen-day-old embryo averaged 72 beats per minute at a temperature of 30 degrees. Embryos at fifteen days, measured in a straight line from cephalic flexure to posteriormost portion of body, were approximately nine to ten millimeters long and at 22 days were 14 millimeters long. At approximately 35 days the eggs became dark red; embryonic structures were discernible thereafter only in eggs that had embryos situated at one end, close to the shell.

Incubation periods for 49 eggs (representing 12 clutches) kept in the laboratory ranged from 56 to 127 days, depending on the temperature of the air during the incubation period. In 1955, eggs were kept at my home in Lawrence where air temperatures were uncomfortably hot in summer and fluctuations of 20 degrees (Fahrenheit) or more in a 24-hour period were common. The following summer eggs were kept in my office at the Museum where temperatures were but slightly cooler than in my home and subject also to wide variation. In 1957 this part of the Museum was air-conditioned and kept at approximately 75 degrees. The greater lengths of incubation periods at lower temperatures are shown in Table 1. Risley (1933:698) found the incubation period of Sternotherus odoratus to be longer at lower temperatures; corresponding observations were made by Allard (1935:332) and Driver (1946:173) on the eggs of Terrapene carolina. Cagle (1950:40) and Cunningham (1939) found no distinct differences in length of incubation period for eggs of Pseudemys scripta and Malaclemys terrapin, respectively, at different temperatures within the range tolerated by the eggs.

Most nests observed in the field were in open situations where they would receive the direct rays of the sun for at least part of the day; the shorter average incubation periods (59 and 70 days, respectively), observed in 1955 and 1956, therefore, more nearly reflect the time of incubation under natural conditions than does the excessively long period (125 days at 75 degrees) observed in 1957 under cooler, more nearly even temperatures.

Sixty-five days seems to be a realistic estimate of a typical incu bation period under natural conditions; eggs laid in mid-June would hatch by mid-August. Even in years when summer temperatures are much cooler than normal, eggs probably hatch by the end of October. Hatchlings or eggs would have a poor chance of surviving a winter in nests on exposed cut-banks or in other unprotected situations. Overwintering in the nest, hatchlings might survive more often than eggs, since hatchlings could burrow into the walls and floor of the nest cavity. Unsuitable environmental conditions that delay the nesting season and retard the rate of embryonic development may, in some years, be important limiting factors on populations of ornate box turtles.

In areas where T. ornata and T. carolina are sympatric (for example, in Illinois, Kansas, and Missouri) the two species occupy different habitats, ornata preferring open grassland and carolina wooded situations. Under natural conditions, the average incubation periods of these two species can be expected to differ, T. carolina having a somewhat longer period due to lower temperatures in nests that are shaded. In the light of these speculations, the remark of Cahn (1937:102)—that T. ornata nested later in the season (in Illinois) and compensated for this by having a shorter incubation period—is understandable.

The range of temperatures tolerated by developing eggs probably varies with the stage of embryonic development. When temperatures in the laboratory were 102 to 107 degrees Fahrenheit for approximately eight hours, due to a defect in a thermostat, the young in two eggs of T. ornata, that had begun to hatch on the previous day, were killed, as were the nearly full-term embryos in a number of eggs of T. carolina (southern Mississippi) kept in the same container. A five-day-old hatchling of T. ornata, kept in the same container, survived the high temperatures with no apparent ill effects. Cagle (1950:41) found that eggs of Pseudemys scripta could not withstand temperatures of 10 degrees for two weeks nor would they survive if incubated at 40 degrees. Cunningham (1939) reported that eggs of Malaclemys terrapin could not survive prolonged exposure to temperatures of 35 to 40.6 degrees but tolerated temporary exposure to temperatures as high as 46 degrees.

In the summer of 1955, a clutch of three eggs, all of which contained nearly full-term embryos, was placed in a refrigerator for 48 hours. The temperature in the refrigerator was maintained at approximately 4.5 degrees; maximum and minimum temperatures for the 48 hour period were 2.8 and 9.5 degrees, respectively. When the eggs were removed from the refrigerator they showed gains in weight and increases in size comparable to eggs, containing embryos of the same age, used as controls. The experimental eggs began to hatch two days after they were removed to normal temperatures—approximately 24 hours later than the controls.

In the late stages of incubation, the outer layer of the shell becomes brittle and is covered with a mosaic of fine cracks or is raised into small welts. Several days before hatching, movements of the embryo disturb the surface of the shell and cause the outer layer to crumble away, especially where the head and forequarters of the embryo lie against the shell. Some embryos could be seen spasmodically thrusting the head and neck dorsally against the shell.

The role of the caruncle in opening the shell seems to vary among different species of turtles. Cagle (1950:41) reported that it was used only occasionally by Pseudemys scripta; Allard (1935:332) thought that it was not used by Terrapene carolina; and, the observations of Booth (1958:262) and Grant (1936:228) indicate that embryos of Gopherus agassizi use the caruncle at least in the initial rupturing of the shell.

In the three instances in which hatching was closely observed in T. ornata, the caruncle made the initial opening in the shell; claws of the forefeet may have torn shells in other hatchings that were not so closely observed. In all observed instances, the shell was first opened at a point opposite the anterior end of the embryo. The initial opening had the appearance of a three-cornered tear. A quantity of albuminous fluid oozed from eggs as soon as the shells were punctured.

The initial tear is enlarged by lateral movements of the front feet, and later the hind feet reach forward and lengthen the tear farther posteriorly. In many instances a tear develops on each side and the egg has the appearance of being cleft longitudinally. The young turtle emerges from the anterior end of the shell or backs out of the shell through a lateral tear.

The process of hatching, from rupture of shell to completion of emergence, extended over three to four days in the laboratory. Many hatchlings from time to time crawled back into the shell over a period of several days after hatching was completed. In a clutch of eggs kept in a pail of earth, by William R. Brecheisen, eight days elapsed between onset of hatching and appearance of the first hatchling at the surface.

A nest in an outdoor pen at the Reservation was discovered in early October. The cap had been recently perforated and the hatchlings had escaped. One of them, judged to be approximately two weeks old, was found in a burrow nearby. The cavity of the nest appeared to have been enlarged by the young. The eggs were probably laid in early July. Emergence of young from the nest had been delayed for a time after hatching, until rain softened the ground in late September and early October.

Fertility and Prenatal Mortality

Eggs were incubated in the laboratory at more nearly optimum temperature and humidity than were eggs in natural nests. Percentage of prenatal mortality probably was lower in laboratory-incubated eggs than in those under natural conditions.

Of sixty eggs studied in the laboratory, 45 (75 per cent) were fertile; 36 (80 per cent) of the fertile eggs (those in which the blastodisc was at some time discernible by transmitted light) hatched successfully. In six clutches all the eggs were fertile and five of these clutches hatched with 100 per cent success. One clutch contained eggs that were all infertile and another clutch had four infertile eggs and two fertile eggs that failed to hatch. Among nine fertile eggs that failed to survive, four casualties occurred in the late stages of incubation or after hatching had begun, indicating that these are probably critical periods.

Fertility of eggs was not correlated with size or age of female, with size of clutch, or with size of egg. Eggs laid in the laboratory had higher rates of infertility and prenatal mortality than did eggs dissected from oviducts. Handling of eggs in removing them from nests to incubation dishes, after embryonic development had begun, might have been responsible for reduced viability ([Table 2]).

Reproductive Potential

Assuming that 4.7 eggs are laid per season, that all eggs are fertile and all hatch, that all young survive to maturity, that half the hatchlings are females, and that females first lay eggs in the eleventh year, the progeny of a single mature female would number 699 after twenty years. Considering that infertility and prenatal mortality eliminate approximately 40 per cent of eggs laid (according to laboratory findings) the average number of surviving young per clutch would be 2.8 and the total progeny, after 20 years, would be 270, provided that only one clutch of eggs was laid per year. But it is thought that, on the average, one third of the female population produces two clutches of eggs in a single season. If the second clutch contains 3.5 eggs (resulting in 2.1 surviving young when factors of infertility and prenatal mortality are considered), the progeny of a single female, after 20 years, would number approximately 380. Postnatal mortality reduces the progeny to a still smaller number.

The small number of eggs laid each year and the long period required to reach sexual maturity make the reproductive potential of T. ornata smaller than that of the other turtles of the Great Plains, and much smaller than nearly any of the non-chelonian reptiles of the same region.

Number of Reproductive Years

The total span of reproductive years is difficult to determine; I am unable to ascertain the age of a turtle that has stopped growing. No clearly defined external characteristics of senility were discovered in the populations studied. A male that I examined had one atrophied testis. In another male both testes were shrunken and discolored and appeared to be encased by fibrous tissue. Both males were large, well past the age of regular growth, and had smoothly worn shells. Several old females had seemingly inactive ovaries. Reproductive processes probably continue throughout life in most members of the population, although possibly at a somewhat reduced rate in later life.

GROWTH AND DEVELOPMENT

Initiation of Growth

Young box turtles became active and alert as soon as they hatched, and remained so until low temperatures induced quiescence. If sand or soil was available, hatchlings soon burrowed into it and became inactive. Covering containers with damp cotton also induced inactivity; the hatchlings usually made no attempt to burrow through the confining layer. Desire to feed varied in hatchlings of the same brood and seemed not to be correlated with retraction of the yolk sac or retention of the caruncle. Some hatchlings actively pursued mealworms; on subsequent feedings they learned to associate my presence with food and eagerly took mealworms from forceps or from my hand. Meat, vegetables, and most other motionless but edible objects were ignored by hatchlings but some individuals learned to eat meat after several forced feedings. Hatchlings that regularly took food in the first month of life ordinarily grew faster than hatchlings that did not eat. Many of the hatchlings in the laboratory showed no areas of new epidermal growth on the shell in the time between hatching and first (induced) hibernation.

Size and Appearance at Hatching

The proportions of the shell change somewhat in the first few weeks of life. At hatching the shell may be misshapen as a result of confinement in the egg. Early changes in proportions of the shell result from expansion—widening and, to a lesser degree, lengthening of the carapace—immediately after hatching. Subsequent retraction or rupture of the yolk sac and closure of the navel are accompanied by a decrease in height of shell and slight, further widening of the carapace.

The yolk sac retracts mainly between the time when the egg shell is first punctured and the time when the turtle actually emerges from the shell. When hatching is completed, the yolk sac usually protrudes no more than two millimeters, but in some individuals it is large and retracts slowly over a period of several days.

One individual began hatching on November 11 and was completely out of the egg shell next day; the yolk sac was 15 millimeters in diameter, protruded six millimeters from the umbilical opening, and hindered the hatchling's movements. The sac broke two days later, smearing the bottom of the turtle's dish with semifluid yolk. The hatchling then became more active. Twenty-six days later the turtle was still in good condition and its navel was nearly closed. A turtle that hatched with a large yolk sac in a natural nest possibly would benefit, through increased ease of mobility, if the yolk sac ruptured.

A recently hatched turtle was found at the Reservation in October, 1954, and was kept in a moist terrarium in the laboratory where it died the following May. The turtle was sluggish and ate only five or six mealworms while in captivity; no growth was detectable on the laminae of the shell. Autopsy revealed a vestige of the retracted yolk sac, approximately one millimeter in diameter, on the small intestine.

The navel ("umbilical scar") of captive hatchlings ordinarily closed by the end of the second month but in three instances remained open more than 99 days. The position of the navel is marked by a crescent-shaped crease, on the abdominal lamina, that persists until the plastron is worn down in later years ([Pl. 24, Fig. 1]).

Fig. 7. A hatchling of T. o. ornata (× 2) that still retains the caruncle ("egg tooth"). A distinct boss will remain on the maxillary beak after the caruncle is shed.

The caruncle ("egg tooth") ([Fig. 7]) remains attached to the horny maxillary beak for a variable length of time; 93 per cent of the live hatchlings kept in the laboratory retained the caruncle on the tenth day, 71 per cent on the twentieth day, and only 10 per cent on the thirtieth day of life. Few individuals retained the caruncle when they entered hibernation late in November, and none retained it upon emergence from hibernation. Activities in the first few days or weeks of life influence the length of time that the caruncle is retained; turtles that begin feeding soon after hatching probably lose the caruncle more quickly than do those that remain quiescent. The caruncles of some laboratory specimens became worn before finally dropping off. Almost every caruncle present after 50 days could be flicked off easily with a probe or fingernail. The initiation of growth of the horny maxillary beak probably causes some loosening of the caruncle. The caruncle may aid hatchlings in escaping from the nest.

After the caruncle falls off, a distinct boss remains, marking its former place on the horny beak ([Pl. 25, Fig. 1]); this boss is gradually obliterated over a period of weeks by wear and by differential growth, and is seldom visible in turtles that have begun their first full year of growth. The "first full year of growth" is here considered to be the period of growth beginning in the spring after hatching.

Growth of Epidermal Laminae

Growth of ornate box turtles was studied by measuring recaptured turtles in the field, by periodically measuring captive hatchlings and juveniles, and by measuring growth-rings on the epidermal laminae of preserved specimens. Studies of growth-rings provided by far the greatest volume of information on growth, not only for the years in which field work was done, but for the entire life of each specimen examined.

It was necessary to determine the physical nature of growth-rings and the manner in which they were formed before growth could be analyzed. Examination of epidermal laminae on the shell of a box turtle reveals that each has a series of grooves—growth-rings—on its surface. The deeper grooves are major growth-rings; they occur at varying distances from one another and run parallel to the growing borders of the lamina. Major growth-rings vary in number from one to 14 or more, depending on the age of the turtle ([Pl. 22]). In juvenal turtles and in young adults, major growth-rings are distinct and deep. Other grooves on the shell—minor growth-rings—have the same relationship to the borders of the laminae but are shallower and less distinct than major growth-rings. One to several minor growth-rings usually occur on each smooth area of epidermis between major growth-rings. As the shell of an adult turtle becomes worn, the minor growth-rings disappear and the major rings become less distinct. Both sets of rings may be completely obliterated in old turtles but the major rings usually remain visible until several years after puberty.

In cross section, major growth-rings are V- or U-shaped. The inner wall of each groove is the peripheral edge of the part of the scute last formed whereas the outer wall represents the inner edge of the next new area of epidermal growth. The gap produced on the surface of the lamina (the open part of the groove) results from cessation of growth at the onset of hibernation. Minor growth-rings are shallow and barely discernible in cross-section ([Fig. 8]). It may therefore be understood that growth-rings are compound in origin; each ring is formed in part at the beginning of hibernation and in part at the beginning of the following growing season.

The few publications discussing growth in turtles express conflicting views as to the exact mode of growth of epidermal laminae. Carr (1952:22) briefly discussed growth of turtle scutes in general and stated that eccentric growth results from an entirely new laminal layer forming beneath, and projecting past the edges of the existing lamina. Ewing (1939) found the scutes of T. carolina to be the thickest at the areola and successively thinner in the following eight annual zones of growth; parts of scutes formed subsequent to the ninth year varied irregularly in thickness. He stated that epidermal growth took place at the margins of the laminae rather than over their entire under-surfaces.

It is evident that the mode of scutular growth described by Carr (loc. cit.) applies to emyid turtles that shed the epidermal laminae more or less regularly (for example, Chrysemys and Pseudemys). In these aquatic emyids a layer of the scute, the older portion, periodically becomes loose and exfoliates usually in one thin, micalike piece; since the loosened portion of the scute corresponds in size to the scute below, it must be concluded that a layer of epidermis is shed from the entire upper surface of the scute, including the area of new epidermal growth. Box turtles ordinarily do not shed the older parts of their scutes; the areola and successively younger portions of the lamina remain attached to the shell until worn off. The appearance of a single unworn scute, especially one of the centrals or the posterior laterals, closely resembles a low, lopsided pyramid.

Examination of parasagittal sections of scutes revealed that they were composed of layers, the number of layers varying with the age of the scute. A scute from a hatchling consists of one layer. A scute that shows a single season of growth has two layers; a new layer is added in each subsequent season of growth. Stratification is most evident in the part of the scute that was formed in the first three or four seasons and becomes increasingly less distinct in newer parts of the scute. It may further be understood that scutes grow in the manner described by Carr (loc. cit.).

When the epidermal laminae are removed, a sheet of tough, pale grayish tissue remains firmly attached to the bones of the shell beneath. This layer probably includes, or consists of, germinal epithelium. Contrasting pale and dark areas of the germinal layer correspond to the pattern of markings on the scute removed.

Growth of epidermal laminae is presumably stimulated by growth of the bony shell. As the bone grows, the germinal layer of the epidermis grows with it. When growth ceases at the beginning of hibernation, the thin edges of the scutes are slightly down-turned where they enter the interlaminal seams ([Fig. 8]). When growth is resumed in spring, the germinal layer of the epidermis, rather than continuing to add to the edge of the existing scute, forms an entirely new layer of epidermis. The new layer is thin and indistinct under the oldest part of the scute but becomes more distinct toward its periphery. Immediately proximal to the edge of the scute, the new layer becomes greatly thickened, and, where it passes under the edge, it bulges upward, recurving the free edge of the scute above. At this time the formation of a major growth-ring is completed. The newly-formed epidermis, projecting from under the edges of the scute, is paler and softer than the older parts of the scute; the presence or absence of areas of newly formed epidermis enables one to determine quickly whether a turtle is growing in the season in which it is captured. There is little actual increase in thickness of the scute after the first three or four years of growth. The epidermal laminae are therefore like low pyramids only in appearance. This appearance of thickness is enhanced by the contours of bony shell which correspond to the contours of the scutes.

Minor growth-rings differ from major growth-rings in appearance and in origin. Ewing (op. cit.: 91) recognized the difference in appearance and referred to minor growth-rings as "pseudoannual growth zones." Minor growth-rings result from temporary cessations of growth that occur in the course of the growing season, not at the onset of hibernation. They are mere dips or depressions in the surface of the scute. The occurrence of minor growth-rings indicates that interruptions in growth of short duration do not result in the formation of a new layer of epidermis. Slowing of growth or its temporary cessation may be caused by injuries, periods of quiescence due to dry, hot, or cold weather, lack of food, and possibly by physiological stress, especially in females, in the season of reproduction. Minor growth-rings that lie immediately proximal to major growth-rings ([Pl. 22, Fig. 2]), are the result of temporary dormancy in a period of cold weather at the end of a growing season, followed by nearly normal activity in a warmer period before winter-long hibernation is begun. Cagle (1946:699) stated that sliders (Pseudemys scripta elegans) remaining several weeks in a pond that had become barren of food would stop growing and develop a growth-ring on the epidermal laminae; he did not indicate, however, whether these growth-rings differ from those formed during hibernation.

In species that periodically shed scutes a zone of fracture develops between the old and new layers of the scute as each new layer of epidermis is formed, and the old layer is shed. Considering reptiles as a group, skin shedding is of general occurrence; the process in Pseudemys and Chrysemys differs in no basic respect from that in most reptiles. Retention of scutes in terrestrial emyids and in testudinids is one of many specializations for existence on land. Retention of scutes protects the shell of terrestrial chelonians against wear. Some box turtles were observed to have several scutes of the carapace in the process of exfoliation but no exfoliation was observed on the plastron. Exfoliation ordinarily occurred on the scutes of the carapace that were the least worn; the exfoliating portion included the areola and the three or four oldest (first formed) layers of the scute. The layer of scute exposed was smooth and had yellow markings that were only slightly less distinct than those on the portion that was exfoliating.

Wear on the shell of a box turtle reduces the thickness of scutes, as does the shedding of scutes in the aquatic emyids mentioned. It is noteworthy that any of the layers in the scute of a box turtle can form the cornified surface of the scute when the layers above it wear away or are shed.

It is uncertain whether turtles that have ceased to grow at a measurable rate continue to elaborate a new layer of epidermis at the beginning of each season. Greatly worn shells of ornate box turtles, particularly those of the subspecies luteola, have only a thin layer of epidermis through which the bones of the shell and the sutures between the bones are visible. I suspect that, in these old individuals, the germinal layer of the epidermis does not become active each year but retains the capacity to elaborate new epidermis if the shell becomes worn thin enough to expose and endanger the bone beneath it. The germinal layer of old turtles loses the capacity to produce color.

Major growth-rings constitute a valuable and accurate history of growth that can be studied at any time in the life of the turtle if they have not been obliterated. They are accurate indicators of age only as long as regular growth continues but may be used to study early years of growth even in turtles that are no longer growing. Minor growth-rings, if properly interpreted, provide additional information on growing conditions in the course of each growing season.

Nichols (1939a: 16-17) found that the number of growth-rings formed in marked individuals of T. carolina did not correspond to the number of growing seasons elapsed; he concluded that growth-rings were unreliable as indicators of age and that box turtles frequently skipped seasons of growth. Woodbury and Hardy (1948:166-167) and Miller (1955:114) came to approximately the same conclusion concerning Gopherus agassizi. It is significant that these workers were studying turtles of all sizes and ages, some of which were past the age of regular, annual growth. Cagle's review of the literature concerning growth-rings in turtles (1946) suggests that, in most of the species studied, growth-rings are formed regularly in individuals that have not attained sexual maturity but are formed irregularly after puberty.

Cagle's (op. cit.) careful studies of free-living populations of Pseudemys scripta showed that growth-rings, once formed, did not change in size, that the area between any two major growth-rings represented one season of growth, and that growth-rings were reliable indicators of age as long as the impression of the areola remained on the scutes studied. Cagle noted decreasing distinctness of growth-rings after each molt.

The relative lengths of the abdominal lamina and the plastron remain approximately the same throughout life in T. ornata. Measurements were made of the plastron, carapace, and abdominal lamina in 103 specimens of T. o. ornata from Kansas and neighboring states. The series of specimens was divided into five nearly equal groups according to length of carapace. Table 3 summarizes the relationship of abdominal length to plastral length, and of carapace length to plastral length. The mathematical mean of the ratio, abdominal length/plastral length, in each of the four groups of larger-sized turtles, was not significantly different from the same ratio in the hatchling group. The relative lengths of carapace and plastron are not so constant; the carapace is usually longer than the plastron in hatchlings and juveniles, but shorter than the plastron in adults, especially adult females.

Table 3.—The Relationship of Length of Abdominal Scute to Plastral Length, and of Plastral Length to Length of Carapace, in 103 Specimens of T. o. ornata Arranged in Five Groups According to Length of Carapace. The Relative Lengths of Abdominal Scute and Plastron are not Significantly Different in the Five Groups. The Plastron Tends to be Longer than the Carapace in Specimens of Adult or Nearly Adult Size.

The length of any growth-ring on the abdominal lamina can be used to determine the approximate length of the plastron at the time the growth-ring was formed. Actual and relative increases in length of the plastron can be determined in a like manner. For example, a seven-year-old juvenile (KU 3283) with a plastron 74.0 millimeters long had abdominal growth-rings (beginning with areola and ending with the actual length of the abdominal) 5.9, 7.8, 9.5, 10.7, 12.0, 12.5, 14.3, and 14.9 millimeters long. Using the proportion,

where AB is the abdominal length, PL the plastral length, AB¹ the length of any given growth-ring, and X the plastral length at the time growth-ring AB^1 was formed, the plastral length of this individual was 29.3 millimeters at hatching, 38.8 at the end of the first full season of growth, and 47.2, 53.2, 59.6, 62.1, and 71.0 millimeters at the end of the first, second, third, fourth, fifth, and sixth seasons of growth, respectively. The present length of the abdominal (14.9 mm.) indicates an increment of three millimeters in plastral length in the seventh season, up to the time the turtle was killed (June 25). This method of studying growth in turtles was first used by Sergeev (1937) and later more extensively used by Cagle (1946 and 1948) in his researches on Pseudemys scripta. Because the plastron is curved, no straight-line measurement of it or its parts can express true length. Cagle (1946 and 1948) minimized error by expressing plastral length as the sum of the laminal (or growth-ring) lengths. This method was not possible in the present study because growth-rings on parts of one or more laminae (chiefly the gulars and anals) were usually obliterated by wear, even in young specimens. It was necessary to express plastral length as the sum of the lengths of forelobe and hind lobe.

The abdominal lamina was selected for study because of its length (second longest lamina of plastron), greater symmetry, and flattened form. Although the abdominal is probably subject to greater, over-all wear than any other lamina of the shell, wear is even, not localized as it is on the gulars and anals.

In instances where some of the growth-rings on an abdominal lamina were worn but other rings remained distinct, reference to other, less worn lamina permitted a correct interpretation of indistinct rings.

Abdominal laminae were measured at the interlaminal seam; since the laminae frequently did not meet perfectly along the midline (and were of unequal length), the right abdominal was measured in all specimens. Growth-rings on the abdominal laminae were measured in the manner shown in [Plate 22].

Data were obtained for an aggregate of 1272 seasons of growth in 154 specimens (67 females, 48 males, and 39 of undetermined sex, chiefly juveniles). Averages of calculated plastral length were computed in each year of growth for specimens of known sex (Figs. [9] and [10]) and again for all specimens examined. Annual increment in plastral length was expressed as a percentage of plastral length at the end of the previous growing season ([Fig. 11]). Increment in plastral length for the first season of growth was expressed as a percentage of original plastral length because of variability of growth in the season of hatching; growth increments in the season following hatching are, therefore, not so great as indicated in [Figure 11].

Growth of Juveniles

Areas of new laminal growth were discernible on laboratory hatchlings soon after they ate regularly. Hatchlings that refused to eat or that were experimentally starved did not grow. The first zone of epidermis was separated from the areola by an indistinct growth-ring (resembling a minor growth-ring) in most hatchlings, but in a few specimens the new epidermis appeared to be a continuation of the areola. Major growth-rings never formed before the onset of the first hibernation.

Growth in the season of hatching seems to depend on early hatching and early emergence from the nest. Under favorable conditions hatchlings would be able to feed and grow eight weeks or more before hibernation. Hatchlings that emerge in late autumn or that remain in the nest until spring are probably unable to find enough food to sustain growth.

Sixty-four (42 per cent) of the 154 specimens examined showed measurable growth in the season of hatching. The amount of increment was determined in 36 specimens having a first growth-ring and an areola that could be measured accurately. The average increment of plastral length was 17.5 per cent (extremes, 1.8-66.0 per cent) of the original plastral length. Ten individuals showed an increment of more than 20 per cent; the majority of these individuals (8) were hatched in the years 1947-50, inclusive.

Fig. 9. See legend for [Fig. 10]

Fig. 10. The relationship of size to age in T. o. ornata, based on studies of growth-rings in 115 specimens of known sex (67 females and 48 males) from eastern Kansas. Size is expressed as plastral length at the end of each growing season (excluding the year of hatching) through the twelfth and thirteenth years (for males and females, respectively) of life. Vertical and horizontal lines represent, respectively, the range and mean. Open and solid rectangles represent one standard deviation and two standard errors of the mean, respectively. Age is expressed in years.

Some hatchlings that grow rapidly before the first winter are as large as one- or two-year-old turtles, or even larger, by the following summer. Individuals that grew rapidly in the season of hatching tended also to grow more rapidly than usual in subsequent seasons; 80 per cent of the individuals that increased in plastral length by 20 per cent or more in the season of hatching, grew faster than average in the two seasons following hatching. Early hatching and precocious development presumably confer an advantage on the individual, since turtles that grow rapidly are able better to compete with smaller individuals of the same age. Theoretically, turtles growing more rapidly than usual in the first two or three years of life, even if they grew subsequently at an average rate, would attain adult size and sexual maturity one or more years before other turtles of the same age. A few turtles (chiefly males) attain adult size (and presumably become sexually mature) by the end of the fifth full season of growth (Figs. [9] and [10]). These individuals, reaching adult size some three to four years sooner than the average age, were precocious also in the earlier stages of post-natal development.

Young box turtles reared in the laboratory grew more slowly than turtles of comparable ages under natural conditions; this was especially evident in hatchlings and one-year-old specimens. Slower growth of captives was caused probably by the unnatural environment of the laboratory. Captive juveniles showed a steady increase in weight (average, .52 grams per ten days) as they grew whereas captive hatchlings tended to lose weight whether they grew or not.

Growth in Later Life

After the first year growth is variable and size is of little value as an indicator of age. Although in the turtles sampled variation in size was great in those of the same age, average size was successively greater in each year up to the twelfth and thirteenth years (for males and females, respectively), after which the samples were too small to consider mathematically.

Increments in plastral length averaged 68.1 per cent in the year after hatching, 28.6 per cent in the second year and 18.1 per cent in the third year. From the fourth to the fourteenth year the growth-rate slowed gradually from 13.3 to about three per cent ([Fig. 11]). These averages are based on all the specimens examined (with no distinction as to sex); they give a general, over-all picture of growth rate but do not reflect the changes that occur in growth rate at puberty (as shown in Figs. [9] and [10]).

Rate of growth and, ultimately, size are influenced by the attainment of sexual maturity. Adult females grow larger than adult males. Males, nevertheless, grow faster than females and become sexually mature when smaller and younger. Examination of gonads showed 17 per cent of the males to be mature at plastral lengths of 90 to 99 millimeters, 76 per cent at 100 to 109 millimeters, and 100 per cent at 110 millimeters, whereas the corresponding percentages of mature females in the same size groups were: zero per cent, 47 per cent, and 66 per cent. Of the females, 97 per cent were mature at 120 to 129 millimeters and all were mature at 130 millimeters ([Fig. 13]). Because growth slows perceptibly at sexual maturity, it is possible, by examination of growth-rings, to estimate the age of puberty in mature specimens.

Fig. 11. Average increment in plastral length (expressed as a percentage of plastral length at the end of the previous season of growth) in the season of hatching (H) and in each of the following 14 years of life, based on 1073 growth-rings. The number of specimens examined for each year of growth is shown in parentheses. Records for males and females are combined.

Attainment of sexual maturity, in the population studied, was more closely correlated with size than with age. For example, nearly all males were mature when the plastron was 100 to 110 millimeters long, regardless of the age at which this size was attained. The smallest mature male had a plastral length of 99 millimeters; according to the data presented in Figures [9] and [10], therefore, a few males reach sexual maturity in the fourth year, and increasingly larger portions of the population become mature in the fifth, sixth, and seventh years. The majority become mature in the eighth and ninth years. Likewise, females (smallest mature specimen, 107 mm.) may be sexually mature at the end of the sixth year but most of them mature in the tenth and eleventh years.

Annual Period of Growth

In growing individuals, narrow zones of new epidermis form on the laminae in spring. Nearly all the growing individuals collected in May of 1954 and 1955 had zones of new epidermis on the shell but those collected in April did not. Activity in the first week or two after spring emergence is sporadic and regular feeding may not begin until early May. Once begun, growth is more or less continuous as long as environmental conditions permit foraging. The formation of minor growth-rings and adjacent growth-zones in autumn, provides evidence that growth commonly continues up to the time of hibernation. The number of growing days per year varies, of course, with the favorableness of environmental conditions. The length of time (162 days) given by Fitch (1956b:438) as the average annual period of activity for T. ornata is a good estimate of the number of growing days per season.

Environmental Factors Influencing Growth

Zones of epidermis formed in some years are wider or narrower than the zones bordering them ([Pl. 22]). Zones notably narrower or wider than the average, formed in certain years, constituted distinct landmarks in the growth-histories of nearly all specimens; for example, turtles of all ages grew faster than average in 1954 and zones of epidermis formed in this year were always wider than those formed in 1953 and 1955.

An index to the relative success of growth in each calendar year was derived. Records of growth for all specimens in each age group were averaged; the figure obtained was used to represent "normal" or average growth rate in each year of life ([Fig. 12]). The over-all averages for the various age groups were then compared with records of growth attained by individuals of corresponding age in each calendar year, growth in a particular year being expressed as a percentage of the over-all average. The percentages of average growth for all ages in each calendar year were then averaged; the mean expressed the departure from normal rate of growth for all turtles growing in a particular calendar year. For example, the over-all average increment in plastral length in the fifth year of life was 12.1 per cent, the increment in the sixth year was 10 per cent, and so on ([Fig. 11]). In 1953, turtles in their fifth and sixth years increased in plastral length by 11.4 and 9.1 per cent, or grew at 94.2 and 91.0 per cent of the normal rate, respectively. The percentages of normal growth rate for these age groups averaged with percentages of the other age groups in 1953 revealed that turtles grew at approximately 86 per cent of the normal rate in 1953.

Growth rates were computed for the twelve-year period, 1943-1954, because of the concentration of records in these years. Scattered records also were available for many of the years from 1901-1942. Records for individuals in the season of hatching and the first full season of growth were not considered.

Direct correlation exists between growth rate and average monthly precipitation in the season of growth (April to September) ([Fig. 12]). In nine of eleven years, the curve for growth rate followed the trend of the curve for precipitation; but because other climatic conditions also influenced growth, the fluctuations in the two curves were not proportional to one another.

Grasshoppers form an important element in the diet of box turtles. Smith (1954) traced the relative abundance of grasshoppers over a period of 100 years in Kansas, and this information is of significance for comparison with data concerning growth of box turtles. In general, the growth index was higher when favorable weather and large populations of grasshoppers occurred in the same year.

In the following summary, the numbers (1 to 5) used to express the relative abundance of grasshoppers are from Smith (op. cit.). Maxima and minima refer to the twelve-year period, 1943-1954. The growth index for each year (shown as a graph in [Fig. 12]) appears in brackets and indicates the percentage of normal growth attained by all turtles in that year.

Years Favorable for Growth

1954 [126.3]: Growth was better than average for turtles of all ages. Grasshopper populations were highest (4+) since 1948. Continuously warm weather, beginning in the last few days of March, permitted emergence in the first week of April; thereafter conditions were more or less continuously favorable for activity until late October. Although there was less than an inch of precipitation in September, precipitation in August and October was approximately twice normal and more or less evenly distributed. Warm weather in early November permitted an additional two weeks of activity.

1945 [125.5]: This was the second most favorable year for growth and the second wettest year. Records of growth are all from young turtles (one to four years old), all of which grew more than average. Daily maximum temperatures higher than 60 degrees Fahrenheit on 18 of the last 19 days of March, combined with twice the normal amount of precipitation in the same period, stimulated early emergence. August and October were both dry (each with less than one inch of precipitation) but diurnal temperatures remained warm through the first week in November and probably prolonged activity of box turtles at least until then. Grasshoppers were more abundant (3.7) than normal.

Years Unfavorable for Growth

1944 [83.1]: This was the poorest growing year for the period considered. The lack of a continuously warm, wet period in early spring probably delayed emergence until the last week in April. Temperatures remained warm enough for activity until early November, but dry weather in September and October probably curtailed activity for inducing long periods of quiescence; most of the precipitation that occurred in the latter two months fell in a one-week period beginning in the last few days of September. Grasshopper populations were higher (4.0) than normal.

1953 [85.6]: This was the second poorest growing year and the driest year in the period considered. Intermittently cold weather in spring delayed emergence until the last week in April when nearly an inch of rain fell. Temperatures were higher than normal from June to October. The period from September to the end of October was dry and the small amount of precipitation that occurred was concentrated chiefly at the beginning and end of that period. Temperatures in late October and early November were lower than normal. Grasshopper populations were low (2.2).

1952 [88.3]: Environmental conditions were poor for growth and much like the conditions described for 1953. In both years growth was much less than normal in turtles of all ages except for one group (adults that were 10 and 11 years old in 1952 and 1953, respectively) that was slightly below normal in 1952 and slightly above normal in 1953.

The small number of records for 1955 were not considered in [Figure 12]. Warm weather in the last half of March lengthened the growing season, and environmental conditions, as in 1954, were more or less favorable throughout the rest of the summer; 1955 probably ranks with 1954 as an exceptionally good year for growth of box turtles.

Although the number of records available for turtles hatched in the period from 1950 to 1954 is small, a few records are available for all these years except 1951. In general, small samples of turtles hatched in these years reflect only the difficulty of collecting hatchlings and juveniles. In 1951, conditions for incubation and hatching were poor and the lack of records for that year actually represents a high rate of prenatal and postnatal mortality. Rainfall in the nesting season was two to three times normal and temperatures were below normal. Flooding occurred in low areas of Douglas County and many eggs may have been destroyed when nests were inundated. Cold weather probably increased the time of incubation for surviving eggs so that only a few turtles could hatch before winter. Flooding and cold, wet weather in the season of growth and reproduction, affecting primarily eggs and hatchlings, may act as checks on populations of T. ornata in certain years.

Fig. 12. The relation of growth rate in Terrapene o. ornata (solid line) to precipitation (dotted line) in eastern Kansas. "Normal" rate of growth was determined by averaging records of increase in length of plastron for turtles in each age group. The growth index is expressed as a percentage of normal growth and is the mean departure from normal of all age groups in each calendar year. Precipitation is for the period, April to September (inclusive), at Lawrence, Douglas Co., Kansas. The means for precipitation (4.3) and growth index (100) are indicated by horizontal lines at the right of the graph.

The environmental factors governing activity of terrestrial turtles seem to differ at least in respect to threshold, from the factors influencing the activity of aquatic turtles. A single month that was drier or cooler than normal probably would not noticeably affect growth and activity of aquatic emyids in northeast Kansas, but might greatly curtail growth of box turtles.

Cagle (1948:202) found that growth of slider turtles (Pseudemys scripta) in Illinois paralleled the growth of bass and bluegills in the same lake; in the two years in which the fish grew rapidly, the turtles did also, owing, he thought to "lessened total population pressure" and "reduced competition for food." Growth of five-lined skinks (Eumeces fasciatus) on the Natural History Reservation paralleled growth of box turtles, probably because at least some of the same environmental factors influence the growth of both species. Calculations of departure from normal growth in E. fasciatus, made by me from Fitch's graph (1954:84, Fig. 13), show that relative success of growth in the period he considered can be ranked by year, in descending order, as: 1951, 1949, 1948, 1950, 1952. This corresponds closely to the sequence, 1951, 1948, 1949, 1950, 1952, for T. ornata.

Number of Growing Years

Growth almost stops after the thirteenth year in females and after the eleventh or twelfth year in males, approximately three years, on the average, after sexual maturity is attained. The oldest individuals in which plastral length had increased measurably in the season of capture were females 14 (2 specimens) and 15 (1) years old. The age of the oldest growing male was 13 years.

The germinal layer of the epidermis probably remains semiactive throughout life but functions chiefly as a repair mechanism in adults that are no longer growing. Growth-rings continue to form irregularly in some older adults. Growth-rings formed after the period of regular growth are so closely approximated that they are unmeasurable and frequently indistinguishable to the unaided eye. If the continued formation of growth-rings is not accompanied by wear at the edges of the laminae, the laminae meeting at an interlaminal seam descend, like steps, into the seam ([Pl. 22, Fig. 2]). Interlaminal seams of the plastron deepen with advancing age in most individuals.

Some individuals that are well past the age of regular growth show measurable increments in years when conditions are especially favorable. The three oldest growing females were collected in 1954—an exceptionally good year for growth. Allowing some latitude for irregular periods of growth in favorable years subsequent to the period of regular, more or less steady growth, 15 to 20 years is a tenable estimate of the total growing period.

Longevity

Practically nothing is known about longevity in T. ornata or in other species of Terrapene although the several plausible records of ages of 80 to more than 100 years for T. carolina (Oliver, 1955:295-6) would indicate that box turtles, as a group, are long-lived. There is no known way to determine accurately the age of an adult turtle after it has stopped growing. It was possible occasionally to determine ages of 20 to 30 years with fair accuracy by counting all growth-rings (including those crowded into the interabdominal seam) of specimens having unworn shells. Without the presence of newly formed epidermis as a landmark, however, it was never certain how many years had passed since the last ring was formed.

Fig. 13. The relationship of sexual maturity to size in 164 specimens (94 females and 70 males) of Terrapene o. ornata, expressed as the percentage of mature individuals in each of five groups arranged according to plastral length. Sexual maturity was determined by examination of gonads. Solid bars are for males and open bars for females. The bar for males in the largest group is based on assumption since no males in the sample were so long as 130 mm. Males mature at a smaller size and lesser age (see also Figs. [9] and [10]) than females. Plastral lengths of the smallest sexually mature male and female in the sample were, respectively, 99 and 107 mm.

Mattox (1936) studied annual rings in the long bones of painted turtles (Chrysemys picta) and found fewer rings in younger than in older individuals but, beyond this, reached no important conclusion. In the present study, thin sections were ground from the humeri and femurs of box turtles of various ages and sizes; the results of this investigation were negative. Distinct rings were present in the compact bony tissue but it appeared that, after the first year or two, the rings were destroyed by encroachment of the marrow cavity at about the same rate at which they were formed peripherally.

The only methods that I know of to determine successfully the longevity of long-lived reptiles would be to keep individuals under observation for long periods of time or to study populations of marked individuals. Both methods have the obvious disadvantage of requiring somewhat more than a human lifetime to carry them to completion. Restudy, after one or more decades, of the populations of turtles marked by Fitch and myself may provide valuable data on the average and maximum age reached by T. ornata.

Ornate box turtles probably live at least twice as long as the total period of growing years. An estimated longevity of 50 years would seem to agree with present scant information on age. Considering environmental hazards, it would be unusual for an individual to survive as long as 100 years in the wild.

Weight