Chapter II: Part 2

A marked reduction in the percentage of females having vaginal orifices was observed in the unusually dry summer of 1952. The rate of reproduction was found to be positively correlated with rainfall (Fig. 9). Correlation coefficients were higher in each case when the amount of rainfall in the month preceding each sampling period was used instead of that in the month of the sample. This suggested that the rainfall exerted its influence indirectly through its effect on plant growth. Bailey (1924:530) reported that a reduction in either the quantity or quality of food had a depressing effect on reproduction. Drought, such as occurred in 1952, would certainly have a depressing effect on both. The critical factor seems to be the supply of new, actively growing shoots available to the voles for food rather than the total amount of vegetation. As far as could be determined from the small sample of males examined, their fecundity was not affected by rainfall. Some decrease in the percentage of males that were fecund was noted in the winter and was reported also by Jameson (1947:145) but most of the males in any sample were fecund. Thus any depression in the reproductive rate was due to loss of fecundity by females. This was in agreement with reports in the literature on the subject (Baker and Ransom, 1932a:320; 1932b:43).

The correlation coefficient between rainfall and the percentage of adult females with a vaginal orifice was 0.53. This was considered to be surprisingly high in view of the expected effects on the breeding rate of temperature, seasonal diet variations and whatever rhythms were inherent in the voles. When only the summer months were considered the correlation coefficient between rainfall and the percentage of adult females with a vaginal orifice was 0.84. This indicated that, during the season when breeding was at its height, rainfall was a factor in determining the rate of reproduction and when rainfall was scarce, as in the summer of 1952, it seemed to be a limiting factor (Fig. 9).

Of the total captures 20.6 per cent involved more than one individual. When the distribution of these multiple captures was graphed for the period of study, a high correlation between the percentage of captures that were multiple and the percentage of females with a vaginal orifice (r = 0.70) was found. An even higher correlation (r = 0.76) was observed between the percentage of captures that were multiple and the population density. The higher percentage of multiple captures may have been largely a result of fewer available traps per individual on the area and thus only indirectly related to the rate of reproduction.

Of the multiple captures, 66 per cent involved both sexes. The correlation coefficient between the percentage of captures involving both sexes and the level of reproductive activity was 0.58. Among those pairs of individuals caught together more than once, 61 per cent were composed of both sexes. Among those pairs taken together three or more times 76 per cent were male and female and among those pairs taken together four or more times 80 per cent were male and female. When adult voles stayed together any length of time their relationship usually appeared to be connected with sex. Family groups were also noted, as pairs were often trapped which seemed to be mother and offspring. A lactating female would sometimes enter a trap even after it had been sprung by a juvenile, presumably her offspring, or a juvenal vole would enter a trap after its mother had been captured. Such family groups persisted only until the young voles had been weaned.

The youngest female known to be gravid was 26 days old and weighed 28 grams. During summer most of the females were gravid before they were six weeks old, although females born in October and after were often more than 15 weeks old before they became gravid. The youngest male known to be fecund was approximately six weeks old. Male fecundity was determined as described by Jameson (1950). Difference in the age of attainment of sexual maturity serves to reduce the mating of litter mates (Hamilton, 1941:7) and has been noticed in various species of the genus _Microtus_ by several authors (Bailey, 1924:529; Hatfield, 1935:264; Hamilton, _loc. cit._; Leslie and Ransom, 1940:32).

For 35 females, each of which was caught at least once each month for ten consecutive months or longer, the mean number of litters per year was 4.07. Certain of the more productive members of the group produced 11 litters in 16 months. _M. ochrogaster_ seems to be less prolific than _M. pennsylvanicus_. Bailey (1924:528) reported that one female meadow vole delivered 17 litters in 12 months. Hamilton (1941:14) considered 17 litters per year to be the maximum and stated that in years when the vole population was low the females produced an average of five to six litters per year. In "mouse years" the average rose to eight to ten litters per year. During this study several females delivered two or more litters in rapid succession. This was noted more frequently in spring and early summer than in other parts of the year. Those females which produced two or three litters in rapid succession in spring and early summer often did not litter again until fall. Post-parous copulation has been observed in _M. pennsylvanicus_ by Bailey (1924:528) and Hamilton (1940:429; 1949:259) and probably occurs also in _M. ochrogaster_.

The gestation period was approximately 21 days, the same as reported for _M. pennsylvanicus_ (Bailey, _loc. cit._; Hamilton, 1941:13) and _M. californicus_ (Hatfield, 1935:264). A more precise study of the breeding habits of _M. ochrogaster_ failed to materialize when the voles refused to breed in captivity. Fisher (1945:437) also reported that _M. ochrogaster_ failed to breed in captivity although _M. pennsylvanicus_ (Bailey, 1924) and _M. californicus_ (Hatfield, 1935) reproduced readily in the laboratory.

Litter Size and Weight

In the course of this study 65 litters were observed. The mean number of young per litter was 3.18 +- 0.24 and the median was three (Fig. 10). Three litters contained but one individual and the largest litter contained six individuals. Other investigators have reported the number of young per litter in _M. ochrogaster_ as three or four (Lantz, 1907:18) and 3.4 (1-7) (Jameson, 1947:146). _M. pennsylvanicus_ seems to have larger litters. Although Poiley (1949:317) found the mean size of 416 litters to be only 3.72 +- 0.18, both Bailey (1924:528) and Hamilton (1941:15) found five to be the commonest number of young per litter in that species. Leslie and Ransom (1940:29) reported the average number of live births per litter to be 3.61 in the British vole, _M. agrestis_. Selle (1928:96) reported the average size of five litters of _M. californicus_ to be 4.8. Hatfield (1935:265), working with the same species, found that litter size varied directly with the age of the female producing the litter. He reported litters of young females as two to four young per litter and of older females as five to seven young per litter. In the litters of _M. ochrogaster_ that I examined, young females did not have more than three young and usually had but two. However, older females had litters of one, two and three often enough so that no relationship, as described above, was indicated clearly.

No seasonal variation in litter size was noted. The mean size of the litters in 1950, 2.68 +- 0.30, was significantly lower than that found in 1951 (3.76 +- 0.20) but neither differed significantly from the mean size of litters in 1952 (3.35 +- 0.66). The lower mean size of litters was in part coincidental with a high population level and the higher mean of the two later years was in part coincidental with a low population level. Since a sharp break in the curve for population density occurred after the flood in July, 1951, the litters were arranged in pre-flood and post-flood categories for study. Pre-flood litters averaged 3.07 +- 0.28 young per litter whereas post-flood litters averaged 3.34 +- 0.48. This difference was not significant. Increase in litter size, if it had actually occurred, might have been a response to the increasing food supply and lower population density after the flood.

A difference in the mean number of young per litter was noted for those litters delivered in traps as compared with those delivered in captivity and the numbers of embryos examined in the uterus. The mean number of embryos per female was higher than the mean number of young per litter delivered in captivity and the mean number of young per litter delivered in traps was lower than in those delivered in captivity. The differences were not statistically significant. In some instances females that delivered young voles in traps may have delivered others prior to entering the trap or the mother or her trapmates may have eaten some of the newborn voles before they were discovered.

The mean weight of 16 newborn (less than one day old) individuals was 2.8 +- 0.36 grams. No other data on the weight of newborn _M. ochrogaster_ were found in the literature but this mean was close to the 3.0 grams (Bailey, 1924:530) and 2.07 grams (Hamilton, 1937a:504; 1941:10) reported for _M. pennsylvanicus_ and to the 2.7 grams (Selle, 1928:97) and 2.8 grams (Hatfield, 1935:268) reported for _M. californicus_. No correlation between the weight of the individual newborn vole and the number of voles per litter was observed.

Although the ratio of the average weight of newborn voles to the average weight of an adult female was approximately equal for _M. pennsylvanicus_ and _M. ochrogaster_, the ratio of the weight of a litter to the average weight of an adult female was larger in the eastern meadow vole because the mean litter size was larger. Perhaps this is related to the more productive habitat in which the eastern meadow vole is ordinarily found.

Size, Growth Rates and Life Spans

The mean weight of adult voles during the period of study was 43.78 grams. The females averaged slightly heavier than the males but the overlapping of weights was so extensive that sexual difference in weight could not be affirmed. The difference observed was less in December and January when gravid females were rare, suggesting that the difference was due, at least in part, to pregnancy. Jameson (1947:128) found, for a sample of 50 voles, a mean weight of 44 grams and a range of 38 to 58 grams. The range in the adult voles I studied was much greater, from 25 to 73 grams. In part, this increase in the range of adult weights was due to a much larger sample.

During the unusually dry summer of 1952, a notable reduction in the mean weight of adults was recorded (Fig. 11). The correlation coefficient between the mean weight of adults and the amount of rainfall for the summer months was 0.68. It seems reasonable to attribute the drop in mean weight to an alteration of plant growth due to decreased rainfall. Some of the reduction in mean weight was due to the loss of weight in older individuals but most of it was due to the failure of voles born in the spring to continue growing.

No data on the growth rate of _M. ochrogaster_ were found in the literature. According to the somewhat scanty data from my study, secured from observations of individuals born in the laboratory, young voles gained approximately 0.6 of a gram per day for the first ten days, approximately one gram per day up to an age of one month, and approximately 0.5 of a gram per day from an age of one month until growth ceases. This growth rate was especially variable after the voles reached an age of thirty days. The growth rate approximates those described for _M. pennsylvanicus_ (Hamilton, 1941:12) and for _M. californicus_ (Hatfield, 1935:269; Selle, 1928:97). Although the data were inadequate for a definite statement, I gained the impression that there was no difference between the sexes in growth rate. In general, young voles grow most rapidly in the April-May-June period and least rapidly in mid-winter. Several voles, born in late autumn, stopped growing while still far short of adult size and lived through the winter without gaining weight, then gained as much as 30 per cent after spring arrived (Fig. 12).

The recorded life spans of most voles studied were less than one year. No accurate mean life span could be determined. Leslie and Ransom (1940:46), Hamilton (1937a:506) and Fisher (1945:436) also found that most voles lived less than one year. Leslie and Ransom (_op. cit._: 47) reported a mean life span of 237.59 +- 10.884 days in voles of a laboratory population. In the present study one female was trapped 624 days after first being captured; another female was trapped 617 days after first being captured; and a male was trapped 611 days after first being captured. The two females were subadults when first captured. The male was already an adult when first captured; consequently its life span must have exceeded 650 days. No evidence of any decrease in vigor or fertility was observed to accompany old age.

Of the 45 marked voles snap-trapped in August of 1952, 21 had been captured first as juveniles. The ages of these voles could be estimated within a few days, and the series presented a unique opportunity for studying individual and age variation. Only individuals weighing less than 18 grams when first captured were used, and their ages were estimated according to the growth rate described above. Howell (1924) reported an analysis of individual and age variation in a series of specimens of _Microtus montanus_, and Hall (1926) studied the changes due to growth in skulls of _Otospermophilus grammarus beecheyi_. The series of specimens described here differs from those of Hall and Howell, and from any other collection known to me, in the fact that the specimens are of approximately known age and drawn from a wild population.

Unfortunately, this sample was small, and the distribution of the specimens among age groups left much to be desired. No specimens less than one and one-half months old were taken and only a few individuals older than four and one-half months. Table 3 shows the age distribution. The small size of the sample and the absence of juveniles were due, partly, to the unusually dry weather in the summer of 1952. The reduction in the rate of reproduction, caused by drought (as described elsewhere in this paper), reduced the populations and the percentage of juveniles to low levels.

TABLE 3. DISTRIBUTION AMONG AGE GROUPS OF 21 VOLES USED IN THE STUDY OF
VARIATION DUE TO AGE

======================================================================
Age in months 1-1/2 2 2-1/2 3 3-1/2 4 4-1/2 6 12
----------------------------------------------------------------------
No. of individuals 1 4 5 1 3 2 3 1 1
----------------------------------------------------------------------

In the series of voles studied, ten individuals were in the process of molting from subadult to adult pelage. Jameson (1947:131) reported the molt to occur between eight and 12 weeks of age and selected 38 grams as the lower limit of weight of adults. I also found all voles molting to be between eight and 12 weeks old but found none so large as 38 grams without full adult pelage. This may have been, in part, due to the dry weather delaying or inhibiting growth. Because of the small size of the sample and the influence of the unusual weather conditions, no conclusions concerning normal molting were drawn from the data described below. They are presented only as a description of a small sample drawn from a single population at one time. Table 4 summarizes these data.

TABLE 4. MEAN SIZES AND AGES OF VOLES MOLTING FROM SUBADULT TO ADULT
PELAGE

=====================================================================
Body length Condylo-basilar
Weight minus tail length Age
---------------------------------------------------------------------
Six males 32.67 gms. 106.16 mm. 23.78 mm. 9.67 wks.
(30-36) (96-116) (23.2-24.4) (8-12)
Four females 29.0 gms. 100.25 mm. 23.45 mm. 10.5 wks.
(28-30) (98-102) (23.5-23.8) (8-12)
Ten voles 31.2 gms. 103.8 mm. 23.73 mm. 10.0 wks.
(28-36) (96-116) (23.2-24.4) (8-12)
---------------------------------------------------------------------

The mean age of the ten voles molting was ten weeks (8-12). Six males averaged 9.67 weeks, almost a week younger than four females, who averaged 10.5 weeks. The difference in age at time of molting between the sexes was not significant. Differences between the sexes in other characteristics to be described also lacked significance. Mean weights at the time of molting were: males, 32.67 gms. (30-36); females, 29.0 gms. (28-30); and all individuals, 31.2 gms. (28-36). Because a piece of the tail of each vole had been removed in marking, the total length of the voles could not be determined. Body length, excluding tail, was used. Howell (1924:986) found this measurement subject to less individual variation than total length and thought body length was probably a better indicator of age. Mean body length at the time of molting was 103.8 mm. (96-116). Males averaged longer than females and were also more variable. The mean body length of males was 106.16 mm. (96-116) and that of females was 100.25 mm. (98-102).

Of the subadults showing no signs of molting, none was above the mean age of molting. Twenty-five per cent of them were longer and heavier than the mean length and weight of those that were molting. Of the 20 adults in the series, one was below the mean weight of molting and one was shorter than the mean length of molting.

When Howell (_op. cit._:1014) studied skulls of _Microtus montanus_ he found that the condylobasilar length was the most satisfactory means for arranging his series of specimens according to their age. When the skulls of my series were arranged according to their age (as determined from trapping records) the graph of the condylobasilar lengths showed a clear, though not perfect, relationship to age (Fig. 13). No separation of sexes was made because the sample did not permit it. In Fig. 13 graphs of weight, as determined in the field, and of length (excluding tail) also were included because they are the most easily measured characters of live voles. The graphs indicate individual variation in these characters which limits their usefulness in determining age.

When other cranial measurements, and ratios of pairs of measurements, were plotted in the same order, individual variation obscured some of the variation due to age and the curves resembled those of weight and length of body rather than that of condylobasilar length. When the cranial measurements were averaged for the age groups the curves showed a relationship to age but the relationship of mean measurements is of little use in determining the age of individual specimens. The data described above indicated that a study of the relationship of the condylobasilar length and age in a large sample might provide useful information.

Anyone who has examined mammalian skulls knows of many other characters which vary with age but which are difficult to measure and describe with precision. Figs 14 and 15 are drawings of skulls of voles of known age. The most obvious change, related to aging, evident in the dorsal view of the skulls (Fig. 14) is the increasing prominence and closer approximation of the temporal ridges in older specimens. The lambdoidal ridge is also more prominent in older voles, and their skulls have a generally rougher and more angular appearance. The individual variation evident in these ridges is probably due to variations in the development of the muscles operating the jaws (Howell, 1924:1003). There is an increased flattening of the roof of the skull of older voles.

4 months 4-1/2 months 6 months 12 months

All x 3.

FIG. 14. Dorsal views of skulls of voles of known age.]

4 months 4-1/2 months 6 months 12 months

All x 3.

FIG. 15. Palatal views of skulls of voles of known age.]

From a palatal view (Fig. 15) the skulls of voles also showed age variation which was apparent but not easily correlated with precise age. The median ridge on the basioccipital bone increases in prominence in older voles. The shape of the posterior margin of the palatine bones changes from a V-shape to a U-shape. On the skull of the oldest (12 months) vole the pterygoid processes are firmly fused to the bullae, a condition not found in any of the other specimens. The anterior spine of the palatine approaches the posterior projection of the premaxillae more closely as age increases and, in the oldest vole is firmly attached and forms a complete partition separating the incisive foramina.

Tooth wear during the life of a vole causes a considerable variation in the enamel patterns, especially of the third upper molar. Howell (1924:1012) considered such variation to be independent of age, but Hinton (1926:103) related the changes to age and interpreted them as a recapitulation of the evolution of microtine molars. In my series, an indentation on the medial margin of the posterior loop of the third upper molar seemed to be related to age. This indentation was absent in the youngest vole (one and one-half months), absent or indefinite in those voles less than 3-1/2 months of age, and progressively more marked in the older voles.

Food Habits

The prairie vole, like other members of the genus _Microtus_, feeds mostly on growing grass in spring and summer. Piles of cuttings in the runways are characteristic sign of the presence of voles. The voles cut successive sections from the bases of grasses until the young and tender growing tips are within reach. The quantity of grass destroyed is greater than that actually eaten, a fact which will have to be considered in any attempt to evaluate the effects of voles upon a range.

In all piles of cut plants that were examined, _Bromus inermis_ was the most common grass, and _Poa pratensis_ was the grass second in abundance. These were, by far, the most common grasses present on the areas studied; in most places, _B. inermis_ was dominant. Other grasses present on the areas were occasionally found in the piles of cuttings. Jameson (1947:133-136) found no utilization of _B. inermis_ by voles but that grass was present in a relative abundance of only one per cent in the areas studied by him. The voles that he studied ate alfalfa in large amounts and alfalfa was, perhaps, the most common plant on the particular areas where his voles were caught. Seemingly, the diet of voles is determined mostly by the species composition of the habitat.

Other summer foods included pokeberries, blackberries and a few forbs and insects. Forbs most commonly found in the piles of cuttings were the leaves of the giant ragweed (younger plants only) and dandelion. Insect remains were found in the stomachs of voles killed in summer and occurred most frequently in those killed in August and September. At no time did insects seem to be a major part of the diet but they were present in most vole stomachs examined in late summer. Laboratory experiments with summer foods gave inconclusive results but suggested that the voles chose grasses on the basis of their growth stage rather than according to their species. Young and tender grasses were chosen, regardless of species, when various combinations of _Triodia flava_, _Bromus inermis_ and _Poa pratensis_ were offered to the voles. At times the voles showed a marked preference for dandelion greens, perhaps because of their high moisture content; the voles' water needs were satisfied mostly by eating such succulent vegetation.

Winter foods consisted of stored hay and fruits and of underground plant parts. _Bromus inermis_ made up nearly all of the hay and was stored in lengths of up to ten inches in underground chambers specially constructed for storage. Underground parts of plants were reached by tunnelling and were an especially important part of the voles' diet in January and February. The fruit of _Solanum carolinense_ was eaten throughout the winter and one underground chamber, opened in February, 1952, was packed full of these seemingly unsavory fruits. Fisher (1945:436), in Missouri, found this fruit to be an important part of the winter diet of voles. An occasional pod of the honey locust tree was found partly eaten in a runway. Fitch (1953, _in litt._) often observed girdling of honey locust and crab apple (_Pyrus ioensis_) root crowns on the Reservation but I saw no evidence of bark eating, perhaps because my study plots were mostly grassland. On two occasions when two voles were in the same trap one of them was eaten. In both traps, all of the bait had been eaten and the captured voles probably were approaching starvation. Because the trapping procedure offered abundant opportunity for cannibalism, the low frequency of its occurrence suggested that it was not an important factor in satisfying food requirements under normal conditions.

Runways and Nests

Perhaps the most characteristic sign of the presence of _Microtus ochrogaster_ were their surface runways and underground tunnels. Only rarely was a vole observed to expose itself to full view. When a trapped vole was released it immediately dove out of sight into a runway. Once in a runway, the vole showed no further evidence of alarm and was usually in no hurry to get away. The runways seemed to provide a sense of security and the voles were familiar with their range only through runway travel. The urge to seek a runway immediately when exposed has obvious survival value.

Surface runways were usually under a mat of debris. In areas where debris was scanty or lacking, runways were usually absent. Jameson (1947:136) reported that in alfalfa and clover fields the voles did not make runways as they did in grassland, even in fields where trapping records showed voles to be abundant. Typical surface runways are approximately 50 mm. wide, only slightly cut into the ground and bare of vegetation while in use. Usually they could be distinguished from the runways of the pine vole, which were cut more deeply into the ground, and those of the cotton rat which were wider and not so well cleared of vegetation. Some runways ended in surface chambers and some of these were lined with grass. Their size varied from a diameter of 90 mm. to 250 mm. and they seemed to be used primarily for resting places.

A runway system usually consisted of a long, crooked runway and several branches. Two typical systems are illustrated in Fig. 16. The runway systems often were not clearly limited; they merged with other systems more or less completely. One map showed a runway system extending across 140 square meters and including 12 underground burrows. All of these runways seemed to be part of a single runway system but the system probably was used by more than one vole or family group of voles. Sixteen of the 22 maps that were made extended across areas between 50 and 90 square meters. One map, mentioned above, was larger and the remaining five smaller. The smallest extended across only 20 square meters. Of course, the area encompassed by a set of runways changed almost daily, as the voles extended some runways, added some and abandoned others in the course of their daily travels.

Each runway system contained underground nests. These were in chambers from 70 mm. to 200 mm. below the surface and were up to 200 mm. in diameter. Most systems that were mapped had from two to six of these burrows. Most of these were lined with dried grass and seemed to be used for delivering and nursing litters. Each burrow was connected to a surface runway by a tunnel. Often the tunnel was short and the hole opened almost directly into the burrow from the surface runway. Others had tunnels several meters long. Jameson (1947:137) reported every burrow to have two connections with the surface. In the present study, however, I found three arrangements in approximately equal frequency of occurrence: (1) one hole to one tunnel leading to a burrow; (2) two holes to two short tunnels which joined a long tunnel leading to a burrow; and (3) two separate tunnels from the surface to a burrow. The size, depth and number of underground burrows in the systems that I studied varied and so did those reported in the literature. Jameson (_loc. cit._) found burrows in eastern Kansas as deep as 18 inches, far deeper than any found in my study. Fisher (1945:435) reported none deeper than five inches in central Missouri. The soil data in my study, as well as in the two reports cited immediately above, were not adequate to permit conclusions, but the type and condition of the soil probably determine the extent of burrowing by the voles of any given locality.

The number of voles using a runway system at one time was difficult to ascertain. In one system, however, four adult individuals were trapped in a ten day period. In August, 1952, at the conclusion of the live-trapping program, a runway system was mapped which had included two trapping stations. In the preceding ten days, four adult voles (three males and one female) had been taken in both traps. During that time, therefore, the runway system was shared by at least four voles. The voles used an area that was considerably larger than that encompassed by any one runway system, a fact obvious when the sizes of home ranges as computed from trapping data were compared with the sizes of the runway systems mapped. A runway system seemed not to be a complete unit, but was only a part of the network of runways used by a single individual.

Activity

Although no special investigation of activity was made, some conclusions concerning it were apparent in the data gathered. There have been a few laboratory studies of the activity pattern of _Microtus_ by various methods. Calhoun (1945:256) reported _M. ochrogaster_ to be mainly nocturnal with activity reaching a peak between dark and midnight and again just before dawn. Davis (1933:235), working with _M. agrestis_, and Hatfield (1935:263), working with _M. californicus_, both found voles to be more nocturnal than diurnal. In a field study of _M. pennsylvanicus_, Hatt (1930:534) found the species to be chiefly nocturnal, although some activity was reported throughout the day. Hamilton (1937c:256-259), however, reported the same species to be more active in the daytime. Agreement on the activity patterns of these species of _Microtus_ has not yet been attained.

From occasional changes in the time of tending a trap line, and from running lines of traps at night a few times in the summer of 1951, I gained the impression that these voles were primarily diurnal. Relatively few of them were caught in the hours of darkness. In summer, however, their activity was mostly limited to the periods between dawn and approximately eight o'clock and between sunset and dark. In colder weather, there was increased activity on sunny days.

PREDATION

Although voles were a common item of prey for many species of predators on the Reservation, no marked effect on the density of the population of this vole could be attributed to predation pressure. Only when densities reached a point that caused many voles to expose themselves abnormally could they be heavily preyed upon. Their normally secretive habits, keeping them more or less out of sight, suggest that they are an especially obvious illustration of the concept that predation is an expression of population vulnerability, rising to high levels only when a population is ecologically insecure, rather than a major factor regulating population levels (Errington, 1935; 1936; 1943; Errington _et al_, 1940).

Scats from predatory mammals and reptiles and pellets from raptorial birds were examined. Most of these materials were collected by Dr. Henry S. Fitch, who kindly granted permission to use them. The results of the study of the scats and pellets are summarized in Table 5. Remains of voles were identified in 28 per cent of the scats of the copperhead snake (_Ancistrodon contortix_) examined. Copperheads were moderately common on the Reservation (Fitch, 1952:24) and were probably important as predators on voles in some habitats. Uhler _et al_ (1939:611), in Virginia, reported voles to be the most important prey item for copperheads. A vole was taken from the stomach of a rattlesnake (_Crotalus horridus_) found dead on a county road adjoining the Reservation. Rattlesnakes were present in small numbers on the Reservation but were usually found along rocky ledges rather than in areas where voles were common (Fitch, _loc. cit._). The rattlesnakes probably were less important as predators on voles than on other small mammals more common in the usual habitat of these snakes. The blue racer (_Coluber constrictor_) was common in grassland situations on the Reservation (Fitch, 1952:24) and twice was observed in the role of a predator on voles; one small blue racer entered a live-trap in pursuit of a vole and another blue racer was observed holding a captured vole in its mouth. The blue racer seems well adapted to hunt voles and probably preys on them extensively. The pilot black snake (_Elaphe obsoleta_) has been reported as a predator on _M. ochrogaster_ in the neighboring state of Missouri (Korschgen, 1952:60) and was moderately common on the Reservation (Fitch, _loc. cit._). _M. pennsylvanicus_, with habits similar to those of _M. ochrogaster_, has been reported as a prey for all of the above snakes (Uhler, _et al_, 1939).

TABLE 5. FREQUENCY OF REMAINS OF VOLES IN SCATS AND PELLETS

=========================================================================
No. of scats or No. containing
Predator pellets examined remains of voles Percentage
-------------------------------------------------------------------------
Copperhead 25 7 28
Red-tailed hawk 25 3 12
Long-eared owl 25 18 72
Great horned owl 32 6 19
Crow 25 4 16
Coyote 25 3 12
-------------------------------------------------------------------------

The red-tailed hawk (_Buteo jamaicensis_), the long-eared owl (_Asio otus_), the great horned owl (_Bubo virginianus_) and the crow (_Corvus brachyrhynchos_) fed on _Microtus_. All four birds were fairly common permanent residents on the Reservation (Fitch, 1952:25). The low density and the strict territoriality of the red-tailed hawk (Fitch, _et al_, 1946:207) prevented it from exerting any important influence on the population of voles, even though individual red-tailed hawks ate many voles. Predation by the long-eared owl was especially heavy; remains of voles were identified in 72 per cent of its pellets examined. Korschgen (1952:39) found remains of voles in 70 per cent of 704 pellets of the long-eared owl. The reason for the heavy diet of _Microtus_ seems to be that both the owl and the vole are especially active at dusk. A group of long-eared owls, living near the edge of Quarry Field, probably exerted an influence on the density of the local population of voles because of the high ratio of predator to prey animals. The crows ate some, and perhaps most, of their voles after the animals had died from other causes. Other birds, mostly raptors, occurring in northeastern Kansas and reported to prey on voles include the sharp-shinned hawk (_Accipiter striatus_), Cooper's hawk (_A. cooperi_), red-shouldered hawk (_Buteo lineatus_), broad-winged hawk (_B. platypterus_), American rough-legged hawk (_B. lagopus_), ferruginous rough-legged hawk (_B. regalis_), marsh hawk (_Circus cyaneus_), barn owl (_Tyto alba_), screech owl (_Otus asio_), barred owl (_Strix varia_) and shrike (_Lanius excubitor_) (Korschgen, 1952:26; 28; 34; 35; 37; McAtee, 1935:9-27; Wooster, 1936:396).

Coyotes, house cats and raccoons were identified as predators on voles in the study areas. Remains of voles were present in 12 per cent of the scats of the coyote (_Canis latrans_) examined. In Missouri, Korschgen (1952:40-43) reported remains of voles in slightly more than 20 per cent of the coyote stomachs that he examined. Fitch (1948:74), Hatt (1930:559) and others have reported other species of _Microtus_ as eaten by the coyote. Although coyotes were rarely seen on the Reservation, coyote sign was abundant (Fitch, 1952:29) and coyotes probably ate large numbers of voles. House cats (_Felis domesticus_), seemingly feral, were observed to tour the trap lines on several occasions and were noted by Fitch (_loc. cit._) as important predators on small vertebrates. Four cats were killed in the course of the study and remains of voles were found in the stomachs of all of them. On several occasions, raccoon tracks were noted following the trap line when the traps had been overturned and broken open, suggesting that raccoons are not averse to eating voles although no further evidence of predation on voles by raccoons was obtained. Fitch (_loc. cit._) reported raccoons (_Procyon lotor_) to be moderately common on the Reservation. Reports of predation by raccoons on voles are numerous (Hatt, 1930:554; Lantz, 1907:41). The opossum (_Didelphis marsupialis_), common on the Reservation, occasionally eats voles (Sandidge, 1953:99-101). Other mammals which are probably important predators on voles on the Reservation, though no specific information is available, are the striped skunk (_Mephitis mephitis_), spotted skunk (_Spilogale putorius_), weasel (_Mustela frenata_) and the red fox (_Vulpes fulva_). Eadie (1944; 1948; 1952), Shapiro (1950:360) and others have reported that the short-tailed shrew (_Blarina brevicauda_) was an important predator on _Microtus_. Shrews were present on the Reservation but were not trapped often enough to permit study.

The variety of vertebrates preying on voles suggests that they occupy a position of importance in many food chains. Errington (1935:199) and McAtee (1935:4) refer to voles as staple items of prey for all classes of predatory vertebrates. An attempt to evaluate prey species was made by Wooster (1939). He proposed a formula which involved multiplying the density of a species, its mean individual weight, the fraction of the day it was active and the fraction of the year it was active to give a numerical index of prey value. Although his methods of determining population densities would now be considered questionable, the purpose of his investigation merits further consideration. He reported _M. ochrogaster_ to be second only to the jack-rabbit (_Lepus californicus_) as a prey species in west-central Kansas.

MAMMALIAN ASSOCIATES

In the course of live-trapping operations several species of small mammals other than _Microtus ochrogaster_ were taken in the traps. Also, from time to time, direct observations of certain mammals were made and various types of sign of larger mammals were noted. These records gave a picture of the mammalian community of which the voles were a part. The three associated species which were most commonly trapped were _Sigmodon hispidus_, _Reithrodontomys megalotis_ and _Peromyscus leucopus_. These three species have been commonly found associated with _Microtus_ in this part of the country (Fisher, 1945:435; Jameson, 1947:137).

The Texas cotton rat, _Sigmodon hispidus_, was the most commonly trapped associate of the voles between November, 1950, and February, 1952. Although a greater number of individuals of the harvest mouse were taken in a few months, the cotton rat had a greater ecological importance because of its larger size (Figs. 17, 18, 19). The cotton rat was an especially noteworthy member of the community for two reasons. It has arrived in northern Kansas only recently and its progressive range extension northward and westward has attracted the attention of many mammalogists (Bailey, 1902:107; Cockrum, 1948; 1952:183-187; Rinker, 1942b). Secondly, _Sigmodon_ has long been considered to be almost the ecological equivalent of _Microtus_ and to replace the vole in the southern United States (Calhoun, 1945:251; Svihla, 1929:353). Since the two species are now found together over large parts of Kansas their relationships in the state need careful study.

Both this study and the literature (Black, 1937:197; Calhoun, _loc. cit._; Meyer and Meyer, 1944:108; Phillips, 1936:678; Rinker, 1942a:377; Strecker, 1929:216-218; Svihla, 1929:352-353) showed that, in general, the habitat needs of _Microtus_ and _Sigmodon_ were similar. Studies on the Natural History Reservation, both in connection with my problem and otherwise, suggested, however, that _Sigmodon_ occurred in only the more productive habitat types used by voles, where the vegetation was relatively high and rank. On the Reservation the cotton rat was found mostly in the lower meadows; they were more moist and had a more luxuriant vegetation than the higher fields. Although a few cotton rats were taken in Quarry Field and still fewer in Reithro Field, the population of those hilltop areas did not approach, at any time, the levels reached on House Field, which produced a more luxuriant cover. Only when the levels of population were exceptionally high did the cotton rats spread into less productive habitats. At all times, there were areas on the Reservation used by _Microtus_ which could not support a population of _Sigmodon_.

The cotton rats reacted differently to the floods of July, 1951, than did the voles. Although the population of the cotton rat decreased slightly immediately after the wet period, this decrease was insignificant when compared with the drop in population level of other species of small mammals on the same area. During the autumn of 1951 and until March, 1952, the cotton rat became the most important mammal on the House Field study area in terms of grams per acre (Fig. 17), although the number of cotton rats per acre never matched the density of the voles. A similar, though less pronounced, trend was observed on the Quarry Field study area (Fig. 18). One factor in the success of the cotton rat at this time seemed to be the greater resistance to wetting shown by very young individuals. Few adults (of any species) marked before the heavy rains of July, 1951, were trapped in September, 1951, when trapping was resumed after a lapse of one month. Several subadults and some juvenal cotton rats did survive, however, and provided a breeding population from which the area was repopulated. Cotton rats are born fully furred and able to move well, and are often weaned at ten days (Meyer and Meyer, 1944:123-124). Voles, on the other hand, are born naked and helpless and are often not weaned for three weeks. It seems, therefore, that extremely wet soil would harm the voles more than it would the cotton rats.

Several instances of cotton rats eating voles, caught in the same live-trap, were noted. There is reason to believe that young voles, unable to leave the nest, are subject to predation by cotton rats. This would accentuate any competitive advantage gained otherwise by the cotton rats.

The population of _Sigmodon_ retained its high level, relative to _Microtus_, until February, 1952. In March only one individual was captured and after that none was trapped until August, 1952, when a single subadult male was captured. Early in March, 1952, before the trapping period for the month had begun, the area suffered three successive days of unusually low temperature, with snow, which lay more than six inches deep in places. As suggested by Cockrum (1952:185), such conditions proved detrimental to the cotton rats and, at least to the end of the study period in August, 1952, the population of cotton rats had failed to recover. Perhaps the extremely dry weather which followed the heavy winter mortality delayed the recovery of the population.

These limited data seem to indicate competition between _Sigmodon_ and _Microtus_ in Kansas. Extremely wet conditions seem to give _Sigmodon_ a competitive advantage whereas _Microtus_ is better able to survive dry summers and severe winters. However, these relationships need further clarification by an intensive study of the life history of _Sigmodon_ in Kansas (especially the more arid western part), including its coactions with the communities it has invaded successfully recently.

The harvest mouse (_Reithrodontomys megalotis_) also was a common inhabitant of the study plots, but this small rodent seemed not to be a serious competitor of the voles, as its food consists almost entirely of seeds (Cockrum, _op. cit._:165) not usually used by voles. In this study, at least, no conflict over space was apparent. Harvest mice frequently were taken in the runways of voles and even in the same trap with voles. Reithro Field, the part of the Reservation having the heaviest population of the harvest mouse, differed from the habitats that were better for voles in being higher, drier and less densely covered with vegetation. However, during the summer of 1951 when the voles were most abundant, Reithro Field supported a large population of voles. Estimates of population of the harvest mouse were of doubtful validity in summer because it was readily trapped only in winter and early spring. Many individuals marked in late spring were not trapped again until late autumn although presumably they remained on the area. This seasonal variation in trapping success seemed to be a matter of acceptance and refusal of bait (Fitch, 1954:45).

The presence of the wood mouse (_Peromyscus leucopus_) on the study plots indicated an overlapping of habitats. Both House and Quarry Fields were on the ecotone between forest and meadow and a mixture of mammals from both types of habitat occurred. No sign of the homes of the wood mouse was found on the study plots, and on the larger trap line, operated by Fitch, wood mice were captured only near the edge of the woods.

Only six deer mice (_Peromyscus maniculatus_) were taken on the study plots. This small number probably provided an inaccurate index of the association of the deer mouse and the prairie vole, because samples from snap-traps and the data of other workers on the Reservation showed a more common occurrence of the two species together. The deer mice seemed to prefer a sparser vegetation and did not approach so closely to the forest edge as did the voles. It may have been, in part, the presence of _P. leucopus_ in the ecotonal region which made it unsuitable for _P. maniculatus_.

Other mammals noted on the study areas were the following: _Didelphis marsupialis_, _Blarina brevicauda_, _Scalopus aquaticus_, _Canis familiaris_, _Canis latrans_, _Procyon lotor_, _Felis domesticus_, _Sylvilagus floridanus_, _Microtus pinetorum_, _Mus musculus_ and _Zapus hudsonius_.

SUMMARY AND CONCLUSIONS

In the 23-month period from October, 1950, to August, 1952, the ecology of the prairie vole, _Microtus ochrogaster_, was investigated on the Natural History Reservation of the University of Kansas. In all, 817 voles were captured 2941 times in 13,880 "live-trap days." For some aspects of this study, Dr. Henry S. Fitch, resident investigator on the Reservation, permitted the use of his trapping records. He had captured 1416 voles 5098 times. The total number of live voles used in the study was thus 2233, and they were captured 8039 times. In addition to the voles, I caught 96 cotton rats, 108 harvest mice, 29 wood mice, 2 pine voles and 6 deer mice in live traps. When Fitch's records were used, the live-trapping data covered a thirty-month period and general field data were available from July, 1949, to August, 1952.

Hall and Cockrum (1953:406) stated that probably all microtine rodents fluctuate markedly in numbers. Certainly the populations I studied did so, but the fluctuations were not regularly recurring for _M. ochrogaster_ as they seem to be for some species of the genus in more northern life zones. The changes in the density of populations described in this paper can be explained without recourse to cycles of long time-span and literature dealing specifically with _M. ochrogaster_ makes no references to such cycles. There is, however, an annual cycle of abundance: greatest density of population occurs in autumn, and the least density in January.

This annual pattern is often, perhaps usually, obscured because of the extreme sensitivity of voles to a variety of changes in their environment. These changes are reflected as variations in reproductive success. In this study, some of these changes were accentuated by the great range in annual precipitation. Annual rainfall was approximately average in 1950 (36.32 inches, 0.92 inches above normal), notably high in 1951 (50.68 inches, 15.28 inches above normal) and notably low in 1952 (23.80 inches, 11.60 inches below normal).

Among the types of environmental modification to which the populations of voles reacted were plant succession, an increase in competition with _Sigmodon_, abnormal rainfall and concentration of predators. In the overgrazed disclimax existing in 1948 when the study areas were reserved, no voles were found because cover was insufficient. After the area was protected a succession of good growing years hastened the recovery of the grasses and the populations of voles reached high levels. In areas where the vegetation approached the climax community, the densities of voles decreased from the levels supported by the immediately preceding seral stages. The higher carrying capacity of these earlier seral stages was probably due to the greater variety of herbaceous vegetation which tended to maintain a more constant supply of young and growing parts of plants which were the preferred food of voles. Later in the period of study the succession from grasses to woody plants on parts of the study areas also affected the population of voles. Not only did the voles withdraw from the advancing edge of the forest, but their density decreased in the meadows as the number of shrubs and other woody plants increased. These influences of the succession of plants on the population density of voles were exerted through changes in cover and in the quality, as well as the quantity, of the food supply.

Whenever voles were in competition with cotton rats, there was a depression in the population levels of voles. Primarily, the competition between the two species is the result of an extensive coincidence of food habits, but competition for space, cover and nesting material is also present. There was one direct coaction between these two species observed. Cotton rats, at least occasionally, ate voles, especially young individuals. In extremely wet weather, as in the summer of 1951, the high survival rate of newborn cotton rats resulted in an increase in their detrimental effect on the population of voles. However, cotton rats proved to be less well adapted to severe cold or drought than were voles.

Heavy rainfall reduced the densities of populations of voles by killing a large percentage of juveniles. During the summer of 1951 the competition of cotton rats further depressed the population level of the voles, but the relative importance of competition with cotton rats and superabundant moisture in effecting the observed reduction in population density is difficult to judge. Perhaps most of the decrease in population which followed the heavy rains was due to competition rather than to weather. Subnormal rainfall, as in 1952, reduced the population of voles by inhibiting reproduction. Presumably because of an altered food supply, reproduction almost ceased during the drought. Utilization of the habitat was further reduced in the summer of 1952 because the voles did not grow so large as they otherwise did.

Predation, as a general rule, does not significantly affect densities of populations, but large numbers of predators concentrating on small areas may rapidly reduce the numbers of prey animals. In the course of my study, such a situation occurred but once, when a group of long-eared owls roosted in the woods adjacent to Quarry Field. The population of voles in that area was probably reduced somewhat as a result of predation by owls.

Population trends in either direction may be reversed suddenly by changes in the factors discussed above. In the fall of 1951, a downward trend in the density of the voles was evident. At this time, populations of cotton rats were increasing rapidly and competition between cotton rats and voles was intensified. In February, 1952, the population of cotton rats was decimated suddenly by a short period of unusually cold weather. The voles were suddenly freed from the stress of competition and the population immediately began to rise. The upward trend began prior to the annual spring increase and was subsequently reinforced by it. In the last part of May, 1952, the upward trend of the population was reversed, as the drought became severe, and the density of the population decreased rapidly. This drop was too sudden and too extreme to be only the normal summer slump. The relatively rapid response of voles to a heavy rain after a dry period, first by increased breeding and later by increases in density, is one more example of abrupt changes in population trends caused by altered environmental conditions.

In the population changes that I observed, no evident "die-off" of adults accompanied even the most drastic reductions in population density. The causative factor directly influences the population either by inhibiting reproduction or by increasing infant and prenatal mortality. The net reduction is due to an inadequate replacement of those voles lost by normal attrition.

Most voles, under natural conditions, live less than one year. Those individuals born in the autumn live longer, as a group, than those born at any other time. Since the heaviest mortality is in young voles, adults which become established in an area may live more than 18 months and, if they are females, may produce more than a dozen litters. No decrease in vigor and fertility was found to accompany aging. A relationship between the condylobasilar length of the skull and the age of a vole was discovered and, with further study, may yield a method of aging voles more accurately than has been possible heretofore. Other characteristics, varying with age, were described. The most reliable indicator of age seemed to be the prominence of the temporal ridges.