Population densities were ascertained on the study areas by means of the live-trapping program. Blair (1948:396) stated that almost all small mammals old enough to leave the nest (except shrews and moles) are captured by live-trapping. My experience, and that of other workers on the Reservation, requires modification of such a statement. The distance between traps is an important factor in determining the efficiency of live-trapping. As mentioned earlier, when House Field and Quarry Field were trapped out at the conclusion of the live-trapping program no unmarked voles were taken. This showed that the 30 foot interval between traps was short enough to cover the area as far as Microtus was concerned. The fact that unmarked adults were caught almost entirely in marginal traps is additional evidence. On the other hand, the Fitch traps were 50 feet apart and voles seemed to have lived within the grid for several months before being captured. Fitch (1954:39) has shown that some kinds of small mammals are missed in a live-trapping program because of variation in bait acceptance, both seasonal and specific.

A few individuals, missed in a trapping period, were captured again in subsequent months. These voles were assumed to have been present during the month in which they were not caught. The area actually trapped each month was estimated by a modification of the method proposed by Stickel (1946:153). The average maximum move was calculated each month and a strip one half the average maximum move in width was added to each side of the study area actually covered by traps. The study plots were bounded in part by gravel roads and forest edge acting as barriers, and for these parts no marginal strip was added. Trap lines on the opposite sides of these roads rarely caught marked voles that had crossed in either direction. It is perhaps advisable to say here that the size of House Field and Quarry Field study plots (0.56 acres) was too small for best results in estimating population levels (Blair, 1941:149). In the computations of population levels the data for males and females were combined, because no significant difference between the average maximum move of the sexes was apparent.

Fluctuations of the populations were graphed in terms of individuals per acre (Fig. 5). The variation was great in the 30 month period for which data were available, and was both chronological and topographical. The lowest density recorded was 25.2 individuals per acre and the highest density was 145.8 individuals per acre. The weight varied from a low of 847 grams per acre to a high of 5275 grams per acre.

There are few records of density of M. ochrogaster in the literature. Brumwell (1951:213) found nine individuals per acre in a prairie on the Fort Leavenworth Military Reservation and Wooster (1939:515) reported 38.5 individuals per acre for M. o. haydeni in a mixed prairie in west-central Kansas. High densities for M. pennsylvanicus reported in the literature include 29.8 individuals per acre (Blair, 1948:404), 118 individuals per acre (Bole, 1939:69), 160-230 individuals per acre (Hamilton, 1937b:781) and 67 individuals per acre (Townsend, 1935:97).

Because the study period included one period of unusually high rainfall and one year of unusually low rainfall, the normal pattern of seasonal variation of population density was obscured. An examination of the data suggested, however, that the greatest densities were reached in October and November with a second high point in the April-May-June period. These high points generally followed the periods of high levels of breeding activity (Fig. 8). The autumn rise in population may have been due, in part, to the addition of spring and early summer litters to the breeding population, but the rise occurred too late in the year to be explained by that alone. Another factor may have been the spurt in growth of grasses occurring in Kansas in early autumn, in September and October. There was a seeming correlation between high rainfall with rapid growth of grasses and reproductive activity, and, secondarily with high population densities of voles. These relationships are discussed in connection with reproduction. Lowest annual densities were found to occur in January when there is but little breeding activity and when rainfall is low and plant growth has ceased.

Marked deviation from the usual seasonal trends accompanied flood and drought. In the flood of July, 1951, although the study areas were not inundated, the ground was saturated to the extent that every footprint at once became a puddle. Immediately after the floods, on all three areas studied, populations were found to have been drastically reduced. The effect was most severe on the population of House Field, the lowest area studied, and the recovery of the population there was much slower than that of those on the other study areas (Fig. 5). Newborn voles were killed by the saturated condition of the ground in which they lay. The more precocious young of Sigmodon hispidus survived wetting better. They thus acquired an advantage in the competitive relationship between cotton rats and voles. These relationships are discussed more fully in the section on mammalian associates of Microtus.

Adverse effects of heavy rainfall on populations of small mammals have been reported by Blair (1939) and others. Goodpastor and Hoffmeister (1952:370) reported that inundation sharply reduced populations of M. ochrogaster for a year after flooding but that the area was then reoccupied by a large population of voles. Such a reoccupation may have begun on the areas of this study in the spring of 1952 when the upward trend of the population was abruptly reversed by drought. While cotton rats were abundant their competition may have been an important factor in depressing population levels of voles. The population of voles began to rise only after the population of cotton rats had decreased (Fig. 19).

In the unusually dry summer of 1952, there was a marked decline of population levels beginning in June and continuing to August when my field work was terminated. Dr. Fitch (1953, in litt.) informed me that the decline continued through the winter of 1952-53 and into the summer of 1953, until daily catches of Microtus on the Reservation were reduced to 2-10 per cent of the number caught on the same trap lines in the summer of 1951. The drought seemed to affect population levels by inhibiting reproduction, as described elsewhere in this report. A similar sensitivity to drought was reported by Wooster (1935:352) who found M. o. haydeni decreased more than any other species of small mammal after the great drought of the thirties.

No evidence of cycles in M. ochrogaster was observed in this investigation. All of the fluctuations noted were adequately explained as resulting from the direct effects of weather or from its indirect effect in determining the kinds and amounts of vegetation available as food and shelter.

The differences in densities supported by the various habitats were discussed earlier in connection with the analysis of habitats.