While general observation has indicated that annual fluctuations in population density of rodents in southern United States are less “cyclic,” i.e., peaks less regular, than is the case in the classic northern four-year cycles, few long-term observations have been made on populations in stable habitats. The possibility of a north-south difference in the nature of population oscillations has an important bearing on current theories of “cycles.” For the past eleven years trapping has been carried out according to a consistent plan in an old field where habitat features have been relatively constant. The field is located near Athens, Clarke County, Georgia in the Piedmont section of the state. The cotton rat, Sigmodon hispidus, comprised the great bulk of the readily trappable small mammal population in the old field, which is in a stage of vegetative succession favorable to the species. (Cotton rats occurring in Georgia except for the coastal strip, below 100 foot contour line, have been assigned to the subspecies Sigmodon hispidus komareki by Gardner, 1948.) Results of this study are presented simply as evidence of changes in density, natality, age distribution and other population characteristics that apparently have occurred on a limited area. No attempt will be made at this time to interpret these data in terms of various theories that have been advanced to explain population cycles.

The writer is indebted to the following students and former students who helped with trap setting and specimen preparation at various times during the period: Drs. Robert A. Norris and David W. Johnston, James C. Major, H. C. Robert, James O. Harrison, Edward J. Kuenzler, Jack Lowe, and J. B. Gentry.

The study area.—The old field selected for this study (PL I) is one of a series of abandoned fields lying on a slope a quarter of a mile east of the North Oconee River at a point opposite the Dairy Farm of the University of Georgia, which lies on the west side of the river. The fields are typical of abandoned cotton lands of the Piedmont region. At the start of the study the field had been abandoned about eight years and was in the broomsedge-shrub (Andropogon-Rubus) stage of succession. Both gully and sheet erosion are evident throughout the area. Terraces constructed along contours during previous cultivation remain prominent and are an important habitat feature. The particular field was selected because both previous erosion and the absence of a nearby source of pine seeds would operate to keep the field in virtually the same stage of succession for many years. Except for continued growth of small trees and shrubs already present, especially on the old terraces, and the invasion of a few new pine seedlings, the area has remained remarkably stable. The last peak in cotton rat abundance in 1952–3 was about the same order of magnitude as the first peak observed in 1946 (Fig. 1), which would indicate that the carrying capacity of the area for cotton rats has indeed remained about the same over the eleven year period.

Plate 1

The study area as it appeared in the early spring of 1952. The white flag marks the beginning of trapline no. 1, which continues over the hill in the background. Line no. 2 is to the right just out of the picture. The first row of trees and shrubs (in the foreground) marks the first low terrace (see Fig. 3). A second and larger terrace is mostly out of view over the hill, but a portion of it may be seen in the left background. The light-colored stems in the foreground are broomsedge (Andropogon) and the dark stems are canes of blackberries (Rubus).

Plate 1

The study area as it appeared in the early spring of 1952. The white flag marks the beginning of trapline no. 1, which continues over the hill in the background. Line no. 2 is to the right just out of the picture. The first row of trees and shrubs (in the foreground) marks the first low terrace (see Fig. 3). A second and larger terrace is mostly out of view over the hill, but a portion of it may be seen in the left background. The light-colored stems in the foreground are broomsedge (Andropogon) and the dark stems are canes of blackberries (Rubus).

Fig. 1

#x2014;Autumn (November) densities of Sigmodon in number of individuals caught in the double trapline (120 traps in two lines of 20 stations each) between 1948 and 1954 and in equivalent quadrats 1944–1947 (see Table 1).

Fig. 1

#x2014;Autumn (November) densities of Sigmodon in number of individuals caught in the double trapline (120 traps in two lines of 20 stations each) between 1948 and 1954 and in equivalent quadrats 1944–1947 (see Table 1).

Table 1

Numbers of small mammals taken per double trapline in the same field in fall (F) and spring (S) of 1944 through 1954

Species 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 Total 
Sigmodon hispidus20 10 56 15 59 29 17 51 42 333 
Reithrodontomys humulis…              
Peromyscus polionotus…                 
Cryptotis parva…             10 
Microtus pennsylvanicus…                 
Mus musculus…              
Peromyscus nuttalli…                
Peromyscus leucopus…                 
Species 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 Total 
Sigmodon hispidus20 10 56 15 59 29 17 51 42 333 
Reithrodontomys humulis…              
Peromyscus polionotus…                 
Cryptotis parva…             10 
Microtus pennsylvanicus…                 
Mus musculus…              
Peromyscus nuttalli…                
Peromyscus leucopus…                 

Broomsedge, Andropogon virginicus, forms an almost continuous cover over the area (PL I) except for a few very badly eroded spots which have remained bare. Old field forbs such as species of Gnaphalium, Solidago, Lespedeza, and Aster are secondary seasonal dominants. Scattered throughout the grass are numerous small and a few large clumps of blackberry (Rubus) and greenbrier (Smilax). There are a few pine seedlings (Pinus taeda and P. echinata) here and there as well as a few small deciduous trees and shrubs, chiefly persimmon (Diospyros virginiana), sumac (Rhus glabra), water oak (Quercus niger), and plum (Prunus). Cover is more dense on the low terraces where the growth of grasses, blackberries, trumpet vines, shrubs and small trees is thicker (PL I).

The study area is surrounded by similar fields a few of which have been cultivated more recently, or are mowed for hay. Narrow hedgerows or thickets separate the study area from other fields. Thus, the study area is in contact with other fields where cotton rats are numerous.

Methods.—During the first four years (1944–47) the population was sampled each fall by means of a half-acre quadrat following the method of Bole (1939). When Dr. John B. Calhoun organized the North American Small Mammal Census in 1948 we switched to the use of a double trapline and sampled the population both spring and fall of each year in order that our data might be comparable with that obtained by other workers following Calhoun's plan (Calhoun, 1948). Spring trapping was done during the first two weeks in May and fall trapping between October 20 and November 20. Traplines were run at right angles to terraces as shown in Fig 3. Each line consisted of 20 stations 25 feet apart with three traps (two regular commercial rat traps and one mouse or museum special trap) at each station. The lines were placed 200 feet apart as originally standardized by Calhoun (1948). The two lines contained 40 stations with a total of 120 traps, which were operated for three successive days (totaling 360 trap nights). On one or two occasions heavy rains sprung traps and the line was run for an extra day, in order to obtain three effective 24 hour periods of trapping. During the first four years only number, sex and approximate age of specimens were recorded, but beginning in 1948 all animals were weighed, measured and examined for reproductive condition; some were kept as study skins. The same bait was used throughout, i.e., a half of a large peanut tightly impaled on the trigger release. It was possible to compare the population level on the study field with the general level in the region for some of the years when some other trapping was done in the Athens vicinity. Also, in 1949, a marking-recapture study was carried out on another, similar field. Animals were marked in February, May, August, and November; data on home range and longevity were useful in the interpretation of trapline results. In 1948 Robert Norris and David Johnston set out half-acre quadrats and double traplines in the same area simultaneously. They found that the double trapline caught approximately twice as many animals as the quadrat. On this basis the quadrat data of 1944–47 were converted into trapline density.

Fig. 2

Spring (May) and autumn (November) densities in number of individuals caught in the double trapline (see Table 2).

Fig. 2

Spring (May) and autumn (November) densities in number of individuals caught in the double trapline (see Table 2).

Table 2

Numbers, biomass and age distribution in population samples of Sigmodon, 1948–1954

 Number Total Average wt., gms. Estimated population biomass, kgms. per acre Per cent adult 
♂ ♀ 
1948: Spring 36.9 0.22 17 
Fall 28 31 59 57.7 3.40 16 
1949: Spring 66.8 0.53 50 
Fall 10 19 29 76.4 2.21 60 
1950: Spring — — — 
Fall 17 42.0 0.71 19 
1951 : Spring — — — 
Fall 64.3 0.51 50 
1952: Spring — — — 
Fall 33 18 51 50.4 2.57 24 
1953: Spring 36.1 0.29 25 
Fall 24 20 44 63.2 2.72 30 
1954: Spring 103.9 0.21 — 
Fall 79.5 0.16 — 
Totals 116 118 234 — — — 
 Number Total Average wt., gms. Estimated population biomass, kgms. per acre Per cent adult 
♂ ♀ 
1948: Spring 36.9 0.22 17 
Fall 28 31 59 57.7 3.40 16 
1949: Spring 66.8 0.53 50 
Fall 10 19 29 76.4 2.21 60 
1950: Spring — — — 
Fall 17 42.0 0.71 19 
1951 : Spring — — — 
Fall 64.3 0.51 50 
1952: Spring — — — 
Fall 33 18 51 50.4 2.57 24 
1953: Spring 36.1 0.29 25 
Fall 24 20 44 63.2 2.72 30 
1954: Spring 103.9 0.21 — 
Fall 79.5 0.16 — 
Totals 116 118 234 — — — 

Because our studies on home range size in Sigmodon are incomplete at this time, trapline animal numbers can not be converted into density in terms of number per acre with any degree of confidence. However, it appears from home range data at hand that for Sigmodon the double trapline samples approximately one acre, as is also indicated by the quadrat-trapline comparisons mentioned above.

Results

Numerical density.—The total catch is shown in Table 1, and the annual fluctuations are diagramed in Figs 1 and 2. Seven species of small mammals other than cotton rats were taken, but only in small numbers and at irregular intervals. The habitat was somewhat far advanced, succession-wise, to be favorable for Peromyscus polionotus, Cryptotis, and Reithrodontomys while it was too “young” for P. nuttalli and P. leucopus. Microtus is rare in the region and was taken in the study area only in 1946. As shown in Fig 2 there seemed to be a tendency, perhaps only coincidental, for these “marginal” species to invade the field during periods when Sigmodon was less common.

Fig. 3

Total number of Sigmodon caught at each station on the double line between 1948 and 1954 in relation to major habitat features.

Fig. 3

Total number of Sigmodon caught at each station on the double line between 1948 and 1954 in relation to major habitat features.

As shown in Fig 1, autumn densities of Sigmodon oscillated rather sharply between high levels of 50 or more per trapline to less than 20, with a high density occurring in four of the eleven years. There were three major peaks if we consider 1952 and 1953 as one peak. The only year with a fall catch of intermediate size was 1949, and this could well be considered part of the second peak. Thus, it would appear that there have been three well marked “cycles” with major peaks in 1946, 1948, and 1952. The interval between peaks was two and four years. The peaks that occurred on the study area did not always synchronize with those on other local areas, but as nearly as could be determined from somewhat sporadic random trapping in Sigmodon habitat, peaks for the Athens region as a whole coincided within a year or so with peaks on the study area.

Spring and fall density since 1948 is compared in Fig 2. In each year, except for 1954, fall density was higher than spring density. While a low density was to be expected in 1954 following two years of abundance, the exceedingly dry summer and fall of 1954 may have depressed the population so that the usual recovery from the spring low did not occur, at least not by the time of the November trapping period. A seasonal cycle with a low point in the spring and a high point in the fall appears to be the usual rule in northern areas since the major part of the breeding season occurs between these two periods. However, we are finding that the seasonal cycle in Peromyscus polionotus on the Coastal Plain is often reversed, with the greatest abundance coming in the late winter or spring and the least abundance in late summer or fall. Pournelle (1952) and McCarley (1954) have reported the same thing for Peromyscus gossypinus in Florida and Texas. At present this reversed rhythm is interpreted as due to the depressing effects of hot weather in midsummer, which results in the highest natality occurring in the more favorable periods of late winter and early spring. Since cotton rats breed equally well throughout the year when kept in captivity under constant temperature (Meyer and Meyer, 1944) it would appear that photoperiodicity is not a major regulating factor as is true for some mammal species. Consequently, temperature is undoubtedly important. It is apparent from Fig 2 that, in most years, Sigmodon in northern Georgia may be expected to exhibit a “northern type” seasonal cycle of abundance. Since Komarek (1937) has reported that cotton rats in southern Georgia also reach a peak in the fall, it may be that reproduction in this well adapted southern species is not curtailed by high temperatures in summer, as is the case with P. gossypinus, but instead reproduction is more likely to be inhibited by low temperatures in winter.

In Fig 3 the total number of cotton rats caught at each of the twenty stations on the two lines is shown in relation to the bushy terraces and the more grassy inter-terrace areas. As might be expected, the catch was greatest in the vicinity of terraces and thickets and least in eroded areas. While animals were often unevenly distributed along the line in a particular season, the total catch over the period of years was remarkably uniform from station to station (except as influenced by obvious habitat features), indicating that a consistent pattern of aggregation or “clumping” of individuals did not persist from one year to the next.

Weights and biomass.—As shown in Table 2 the average weights of individuals varied due to differences in the age structure of the population as discussed below. In general, individual size tended to increase when numbers were low and to decrease when numbers were high. This was especially true when peak years are compared with the year following. Thus, the average weights increased from 57.7 gms. in the fall of 1948 to 76.4 gms. in fall of 1949, and from 50.4 gms. in 1952 to 63.2 gms. in 1953. When the population density is expressed in terms of total weight or biomass, as shown in column 5, Table 2, fluctuations from year to year were less marked than when numbers alone are considered.

Natality.—The relationship between weight and reproductive activity in all females trapped since 1948 is shown in Table 3. No females weighing less than 60 gms. were found with embryos, nor were placental scars or other evidence of breeding discovered. This was also true of all other specimens taken in the region for which we have records. Meyer and Meyer (1944) found that in captivity Sigmodon began breeding between 40 and 50 days of age when females weighed 62–87 grams. As shown in Table 3 the per cent pregnancy and the average number of embryos in our field populations were much less in females between 60 and 80 gms. than in larger animals. We may conclude that females in the field, as well as in the laboratory, may breed when about 60 gms. in weight but are not fully mature until they reach about 80 gms. in body weight. No examinations of testes were made but since Meyer and Meyer (1944) found that males begin breeding at the same age as females, but at a slightly larger body weight, it is likely that few males in the field would begin breeding before attaining about 65–70 gram size.

Table 3

Frequency distribution of pregnant females by weight classes

Weight, gms. Number females Number with embryos Per cent pregnant Average no. embryos 
0–20 11 
20–40 26 
40–60 17 
60–80 23 17 4.5 
80–100 71 5.4* 
100–120 15 10 67 4.9* 
120–140 88 5.6* 
Weight, gms. Number females Number with embryos Per cent pregnant Average no. embryos 
0–20 11 
20–40 26 
40–60 17 
60–80 23 17 4.5 
80–100 71 5.4* 
100–120 15 10 67 4.9* 
120–140 88 5.6* 
*

Average for females between 80 and 140 grams is 5.2.

Breeding occurred during both the May and the November trapping periods, but the pregnancy rate was less in the fall. This would indicate, as do other records, that breeding tapers off in November, and hence late November or early-December would generally be the correct period to sample the seasonal high. Of seven mature females taken in spring, six were pregnant (86 %); while of 46 mature females taken in fall, 20 were pregnant (43%).

The number of embryos varied from 3 to 8 with the average in 30 fully adult females (80 gms. or more) being 5.2. It is interesting to compare this with Meyer and Meyer's laboratory colony where 44 litters averaged 5.6 with extremes being 2–10. Svihla (1929) reported an average of 4.75 young per litter in a small sample of Louisiana cotton rats. More interesting than averages, however, is the comparison of litter size in low and high populations years as shown in Table 4. The number of embryos was greater in years of high population density. These averages are not unduly weighted by the four young females (Table 3), since two occurred in the “low” year series and two in the “high.” The difference (1.6 ± 0.49) is significant at the 1 per cent level.

Table 4

Comparison of natality in low and high population years

 Mature females, number Pregnant females Embryos 
Number Per cent Av. no. Extremes 
Four low years (1949–50–51–54) 24 10 42 4.1 (3–6) 
Three high years (1948–52–53) 29 16 53 5.7* (3–8) 
 Mature females, number Pregnant females Embryos 
Number Per cent Av. no. Extremes 
Four low years (1949–50–51–54) 24 10 42 4.1 (3–6) 
Three high years (1948–52–53) 29 16 53 5.7* (3–8) 
*

The difference, 1.6 ± 0.49, is significant at 1% level (t = 3.26).

Age distribution.—Comparison of age distribution of populations during low and high density years is shown by means of age pyramids in Fig 4. As has already been indicated the pattern of age distribution is strikingly different with three-fourths of the individuals being non-breeding juveniles (less than 60 gms.) during high density years as compared with only 55 per cent during low years. During the live-trapping study in 1949 when cotton rats were marked in February, May and June, August, and November very few individuals survived from one trapping period to another, and only one individual was recaptured in three successive periods. The record of this individual, a female, is suggestive of the growth rate in the wild. When captured on June 6 it weighed 15.0 gms. at 7 days of age; on August 21, 87.4 gms. at 76 days; and on November 12, 96.1 gms. at 159 days. Another female weighed 74.5 gms. on August 21 and 103.0 gms. on November 9, two and one-half months later. Since cotton rats weigh 6–8 gms. at birth (Meyer and Meyer, 1944) and increase in weight at the rate of a gram or better a day for the first few weeks, the 15 gm. female was probably about 7 days old. The smallest individuals caught on the trapline weighed between 10 and 12 gms. suggesting that young may leave the nest as early as the fourth day. The largest male was 280 mm. long and weighed 177.6 gms., while the largest female measured 274 mm. in total length and weighed 136 gms. (without embryos). Only a few individuals of either sex exceeded 125 gms. in weight.

Fig. 4

Age pyramids for high density years (1948–52–53) and low density years (1949–50–51–54). Juveniles (non-breeding) represent animals weighing between 10 and 60 gms., young adults, 60–110 gms., and old adults, 110 gms. and over.

Fig. 4

Age pyramids for high density years (1948–52–53) and low density years (1949–50–51–54). Juveniles (non-breeding) represent animals weighing between 10 and 60 gms., young adults, 60–110 gms., and old adults, 110 gms. and over.

To summarize, cotton rats in the field appear on the runways when a week or less old (weight 10–20 gms., total length 120–130 mm.), and begin to breed when two months or less in age (weight 60–80 gms., total length about 200 mm.). Females over 100 gms. and males over 110 gms. are probably five months or older. Since few individuals may be expected to live more than six months, the population turnover may often be complete every six months. With a species such as the cotton rat which begins to breed at an early age and rarely lives long enough to reach maximum size it is clear that average measurements of “adults” will vary according to the stage in the population oscillation (see also Komarek, 1937). Therefore, size should not be used as a criterion for the establishment of geographic races or subspecies unless the sample is very large and the stage in the population cycle from which it was taken is known.

Summary

Autumn densities (Nov.) of Sigmodon hispidus (Fig 1) oscillated between 50 or more and 20 or less per double trapline during eleven consecutive years of sampling in an old field (Pl. I) that remained relatively stable in major habitat features. There were three well-marked “cycles” with major peaks in 1946, 1948, and 1952. Seven other species were taken but in small numbers (Table 1).

In six of the seven years in which spring (May) samples were also taken, the fall density was markedly greater than the spring density (Fig 2) suggesting a “northern type” seasonal cycle for Sigmodon in the Piedmont of Georgia as contrasted with a reversed seasonal rhythm (high in late winter and spring) that has been reported for Peromyscus in the Coastal Plain.

Except for the influence of obvious habitat features, such as old terraces, the total catch for the seven-year period was remarkably uniform from station to station in the 20-station lines indicating that there was no consistent pattern of aggregation that persisted from season to season and year to year (Fig 3).

In general, average age and weight of individuals tended to increase when the numbers were low and decrease when the numbers were high. Therefore, total biomass fluctuated from year to year to a lesser extent than total numbers (Table 2).

A small percentage of females between 60 and 80 gms. in weight contained embryos (average 4.5 per female); a much larger percentage of females over 80 gms. were pregnant with an average of 5.2 embryos (Table 3). The number of embryos was significantly greater in years of high population density (5.7) than in years of low density (4.1) (Table 4). A larger percentage of mature females were pregnant in the spring (May) than in late fall (November).

The age distribution in the population was quite different in low and high density years (Fig 4). In high years, 76 per cent of individuals were juvenile (10 to 60 gms.), 18 per cent young adults (60 to 110 gms.) and 6 per cent old adults (above 110 gms.). In low years the percentages were: 55 per cent, 30 per cent and 15 per cent.

On the basis of recapture data and Meyer and Meyer's (1944) study of captive animals it is estimated that cotton rats leave the nest at four to seven days of age (weight 10–20 gms.), begin to breed at two months or less (60–80 gms.), and reach 100 gms. (females) to 110 gms. (males) at five months of age. Very few individuals on a 1949 study plot survived longer than six months and few of the total catch (333 individuals) exceeded 130 gms. in weight, even though this is well below the maximum size recorded (177 gms.).

In a species such as Sigmodon hispidus that begins breeding at an early age and rarely lives long enough in the field to reach maximum size, it is clear that average weights and measurements of “adults” will vary according to the stage in population oscillation. Therefore, size should not be used in subspecific determination unless the sample is very large and the stage in the seasonal and annual population cycle known.

Literature Cited

Bole
B. P.
Jr.
1939
.
The quadrat method of studying small mammal populations
.
Sci. Pub. Cleveland Mus. Nat. Hist.
 ,
5
:
15
77
.
Calhoun
J. B.
1948
.
North American census of small mammals. Release No. 1. Announcement of program
 .
Rodent Ecology Project, Johns Hopkins University
.
Mimeo
.
8
p.
Gardner
M. C.
1948
.
An undescribed eastern cotton rat
.
Proc. Biol. Soc. Wash.
 ,
61
:
97
98
.
Komarek
E. V.
1937
.
Mammal relationships to upland game and other wildlife
.
Trans. 2nd N. A. Wildlife Conf.
 ,
561
569
.
Meyer
B. J.
Meyer
R. K.
.
1944
.
Growth and reproduction of the cotton rat, Sigmodon hispidus hispidus, under laboratory conditions
.
Jour. Mamm.
 ,
25
:
107
129
.
McCarley
W. H.
1954
.
Fluctuations and structure of Peromyscus gossypinus populations in eastern Texas
.
Jour. Mamm.
 ,
35
:
526
532
.
Pournelle
G. H.
1952
.
Reproduction and early post-natal development of the cotton mouse, Peromyscus gossypinus gossypinus
.
Jour. Mamm.
 ,
33
:
1
20
.
Svihla
A.
1929
.
Life history notes on Sigmodon hispidus
.
Jour. Mamm.
 ,
10
:
352
353
.