A COMPARISON OF DIFFERENT METHODS OF CAPTURING AND ESTIMATING NUMBERS OF MICE MICHAEL H. SMITH ABSTRACT.-The number of mice of two species, Mus musculus and Perotrujscus polionotus, on an abandoned peanut field was determined by live-trapping using the mark-release technique and also by capturing the mice by digging out their burrows. The two sampling methods gave results that were significantly different. Available evidence suggests that the latter technique revealed the exact structure of the population. The mark-release method overestimated the male population of M. musculus as calculated by the Lincoln index but not by the Hayne method, and underestimated the female population of P. polionotus as calculated by either method. Males of both species were more readily trapped than were females, especially when the females were pregnant or lactating or both. These results are best interpreted in terms of the relative amount of movement shown by the different types of mice. The size of small mammal populations is often estimated by live trapping and the use of the Lincoln index (Lincoln, 1930) or the Hayne method (Hayne, 1949). These indirect methods assume that each animal in a population has an equal chance of being recorded in a given sample. This assumption is rarely tested and may be in error for certain species in which individuals differ in their response to the method of capture. While engaged in an overall study of the biology of the oldfield mouse, Peromsjscus polionotus, I tested this assumption and compared the two methods for estimating population size. In the autumn of 1964, an abandoned peanut field in central Florida, which had been planted but not harvested, was found to contain a dense population of mice. The burrows of the oldfield mouse are easily located because of the characteristic mound of dirt around the entrance and are also easily excavated because they are always found at relatively shallow depths in loose sandy soils (Smith and Criss, 1967). There was almost no ground cover in the field, and it seemed probable that all of the mice were living in underground burrows. For these reasons, it was thought that the number of mice in the field could be determined by digging out all of the burrows and capturing the inhabitants. The purpose of this study was to estimate the size of the population by a mark-release technique using two different mathematical formulae and then compare these estimates with the actual number of mice on the field. STUDY AREA AND METHODS The field, which was 3908 m sq, was located 4.8 km W Williston, Levy Co., Florida. The vegetation surrounding the field was typical of sandhills, with oak predominating. Traps were set in a grid with 5 m between each station. The grid included the field plus a 8-m border on all sides. There were two traps per station in the field and one per station in the border zone; 394 Sherman live traps were used. They were freshly baited with rolled oats and peanut butter in the late afternoon, checked in the early morning, and shut during the day. Trapping was carried out from 11 October to 25 October 1964. Traps
456 JOURNAL OF MAMMALOGY Vol. 49, No.3 TABLE I.-Summary of the number of mice captured in burrows, in traps or in both. 1964 1965 Mus musculus Peromyscus polionotus Peromuscus polionotus adult juvenile adult juvenile adult a ~ 0 ~ a ~ a ~ 0 ~ Trapped on night 1, 2, 3, or 4 13 10 2 2 9 2 0 0 9 4 Trapped on night 10, 11, 12, or 13 15 11 4 2 11 3 0 0 9 6 Previously marked 7 8 2 2 5 2 0 0 5 3 Estimated population size, Lincoln index 27.9** 13.9 4.0 2.0 19.8 3.0** 0 0 16.2 8 Estimated population size, Hayne method 18.2, 12.2. 3.7 1.9 13,.7 2.8** 0 0 11.5 6.5 Caught in burrows 16 20 18 19 13 16 10 6 14 17 Trapped after excavation of burrows 0 0 0 0 0 0 0 0 0 0 Caught in burrows but not trapped 0 7 14 17 0 13 10 6 1 10 Trapped but not caught in a burrow 5 0 0 0 2 0 0 0 1 0 Number on field at some time during study 21 20 18. 19 15 16 10 6 15 17 Number/hectare* 53.8 51.1 45.9 48.9 38.0 41.0 25.7 15.3 38.3 43.5 * Calculated from the number of mice captured by both methods. ** Percentages of this sex in the estimate differed significantly at.05 level from the percentages based on the numbers caught in the burrows or captured by both methods. were set on nights one, two, three, four, 10, 11, 12, 13 and 14. Each captured mouse was toe-clipped for later identification. Its sex, approximate age as indicated by pelage, and any other identifying marks were recorded. The mouse was then released and visually followed back to its burrow. All of the burrows were excavated on the 14th day. Traps were set again the following night to determine if any mice avoided capture during excavation; animals were sacrificed, and the presence of sperm in the epididymides of males and embryos in the uteri of the females were noted. The entire procedure was repeated once after the area was repopulated. The second trapping period started on 10 July 19650 and ended on 24 July 1965. Observations on the number and location of new burrows were made at monthly intervals during the time between the first and second trapping periods. RESULTS A summary of the captures is given in Table 1. Two species, P. polionotus and Mus musculus, were trapped in 1964, but only the former was found in 1965. There were no juvenile ~1. musculus in the 1964 sample. Data for juve-
August 1968 SMITH-ESTIMATING NUMBERS OF MICE 457 nile P. polionotus were excluded from the statistical analysis because most of these mice could not be considered part of the trappable population. Population size was estimated mathematically by the Lincoln index (Lincoln, 1930) and the Hayne method (Hayne, 1949). The formula for the Lincoln index is P == N M/R, where }';J is the total number of mice marked and released on the first four nights, R the number of mice that were initially marked and then recaptured on the last four nights, and N the total number captured during the latter period. For the Hayne method, P == ~tvx~/~tvxy, where tv is the number of mice captured each night, y is the proportion of the catch each night that had been previously marked, and x is the accumulative total of the number marked each night. Lincoln index values were higher for males and lower for females than the number of mice actually observed on the field (Table 1). The Hayne method underestimated the size of the population in all categories (Table 1). Estimates from the latter method were also lower for each species, year, and age subgroup than those calculated by the Lincoln index. Thirty-one adults were captured for the first time in a burrow, whereas eight were recorded only in the trapping data (Table 1). Six of the eight were captured only during the first four nights and had apparently disappeared from the field on the 14th day. Of the eight, six were captured once, one two times, and one three times. No mice were captured on the night after burrows had been excavated. The eight mice could have been transients or residents that were removed by predators during the study. However, short term changes in population number increase the error associated with population estimates as calculated by both mathematical methods. Percentages of each sex within each species for each year estimated by the Hayne method or Lincoln index were compared with the percentages from the number of mice caught in the burrows or by both collecting methods (Table 1). The following formula (Davis and Emlen, 1956) was used: P t-p2 "nl n~ -6/ PQ + PQ. Subscripts refer to the two samples being compared, and P and Q refer to the pooled percentages for each sex for both samples. Both the Lincoln index and Hayne method significantly underestimated the percentage of P. polionotus females in 1964. The percentages of male M. musculus in the Lincoln index estimate and in the mice caught in burrows or captured by both methods also were different. Most of the adults of both species were sexually active; 75% of the male NI. musculus and 81% of the male P. polionotus had sperm in their epididymides; 25% of the female M. musculus were pregnant but none was lactating, and 70% of the female P. polionotus were pregnant or lactating or both. Of the P. polionotus females, 91ro were in the group that was recorded only from cap-
458 JOURNAL OF MAMMALOGY Vol. 49, No.3 60 50 - ~ 40 - o a:: a:: ::> ro ~ 30 - a:: L&J CD ~ ::> z 20-10 - I ",d r I I I. o I I Nov. Dec. Jan. 1964 I I I! I! Feb. Mar. Apr. May Jun. Jul. 1965 FIG. I.-Gradual repopulation of the study area as indicated by the steady increase in the number of burrows (broken line). There was a significant positive linear correlation between the number of burrows (Y) and the time in days since the end of the first sampling period (solid line). tures made in burrows. All females that were both pregnant and lactating also were in this group. Sixty-four burrows were excavated during the first sampling period. The number of burrows on the field then gradually increased from none on 25 October 1964 to 54 on 24 July 1965. There was a significant positive correlation between the number of burrows (Y) and the number of days (X) since the end of the first sampling period (r ==.98, df == 8 and P <.01); the linear relationship was given by Y ==.19X -.8 (Fig. 1). Densities can be calculated in several different ways from the data in Table 1. The maximum value, which includes the number of juveniles and adults of both species that were on the field at some time during the first sampling period, was 278.7 mice per hectare. The densities of adult P. polionotus in 1964 (79.0 mice per hectare) and 1965 (81.8 mice per hectare) were essentially the same, but the difference between the overall densities, 173.8 mice per hectare in 1964 and 122.8 mice per hectare in 1965, approached significance (Chi square == 3.45, df == 1 and P ==.07). The latter was probably due to relatively high level of reproduction of P. polionotus during October 1964. The average number of embryos per female was 3.2 (range 3-4) during the first sampling period and 1.8 (range 1-2) during the second sampling period. Densities as calculated from the Lincoln index estimates were 153.9 adult
August 1968 SMITH-ESTI~1ATINGNUMBERS OF ~1ICE 459 mice per hectare in 1964 and 61.9 adult mice per hectare in 1965. These values for the Hayne method were 120.0 and 46.1 adult mice per hectare in 1964 and 1965, respectively. The estimates of the total size of the adult population made by the Lincoln index were closer to the actual number of mice captured by both methods than those made by the Hayne method. DISCUSSION Although the two methods of sampling gave results that were significantly different in several respects, seemingly the exact structure of the population was revealed by the technique of capturing mice by digging out their burrows. This knowledge allows determination of the biases inherent in the method of live trapping. Certain types of mice were less susceptible to trapping than others. For this reason, both the Lincoln index and the I-Iayne method gave an erroneous picture of the population. The probability that an animal will be caught in a given trap is the product of three separate" probabilities. The first depends upon the animal's chance encounter with the trap, the second on its response to the trap, and the third on the efficiency of the trap. The latter was essentially constant and cannot be used to explain the significant differences between the several types of mice. Variations in the other two probabilities merit further consideration. Pregnant or lactating female P. polionotus as well as juveniles of both sexes were not caught as readily as were other mice. Most of the juveniles were not old enough to have left their nest and thus were not exposed to traps. This is normally taken into account in similar studies by considering only the "population at risk of capture" (Leslie, 19'52). The pregnant and/or lactating females were exposed to traps less often, or they behaved differently than other adults with respect to the traps, or both. Knowledge concerning the natural history of this species can be useful in interpreting these data. An adult pair with or without young is the basic social unit in P. polionotus. Post-partum estrus is common, and the species breeds at all times of year. During certain critical periods, death of a female that is pregnant and lactating can result in an overall loss of up to 13 individuals from the population (Smith, 1966). It would be advantageous for the female to remain in the sealed burrow with her young if she could get sufficient food. This would reduce her exposure to predators and also to the traps. Lactating females of this and a closely related species do spend large amounts of time in their nests (King, 1963; Smith, 1966). I-Iowever, this would require that the males bring food to the females, or that food was stored in some convenient spot at an earlier time. Caches of stored food are frequently found in adjacent burrows (Smith, 1966). Since nothing is known about the possibility of males providing food for females or about the response(s) of females to traps, it is not possible to reach a definite conclusion. Many studies have shown that in small rodents males tend to move farther than do females (e.g., Blair, 1951; Howard, 1960; Davenport, 1964; Smith,
460 JOURNAL OF MAMMALOGY Vol. 49, No.3 1965, 1968). Greater movement increases the probability of encountering. a trap as long as the animal stays within the study area. Forthis reason, more males than females should be caught in traps, and the.estimates of the male population should be better than those of the female population. These generalizations are supported by the data, the only exceptions being the overestimate of the M. musculus male population by the Lincoln index and the close estimates of the M. musculus female population by both methods. Trapping resulted in an almost complete sample of the adult male population. Only one of the males captured in the burrows had not been previously caught in a trap. The major difference between the males of the two species as shown by trapping was the number that were trapped but not found in the burrows (24 percent for M, musculus and 10 percent for P. polionotus). The former shows a greater tendency to wander and to form temporary populations than does P. polionotus (Caldwell, 1964; Caldwell and Gentry, 1965). Some M. musculus males probably left the study area after they had been trapped. This decreased the probability of their encountering a trap to zero and explains why the Lincoln index overestimated this segment of the population. The accurate estimates of the population size of the M. musculus females probably were due also to the greater mobility of this type of mouse. None of these females was in the late stages of pregnancy or lactating, and they presumably would not have to spend a great deal of time in their nests as was the case for P. polionotus. The absence of house mice from the field during 1965, was unexpected. They were known to inhabit several nearby farm buildings and should have been able to move the short distance easily. Lack of extra or abandoned burrows during early phases of repopulation may have been the factor that stopped the house mice from establishing a new population. House mice usually live in burrows dug by P. polionotus when the two species are found together. Successful colonization of a sparsely covered field by small rodents undoubtedly depends upon the availability of suitable underground homesites, especially if the rodents do not readily dig their own burrows. Food was abundant during both sampling periods and was probably not important as a limiting factor. The simplest explanation of the differences in the trapping data seemingly depends upon the relative amount of movement shown by mice in the various groups. In this study an animal could not move far without encountering a trap because the distance from most burrows to the nearest trap was much less than 5 m. Since it is unlikely that any healthy adult mouse would occupy a burrow for several days without leaving, some mice must have encountered a trap(s) several times without entering it. Response of the animals to traps probably was the most important factor in determining the results of this experiment, but this does not mean that the probability of encountering a trap was unimportant. Multiple encounters with a trap(s) probably modify the
August 1968 S?\11TH-ESTI?\1ATING NUl\1BERS OF l\/iice 461 responses of animals; familiarity may increase the chance that an animal will enter a trap. These results are important because of the large number of studies in which conclusions have been based on data from trapping. Most investigations that require an exact knowledge of population structure in small rodents will suffer from relying too heavily upon capture data collected by live trapping. This is especially true for most species if only one sampling method is used. Greater movement by certain types of animals in the population will usually bias the results. Part of this problem can be alleviated by extending the sampling over a longer period of time. However, a species such as P. polionotus, with a short life span and large dispersal capacity (Smith, 1966, 1968), may experience considerable variation in population structure within a short time span; thus an average value could be quite misleading. Variations in density and extensive trapping probably also influence the population structure (Emlen et al., 1949). Clearly, conclusions made without taking these considerations into account may be unreliable. Litter size in P. polionotus was significantly different during the t\vo sampiing periods (3.2 in 1964 and 1.8 in 1965). Smith (1966) has previously shown that in north-central Florida litter size varies as a function of season, larger litters being born in the winter. ACKNOWLEDGMENTS Partial support was obtained from the American Museum of Natural History through the Theodore Roosevelt Memorial Award and an NIH predoctoral fellowship. The preparation of this paper was supported by AEC Grant AT(38-1)-310. I am grateful to R. J. Beyers, F. B. Golley, J. B. Gentry, and J. W. Gibbons for critically reading the manuscript. LITERATURE CITED BLAIR, W. F. 1951. Population structure, social behavior, and environmental relations in a natural population of the beach mouse (Peromyscus polionotus leucocepholus). Contrib. Lab. Vert. BioI., Univ. Michigan, 48: 1-47. CALDWELL, L. D. 1964. An investigation of competition in natural populations of mice. J. Mamm., 45: 12-30. CALDWELL, L. D., AND J. B. GENTRY. 1965. Interactions of Peronujscus and "A/us in a one-acre field enclosure. Ecology, 46: 189-192. DAVENPORT, L. B., JR. 1964. Structure of two Peronujscus polionotus populations in old-field ecosystems at the AEC Savannah River Plant. J. Mamm., 45: 95-113. DAVIS, D. E., AND J. T. EMLEN. 1956. Differential trapability of rats according to size and sex. J. Wildlife Mgt., 20: 326-327. E~ILEN, J. T., A. W. STOKES, AND D. E. DAVIS. 1949. Methods for estimating populations of brown rats in urban habitats. Ecology, 30: 430-442. HAYNE, D. W. 1949. Two methods for estimating population fro III trapping records. J. Mamm., 30: 399-411. HOWARD, W. E. 1960. Innate and environlnental dispersal of individual vertebrates. Amer, Midland Nat., 63: 152-161. KING, J. A. 1963. Maternal behavior in Peronujscus. Pp. 58-93, in Maternal behavior in mammals (H. L. Rheingold, ed.), John Wiley and Sons, Inc., New York, 349 pp. LESLIE, P. H. 1952. The estimation of population parameters from data obtained by
462 JOURNAL OF MAMMALOGY Vol. 49, No.3 means of the capture-recapture method. II. The estimation of total numbers. Biometrika, 39: 363-388. LINCOLN, F. C. 1930. Calculating waterfowl abundance on the basis of banding returns. Circ. U. S. Dept. Agric., 118: 1-4. Sl\fITH, M. H. 1965. Dispersal capacity of the dusky-footed woodrat, Neotoma fuscipes. Amer. Midland Nat., 74: 457-463. 1966. The evolutionary significance of certain behavioral, physiological, and morphological adaptations of the old-field mouse, Peromuscus polionotus. Ph.D. dissertation, Univ. Florida, 187 pp. 1968. Dispersal of the old-field mouse, Peromuscus polionotus. Bull. Georgia Acad. Sci., in press. Sl\UTH, M. H., AND W. C. CRISS. 1967. Effects of social behavior, sex and ambient temperature on the endogenous diel body temperature cycle of the old-field mouse, Peromuscus polionotus. Physiol, Zool., 40: 31-39. Department of Zoology, University of Georgia, Athens (mailing address: Savannah River Ecology Laboratory, Aiken, South Carolina 29801). Accepted 24 April 1968.