ABSTRACT MICHIGAN DEPARTMENT OF NATURAL RESOURCES Wildlife Report No. 3376 THE 2002 DEER PELLET GROUP SURVEYS by Harry R. Hill During the spring of 2002, a deer pellet group survey was conducted in the western Upper Peninsula (UP) and in DMU 452 (Core Area) of the northern Lower Peninsula (NLP) (Fig. 1). The units surveyed in the western UP were divided into the high, medium, and low snowfall units. The high snowfall unit consists of deer management units (DMUs) 007, 042, and 066. The medium snowfall unit is comprised of DMUs 036 and 052. The low snowfall unit consists of DMUs 022 and 055. The pellet group survey was not conducted in the eastern UP. During the spring of 2002, the deer pellet survey was conducted in the 562 square mile core area of the bovine tuberculosis surveillance area. The core area has been surveyed since 1996. INTRODUCTION Free-ranging wildlife populations are difficult to sample. The animals abilities to run and to hide not only make them difficult to capture, but also make them very difficult to observe. Therefore, wildlife biologists have determined that the best method for judging the density of white-tailed deer (Odocoileus virginianus) is to make counts of some sign that the deer leave. Deer droppings or pellet groups seem to be best suited for systematic appraisal. This technique is simply a formalized extension of methods used by experienced hunters to gauge the abundance of animals. RATIONALE The information about the density of deer pellet groups is primarily useful as an index to the abundance of deer. Bennett, English, and McCain (1940) originally used estimates of pellet group density to compare deer use on different areas. However, it may have seemed unsatisfactory to talk about pellet groups when they were actually interested in deer. Therefore, biologists defined a simple relationship between the number of pellet groups on an area and the number of deer necessary to produce those groups. This relationship was then used to mathematically convert pellet groups per acre into deer per square mile or deer per section. In order to estimate a deer population, we need to know (1) the rate that pellet groups are produced, (2) the number of pellet groups present, and (3) the period during which they were deposited. Studies with penned deer indicated that deer defecate on the average about 13.4 times in a 24-hour period. Thus, the total pellet groups on an area divided by 13.4 provides an estimate of the number of deer-days of use. Then if we can determine the period over which groups were deposited, we can estimate the number of deer present (McCain 1948). Printed by Authority of: P.A. 451 of 1994 Total Number of Copies Printed: 150 Cost per Copy:.....$.513 Total Cost:....$77.09 A contribution of Federal Aid in Wildlife Restoration, Michigan Project W-127-R. Michigan Department of Natural Resources IC 2578-59 (08/01/2002)
The period of time over which the groups have been deposited is defined to begin after leaf fall in the autumn and to end on the average date that the sample plots were searched. Usually, the leaves in any particular area are knocked from the trees during a relatively short period because of heavy winds, rain, snow, or a combination of these. Fallen leaves form a mat that hides groups dropped earlier. Thus, only those groups dropped after leaf fall are visible. Leaf fall is noted for each part of the state by the local wildlife biologists. However, in some areas the leaf cover is sparse, leaving it up to the biologists to separate the new pellet groups from the old. The actual relationship between deer density and pellet group density is much more complex. It is affected by many factors, including weather, diet, and composition of the deer herd. Since the exact form of this relationship is unknown, the new figure which is calculated is not an exact number of deer, but it is, at least, an improved population index (Overton 1969). However, experiments have shown that the simple relationship is a reasonable approximation of the true relationship between pellet group density and actual deer density (Eberhardt and VanEtten 1956). SAMPLE The northern deer range in Michigan encompasses an area of over 30,000 square miles. Since it is virtually impossible, and far too expensive, to search the entire area, a sampling plan must be formulated. The area is classified by the field biologists into three categories of deer abundance. Then the three categories (strata) are separately sampled. The number of samples allocated to each category depends on the area included in the category and the variability observed within the strata. The survey is primarily designed to produce estimates for areas greater than the size of two counties. This sampling design is called stratified sampling. Stratified sampling does not introduce any personal bias into the survey but does provide more precise estimates with less effort. Stratification is merely a method of assuring that the greatest effort is spent where it will do the most good. The entire deer range is divided into over 30,000 first stage sampling units or sections. Typically, sections are one square mile. Due to land survey corrections and lake shores, some sections may contain less than one square mile of land area. This survey design assumes that only sections that contain more than one-half square mile of land area are available for sampling. Therefore, the number of sections in a district may not agree exactly with the district s land area. The sections to be sampled are determined by random selection. Each section within a stratum within a district has the same chance of being selected. Second stage samples are areas located in each randomly selected section. These are a series of five 1/50-acre rectangular plots (12 x 72.6 ). The five plots together make up a course. The midpoint of each end of the plot is marked with a wooden stake. COMPUTATIONS The estimates are summarized in Table 1 and Figure 1. The estimated spring deer densities by stratum for each unit are found in Table 2. The survey estimates the average number of deer pellet groups to be found on any randomly selected course in a unit (Table 3). In Table 4, the average groups per course are changed to groups per section and then this estimate is converted to total deer. Dividing the average groups per course by five gives the average groups per plot (a course consists of five plots). This value is multiplied by 50 to calculate average groups per acre (the plots are 1/50-acre) and then multiplied by 640 acres per section to estimate the number of groups per section. This is converted to the number of deer by dividing the groups by the deposition rate to give deer-days and then by the number of days to give the average number of deer present for the period. These figures are termed unadjusted deer in unit. The 2
calculations are shown in a simplified formula at the bottom of the table. The unadjusted figure is an estimate of the average overwinter deer population. Fewer deer than the unadjusted estimate (Table 4) are actually present in the spring. Deer that die during the deposition period also contribute pellet groups. For example, four deer which each live for one month will contribute about the same number of pellet groups as two deer which each live for two months or one deer which lives for four months. To correct for this, we subtract the pellet groups deposited by deer that do not survive the entire pellet deposition period. Deer killed during the regular firearm deer season dropped pellet groups for about a month before being shot, and these groups were included in the total estimate of pellet groups. Their contributions must be deducted for the period they were present. This same process is used for other losses to calculate the spring herd estimate. Then the estimated number of deer lost during the deposition period, not their contributions, are added to the estimated spring herd to get the previous fall s herd estimate. Table 5 contains these calculations. This does not account for deer illegally killed and removed during the pellet deposition period. The illegal removal is unknown, but perhaps sizable. This means that the actual fall population is somewhat larger than the figure given, and the spring population is somewhat smaller. The estimates of legal deer kill are from annual mail surveys (Frawley 2001). In years when losses are expected to be high, dead deer surveys are undertaken to quantitatively estimate these overwinter losses. SOURCES AND CONTROL OF ERROR The number of deer is not constant from section to section across the state. Likewise, the number of pellet groups also varies from one course to another. The amount and direction of these variations are due to chance and are termed sampling errors. The deer pellet group survey, like all sampling techniques, is subject to many sources of error in addition to chance or sampling errors. These non-random errors or biases arise in counting and aging the pellet groups, in estimating deer defecation rates and leaf fall dates, and in sampling error. Ryel (1959), Eberhardt (1960), and VanEtten and Bennett (1965) discuss these problems in some detail. It appears that the actual determination of (1) the number of pellet groups present on a sample plot and (2) their relative age, is responsible for a large share of the variations in survey results. Errors in estimating the defecation rate and in estimating the leaf fall date does not appear to contribute major errors to the final estimate. Ideally, to reduce counting errors and misidentification of pellet age to a minimum, a few experienced individuals should search all of the courses each year. This is not possible because of the effort involved and the time limits imposed. Therefore, a recheck system is used. On the 1956, 1957, and 1958 pellet group surveys, all plots were originally searched by one person. As a recheck, 20 percent of the courses were randomly selected and searched a second time by an experienced biologist. Discrepancies between the original counts and the rechecks resulted in increasing the estimates of the total deer population by 30 percent in the northern Lower Peninsula in 1956, by 1 percent in this area in 1957, and by 16 percent in the Upper Peninsula in 1957 (Eberhardt 1957). A system of concurrent rechecks is used on the surveys. This involves making independent counts on all plots by members of two-person crews and then arriving at a composite count. Crew members start at opposite ends of the plot and count pellet groups on the half plot to their right. Metal disks are used to mark all groups found. Searchers then switch sides and check their partner s work. The biologists on the crews are responsible for classifying pellets into age categories and making final decisions on the number of groups present. Where a crew has only one biologist (the usual case), there is no real check on his/her identification of old and new groups. We have not found consistent characteristics to distinguish pellet groups dropped prior to leaf fall from those dropped after leaf fall. This means that we must use the relationship of the groups to fallen leaves and ground 3
vegetation whenever possible. Where this is not feasible, as in grasslands, oak stands, conifer swamps, etc., we must depend on the searcher s ability to make correct judgments on the age of questionable groups. Some notion of the magnitude of these errors can be obtained by comparing the ratios of old to new pellet groups between districts within the various strata. Unfortunately, such comparisons are complicated by changes in weather, changes in deer use, and changes in deer foods among the various districts and between years. Thus, we cannot be sure that any differences that we find are due to human error alone. STATISTICAL CALCULATIONS Table 6 contains the summaries of the statistical analyses by district, giving stratum averages ( x j ), district average ( x st ), and stratum standard deviation s j. Computations for each district were made as suggested by Cochran (1953) for stratified random sampling: x = w x w x w x st 1 1 + 2 + K + 2 n n where the wj ' s are the proportion of the total number of sections in each stratum and the x j ' s are the stratum averages. 2 2 2 ( ) = st 1 ( 1 ) + 2 ( 2 ) + + n ( n ) vx w vx w vx K w v x where the w s ' are as above and j vx ( j ) ' sare variances of stratum averages = s 2 j. n In a similar fashion, estimates of the averages and variances were made for the Upper Peninsula. Here the w ' s j ' become the proportion of the Upper Peninsula occupied by each district, the xj ' s are the district averages, and the vx ( )' stheir corresponding variances. With a systematic sub-sample, the model is equivalent to a simple random sample with one element per sample. Therefore, there is no estimate provided for a component of variation from plot to plot within each first stage sampling unit (section). Cochran indicates the variance estimate based on the first stage sampling unit, as derived from the second stage samples, is a valid approximation as long as n/n is small (less than.05 is suggested). Here n refers to the number of sections selected out of the total possible, N, in each stratum. Equal Rights for Natural Resources Users The Michigan Department of Natural Resources (MDNR) provides equal opportunities for employment and access to Michigan s natural resources. Both State and Federal laws prohibit discrimination on the basis of race, color, national origin, religion, disability, age, sex, height, weight or marital status under the Civil Rights Acts of 1964 as amended (MI PA 453 and MI PA 220, Title V of the Rehabilitation Act of 1973 as amended, and the Americans with Disabilities Act). If you believe that you have been discriminated against in any program, activity, or facility, or if you desire additional information, please write the MDNR, Human Resources, P.O. Box 30028, Lansing, MI 48909-7528, or the Michigan Department of Civil Rights, State of Michigan Plaza Building, 1200 6th Street, Detroit, MI 48226, or the Office for Diversity and Civil Rights, U.S. Fish and Wildlife Service, 4040 North Fairfax Drive, Arlington, VA 22203. For information or assistance on this publication, contact: MDNR, Wildlife Division, P.O. Box 30444, Lansing, MI 48909-7944, http://www.dnr.state.mi.us. This publication is available in alternative formats upon request. TTY: Michigan Relay Center 1-800-649-3777 4
LITERATURE CITED Bennett, L. J., P. F. English, and R. McCain. 1940. A study of deer populations by use of pellet-group counts. J. Wildl. Mgmt. 4(4):398-403. Cochran, W. G. 1953. Sampling techniques. John Wiley and Sons, Inc., New York. Eberhardt, L. L. 1957. The 1956 and 1957 pellet-group surveys. Game Div. Rept. No. 2133, Michigan Dept. Conserv., Lansing. 37pp.. 1960. Estimation of vital characteristics of Michigan deer herds. Game Div. Rept. No. 2282, Michigan Dept. Conserv., Lansing. 192pp.. R. C. VanEtten. 1956. Evaluation of the pellet-group count as a deer census method. J. Wildl. Mgmt. 20(1):70-74. Frawley, B. J. 2002. Michigan Deer Harvest Survey Report - 2001 Seasons. Wildlife Report No. 3344. Michigan Dept. Natural Resources, Lansing. 32pp. McCain, R. 1948. A method for measuring deer range use. Trans. N. Am. Wildl. Conf. 13:431-441. Overton, W. S. 1969. Estimating the numbers of animals in wildlife populations, pp. 403-455 in Giles, R. J. (ed.), Wildlife Management Techniques, Third Edition:Revised. Ryel, L.A. 1959. Deer pellet-group surveys on an area of known herd size. Game Div. Rept. No. 2252, Michigan Dept. Conserv., Lansing. 26pp. VanEtten, R. C., and C. L. Bennett, Jr. 1965. Some sources of error in using pellet group counts for censusing deer. J. Wildl. Mgmt. 29(4):723-29. 5
Figure 1 HOUGHTON KEWEENAW High Snowfall Unit 2002 Pellet Survey Units GOGEBIC ONTONAGON BARAGA MARQUETTE Medium Snowfall IRON Unit DI CK INSON Low Snowfall Unit MENOMINEE ALGER DE LTA SCHOOLCRAFT LUCE CHIP PEWA MACKI NAC EMMET Legend Boundary for Pellet Survey Districts/Units County Boundary State Boundary or Great Lakes Shoreline LEELANAU CHARLEVOIX ANTRIM GRAND BENZIE TRAVERSE KALKASKA CHE BOYGA N OTSEGO CRA WFORD MANISTEE WEXFORD MISSAUKEE ROSCOMM ON MASON LAKE OSCEOLA CLARE GLADWIN PRESQUE ISLE DMU 452 MO NTMO RENCY OSCODA OGEMAW Core Area ALPENA ALCONA IOSCO ARENAC HURON OCEANA MECOSTA ISABELLA BAY MIDLAND N 0 20 40 60 Miles NE W AYGO MONTCALM MUSKEGON OTTAWA GRATIOT KENT IONIA CLINTON TUSCOL A SANILAC SAGINAW GENESEE SHIAWASS EE LAPEER ST.CLAIR MACOMB 0 20 40 60 Kilometers ALLEGAN BARRY EATON INGHAM LIVINGSTON OAKLAND VAN BUREN KALAMA ZOO CA LHOUN JACKSON WASHTENAW WAYNE BERRIEN CA SS ST.JOSEPH BRANCH HILLSDALE LENAWEE MONROE DEER PER SECTION: FALL 2001 AND SPRING 2002 Fall Spring Unit 2001 2002 High Snowfall 14.56 10.20 Med. Snowfall 35.75 29.33 Low Snowfall 67.91 54.86 DMU 452* (Core Area) 36.33 23.66 6
Table 1 SUMMARY OF ESTIMATES Units Sections Unadjusted Deer Legal Kill [1] Other Losses [2] Deer Population Fall 2001[3] Estimates Spring 2002 High Snowfall 4,251 49,435 10,385 8,117 61,880 43,378 Medium Snowfall 3,690 115,411 16,307 7,386 131,915 108,222 Low Snowfall 2,579 149,125 26,190 7,456 175,129 141,483 DMU 452 (Core Area) 562 14,541 5,780 1,340 20,418 13,298 [1] Legal kill (all seasons including block permits and deer management assistance permits). [2] Does not include illegal kill completely removed from the field. [3] Fall population estimate prior to October 1, 2001. Figures include adjustments for early seasons, archery, regular firearm, muzzleloader, block permits, and deer management assistance permits. Due to rounding, the figures in this table may not sum exactly to the totals. Table 2 ESTIMATED SPRING POPULATION DENSITY BY STRATUM (Deer Per Square Mile) Stratum High Snowfall Medium Snowfall Low Snowfall DMU 452* I 45.82 33.46 60.31 26.03 II 12.08 33.82 48.47 29.48 III 8.51 25.26 49.95 12.13 Average 10.20 29.33 54.86 23.66 *562-square mile core area. 7
Table 3 STATISTICAL ANALYSIS OF DEER PELLET GROUPS PER COURSE Unit Stratum Sections Number of Samples Average Standard Deviation High Snowfall I 154 8 22.75 21.82 II 406 9 6.00 8.19 III 3,691 31 4.23 6.49 Unit Average 5.06 Standard Error 1.08 Medium Snowfall I 440 13 15.77 14.03 II 1,333 33 15.94 23.06 III 1,917 32 11.91 15.02 Unit Average 13.82 Standard Error 2.05 Low Snowfall I 1,385 40 24.35 19.76 II 1,144 28 19.57 22.58 III 50 6 20.17 25.35 Unit Average 22.15 Standard Error 2.54 DMU 452 (Core area) I 91 17 10.94 8.96 II 301 42 12.38 17.61 III 170 10 5.10 9.10 Unit Average 9.94 Standard Error 1.73 8
Table 4 UNADJUSTED POPULATION ESTIMATES[1] (Average Overwinter Population) Units Days From Leaf Fall Avg. Pellet Groups/Course Average Deer/Section Sections In Unit Unadjusted Deer in Unit High Snowfall 207 5.07 11.63 4,251 49,435 Medium Snowfall 210 13.82 31.28 3,690 115,411 Low Snowfall 182 22.15 57.82 2,579 149,125 DMU 452 (Core Area) 184 9.95 25.87 562 14,541 Upper Peninsula deer per section = (Avg. Pellet Groups/Course) x 50 x 640 (days from leaf fall) x 13.47 x 5 13.47 is average pellet groups deposited per deer day in the U.P. Northern Lower Peninsula deer per section = (Avg. Pellet Groups/Course) x 50 x 640 (days from leaf fall) x 13.37 x 5 13.37 is average pellet groups deposited per deer day in the N.L.P. 50x640 is a constant which converts the counts from per course to per section. 5 [1] Unadjusted means that deer dying during the pellet deposition period have not been taken into account. 9
Table 5 ADJUSTMENTS FOR DEER REMOVALS High Snowfall Unit Deposition period 207 days Unadjusted pellet group estimate (average overwinter population) [1]Legal hunting kill - Archery season--about 2,235 deer contributing for 22 days Firearm season--about 7,563 deer contributing for 31 days Muzzleloader season--about 527 deer contributing for 54 days [2]Fall and early winter losses-- about 1,623 deer contributing for 44 days [2]Late winter and spring losses-- about 6,494 deer contributing for 134 days 49,435-238 -1,133-137 -345-4,204 [1]Block and deer management assistance permit harvests are prorated into each season. [2]Does not include illegal kill completely removed from the field. 2002 Adjusted Spring Population 43,378 Hunting Removal for all Seasons 10,385 Other Losses 8,117 2001 Fall Population 61,880 10
Table 5 (Continued) ADJUSTMENTS FOR DEER REMOVALS Deposition period - 210 days Medium Snowfall Unit Unadjusted pellet group estimate (average overwinter population) [1]Legal hunting kill - Archery season--about 3,532 deer contributing for 30 days Firearm season--about 11,576 deer contributing for 39 days Muzzleloader season--about 1,121 deer contributing for 62 days [2]Fall and early winter losses-- about 1,846 deer contributing for 52 days [2]Late winter and spring losses-- about 5,540 deer contributing for 142 days 115,411-505 -2,150-331 -457-3,746 2002 Adjusted Spring Population 108,222 Hunting Removal for all Seasons 16,307 Other Losses 7,386 2001 Fall Population 131,915 Low Snowfall Unit Deposition period - 182 days Unadjusted pellet group estimate (average overwinter population) [1]Legal hunting kill - Archery season--about 4,579 deer contributing for 15 days Firearm season--about 18,853 deer contributing for 24 days Muzzleloader season--about 1,927 deer contributing for 47 days [2]Fall and early winter losses-- about 1,864 deer contributing for 37 days [2]Late winter and spring losses-- about 5,592 deer contributing for 127 days 149,125-377 -2,486-498 -379-3,902 [1]Block and deer management assistance permit harvests are prorated into each season. [2]Does not include illegal kill completely removed from the field. 2002 Adjusted Spring Population 141,483 Hunting Removal for all Seasons 26,190 Other Losses 7,456 2001 Fall Population 175,129 11
Table 5 (Continued) ADJUSTMENTS FOR DEER REMOVALS DMU 452 (Core Area) Deposition period 184 days Unadjusted pellet group estimate (average overwinter population) [1]Legal hunting kill - Archery season--about 1,030 deer contributing for 18 days Firearm season--about 4,021 deer contributing for 27 days Muzzleloader season about 147 deer contributing for 57 days Late season--about 150 deer contributing for 64 days [2]Fall and early winter losses-- about 1,005 deer contributing for 40 days [2]Late winter and spring losses-- about 335 deer contributing for 130 days 14,541-101 -590-46 -52-218 -237 2002 Adjusted Spring Population 13,298 [3]Hunting Removal for all Seasons 5,780 Other Losses 1,340 2001 Fall Population 20,418 [1]Block and deer management assistance permit harvests are prorated into each season. [2]Does not include illegal kill completely removed from the field. [3]Includes early season harvest of 432 deer. 12