---

---

Abstract: Radio telemetry has been an important research tool in waterfowl studies for >>20 years, yet little effort has been made to evaluate potential effects of transmitters on the birds that carry them. As part of a 4-year mallard (Anas platyrhynchos) study in the prairie pothole region of North Dakota and Minnesota, we compared radio-marked and unmarked female mallards in terms of percent time observed feeding, resting, and preening; nest initiation date; and clutch size and egg volume. Radio-marked females carried a 23-g back-mounted transmitter attached with a 2-loop harness (Dwyer 1972). On average, radio-marked females tended to feed less, rest and preen more, initiate nests later, and lay smaller clutches and eggs than unmarked females. Thus, behavioral and reproductive data from ducks marked with back-mounted harness-attached transmitters may be biased. We recommend that new designs of radio packages be field tested and caution that effects may be masked under extreme environmental conditions.

---

Table of Contents

• Introduction
• Study Areas and Methods
• Results
• Discussion
• Research Implications
• Literature Cited

---

Introduction

Radio telemetry has been used to study waterfowl ecology for >20 years (e.g., Schladweiler and Tester 1972, Ball et al. 1975, Gilmer et al. 1977, Kirby and Cowardin 1986, Longcore et al. 1991). Back-mounted radio transmitters attached with Dwyer (1972) harnesses have been used extensively to study movements, habitat use, survival, and reproduction of ducks (e.g., Derrickson 1978, Talent et al. 1982, Ringelman and Longer 1983, Frazer et al. 1990, Miller et al. 1992).

Relatively little effort has been made to determine if radio packages affect ducks in ways that bias the data collected. Prior to our work, 2 studies compared behavior of female dabbling ducks wearing back-mounted transmitters with Dwyer (1972) harnesses (hereafter "harness transmitters") to those without transmitters. Greenwood and Sargeant (1973) showed that harness transmitters affected body mass and swimming and preening behavior of captive female mallards and blue-winged teal (Anas discors) during winter. Chabaylo (1990) showed that harness transmitters had similar effects on captive female mallards during the first 2 weeks of brood rearing. Apparently, no one had studied long-term effects of harness transmitters on free-ranging female dabbling ducks during the nesting season.

In 1988-91, biologists from Northern Prairie Wildlife Research Center (NPWRC) studied the breeding ecology of mallards in the prairie pothole region of North Dakota and Minnesota. As part of this study, we collected information on behavior and reproduction of wild free-flying mallards with and without harness transmitters to see if these transmitters affected time spent feeding, resting, and preening; nest initiation date; and clutch size and egg volume.

We gratefully acknowledge D. A. Buhl for collecting and processing behavioral data and preparing figures; A. B. Sargeant for providing unpublished data on nesting dates of mallards; R. O. Woodward for leading nest-searching crews; D. J. Rova for entering data in computer files; D. A. Brandt for preparing nest data files for analysis; and W. E. Newton, R. C. Khan-Malek, D. J. Twedt, and T. L. Shaffer for providing statistical advice and analyses. In addition, we thank the host of dedicated seasonal technicians who collected behavioral data, conducted nest searches, and trapped, radio tagged, and monitored mallards. D. L. Larson, W. E. Newton, K. M. Kraft, K. J. Reineeke, and an anonymous reviewer provided helpful comments on the manuscript. We followed the American Ornithologists' Union (1988) guidelines for use of wild birds in research, and our procedures were approved by the NPWRC Animal Care and Use Committee.

Study Areas and Methods

During 1988-91, we obtained data from 9 50-km² study sites in the prairie pothole region of southeastern North Dakota and western Minnesota. Three study sites were located near Kulm, North Dakota; 3 near Jamestown, North Dakota; and 3 near Detroit Lakes, Minnesota. Behavioral data were collected on 2 study sites, 1 near Kulm and 1 near Detroit Lakes. Data on nest initiation date, clutch size, and egg volume were collected on all 9 study sites. In 1989, additional data on nest initiation dates were obtained from 10 more study sites: 6 Waterfowl Production Areas (WPA's) near Detroit Lakes and 4 WPA's near Jamestown. Data collection in North Dakota was reduced in 1990 and further curtailed in 1991 because of continuing drought conditions. In 1991, no behavioral data were collected in North Dakota.

Mallards were captured in April each year using decoy traps (Sharp and Lokemoen 1987) in wetlands occupied by breeding pairs. We attached a back-mounted radio transmitter to each captured female using an adjustable 2-loop harness similar to that described by Dwyer (1972). Radio packages (i.e., transmitter, antenna, and harness) averaged 23 g, roughly 2% of body mass of females at the time of capture. We also attached a U.S. Fish and Wildlife Service leg band to each female and, on the 2 study sites used for behavioral observations, attached nylon nasal markers in unique color and shape combinations (Lokemoen and Sharp 1985).

We observed radio-marked and unmarked female mallards through 20-80 x spotting scopes and l0x binoculars. Instantaneous sampling procedures were used to estimate time budgets (Altmann 1974); behaviors were recorded at 30-second intervals indicated by the sound of a metronome (Wiens et al. 1970). In most cases, only 1 female could be observed at a time, and observation periods on focal individuals lasted up to 4 hours. We collected data during all day-light hours, several days a week. Time observations were not made on radio-marked birds during the first week after capture to minimize disturbance while they recovered from stresses of trapping and handling.

To compare behaviors of radio-marked and unmarked mallards, we used data collected on paired females from late April through June. Behaviors were grouped into 4 categories: feeding, resting, preening, and other; only the first 3 categories were used in analyses. For radio-marked females, the mean percent time observed in each behavioral category was computed by calculating the proportion of all 30-second sampling points assigned to that behavior for each female (combining observation periods within females), weighting each proportion by the square root of the total number of sampling points for that female, and then averaging across females. Weighting by the square root was a compromise between giving each female equal weight and weighting by the number of observations per female. For unmarked females, the number of individuals observed and the distribution of observations among them were unknown. Thus, the percent time observed in each category could not be weighted and averaged, and standard errors could not be calculated. We estimated the percent time unmarked females spent feeding, resting, and preening by dividing the number of 30-second sampling points assigned to each behavioral category by the total number of 30-second sampling points in all their observation periods.

For the statistical comparison of unmarked and radio-marked birds, we assumed that the percent time unmarked females were observed in each behavioral category was measured without error. Although this assumption was somewhat liberal, we believe the standard errors would have been negligible, given the year-to-year consistency in percentages for unmarked females within each study site (Fig. 1). To test for an overall behavioral difference between radio-marked females (sample mean vector) and unmarked females (hypothesized mean vector) within each study site, we performed a 1-sample Hotelling's T ²-test (Johnson and Wichern 1988) using the Interactive Matrix Language procedure of SAS (SAS Inst. Inc. 1985) on arcsine-transformed (Steel and Torrie 1980:236) data. Simultaneous 90% and 95% confidence intervals were calculated for the 3 behavioral categories. All percentages reported are back-transformed values.

We tested for year and site effects on behaviors of radio-marked birds using multivariate analysis of variance (MANOVA) techniques with the General Linear Models (GLM) procedure of SAS (SAS Inst. Inc. 1989b). Year and site effects were tested for unmarked birds with Chi-square statistics using the weighted least squares estimation technique (Grizzle et al. 1969) in the Categorical Data Modeling (CATMOD) procedure of SAS (SAS Inst. Inc. 1989a); observation periods were assumed to be independent.

We obtained data on nests of radio-marked and unmarked females using 2 nest-searching techniques. Nests of most radio-marked birds were found by using telemetry. Females generally were located by triangulation several times each day; if a female was located at a potential nesting site on ≥4 consecutive days, we visited that site and searched for a nest. Nests of unmarked birds and some radio-marked birds were found by searching inside 25-ha predator exclosures (design modified from Greenwood et al. 1990) every 2 weeks and by searching other fields (63-174 ha each) in WPA's every 3 weeks. These searches commenced in late April or early May, after trapping and radio-tagging operations had ended. Searches were made by dragging a chain between 2 vehicles to flush females from nests (Higgins et al. 1969).

Eggs were counted and candled to determine incubation stage each time a nest was visited, and nest initiation dates were estimated from this information (Weller 1956). Clutch size was recorded as the maximum number of eggs found in completed clutches. Length and width of each egg was measured with a vernier caliper to the nearest 0.1 mm, and egg volume was calculated with the formula:

Volume = 0.515(length)(width)² using the shape constant for mallards (0.515) given by Hoyt (1979:74).

To test for a harness-transmitter effect on nest initiation date, we used analysis of variance (ANOVA) following the means model approach described by Milliken and Johnson (1984) and the GLM procedure of SAS (SAS Inst. Inc. 1989b). The data were analyzed for a transmitter effect, accounting for year (1988-91), predator exclosure (inside, outside), and area (Detroit Lakes, Kulm, Jamestown). We assumed that the samples of nests found by telemetry and chain dragging contained comparable proportions of first nesting attempts.

To test for harness-transmitter effects on clutch size and egg volume, we used nested ANOVA procedures following the means model approach (Milliken and Johnson 1984, SAS Inst. Inc. 1989b). The data were analyzed for transmitter effects, accounting for year (1988-91) and area (Detroit Lakes, Kulm, Jamestown). Because nest initiation date can affect clutch size, and possibly egg volume (Krapu et al. 1983, Eldridge and Krapu 1988), we wanted to determine whether there was an additional effect from the transmitters beyond that associated with initiation date. To reduce the confounding effect of transmitters on initiation date, we analyzed a subset of egg and clutch data from nests initiated during the time interval in which both radio-marked and unmarked females started nests. Because this time interval was different for the North Dakota and Minnesota areas, we did not examine this subset for region and year effects.

Results

The combined differences in feeding, resting, and preening time differed (P < 0.0001, Detroit Lakes; P = 0.0018, Kulm) between radio-marked and unmarked female mallards at both study sites ( Table 1). On the Detroit Lakes site, radio-marked females were observed feeding less and resting and preening more than unmarked females (P < 0.05 for all 3 categories) (Fig. 1). On the Kulm site, the same trends were evident, but tests for individual behavioral categories were not significant (P > 0.05). On both study sites, radio-marked females spent only about half as much time feeding as did unmarked females.

Females on the Detroit Lakes site allocated their time differently than those on the Kulm site for the periods they were under observation. Unmarked females on the Detroit Lakes site spent less time feeding than those on the Kulm site (X² = 7.33, 1 df, P = 0.007 for 1989; X² = 5.26, 1 df, P = 0.02 for 1990). Radio-marked females on the Detroit Lakes site fed less (F = 5.32; 1, 48 df; P = 0.025), preened more (F = 5.24; 1, 48 df; P = 0.026), and rested more (F = 3.73; 1, 48 df; P = 0.06) than those on the Kulm site (1989 data available only). Within study sites, neither unmarked nor radio-marked females showed between-year differences in behavioral allocations (P = 0.27-0.92 for 6 comparisons).

Average nest initiation dates of radio-marked and unmarked females (Fig. 2) were different (F = 6.68; 12, 410 df; P = 0.0001). Radio-marked females nested later than unmarked females in all within-year, within-area comparisons. Average nest initiation dates also differed among areas (F = 1.93; 20, 410 df; P = 0.010); females tended to nest later on the Minnesota area than on the North Dakota areas. We found no consistent trends or significant differences in nest initiation dates among years (F = 1.23; 17, 410 df; P = 0.24) or between nests found inside and outside predator exclosures (F = 1.63; 14, 410 df; P = 0.068). For unmarked and radio-marked females, respectively, overall mean nest initiation dates were 13 and 30 May for Detroit Lakes, 8 and 23 May for Kulm, and 11 and 25 May for Jamestown.

Radio-marked females produced smaller clutches (F = 8.55; 9, 274 df; P = 0.0001) and smaller eggs (F = 2.59; 8, 185 df; P = 0.010) than unmarked females. On average, clutch sizes of radio-marked females were 17, 15, and 14% smaller and egg volumes were 3, 6, and 7% smaller than those of unmarked females on the Detroit Lakes, Kulm, and Jamestown areas, respectively ( Table 2). These differences appeared to be related, in part, to nest initiation date. When data were constrained by nest initiation date, radio-marked and unmarked females still differed in clutch size (F = 3.25; 9, 183 df; P = 0.001) but not in egg volume (F = 1.71; 8, 134 df; P = 0.10).

For the unconstrained data set, no differences were detected in clutch size among areas (F = 1.39; 13, 274 df; P = 0.16) or years (F = 0.88; 11, 274 df; P = 0.56). Egg volume differed marginally among areas (F = 1.79; 12, 185 df; P = 0.052) but not among years (F = 1.74; 11, 185 df; P = 0.068).

Discussion

The reduced feeding time of radio-marked females may have been related to their delayed nesting and low reproductive effort. During our study, <2/3 of the resident females (i.e., those that stayed within 12 km of the center of a study site through 15 May) with harness transmitters were known to have initiated nests (G. L. Krapu and R. J. Greenwood, unpubl. data). Female mallards acquire nearly all their protein for egg production from food obtained on the breeding grounds (Krapu 1981). If harness transmitters inhibit feeding behavior, females may be less capable of initiating clutches or renesting. Conversely, if transmitters interfere directly with some aspect of reproductive behavior, radio-marked females may have fed less because they were not nesting.

Mallards with harness transmitters have had much higher nesting rates in wet years than in dry years in the prairie pothole region (Krapu et al. 1983; Cowardin et al. 1985; D. Rave and M. Zicus, Minn. Dep. Nat. Resour., pers. commun.). Drought conditions undoubtedly reduced nesting rates during much of our study.

When ponds are plentiful, females may be able to nest successfully despite negative effects of transmitters. At the other extreme, when ponds are scarce, unmarked as well as radio-marked females may fail to nest. Thus, some transmitter effects may only be detectable when birds experience an intermediate level of environmental stress.

Preliminary analyses of our behavioral data prompted initiation of 3 transmitter evaluation studies in 1991; these compared reproductive efforts of female mallards equipped with several types of radio packages. Houston and Greenwood (1993) studied captive wild-strain mallards with no transmitters, with 4-g sutured-and-glued transmitters, and with 10-g and 18-g harness transmitters. They detected no differences in number of clutches laid, time required to renest, clutch size, egg mass, or percent change in female body mass. In contrast, Rotelia et al. (1993) found that wild free-ranging mallards in Alberta with 9-g harness transmitters initiated fewer nests and devoted fewer days to egg laying and incubation than females with 4- and 9-g sutured-and-glued transmitters, or with 22-g abdominally implanted transmitters. G. L. Krapu and P. J. Pietz (unpubl. data) found that wild free-ranging mallards in Minnesota with 23-g harness transmitters nested about 2 weeks later than those with 3-g sutured-and-glued transmitters (F = 5.49; 1, 6 df; P = 0.058). Differences in results between field and captive studies may reflect the fact that captive birds had unlimited food and expended virtually no energy on flight or predator avoidance. Swanson et al. (1986) demonstrated that reduced food availability can increase renesting intervals and decrease clutch size in captive mallards.

The behavioral and reproductive differences we observed between unmarked and radio-marked females cannot be attributed in any large part to effects of capturing, handling, or nasal tagging the radio-marked birds. Our behavioral observations were timed to exclude the short-term effects of capture and handling (see Methods), and our time budget analyses for male mallards with and without nasal markers detected no decrease in feeding or increase in resting or preening among males with nasal tags. Capture and handling stress may have caused some of the earliest nesters to lose their first clutches before we found them; in those cases the nest initiation dates we recorded would have been for renests. However, 93% of trapping was completed by 24 April, well before most nests in our sample were started; and the average delay between radio-marking and initiation of the first nest found was 33.5 days, far longer than a typical renesting interval (Swanson et al. 1986). Furthermore, field studies comparing mallards with harness transmitters to those with other types of transmitters (Rotelia et al. 1993) found reproductive differences among groups of birds, all of which had been captured and handled.

Wheeler (1991) found higher predation rates on free-ranging blue-winged teal equipped with 18-g harness transmitters than on those with 11-g sutured-and-glued transmitters. However, it is unclear whether transmitter mass or attachment method was responsible for the difference. Results from 2 other studies suggest that mallards are affected more by the harness than by mass of the transmitter. In 1991 field studies, mallards with 9-g harness transmitters in Alberta nested nearly as late (median = 22 May; Rotella et al. 1993) as those with 23-g harness transmitters in Minnesota (median = 26 May; G. L. Krapu and P. J. Pietz, unpubl. data). Greenwood and Sargeant (1973) found that captive mallards equipped with 17-, 26-, or 36-g harness transmitters lost more body mass than did control birds, but there was no correlation between loss of body mass and transmitter mass.

Research Implications

Researchers are cautioned that effects of harness transmitters may bias data in studies of waterfowl behavior and productivity. Potential effects of harness transmitters may be overlooked if assessments are based only on captive ducks with unlimited food, or on free-ranging ducks in extreme environmental conditions.

---

Literature Cited

This resource should be cited as:

Pietz, Pamela J., Gary L. Krapu, Raymond J. Greenwood, and John T. Lokemoen. 1993. Effects of harness transmitters on behavior and reproduction of wild mallards. Journal of Wildlife Management 57(4):696-703. Jamestown, ND: Northern Prairie Wildlife Research Center Online. http://www.npwrc.usgs.gov/resource/1999/wildmal/index.htm (Version 08OCT99).

---

Pamela J. Pietz, Gary L. Krapu, Raymond J. Greenwood, and John T. Lokemoen, U.S. Fish and Wildlife Service, Northern Prairie Wildlife Research Center, Route 1, Box 96C, Jamestown, ND 58401

Downloading Instructions -- Instructions on downloading and extracting files from this site.

wildmal.zip 27K ) -- Effects of Harness Transmitters on Behavior and Reproduction of Wild Mallards

Installation: Extract all files and open index.htm in a web browser.