Influence of upland cover on occupancy rates of nest structures by mallards

University of Montana ScholarWorks at University of Montana Graduate Student Theses, Dissertations, & Professional Papers Graduate School 1999 Influence of upland cover on occupancy rates of nest structures by mallards Michael J. Artmann The University of Montana Let us know how access to this document benefits you. Follow this and additional works at: https://scholarworks.umt.edu/etd Recommended Citation Artmann, Michael J., "Influence of upland cover on occupancy rates of nest structures by mallards" (1999). Graduate Student Theses, Dissertations, & Professional Papers. 6920. https://scholarworks.umt.edu/etd/6920 This Thesis is brought to you for free and open access by the Graduate School at ScholarWorks at University of Montana. It has been accepted for inclusion in Graduate Student Theses, Dissertations, & Professional Papers by an authorized administrator of ScholarWorks at University of Montana. For more information, please contact scholarworks@mso.umt.edu.

Maureen and Mike MANSFIELD LIBRARY The University of IVIONTANA Peraiission is granted by the author to reproduce Üiis material in its entirety, provided that tliis material is used for scholarly purposes and is properly cited in published works and reports. ** Please check "Yes" or "No" and provide signature ** Yes, I gi'ant pennission No, I do not grant permission Author's Signature ^ Date Any copying for commercial purposes or financial gain may be undertaken only with the author s explicit consent.

INFLUENCE OF UPLAND COVER ON OCCUPANCY RATES OF NEST STRUCTURES BY MALLARDS by Michael J.Artmaim B.S., University of Minnesota, 1995 Presented in partial fulfillment of the requirements for the degree of Masters of Science University of Montana 1999 Approved by Chairman, Board o f Examiners Dean, Graduate School J Date

UMI Number: EP37721 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. UMT Disaartalion PubKahing UMI EP37721 Published by ProQuest LLC (2013). Copyright in the Dissertation held by the Author. Microform Edition ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code P r o Q ^ s C ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 48106-1346

Artmann, Michael J. M.S. Spring 1999 Wildlife Biology Influence of upland cover on occupancy rates of nest structures by mallards Director: Dr. Joe B a ll'$ ^ i^ ^ ABSTRACT Artificial nest structures are one of the most cost effective methods for increasing mallard (Anas platyhynchos) nest success; efficacy of structures depends on both nest success and the rate at which they are occupied. Nest success in nest structures is usually high, but occupancy rates are highly variable. The factors causing this variation are largely unknown. The degree of competition between nest structures and surrounding upland cover for nesting mallard hens is unknown, but commonly considered important. To address this question, I compared occupancy rates of structures in areas dominated by grassland (Grassland sites), to rates in areas dominated by cropland (Cropland sites). Mallards occupied 11.4 % (56/492) of nest structures during 1997 and 1998. A significantly higher proportion of nest structures was occupied on Grassland sites than on Cropland sites (17.8 % vs. 3.9 %), Occupancy rates increased on both Grassland (8% to 27%), and Cropland sites (1.5% to 6%) in 1998. Nest success averaged 86 % (48/56), and was similar among Grassland (85%) and Cropland sites (89%). Densities of mallard pairs and occupancy rates were positively correlated both years. Upland cover can significantly affect occupancy rates of nest structures. Contrary to the intuitive relationship between upland cover and occupancy rates, mallards occupied far more nest structures in the areas dominated by attractive nesting cover than in cropland areas. I recommend that managers interested in using nest structures seek areas with high numbers of wetlands surrounded by moderate grass upland cover. Agreements with private landowners, especially those with land enrolled in the Conservation Reserve Program, could led to increased program efficiency. It

ACKNOWLEDGMENTS I am grateful to my graduate committee, Drs. Joe. Ball, Dan Pletscher, and Dave Patterson for thoughtful comments, advice on statistical design, and guidance throughout my graduate career. I would like to thank Bill Sommerfeld for his tireless effort during the construction and installation process. Monica Noble and Mark Howell for their assistance during pair counts. Roger Hollevoet and his staff at the U.S. Fish and Wildlife Service, Devils Lake Wetland Management District, provided tremendous logistical support. Funding for the field research was provided by the Delta Waterfowl Foundation and support during the school year was provided by the Wildlife Biology Program at the University o f Montana. Numerous graduate students provided insight and advice, especially Steve Hoekman, Alison Perkins, Brian Logan, and John Lloyd. To the numerous private landowners, especially M. Barney, D. Beck, L. Borgen, R. Dahlen, S. Eidsness, B. Fossen, S. Jacobson, H. Johnson, V. Leikas, R. Meyer, G. Mork, J. Neidlinger, L. Nelson, A. Polasky, V. Ralston, H. Ruud, C. Sandagar, B. Sateren, D. Sateren, E. Schmidt, A. Shirek, D Shirek, G. Umess, B. Wass, T. Wixo. Thanks to Les Johnson of Holbeck Well-Drillers in Devils Lake, ND, and Doug Bratrud of the Minnesota Department of Transportation for donating material for the project. Finally, to my parents, Bonnie and Joe, for their never-ending support and encouragement. Thank you for instilling the values of hard work and dedication. ui

TABLE OF CONTENTS A BSTRACT... ü ACKNOWLEDGMENTS... iii LIST OF TABLES...v LIST OF ILLUSTRATIONS... v INTRODUCTION...1 M ETHODS... 4 Study R egion... 4 Site Characteristics... 6 Structure Design...8 Placement...10 Pair Counts...11 Nest Checks...12 Statistical M ethods...12 RESULTS... 14 Occupancy...14 Pair Densities... 14 Placement...14 Nest Initiation... 19 Clutch Size...19 DISCUSSION...24 Occupancy...24 Pair Densities... 25 Nest Initiation...26 Clutch Size... 27 MANAGEMENT IMPLICATIONS... 29 LITERATURE C IT E D... 31 APPENDIX...37 IV

LIST OF TABLES Table Page 1. Habitat characteristics of 13, 10.4 km^ study sites located in Ramsey and Nelson counties in northeastern North Dakota. Percent grassland, cropland, and wetland estimated from 1997 aerial photographs...7 2. Reproductive parameters for mallards nesting in nest structures on 13, 10.4 km^ study sites in Ramsey and Nelson counties. North Dakota during 1997 and 1998.. 15 3. Estimated number of mallard pairs, wetland area surveyed, and total wetland area for 7 Grassland and 6 Cropland sites in Ramsey and Nelson counties, North Dakota for 1997-1998 (x ±SE)...16 4. Departure from mean monthly temperature, median nest initiation date, 10 and 90% quantiles, and central span of nest initiations for mallards occupying nest structures in Ramsey and Nelson counties. North Dakota 1997-1998...23 LIST OF ILLUSTRATIONS Figure Page 1. Location of Grassland (squares) and Cropland sites (circles) in Ramsey and Nelson counties in northeastern North Dakota...5 2. Carpet nest structure design. Measurements are in English units (in text also) because material are sold in English units... 9 3. Relationship between estimated pair densities and occupancy rates for 1997 on Grassland sites (closed circles, n = 7), and Cropland sites (open circles, n = 6) R^ = 0.14... 17 4. Relationship between estimated pair densities and occupancy rates for 1998 on Grassland sites (closed circles, n = 7) and Cropland sites(open circles, n = 6). Relationship influenced by outlier (28.8). With outfier included (line shown), R^ = 0.34; with outlier excluded, R^ = 0.22... 18

Figure Page 5. Cumulative nest initiation curves for mallards occupying nest structures during 1997 and 1998 in northeastern North Dakota...20 6. Cumulative nest initiation curves for mallards occupying nest structures in 1998 in northeastern North Dakota... 21 7. Relationship between nest initiation date and clutch size for Grassland and Cropland sites in 1998. Rate of decline is 0.06 eggs / day, and is consistent between treatments. Grassland sites (closed circles: 34 nests), and Cropland sites (open circles: 7 nests)...22 VI

INTRODUCTION Historically, the Prairie Pothole Region (PPR) of North America produced, on average, about half of the continental duck population (Smith et al. 1964). However, extensive anthropogenic changes have created a highly fragmented landscape (Bird 1961, Kiel at al. 1972), most of which is unproductive for nesting ducks (Higgins 1977, Sugden and Beyersbergen 1984, Boyd 1985, Cowardin et al. 1985, Greenwood et al. 1995). Nest success and hen survival during the breeding season are not sufficient to sustain local breeding populations of upland-nesting ducks throughout much of the PPR (Cowardin et al. 1985, Klett et al. 1988, Fleske and Klaas 1991, Greenwood et al. 1995, Beauchamp et al. 1996, McKinnon and Duncan 1999). Areas specifically established to attract nesting ducks, such as Waterfowl Production Areas (WPA), may concentrate nest predators such as red fox (Vulpes vulpes), raccoons (Procyon lotor) and striped skunks {Mephitis mephitis), limiting the ability of these areas to provide safe nesting sites (Clark and Nudds 1991, Kantrud 1993, Sargeant et al. 1995). Managers have implemented several techniques to reduce predation on hens and their nests. Artificial nest structures are one of the most cost-effective methods for increasing nest success, particularly for mallards {Anasplatyrhynchos; Lokemoen 1984, Cowardin et al 1988). Nest success and occupancy rates are the primary factors that determine the effectiveness of nest structures. Occupancy of structures by mallards far exceed other duck species and nest success usually is high, commonly >80% (Bishop and Barratt 1970, Doty and Lee 1974, Doty 1979, Ball and Ball 1991, but see Doty et al.

1975). This represents 4-5 fold increase over nest success for mallards nesting on the ground in most areas (Cowardin et al. 1985, Klett et al 1988, Fleske and Klaas 1992, Greenwood et al. 1995). In addition, nest structures protect females from predators at the nest, thus reducing an important source of mortality during the breeding season (Johnson and Sargeant 1977, Sargeant et al. 1984, Sargeant and Raveling 1992). Although rates of nest success are consistently high in structures, occupancy rates are highly variable (Doty 1979, Sidle and Arnold 1982, Kowalchuk 1996); the causes of this variation are largely unknown. Structure type and their location within wetlands are two factors believed partly responsible for differences in occupancy rates. However, in most cases, these differences tend to be small and inconsistent among regions (e.g., Bishop and Barratt 1970, Doty 1979, Haworth and Higgins 1993). Because mallards prefer tall dense vegetation for nesting and generally avoid cropland (Duebbert and Kantrud 1974, Higgins 1977, Kirsch et al. 1978, Cowardin et al. 1985), high occupancy rates of structures might be expected in cropland where competing nesting cover is scarce (Cowardin et al. 1988). Using a model to simulate the effect of installing culverts on mallard recruitment in the PPR, Shaffer et al. (1996) concluded that intensive techniques such as nest structures compete with large-scale cover management programs such as the Conservation Reserve Program (CRP) for nesting hens, and thus are most useful where upland cover is scarce. This notion has considerable intuitive appeal, but has not been tested. Densities of mallard pairs also could influence occupancy rates of nest structures. Krapu et al. (1997) suggested that density o f mallard pairs was inversely related to

proportion of cropland when effects of wetland density were controlled. If this relationship exists, and if occupancy rates of structures are correlated with pair densities, then higher occupancy rates might occur in areas where nesting cover is abundant. Insight into the relationship between upland cover, densities of mallard pairs, and their influence on occupancy rates will allow managers to locate the most suitable areas for nest structures, with considerable impact on program efficiency. The objectives of my study were: 1) compare occupancy rates of nest structures in areas dominated by grassland with areas dominated by cropland, while controlling wetland density, 2) examine whether densities of mallard pairs are correlated with upland cover and occupancy rates, and 3) compare nest success, nest initiation date, and clutch size between grassland- and cropland-dominated areas.

METHODS Study Region I conducted this study during 1997 and 1998 in Ramsey and Nelson counties in northeastern North Dakota (Fig. 1). The area is in the Drift Plain ecotone (Stewart and Kantrud 1973) and is characterized by flat to gently rolling topography interspersed with numerous wetland basins. Production of small grains (primarily wheat and barley) and row crops (primarily sunflowers and soybeans) was the predominant land-use; small amounts of pasture and hayland were also present. Within the last ten years, many landowners have restored large blocks (some >200 ha) of cropland to perennial grass under the Conservation Reserve Program (CRP). In 1997, approximately 44,200 ha were enrolled in CRP in Nelson and Ramsey counties (NRCS 1998). Selection Criteria 1 used land-use maps provided by local Natural Resource Conservation Service and Farm Service Agency offices to locate 10.4 km^ (4 mi^) study sites differing in the proportion of upland nesting cover (high vs. low), yet were similar in wetland number and total area o f wetlands. Among potential sites, I used 1996 aerial photographs to control for wetland density because pair density correlates with wetland density (Crissey 1969, Stewart and Kantrud 1972, Krapu et al. 1983, Johnson and Grier 1988, Krapu et al. 1997). I defined Grassland sites as sites having ^50% of the total upland area in grassland habitat (see below) and Cropland sites as sites having ^ 20 % of the total upland area in grassland habitat. I located 10 candidate sites of each group satisfying the above

N elson IS km Figure 1 Location of Grassland (squares) and Cropland sites (circles) in Ramsey and Nelson counties in northeastern North Dakota

criteria and randomly selected 7 Grassland and 6 Cropland sites for use in this study. If permission for access was denied on ^25 % of an intended study site, I selected an alternative site from the list of candidate sites. Sites were separated by a minimum of 3.2 km to maintain spatial independence. Site Characteristics I estimated habitat composition (% grassland, % cropland, and % wetland; Table 1) using a planimeter and aerial photographs taken in August 1997. I validated the photographs with extensive ground truthing. I defined grassland habitat as all non-tilled uplands including planted cover, hayland, pasture, road ditches, shelter-belts, farmsteads, and odd areas (i.e., rock piles). Grassland habitat was primarily private land enrolled in CRP, or federally-owned Waterfowl Production Areas. Cropland habitat was defined as annually tilled uplands and cropland set-aside as summer fallow. The majority (>80 %) of cropland habitat was tilled in the fall, hence, residual upland vegetation was minimal. I classified wetlands following Steward and Kantrud (1971). Wetlands densities across all sites were quite high across all sites (Table 1). Number of wetlands and wetland area per site were similar between Grassland and Cropland sites (Table 1). Temporary and seasonal wetlands comprised 37, and 42 % o f the wetland basins, respectively

D OQ. C g Q. D C/) (A o' 3 O 8 ( O ' TABLE 1. Habitat characteristics of 13 10.4 km^ study sites located in Ramsey and Nelson counties in northeastern North Dakota. Percent grassland, cropland, and wetland data estimated from 1997 aerial photographs. Location Cropland Habitat class (%) No. wetlands by class * No. Grassland Cropland Wetland T S SP+P Wetlands 14-152N-60W 4 96 25 118 89 34 241 243 03-153N-60W 6 94 31 56 64 44 164 302 Wetland Area(ha) 06-158N-62W 9 91 13 41 63 28 132 136 03-153N-61W 10 90 34 44 49 32 125 346 3. 3" D OQ. C a o 3 D O Q. 05-154N-60W 14 86 20 89 91 48 228 199 I0-152N-57W 18 82 16 67 42 28 137 161 S ± ( S E ) 10(2.1) 90 (2.1) 23 (3.4) 69(12.1) 66 (8.2) 36 (3.5) 171 (20.5) 231 (33.3) Grassland 36-158N-62W 48 52 22 46 37 52 135 227 34-158N-63W 55 45 22 62 62 24. 148 229 19-153N-57W 56 44 29 48 81 35 164 309 27-155N-61W 56 44 26 49 80 38 167 267 D C/) C/) 35-154N-58W 57 43 28 49 106 44 199 293 10-152N-59W 64 36 25 86 87 40 213 254 21-151N-57W 82 18 23 70 111 42 223 236 s ± ( S E ) 60 (4.1) 40(4.1) 25(1,1) 59 (5.6) 80_(9.5) 39 (3.2) 178 (12.6) 259(12.1) Refers to tlie mile-section occupying the NW comer of the study site. ' Includes planted grass, pasture, hayland, shelter-belts, road ditches, farmsteads and rockpiles. Includes annually tilled land and summer fallow. Note: Percent Grassland and Cropland habitat represent the total proportion of upland habitat only.

8 Less than 1% of the wetlands on my study sites were permanent lakes, so I combined semipermanent and permanent categories. Because I used photographs taken in August to determine wetland presence and extent, most temporary wetlands wére dry. Consequently, I inferred from distinctive bands of coloration marking drowned out areas in crop fields where water had been present in the spring. I used a similar method on the Grassland sites. In general, habitat conditions within sites did not appear to vary between years. Upland cover seemed constant at all sites except on 1 Grassland site where 20 ha of grassland habitat was converted to cropland habitat. Wetland conditions were lower in 1998 compared with 1997; however, the differences were small and generally affected only the number of temporary wetlands and their length of inundation. Most seasonal and semi-permanent wetlands were at or near capacity at the start of both nesting seasons. Because I felt habitat variables remained nearly the same in 1998,1 used the calculated values from the 1997 aerial photographs as representative o f the conditions throughout the study. Structure Design I used a modified version of a cylindrical double-wire nest structure ( hen-house ; Fig.2). I replaced the commonly used flax straw with a flexible astroturf carpet material manufactured by the Monsanto Corporation. I tested this type of nest structure in Manitoba in 1996 and found it to be occupied at nearly the same rate as the flax-straw version.

3 ft. SmtD unount of era» bttfrtta 36"x36" caipttlajrtr and outer layer of wire ns EKb«b rw>3: W M f K B o i n e e i e r s i r j i y p i A r i w l a i b w * < r s * c m». a t» i * w i <^*1 Od ' VPWCTL* i M H B a ^ n a w T r m e #T C '^ # e. a lcx!«fn'>tsmemr.3lhf. wm» a m -<m m L R * M» 4,c w K C M m # m u rn m m > m i > «mrx mc*»n ra get * r a m» r r #*# mm u««r.am j«k œ m/>. & «]«w mad m n * e a he! B#T iihim ors * «.ar ^ mo FV B w Cylinder formed with 7.5 f t of I"x2 x35** welded utility wire. Outside diameter = 1 f t 1 "%fl"%3d" #2 pine board fastened to cylinder with wire ties WmereefW'xlW XllU' flat VM b «t it 8" to 90 degreto 1 t» III 2.5 f t sections of 5" FVC well-pÿc fitted tightly against support board served as predator guard 4 holes drilled in FVC allowed easy installation Figure 2. Carpet nest structure design. Measurements are in English units (in text also) because the materials are commonly sold in English units.

10 The carpet type was used to increase durability and longevity, and for ease of construction. The cylinders were 1* in diameter, 3' in length, and were constructed with the following materials: a 7.5' x 3' piece of welded utility wire (1" x 2" mesh), a 36" x 39" piece of carpet material, a support board (# 3 grade pine; 1" x 8" x 3'), and an iron mounting bracket (1/4" x 1 1/4" X 16"). Wire ties were used to secure the cylinder to the support board via 4 small holes drilled near the outer edges of the boards. I used 2 bolts to fasten the mounting bracket to the support board. Completed nest structures were then bolted to 7 to 10' galvanized metal highway posts which were driven into the wetland bottom. To deter access by predators, 1.5 to 2' sections o f 5" diameter white PVC well pipe were fastened to the posts. Because of extremely low occupancy in 1997, 2 minor changes were made to the nest structures prior to the 1998 nesting season. These changes were done to make the structures appear more natural and hopefully increase occupancy. First, the white PVC predator guards were painted green. Second, a thin layer of grass (~ 2.5 cm) was added between the outer layer of wire and carpet material. Placement Twenty nest structures were installed on each site in February and March of 1997. I selected predominantly semi-permanent (Type IV; Stewart and Kantrud 1971) wetlands for nest structures, primarily to ensure that sufficient water would be present throughout the nesting season and to reduce the risk nest structures would be in farmable wetlands. Wetlands selected ranged in size from 0.26-71.9 ha (x = 13.8 ha). Up to 7 structures were placed in large wetlands, but all structures were >100 m apart. I attempted to standardize installation within

11 wetlands to control factors potentially influencing occupancy such as height above the water (approx. 1.0 m; Doty et al. 1975, Messmer 1986a), distance to emergent vegetation (1 to 5 m; Bishop and Barratt 1970), distance to shore (>5 m), and orientation (northeast-southwest facing). Structures damaged by ice in 1997 (n = 18; 7%) were replaced in 1998; those lost in 1998 (n = 2) were not replaced. Pair Counts I counted breeding pairs to obtain an index of mallard pairs occurring on each site. Pair surveys were conducted once each year, from 5/4-5/23 in 1997, and 4/28-5/20 in 1998, on 8 randomly selected quarter sections (65 ha) per study site. The same quarter sections were counted in both years. The 8 quarter sections per site were divided equally between morning (0730-1 looh) and afternoon (1500-1800h) counting sessions, and were conducted on consecutive days when possible. Although not standard procedure (USFWS and CWS 1987), afternoon counts were conducted for two reasons. First, Barras (1998) found no significant difference between morning and afternoon count periods for mallards in North Dakota, and second, afternoon counts allowed more area within each study site to be surveyed given limited time and resources. Wetlands within the survey area were visited on foot. I recorded breeding pairs based on social group described by Dzubin (1969). Counts were suspended during moderate to heavy rain, and when winds exceeded 40km/h (USFWS and CWS 1987)

12 Nest Checks All structures were checked for evidence of nesting at approximately 25 day intervals in May, June, and July each year. Species were identified by observation of adults or from nest characteristics (Klett et al 1986). Information recorded upon finding a nest included date, species, number of eggs, incubation stage (Weller 1956), height above water, distance to emergent vegetation, and distance to shore I used clutch size and incubation stage to estimate nest initiation date by back-dating, assuming one egg was laid per day through completion of the clutch (Sowls 1955). I defined occupancy rate as the number of structures with > 1 egg present divided by the total number of usable structures per site. Structures were deemed unusable for various reasons including; 1) nesting material absent, 2) structures located on dry land, or submerged, or 3) structures tipped past 45 degree angle. Because virtually all nests were found, I used apparent nest success as the measure of overall nest success (Miller and Johnson 1978). A nest was considered successful if s 1 egg hatched (Klett et. al 1986) and unsuccessful if destroyed or abandoned. Nests were considered abandoned if intact clutches were unattended by females and showed no development between nest checks. Three nests were excluded fi*om clutch size analyses because abandonment occurred prior to laying a full clutch. Statistical methods Because the same study sites were used in consecutive years, variables such as occupancy rates and pair densities could not be considered independent replicates between years. In fact, both occupancy rates (r = 0.47) and pair densities (r = 0.51) of sites were

13 correlated between years. To account for site-specific variation, I used a nested analysis of variance (ANOVA, GLM; SPSS Inc. 1993) with treatment and year as fixed factors, and individual sites as random factors nested within treatment. I used a nested-anova model to assess the main effects of treatment (Grassland or Cropland), year, and their interaction on occupancy rates and on pair densities. I used Pearson correlation coefficients (r) to test for association between occupancy rates and pair density. I used a 2 x 2 contingency table to test whether structures occupied in 1997 were more likely to be occupied in 1998 than those not occupied in 1997. I used t-tests to compare whether variables associated with structure location within wetlands such as height above water, distance to shore, and distance to emergent vegetation influenced occupancy rates. Prior to analyses, I transformed percent occupancy using a modified arcsin transformation to improve normality and stabilize variances (Zar 1984). Nest initiation dates were not normally distributed (P < 0.05; Shapiro-Wilk statistic), so I used Mann-Whitney U tests to compare median nest initiation date between treatments within years (1998 only) and among years. Small sample sizes in 1997 precluded comparisons between treatments. I calculated the difference between the 10 and 90% quantiles of initiation dates as a measure of the length of the nesting period ( central span of Hammond and Johnson 1984). Clutch sizes were not normally distributed, so I used a x^ - transformation to improve normality and homogeneity of variance (Zar 1984). I used analysis of covariance (ANCOVA) with nest initiation date as covariate to test for differences in clutch size with treatment and year as main effects. If the models had similar slopes (interaction term P > 0.05), I tested for differences between intercepts (Zar 1984). I also examined the relationship between clutch size and initiation date using Spearman rank-order correlation coefficients ( r j for 1998.

14 RESULTS Occupancy. I found that mallards occupied a significantly higher proportion of nest structures on Grassland sites than on Cropland sites (17.8 % vs 3.9 %; nested-anova; F,,, = 17.097; P - 0.002; Table 2); overall occupancy rate was 11.4 % (56 o f492). Occupancy rates increased in 1998 across both treatments (Fj jj = 6.98; P = 0.02). The treatment-year interaction was not significant {Fj jj = 2.76; P = 0.13), suggesting increases in occupancy rates were consistent between treatments. Nest structures occupied in 1997 were more likely to be occupied in 1998 than those not occupied in 1997 (75 % of 12 vs. 15 % o f242; P < 0.001). Nest success was high in both years (100 % of 12 nests in 1997, 82 % of 44 nests in 1998), and was similar between treatments (Table 2). All unsuccessful nests (n = 8) were abandoned. Pair Densities. I counted significantly more mallard pairs on Grassland sites than on Cropland sites (nested-anova; Fjjj = 6.499; P = 0.027; Table 3). Pair densities were similar between years, although most sites (10 of 13) tended to increase in 1998 ( F ;= 3.83; P = 0.07). I did not detect a treatment-year interaction (F, j, = 1.59; P = 0.233). Pair densities were correlated with occupancy rates in 1998 (r = 0.59; P = 0.034), but not in 1997 (r = 0.37; P ==0.20). Although positively correlated in both years, pair density explained only 14 % and 34 % of the variation in occupancy rates in 1997 and 1998 (Fig. 3 and 4). Placement - None of the variables tested (i.e., height above water, distance to vegetation, or distance to shore) were significantly different between nest structures occupied and those not occupied (all /* 5 > 0.10), except distance to shore on Grassland sites (mean occupied = 35 m, mean unoccupied = 27 m; F = 0.02).

o I % C/) CO o' 3 8 c5' 3 C p. o OÛ. c a o 3 o o Q. TABLE 2. Reproductive parameters for mallards nesting in nest structures on 13,10.4 km^ study sites in Ramsey and Nelson counties, North Dakota during 1997 and 1998. 1997 1998 Combined Grassland Cropland Totals Grassland Cropland Totals Grassland Cropland Overall No. Structures 126 116 242 137 113 250 263 229 492 No. Occupied 10 2 12 37 7 44 47 9 56 (%) (7.9) (1.7) (4.9) (27.0) (6.1) (17.6) (17.8) (3.9) (11.4) % Nest Success 100 100 100 81.1 85.7 81.8 85.1 88.8 85.7 Eggs Produced 91 15 106 327 59 386 418 74 492 (x) (9.1) (7.5) (8.8) (8.8) (8.4) (8.8) (8.9) (8.2) (8.8) Eggs Hatched 87 15 102 269 49 318 356 64 420 (%y (95.5) (100) (96.7) (93.1) (94.2) (91.3) (93.7) (95.5) (94.0) Includes only nests where 2 I egg hatched (excludes 8 nests abandoned in 1998). O C o CO CO o' 3 VI

D OQ. C g Q. D C/) o" 3 O 8 ci' TABLE 3, Estimated number of mallard pairs, wetland area surveyed, and total wetland area for 7 Grassland and 6 Cropland sites in Ramsey and Nelson counties, North Dakota 1997-1998 (x ± SE). Estimated No. Pairs 1997 1998 Wetland Area Surveyed (ha) Total Wetland Area (ha) Grassland 129.9 ±(12.6) 173.8 ±(23.8)" 133.4 ±(7.8) 259.3 ±(12.1) 33" D O Q. C a O 3 D O Cropland 98.1 ±(14.5) 107.6 ±(11.0) 106.4 ±(12.7) 235.0 ±(33.3) * Represents no. pairs / wetland area surveyed x total wetland area * Includes outlier (300 pairs), withcmt outlier -152.8 ± (11.7). Note: Nested-ANOVA - treatment effect: F, i, = 6.499, P = 0.027; year effect: Fm = 3,83, P = 0.07; treatment-year effect: Fi,, = 1.59, P = 0.223. Q. D C/) C/) ON

17 0.20 0.18 -.1 0.16 - a. 1 E 2 I 0.14-0.12 - y 0.10 - g 0.08 - I 1 0.06 - a. I 0.04-0.02-0.00 4 6 8 10 12 14 16 18 20 Pairs / km Figure 3. Relationship between estimated pair density and occupancy rates for 1997 on Grassland sites (closed circles, n = 7) and Cropland sites (open circles, n = 6). = 0.14.

18 0.5 Ë. 0.4 S w V o E 0.3 - S "S s Ü 0.2 0.1 O 0.0 6 S 10 12 14 16 18 20 22 24 26 28 30 Pairs / km Figure 4. Relationship between estimated pair density and occupancy rates for 1998 on Grassland sites (closed circles, n = 7) and Cropland sites (open circles, n = 6). Relationship influenced by outlier (28.8). With outlier included (line shown), R^ = 0.34; with outlier excluded, R^ = 0.22.

19 Nest Initiation. Median nest initiation date was earlier in 1998 than in 1997 (26 April vs 15 May; U = 118.5, f < 0.005; Fig. 5). In 1998, median nest initiation did not differ between Grassland and Cropland sites (26 April; n = 37 vs. 22 April; n = 7; U = 118.5, P = 0.72). The cumulative distribution of nest initiations during the nesting season was similar among treatments in 1998 (Fig 6). In 1997, 10 % of nests were initiated by May 6 and 90 % by June 9, while in 1998, 10 % were initiated by April 15 and 90 % by May 20 (Table 4). Although the start of the nesting season differed considerably between years, the central span of the nesting season was 35 days in both years. Clutch Size. ~ Clutch size averaged 9.23 ± 1.25 (x ± SE, n = 53 nests) when years were pooled. Clutch size was higher in 1998 than in 1997 (9.41 ± 1.26 vs 8.58 ± 0.99; F, s, = 4.37; P = 0.04). In 1998, clutch size was higher on Grassland sites than Cropland sites (ANCOVA; 9.6 ± 1.2 vs 8.4 ± 1.1; = 14.30; P = 0.001), with nest initiation date controlled. Clutch size declined seasonally in 1998 (r = -0.68, P < 0.005); rate of decline was 0.06 eggs/day and was similar between treatments (ANCOVA; Fjjy = 0.004; P = 0.95; Fig. 7).

20 100 90-80 - SÔ 70-3 1 60 - Z 5 0-40 - o 1998 (n = 44) e 30-1997 (n= 12) 10-10 20 30 10 20 30 9 19 29 April May June Figure 5. Cumulative nest initiation curves for mallards occupying nest structures during 1997 and 1998 in northeastern North Dakota.

21 100 90 - «0 - I 70 - Ii5 60 Iz 5 0-1 H 4 0 - *s I 3 0 - Ik 20- Grassland (n = 34) Cropland (n = 7) 10-10 20 30 10 20 30 9 19 29 April May June Figure 6 Cumulative nest initiation curves for mallards occupying nest structures in 1998 in northeastern North Dakota.

22 13 12 - Grassland (n = 34) O Cropland (n = 7) JS 3 0 # # # # 10 20 30 10 20 30 9 19 29 April M ay Ju n e Figure 7. Relationship between nest initiation date and clutch size for Grassland and Cropland sites in 1998. Rate of decline is 0.06 eggs / day, and is consistent between treatments.

"O OQ. C g Q. "O C/) C/) 8 TABLE 4. Departure from mean monthly temperature, median date, 10 and 90% quantités, and central span of nest initiations for mallards occupying nest structures in Nelson and Ramsey counties. North Dakota 1997-1998. Temperature (C)* Nest Initiation Year No. of nests Mar Apr May Median 10% 90% Central span** 1997 12-3.4-3.7-2.6 15 May 4 May 9 Jun 35 3. 3" "O OQ. C a O 3 "O O 1998 44-0.3 +3.9 +1.7 26 Apr 15 Apr 20 May 35 * Values represent departure from normal (1898-1998). ^ Number of days between 10 and 90% quantités. Q. D C/) C/) K> W

24 DISCUSSION Occupancy Nest success rates for mallards nesting on the ground are usually below the estimated 15% rate needed for population stability (Cowardin et al 1985, Klett et al. 1988, Greenwood et al. 1995). In contrast, mallard nest success in structures commonly exceeds 80% (Bishop and Barratt 1970, Doty 1975, Johnson et al. 1994, this study). Effective use of structures requires knowledge about the factors influencing occupancy. I observed differences in occupancy rates that were substantially greater than in previous studies. Occupancy rates by mallards were about 5 times higher on Grassland sites than on Cropland sites, and the difference was consistent in both years. Therefore, my results strongly suggest the composition of upland cover has a large effect on occupancy rates of nest structures. The overall occupancy rate (11.4%) was affected by the low occupancy rates on the Cropland sites. Grassland sites averaged 18% occupancy which is similar to other studies in the prairie pothole region (Bishop and Barratt 1970, Doty 1979, Haworth and Higgins 1993); these studies were conducted largely on public lands managed for waterfowl production, therefore, upland cover was presumably similar to that of my Grassland sites. In contrast, occupancy rates on Cropland sites averaged only 4% and were among the lowest reported. Occupancy rates increased nearly four-fold in 1998; the increase was similar between Grassland and Cropland sites suggesting that similar factors were operating in both habitats. Occupancy rates increased from 8% to 27% on Grassland sites and from 2 % to 6 % on Cropland sites. I expected occupancy rates to increase given the philopatric tendency of

25 mallards, especially females successful the previous year (Lokemoen et al. 1990, Majewski and Beszterda 1990). Doty and Lee (1974) estimated that 89% of adult mallards believed to be alive one year after nesting in structures returned to nest in the same structure in subsequent years. Although I did not mark females nesting in my structures, the combination of 100% nest success in 1997, and the fact that 75% of structures occupied in 1997 were reoccupied in 1998, suggests that female mallards were philopatric to my structures The minor changes to the appearance of the nest structures prior to the 1998 nesting season may be partly responsible for the increase in occupancy rates between years. However, given overall study objectives, I decided to modify all structures between field seasons and hence could not evaluate the effects of those changes on occupancy. However, the fact that 5 sites exhibited no increase suggests the changes may have played a limited role in the increase in occupancy rates. Pair Densities Densities of mallard pairs tended to be higher on Grassland sites than on Cropland sites in both years. Krapu et al. (1997) suggested that as temporary and seasonal wetland area increased, the number of mallard pairs varied inversely with the percent of landscape in cropland Krapu et al. (1997) hypothesized that declining availability of invertebrates and other reductions in habitat quality made farmed wetlands less attractive as breeding sites. Whether the factors influencing pair densities on my study sites were a function of wetland quality as foraging sites, and (or) the presence of attractive upland nesting cover is unknown. Nonetheless, I found a positive relationship between pair density and occupancy in both years. Clearly, if pair densities were extremely low (i.e., near zero), I would expect occupancy rates to be low. However, pair densities were substantially greater

26 than the number of nest structures on all of my sites. As a result, pair density alone explained little of the total variation in occupancy. This result may reflect the inherent difficulties involved in accurately measuring pairs (i.e., visibility problems, double-counting) and unknown sampling variation across time and space. It also suggests that other factors not related to pair density influence occupancy rates. Mallard pairs on all Cropland sites were 3-6 times more numerous than structures, yet few structures were occupied. Potential reasons for such low occupancy are numerous. Mallards foraging in wetlands on the Cropland sites may not be nesting nearby. Mallards generally avoid tilled cropland, especially early in the nesting season when residual vegetation is minimal, in favor of dense vegetation at nest sites (Higgins 1977, Cowardin et al. 1985, Greenwood et al. 1995). In fact, mallards will travel several kilometers to find suitable nesting h ^ itat (Duebbert et al. 1983, Lokemoen et al. 1984, Cowardin et al. 1985). If females assess nesting habitat and decide that suitable nest sites are not available in the immediate area, yet foraging opportunities are sufficient to meet dietary requirements, nest structures may not be investigated as a potential nest site. The notion that nest structures represent the most attractive nest site in a sea of cropland may not be valid if hens aren t looking for nest sites. Alternatively, because grassland dominated areas tend to attract higher densities o f nesting females (Higgins 1977, Duebbert et al 1983, Cowardin et al. 1985), the likelihood of a nest structure being detected by a female searching for a nest site may be higher. Nest Initiation. The start of the nesting season varied between years. Much of this variation can be attributed to climatic conditions. In 1997, colder than normal spring

27 temperatures coupled with a deep snow pack and a heavy snowfall in early April resulted in unfavorable nesting conditions. I observed few mallard pairs on my study sites before 20 April. In contrast, warmer than normal temperatures and a reduced snow pack created conditions suitable for early nesting in 1998. Mallard pairs began arriving in mid-late March, and the first nest in structures was initiated on 10 April. There was no difference in nest initiation dates between treatments in 1998, although mîdlards on Cropland sites nested slightly earlier than on Grassland sites. The majority of nests were initiated over a rather short period, about 2 weeks during the first half of the nesting season, in both years. Even though wetland conditions were conducive for an extended nesting effort, relatively few mallards nested in structures late in the nesting season. The observation of young broods (Class l-2a ducklings; GoUop and Marshall 1954) in July provided evidence that the nesting effort extended well beyond what I observed in structures. Clutch Size. Mean clutch size varied between years, with mallards tending to lay larger clutches in 1998. The larger clutch sizes in 1998 could be attributable to the early nesting conditions. More interestingly, I found mean clutch size to be about 1 egg smaller on Cropland sites than on Grassland sites in 1998. This difference cannot be explained by seasonal decline in clutch size because nests on Cropland sites were initiated somewhat earlier; the rate of decline (0.06 eggs/day) was similar to other studies for North Dakota (i.e.,cowardin et al. 1985, 0.06 eggs/day; Krapu et al. 1983, 0.05 eggs/day). Females in poor condition tend to lay smaller clutches (Krapu 1981, Eldridge and Krapu 1983, Krapu et al. 1983). Females acquire a relatively large portion of their dietary needs for egg production, especially protein, by consuming invertebrates in wetlands prior to

28 nesting (Krapu 1981, Krapu et al 1983). Invertebrates comprise the majority of the diet during the prelaying period (Swanson et al. 1979, 1985; Krapu 1981), and females tend to spend the majority of their time in shallow seasonal wetlands (Swanson et al. 1985, Krapu et al 1997). If wetlands subject to frequent tillage such as those found in farmed landscapes are less productive in invertebrate biomass as suggested by Swanson et al. (1974) and Krapu et al. (1997), then reduced clutch sizes may have reflected the inability of females to acquire necessary requirements to produce a full clutch. These results, although based on a small sample o f nests in cropland (n=7), tend to support this conclusion.

29 MANAGEMENT IM PLICATIONS Upland cover has large effects on occupancy rates of nest structures. Contrary to expectations of competition between structures and upland cover for nesting hens, mallards occupied far more structures in areas dominated by attractive nesting cover than in cropland areas. The high rate of nest success (85 %) provides additional evidence that nest structures effectively protect both nesting females and nests from the detrimental effects of predators. In terms of net production, the argument could be made that nest structures, even with reduced occupancy rates, provide a larger proportional increase in nest success in intensive farmed landscapes where nest success is typically poor compared to grassland areas (e.g., Higgins 1977, Cowardin et al. 1985). However, if occupancy rates in cropland areas remmn at the low levels observed in this study, this may not be the case. Although occupancy rates did increase slightly in 1998 on Cropland sites, 4 of 6 Cropland sites still had < 1 nest structure occupied. Longer-term monitoring (5+ years) should be conducted to evaluate whether occupancy rates eventually improve with time. The number of ducklings produced for both years combined was estimated to be 420 or 0.85 ducklings/structure. This is comparable to the 1.08 ducklings/basket and 0.81 ducklings/bale found in South Dakota (Haworth and Higgins 1993). Compared to other management techmques, nest structures are quite cost-effective for producing mallards (Lokemoen 1984, Cowardin et al. 1988). Material costs were estimated at about $34-40/structure excluding labor (Appendix 1). The carpet material was fairly expensive

30 per structure), however, the benefits of increased durability, ease of construction, and the realistic possibility of biennial maintenance may make this investment worthwhile. Because much of the prairie pothole region is now, and will remain, in private ownership, access to private land is essential if wide-scale application is to occur. Over 90 % of my study was conducted on private land. From my experience, attitudes of most landowners toward nest structures were positive or at least neutral. This is particularly encouraging given the general animosity towards the wildlife people in this region of North Dakota. The use of nest structures as a means to increase nest success of mallards will likely continue in the prairie pothole region. I recommend that individuals interested in using nest structures seek areas with moderate to high numbers of wetlands surrounded by at least moderate grass upland cover. Landscape cover-maps are available for much of the PPR (Reynolds et al. 1996) and should be used to locate areas with both wetland and grassland components. With continued interest being expressed in the Conservation Reserve Program, innovative agreements should be sought between private landowners and conservation organizations. Such agreements may lead to increased effectiveness of nest structures in the prairie pothole region.

31 T JTERATURE CITED B all, I.J., and S.K. Ball. 1991. Earth-filled culverts as nest sites for waterfowl. Prairie Naturalist 23:85-88. Barras, S C 1998. Evaluation of sampling methods, quantitative indices, and predictive models for estimating Mallard breeding populations. P H D. Dissertation. Utah State University, Logan UT 112 pp. Beauchamp, W.D., R.R. Koford, T.D. Nudds, R.G. Clark, and D.H. Johnson. 1996 Declines in nest success of prairie ducks detected using long-term data. Journal of Wildlife Management 60:247-257. Bishop, R. A., and R. Barratt. 1970. Use of artificial nest baskets by mallards. Journal of Wildlife Management. 34:734-738. Boyd, H. 1985. The large-scale impact of agriculture on ducks in the Prairie Provinces, 1956-81. Can. Wildl. Serv. Prog. Notes 149. 13pp. Clark, R.G,, and T.D. Nudds. 1991. Habitat patch size and duck nest success: the crucial experiments have not been performed. Wildlife Society Bulletin. 19: 534-543. Cowardin, L M, A.B. Sargeant, and H.F. Duebbert. 1983. Problems and potentials for prairie ducks. Naturalist 34(4) 4-11., D.H. Johnson, T L Shaffer, and D W Sparling. 1988. Applications of a simulation model to decisions in mallard management. U.S. Fish and Wildlife Service Publication No. 17. 28 pp., D S Gilmer, and C.W. Shaiffer. 1985. Mallard recruitment in the agricultural

32 environment of North Dakota. Wildlife Monograph 92. 37pp. Crissey. W.F. 1969. Prairie potholes from a continental viewpoint. Pages 161-171 in Saskatoon wetlands seminar. Can. Wildl. Serv. Rep. Serv. 6. Doty H A. 1979. Duck nest structure evaluation in prairie wetlands. Journal of Wildlife Management 43:976-979. Doty, H A, and F.B. Lee. 1974. Homing to nest baskets by wild female mallards. Journal o f Wildlife Management 38:714-719.,, and A D Kruse. 1975. Use of elevated nest baskets by ducks. Wildlife Society Bulletin 3:68-73. Duebbert, H.F., and H A Kantrud. 1974. Upland duck nesting related to land use and predator reduction. Journal of Wildlife Management 44:257-265., J.T. Lokemoen, and D E Sharp. 1983. Concentrated nesting of mallards and gadwalls on Miller Lake Island, North Dakota. Journal of Wildlife Management. 47:729-740. Dzubin, A. 1969. Assessing breeding populations of ducks by ground counts. Pages 178-230, in Saskatoon wetlands seminar. Can. Wildl. Serv Rep. Serv. 6. GoUop, J B, and W.H. Marshall. 1954. A guide for aging duck broods in the field. Flyway Counc. Tech. Sect. 14pp. Greenwood, R J, and M.A. Sovada. 1996. Prairie duck populations and predation management. Transactions North American Wildlife Natural Resource Conference 61: 31-42., J, A..B. Sargeant, D.H. Johnson, L.M. Cowardin, and T.L Shaffer. 1995.

33 Factors associated with duck nest success in the prairie pothole region of Canada. Wildlife Monograph 128. 57pp. Hammond, M C, and D.H. Johnson. 1984. Effects of weather on breeding ducks in North Dakota. U.S. Fish Wildl. Serv., Fish Wildl. Tech. Rep. 1. Haworth, M., and K.F. Higgins. 1993. Waterfowl use and production from nesting baskets and bales in South Dakota wetlands. Prairie Naturist 25 (2): 149-160. Higgins, K. F. 1977. Duck nesting in intensively farmed areas of North Dakota. Journal of Wildlife Management 41: 232-242. Johnson, D.H., and J W Grier. 1988. Determinants of breeding distributions of ducks. Wildlife Monograph 100. 37pp., and A. B. Sargeant. 1977. Impact of red fox predation on the sex ratio of prairie mallards. U.S. Fish and Wildl. Serv. Wildl. Res. Rep. 6. 56pp. Kantrud, H. A. 1993. Duck nest success on Conservation Reserve Program land in the Prairie Pothole Region. Journal o f Soil and Water Conservation. 48:238-242. Kiel, W H, A S. Hawkins, and N.G. Perret. 1972. Waterfowl habitat trends in the aspen parkland of Manitoba. Can. Wildl. Serv. Rep. Ser. 18. 61pp. Kirsch, L M, H.F. Duebbert, and A D Kruse. 1978. Grazing and haying effects on habitats of upland nesting birds. Transactions North American Wildlife Natural Resource Conference 43:486-497. Klett, A T, H.F. Duebbert, C A Faanes, and K.F. Higgins. 1986. Techniques for studying nest success of ducks in upland habitats in the prairie pothole region. U.S. Fish and Wildlife Serv. Resour. Publ. 158. 24pp.

34 T.L. Shaffer, and D.H. Johnson. 1988. Duck nest success in the prairie pothole region o f the United States. Journal o f Wildlife Management 52:431-440. Kowalchuk, T. 1996. Comparison of two nesting structures for mallards in pothole habitat o f Minnesota and Manitoba. Unpublished M.S. Thesis. Univ. Manitoba. 1996. Krapu, G.L., A T Klett, and D.G. Jorde. 1983. The effect of variable spring water conditions on Mallard reproduction. Auk. 100:689-698. Lee, F.B. 1982. Home-made nesting sites for mallards. North Dakota Outdoor Magazine. March. 6-9 Lokemoen, J.T. 1984. Examining economic efficiency of management practices that enhance waterfowl production. Transactions North American Wildlife Natural Resource Conference 49:584-607., H.F. Duebbert, and D E Sharp. 1984. Nest spacing, habitat selection, and behavior of waterfowl on Miller Lake Island, North Dakota. Journal of Wildlife Management 48:309-321.,, and. 1990. Homing and reproductive habits of mallards, gadwalls, and blue-winged teal. Wildlife Monograph 106:1-28. McKinnon, D.T., and D. C. Duncan 1999. Effectiveness of dense nesting cover for increasing duck production in Saskatchewan. Journal of Wildlife Management 63: 382-389. Messmer, T. A., M J Johnson, and F.B. Lee. 1986a. Home-made nest sites for mallards. N.D. State Univ. Coop. Ext. Serv. WL-890. Miller, H.W., and D.H. Johnson. 1978. Interpreting the results o f nesting studies. Journal of