WETLANDS, Vol. 29, No. 2, June 2009, pp. 704–712 ’ 2009, The Society of Wetland Scientists

WETLAND USE AND FEEDING BY DURING SPRING MIGRATION ACROSS THE UPPER MIDWEST, USA

Michael J. Anteau1,2 and Alan D. Afton3 1School of Renewable Natural Resources, Louisiana State University Baton Rouge, Louisiana, USA 70803 2Present address: U.S. Geological Survey, Northern Prairie Wildlife Research Center, Jamestown North Dakota, USA 58401, USA E-mail: [email protected] 3U.S. Geological Survey Louisiana Cooperative Fish and Wildlife Research Unit, Louisiana State University Baton Rouge, Louisiana, USA 70803

Abstract: Low food availability and forage quality and concomitant decreased lipid reserves of lesser scaup (Aythya affinis; hereafter scaup) during spring migration in the upper Midwest may partially explain reductions in the continental population of scaup. In springs 2004–2005, we examined wetland use and feeding activity of scaup on 356 randomly-selected wetlands within 6 regions in Iowa, Minnesota, and North Dakota. We examined wetland characteristics that favor high scaup use in 286 of these wetlands. We found that probabilities of wetland use and feeding by scaup increased with turbidity up to 45 and 30 NTU, respectively, but then declined at higher turbidity levels. Wetland use was positively correlated with size of open-water zone and amphipod densities, but was not correlated with chironomid densities. Feeding increased with amphipod density up to 26 m23 and then declined at higher amphipod densities; scaup seemingly forage most efficiently at amphipod densities above 26 m23. Wetland use was higher in North Dakota than in southern Minnesota and Iowa. Our results indicate that effective wetland restoration efforts to benefit scaup require maintaining abundant populations of amphipods (generally near 26 m23 landscape geometric mean) in wetlands with large (. 500 m diameter) open-water zones throughout the upper Midwest, but especially within Iowa and southern Minnesota.

Key Words: behavioral cue, patch occupancy, optimal foraging, resource selection, staging, stopover areas, waterfowl

INTRODUCTION 2006). Accordingly, restoration, and management of wetland habitats used by scaup during spring The continental scaup population (lesser [Aythya migration is a major goal of recent conservation affinis] and greater scaup [A. marila] combined) has efforts in the upper Midwest (Ducks Unlimited declined markedly since 1978 (Austin et al. 1998, 2004; S. Stephens, Ducks Unlimited Inc., personal Afton and Anderson 2001, Wilkins et al. 2007). The communication; G. Zenner, Iowa Department of largest declines are associated with a major segment Natural Resources, personal communication). of the lesser scaup population that winters in states Three important questions must be addressed to bordering the Gulf of and migrates along target types of wetlands that scaup use and regions the Mississippi valley and through the upper of greatest conservation need. First, what are the Midwest (Afton and Anderson 2001). preferred foods of scaup during spring migration? Food availability and forage quality for female Amphipods and chironomids are principal foods lesser scaup (hereafter scaup) currently are low and consumed by scaup during spring migration (Afton apparently decreased from that historically during et al. 1991, Anteau and Afton 2006, 2008b). spring migration in the upper Midwest (Anteau and Amphipod densities are positively correlated with Afton 2006, 2008a, 2008b, Strand et al. 2008). scaup use on breeding areas (Lindeman and Clark Moreover, scaup currently have low lipid-reserve 1999, Fast et al. 2004), but it is unclear if this is a levels and catabolize lipid reserves during stopover direct link or only because amphipod densities might periods throughout the upper Midwest (Anteau be correlated with densities of other scaup foods. 2006, Anteau and Afton 2004, Badzinski and Petrie Therefore, we simultaneously examined influences

704 Anteau & Afton, WETLAND USE BY LESSER SCAUP 705 of amphipod and chironomid densities on wetland converged basin (Johnson and Higgins 1997) or use and feeding activity of scaup to determine the comparable National Wetland Inventory data (see relative importance of these food items to migrating Anteau 2006). In Iowa, we broadened the constraint to females. Second, what characteristics of wetlands include townships with 200 ha of either semiperma- might scaup use to select foraging wetlands? The size nent or permanent wetlands. Constraining township and class of wetlands used by migrating scaup during selection helped ensure that there were enough migration must be identified so that conservation wetlands to sample within each township. efforts can be efficiently targeted. Moreover, infor- We allocated numbers of sampling clusters among mation on the importance of vegetation structure or the 6 regions (3 to 6 clusters per region; Table 1) water clarity of wetlands as visual cues used by scaup based on region size and number of candidate to select foraging wetlands would be useful for townships available (Table 1, Figure 1). Each region management. Consequently, we examined how var- was divided into sub-regions based on latitude, such ious wetland characteristics (e.g., turbidity, depth, that there was one sampling cluster per sub-region size of open-water zone, water regime, amount and each year (except in North Dakota Glaciated Plains of emergent vegetation, and densities of two [NDGP]). In NDGP, we assigned two sampling principle foods) influenced wetland use and feeding clusters for each sub-region because of the larger by scaup. Third, what areas or regions in the upper width of this region (east to west) relative to other Midwest are in most need of conservation efforts; are regions (Figure 1). there regions that scaup use heavily, but where they Each sampling cluster was comprised of three are unable to acquire nutrient reserves? Accordingly, adjacent 36-square-mile townships (27,972 ha total). we examined inter-regional variation in wetland use The centroids of the second and third townships and feeding by scaup in relation to nutritional and selected were constrained within 50 km of the dietary data from these same areas (Anteau 2006, centroid of the first randomly selected township. Anteau and Afton 2008b). We randomly selected 3 semipermanent or perma- nent wetlands (each . 4 ha) in each township for wetland sampling. We randomly selected new METHODS townships and wetlands annually within each sub- Study Area region from the list of candidate townships to maximize representation of spatial variability in We sampled wetlands and observed scaup behav- wetland characteristics. ior within the Prairie Pothole Regions of Iowa, In springs 2004 and 2005, we observed use and Minnesota, and North Dakota. This area encompasses behavior of scaup on 356 wetlands and conducted the most important spring migration stopover areas complete wetland sampling on a random subset of for scaup in the upper Midwest, based on observations 286 wetlands (Table 1). In 2005, we supplemented of wildlife biologists and managers, band-recovery our sample in Minnesota with 9 additional high data, sightings of color-marked scaup, aerial surveys in scaup use wetlands because preliminary analyses spring, and scaup marked with satellite transmitters indicated that random sampling alone did not (Anteau and Afton 2008b, Afton 2008). provide an optimal balance of high-use wetlands We stratified the three-state area into six eco- for examining factors that influenced wetland use physiographic regions (hereafter regions) based on and feeding by scaup. watershed and groundwater hydrology, geology, and Sampling coincided with estimated middle of the plant communities (Figure 1; Kantrud et al. 1989; scaup migration period, when relatively large num- Minnesota Department of Natural Resources, unpub- bers of migrating scaup were present on our study lished data). The Minnesota Glaciated Plains (MNGP) area (see Anteau 2006). We started sampling in the and Minnesota Morainal (MOR) eco-physiographic southern portion of the study area and worked north. regions included areas outside the traditional Prairie This approach ensured an equal probability of scaup Pothole Region (Figure 1); however, we included these being present within each sub-region, and that regions so that results would be relevant to state- sampling occurred at similar times relative to spring specific management and conservation plans. phenology. Sampling was conducted during 3 April We used a constrained-random clustered sampling to16May2004and30Marchto3May2005. approach to select wetlands; this approach minimized travel time between wetlands. We first estimated Determination of Wetland Use and Feeding numbers of townships within each region that had at least 200 ha of semipermanent wetlands (candidate We assessed wetland use and feeding by scaup on townships for random selection; Table 1), based on each selected wetland with four visits (twice a day 706 WETLANDS, Volume 29, No. 2, 2009

Figure 1. Study area depicting 6 eco-physiographic regions for observations of lesser scaup wetland use and feeding during spring migration (2004 and 2005) in the upper Midwest. Areas in white were not sampled.

Table 1. Numbers of sampling clusters (C), randomly selected wetlands observed for lesser scaup use and feeding (NRandom), wetlands with compete wetland surveys (NComplete) by region in springs 2004 and 2005, and candidate townships (T) available for random selection of sampling clusters.

NRandom NComplete Region C 2004 2005 2004 2005 T North Dakota Missouri Coteau 3 27 26 9 26 208a North Dakota Glaciated Plains 6 42 50 15 50 364a Red River Valley 3 21 21 11 22b 66a Minnesota Morainal 4 28 34 17 39b 348a Minnesota Glaciated Plains 4 31 33 18 36b 285a Iowa Prairie Pothole 3 20 23 20 23 43c Annual wetland totals 169 187 90 196 a Townships that contain at least 200 ha of semipermanent wetlands. b Numbers include supplemental wetlands surveyed (see methods). c Townships that contain at least 200 ha total of semipermanent and permanent wetlands. Anteau & Afton, WETLAND USE BY LESSER SCAUP 707 for two days; each visit was at least 4 hours apart). solution, and then transported to the laboratory for During each visit (approximately 5 minutes) wetland sorting. use was scored as scaup present (1) or absent (0); In the laboratory, each composite sample was scaup feeding also was scored as at least one scaup stained with Rose Bengal (Sigma # R3877, Sigma- feeding (1) or none feeding (0). Lesser and greater Aldrich Corp., St. Louis, Missouri), sieved (500 mm scaup can be difficult to differentiate at long mesh), floated with sugar and water solution in a distances; however, misidentification of these species large dissecting tray, and all plant and benthic should be minimal because observers were experi- material was searched. For this study, we sorted and enced with duck identification and our study area counted principal scaup foods (i.e., amphipods primarily was lesser scaup migration habitat (Bell- [Gammarus lacustris and azteca]and rose 1980, Austin et al. 1998). Indeed, very few chironomids [Chironomidae]) using lighted magni- greater scaup were observed (, 1% of total scaup fying glass and stereomicroscopy. observed; M. J. Anteau and A. D. Afton, US Estimation of Aquatic Vegetation, Open Water Zone, Geological Survey, personal observations; S. Ste- and Water Quality. We visually categorized the phens, Ducks Unlimited, Inc., personal communi- width (0, 1–4, 5–10, 11–30, 31–60, and . 60 m) of cation). the emergent vegetation ring at all 5 transects in each wetland. We also visually categorized (, 100, 100– Wetland Sampling 200, 201–500, and . 500 m) the size of open water zone of each wetland (distance along the transect Macroinvertebrate Sampling. For each wetland, we from each side of the emergent vegetation ring or established 5 transects on a map, radiating from the bank) at all 5 transects in each wetland. We center to the bank at randomly selected bearings (0– classified the dominant plant species in the emergent 359u). Maps then were used to locate transects in the vegetation ring for each wetland as cattail (Typha field. Transects 1–4 had two sampling stations each, spp.), bulrush (Scirpus spp.), other, or no emergent the first was 10 m past the ring of the emergent vegetation. We measured turbidity (6 1 nephelo- vegetation and the second was 50 m away from the metric turbidity unit [NTU]) with portable water first station along the transect (toward the center). quality meters (YSI 6600 sonde with an optical and However, station locations were limited to depths turbidity [YSI 6136] probe; YSI Inc. Yellow Springs, between 0.5–3 m for ease of sampling and because Ohio) at the first 4 transects near the center of the scaup feed at these depths in spring (Austin et al. wetland. 1998). Thus, in some instances (, 5% of transects), station locations were adjusted along transects to accommodate depth requirements. All sampling was Statistical Analyses conducted from a boat to avoid disturbance of Factors Influencing Wetland Use and Feeding.We sediments and macroinvertebrates. calculated the total water volume sweep with sweep At each station, we measured water depth and nets (VS; m3) for each wetland with the equation: sampled macroinvertebrate density with a D- VS ~ 2ðÞS D SN shaped-sweep net (1,200 mm mesh, 0.072 m2 open- i ing, Ward’s Natural Science, Rochester, New York). where Di 5 depth (m) at each sampling station and Each macroinvertebrate sample consisted of a SN 5 sweep-net opening (0.072 m2). We calculated bottom sample (skimming the net along the bottom densities (m23) of each species of amphipod by for a distance equal to the depth at the sampling dividing the count of each species by VS. The water station) and an upward water column sweep. This regime of each wetland was classified as either pattern ensured equal representation of bottom and semipermanent or permanent by converged basin water column in each sample. Submerged aquatic National Wetland Inventory data (Johnson and vegetation (SAV) or other debris brought up with Higgins 1997, Anteau 2006). We used maximum the sweep-net was included in the sample, unless depth (deepest measurement of eight amphipod- greater than 50% of the object hung outside the net. sampling stations) to index the depth of each Although SAV was present (see Anteau and Afton wetland. We indexed emergent vegetation width 2008a), it was never so dense that it impaired our and size of open water zone by attaching ‘‘dummy’’ ability to effectively sample macroinvertebrates with variables (0–5 and 1–4, respectively) to the levels of sweep-nets. In the field, all sweep-net samples (n 5 8 each measurement and averaging all of the dummy per wetland) were combined into one composite variables for each wetland, thus creating separate sample per wetland, preserved in a 95% ethanol covariates for emergent vegetation and open water. 708 WETLANDS, Volume 29, No. 2, 2009

All other measurements were averaged for each 2002) to insure a conservative model because the wetland. deviance/DF (3.36) indicated some overdispersion. We examined factors that influence wetland use and feeding by scaup with separate logistic regres- RESULTS sions (PROC GENMOD; SAS Institute 2002). In both models, we specified year and region as fixed Factors Influencing Wetland Use and Feeding blocking variables, and included water regime and Our final model for wetland use included region emergent vegetation species as nominal class vari- 2 (x 5, 276 5 24.75, P , 0.001), amphipod density ables (in class statement), and amounts of emergent 2 (x 1, 276 5 13.42, P , 0.001), amount of open water vegetation and open water, amphipod and chiron- 2 (x 1, 276 5 5.60, P 5 0.018), and a quadratic term omid density, turbidity, and maximum depth as for turbidity (x2 5 20.71, P , 0.001; x2 2 continuous variables. Both models had a binomial 1[X], 276 1[X ], 276 5 10.53, P , 0.001). Amphipod density (b 5 distribution and a logit link function. We included 0.22, SE 5 0.06) and amount of open water quadratic terms for turbidity and maximum depth (estimate 5 0.28, SE 5 0.12) were positively and interaction terms for region-by-year, region-by- correlated with wetlands use; turbidity was positive- amphipod density, and region-by-chironomid den- ly correlated with wetland use (b 5 1.71, SE 5 0.41) sity. Amphipod and chironomid density, turbidity, at levels up to 45 NTU and then negatively and maximum depth were log transformed to correlated (b 520.22, SE 5 0.07) at higher levels. improve model fit. In the feeding model, we included Our final model for scaup feeding included region quadratic and cubic terms for amphipod and 2 (x 5, 276 5 27.18, P , 0.001), a quadratic term for chironomid densities because density of foods 2 amphipod density (x 1[X], 276 5 14.07, P , 0.001; probably is not linearly correlated to feeding. In 2 2 x 1[X ], 276 5 10.06, P 5 0.002), and a quadratic the wetland use model, we scaled the standard errors 2 term for turbidity (x 1[X], 276 5 24.57, P , 0.001; (scale 5 1.69) for all tests with the DSCALE option 2 2 x 1[X ], 276 5 16.21, P , 0.001). Amphipod density (SAS Institute 2002) to insure a conservative model was positively correlated with scaup feeding (b 5 because the deviance/DF (2.78) indicated some 0.57, SE 5 0.15) up to 26 m23, but was negatively overdispersion (when a model under-predicts the correlated (b 520.09, SE 5 0.03) at higher amount of variation that truly exists). We selected densities. Turbidity was positively correlated with final models for scaup use and feeding using scaup feeding (b 5 1.44, SE 5 0.33) up to 30 NTU, backwards elimination procedures on the type III but was negatively correlated (b 520.21, SE 5 likelihood ratio statistics with a 5 0.05 (Zar 1996). 0.06) at higher levels. When polynomial effects were included in final models, we calculated the level of the explanatory variable at the inflection point of the curve, holding Regional Patterns of Wetland Use and Feeding 2 all other explanatory variables constant at mean Wetland use varied among regions (x 5, 350 5 values. 55.89, P , 0.001). Wetlands in Missouri Coteau of Regional Patterns of Wetland Use and Feeding.We North Dakota (COT) and NDGP had a higher compared wetland use and feeding among regions probability of being used than did those in MNGP, with separate logistic regressions (PROC GEN- MOR, and Iowa Prairie Pothole (IAPP); wetlands in MOD; SAS Institute 2002). The supplemental data MOR had the lowest probability of being used of collected from wetlands with high use by scaup in any region (Figure 2). x2 5 Minnesota (n 5 9) were not included because these Feeding also varied among regions ( 5, 350 29.25, P , 0.001). Scaup had higher probabilities of analyses required a random sample of wetlands. In feeding on wetlands within the Red River Valley both models, we specified region as a class variable, (RRV) and IAPP than within MNGP and MOR; using a binomial distribution and a logit link scaup had the lowest probability of feeding on MOR function. We used the LSMEANS statement to than for any region (Figure 2). calculate least-squares mean probabilities of wetland use and feeding with 95% confidence intervals (CI), and the DIFF option to conduct contrasts between DISCUSSION regions (SAS Institute 2002). We subsequently Factors Influencing Wetland Use and Feeding assigned letters to similar means (P . 0.05) based on contrasts of the DIFF option. In the wetland use Amphipod Densities. Scaup are highly mobile model, we scaled the standard errors (scale 5 1.83) during spring migration, but migrate relatively for all tests with the DSCALE option (SAS Institute slowly across the upper Midwest; the observed Anteau & Afton, WETLAND USE BY LESSER SCAUP 709

Chironomids currently are the predominant food consumed by scaup in RRV, MNGP, MOR, IAPP, and Northwestern Minnesota (Anteau and Afton 2008b). However, like Strand (2005) we found no evidence that chironomid density influenced wetland use by scaup, whereas amphipod densities were positively correlated to wetland use in both studies. These results suggest that scaup prefer amphipods over chironomids during spring migration in the upper Midwest. Similarly, recent diet studies also suggest that scaup prefer amphipods (Anteau and Afton 2006, 2008b), and amphipods historically were their predominant food during spring and early summer throughout the Prairie Pothole Region (Rogers and Korschgen 1966, Bartonek and Hickey 1969, Swanson and Nelson 1970, Swanson and Duebbert 1989, Afton and Hier 1991, Afton et al. 1991). In springs 2004 and 2005, amphipod densities throughout the upper Midwest averaged between 1 and 12 m23, whereas historical average densities generally were over 100 m23 (Anteau and Afton 2008a). Therefore, odds ratios from our analysis, on these amphipod densities, can be informative about the magnitude of the correlation between amphi- pods and wetland use. If amphipod density is 100 m23 in a given wetland, then scaup are 2.8 times more likely to use that wetland than a wetland with no detected amphipods; if amphipod density is Figure 2. Least-squares mean probabilities of wetland 10 m23 then scaup are 1.7 times more likely to use use (6 95% CI) and feeding (6 95% CI) by lesser scaup that wetland than one lacking amphipods. Thus, our among regions in the upper Midwest in springs 2004 and results in the context of those of Anteau and Afton 2005 combined. The regions are depicted as: COT 5 ND (2008a) further illustrate the importance of amphi- Missouri Coteau, NDGP 5 ND Glaciated Plains, RRV 5 Red River Valley of MN and ND, MNGP 5 MN pods to scaup during spring migration (see Anteau Glaciated Plains, MOR 5 MN Morainal, and IAPP 5 IA and Afton 2006 and 2008b). Moreover, the qua- Prairie Pothole. Means with same letter beneath region dratic relationship between amphipod density and labels were similar (P . 0.05). feeding probabilities indicates that feeding efficiency increases appreciably at amphipod densities over 26 m23. Thus, our results suggest that intra-patch middle of spring migration was approximately 45 (within wetland) forage efficiency of scaup on days later in Northwestern Minnesota than in amphipods must be low across the upper Midwest western Illinois (Anteau and Afton 2004). Thus, because current amphipod densities are lower than scaup probably sample food resources in numerous 26 m23 (Anteau and Afton 2008a). wetlands across this large landscape. Wetland use by scaup was positively correlated with relative abun- Wetland Characteristics. The observed positive dance of amphipods on breeding areas in southern correlation between wetland use by scaup and size Saskatchewan (Lindeman and Clark 1999) and in of the open-water zone of wetlands is consistent with the boreal forest near Yellowknife, Northwest that found by Lindeman and Clark (1999) and Territories (Fast et al. 2004). However, amphipods Strand (2005). Lindeman and Clark (1999) specu- are indicators of good wetland and water quality lated that scaup use wetland area as a cue for (Grue et al. 1988, Tome et al. 1995, Murkin and predicting dense populations of amphipods. How- Ross 2000, Besser et al. 2004, Anteau 2006, Anteau ever, larger wetlands or wetlands with a larger open- and Afton 2008a); thus, correlations between water zone may provide better refuge from distur- amphipod densities and scaup use could be driven bances near wetland margins (Korschgen et al. 1985, by high availability of alternate foods. Kahl 1991). We found no evidence that the species 710 WETLANDS, Volume 29, No. 2, 2009 or amount of emergent vegetation influenced transmitters) are needed to determine migration wetland use by scaup. However, the amount of corridors, flight distances, and stopover durations emergent vegetation would have an indirect-negative with in the upper Midwest. influence on the use of wetlands by scaup if the size Feeding probabilities alone cannot directly indi- of the emergent vegetation ring appreciably reduces cate what areas have better forage availability and/ the size of the open-water zone in the wetland. or quality because feeding behavior was related to Amphipod densities were positively correlated amphipod density in a non-linear manner. For with turbidity at lower levels, but negatively example, feeding might be low if high quality food correlated at higher levels (30 to 40 NTU, while is highly available but also if food is so scarce that holding chlorophyll a levels at zero or average levels, birds use more energy feeding than they gain from respectively; Anteau 2006). Moreover, our results the food acquired. Fortunately, interpreting regional indicate that wetland use and feeding by scaup were feeding probabilities in relation to diet and nutri- highest at 30–45 NTU. Wetland use and feeding by tional data of scaup from these regions (Anteau scaup might be high on wetlands that have moderate 2006, Anteau and Afton 2008b) is much more turbidity levels because these wetlands typically informative in relating how foods influence behavior contain the highest amphipod densities. However, and energetics. For example, lower feeding proba- our results suggest that scaup use turbidity as a cue bilities in MOR and MNGP may indicate that to identify wetlands having high amphipod densities energetic costs of foraging outweighed benefits because results indicated that turbidity and amphi- derived from food consumed because diets in these pod densities were independent predictors of wet- regions had lower quality foods (see Anteau and land use and feeding by scaup. Afton 2006, 2008b). Conversely, in NDGP and If scaup use turbidity as a cue for amphipod COT, feeding probabilities were intermediate in abundance, this cue may be less effective than in the relation to other regions, but more scaup were past because other new factors currently influence acquiring higher quality foods there (primarily turbidity in wetlands of the upper Midwest (e.g., amphipods; Anteau and Afton 2008b). Female agricultural sedimentation and detritivorous fish scaup collected in IAPP were over 3 times less likely communities; Swanson and Duebbert 1989, Gleason to be collected with food present in their upper et al. 2003). Thus, inter-patch foraging efficiency digestive tract than were those collected in NDGP or also may have decreased because the presence of COT (Anteau and Afton 2008b); however, feeding anthropogenic turbidity may cause scaup to spend probabilities of scaup were similar between these more time stopping and attempting to feed on regions. Similarly, females in IAPP were cataboliz- wetlands that contain little food. ing lipid reserves at a higher rate than other regions (Anteau 2006). Accordingly, low abundances of high quality foods (amphipods) for scaup across IAPP, Regional Patterns of Wetland Use and Feeding MNGP, and MOR (Anteau and Afton 2008a) Mean amphipod densities currently are higher in apparently have reduced forage efficiency and NDGP and COT than in MNGP, MOR, and IAPP concomitantly lipid reserves of scaup. (Anteau and Afton 2008a); similarly, scaup used wetlands in NDGP and COT more than those in Implications for Migration Habitat Conservation MNGP, MOR, or IAPP. Use of wetlands in Iowa appeared high relative to amphipod densities Our results can inform managers when making (Anteau and Afton 2008a), but low relative to decisions about where and what type of wetlands to historical data for this region (Low 1941). It is target conservation activities focused on restoring possible that some scaup migrate more quickly and improving spring migration habitats for scaup through Iowa, but those that do stop must choose in the upper Midwest. Our data presented here and from relatively few highly altered wetlands (Dahl elsewhere together indicate that increasing the 1990, Anteau and Afton 2008a). Wetland use was abundance of preferred foods in large wetlands, lowest in MOR, perhaps due to poor availability or throughout the upper Midwest, but especially in quality of foods (Anteau and Afton 2008a) or IAPP and MNGP, will benefit migrating scaup by perhaps because MOR is located on the edge of the increasing lipid accumulation and perhaps concom- corridor used by scaup for spring migration. We did itant breeding success and thus population size not record scaup abundance and stopover times of (Anteau 2006, Anteau and Afton 2006, 2008a, individuals; thus, inferences to numbers of scaup 2008b). Further, our results indicate that targeting using each region are not presently possible. Studies conservation efforts on large semipermanent or of individually marked female scaup (e.g., satellite permanent wetlands with open-water zones . Anteau & Afton, WETLAND USE BY LESSER SCAUP 711

500 m (mean diameter), including activities that Canada-through the Bonnycastle Fellowship, Kibbe prevent emergent vegetation from ‘‘choking-out’’ Field Station, Louisiana Department of Wildlife and the open water zones of wetlands, might also be Fisheries, Louisiana State University-through the beneficial to scaup. However, larger wetlands are Bosch Fellowship, Minnesota Department of Nat- not hydrologically independent from smaller ones ural Resources, Minnesota Waterfowl Association, nearby (Euliss et al. 2004). Therefore, a more holistic North Dakota Game and Fish Department, Prairie approach to restoration (i.e., including restoration Pothole Joint Venture, Upper Mississippi River and of adjacent smaller, less-permanent wetlands) may Great Region Joint Venture, USGS-Louisi- be necessary to achieve desired outcomes on larger ana Cooperative Fish and Wildlife Research Unit, wetlands and such an approach may better restore USGS-National Wetland Research Center, USGS- other degraded ecosystem services of wetlands scaup Northern Prairie Wildlife Research Center, and use as well as those used by other wildlife (Euliss et USFWS Regions 3 and 6 HAPET offices. We thank al. 2008, Gleason et al. 2008). R. Gleason, and D. Koons, for providing editorial Several lines of evidence indicate that amphipods help with the manuscript. are important foods for migrating scaup during spring in the upper Midwest. Our results provide a LITERATURE CITED basis for establishing a minimum target mean (geometric) density of amphipods to be generally Afton, A. D. 2008. Chronology and rates of migratory 23 movements, migration corridors and habitats used, and near 26 m for semipermanent and permanent breeding and wintering area affiliations of female lesser scaup wetlands within this landscape. Amphipod densities captured during spring stop-over on Pool 19 of the Mississippi were positively correlated to submerged aquatic River. Online resource: http://www.ducks.org/Conservation/ ScaupResearchProject/3161/ScaupResearchProjectHome.html. vegetation, and width of upland vegetation buffer/ Afton, A. D. and M. G. Anderson. 2001. Declining scaup filter strip, but were strongly, negatively correlated populations: a retrospective analysis of long-term population to fish abundances (Anteau 2006, Anteau and Afton and harvest survey data. Journal of Wildlife Management 2008a). Accordingly, if a conservation goal is to 65:781–96. Afton, A. D. and R. H. Hier. 1991. Diets of lesser scaup breeding increase amphipod density in the upper Midwest, in Manitoba. Journal of Field Ornithology 62:325–34. managers should : 1) target restoration/conservation Afton, A. D., R. H. Hier, and S. L. Paulus. 1991. Scaup diets effort to wetlands that are deep enough to support during migration and winter in the Mississippi Flyway. Canadian Journal of Zoology 69:328–33. over-wintering populations of amphipods, but also Anteau, M. J. 2006. Ecology of lesser scaup and amphipods in the possible to manage fish communities; and 2) provide upper-Midwest: scope and mechanisms of the spring condition abundant submerged aquatic vegetation, and a thick hypothesis and implications for migration habitat conservation. Ph.D. Dissertation. Louisiana State University, Baton Rouge, buffer of upland vegetation around the wetland. LA, USA. http://etd.lsu.edu/docs/available/etd-01242006-093828 Anteau, M. J. and A. D. Afton. 2004. Nutrient reserves of lesser scaup (Aythya affinis) during spring migration in the Mis- ACKNOWLEDGMENTS sissippi Flyway: A test of the spring condition hypothesis. Auk 121:917–29. We thank the following people for assistance with Anteau, M. J. and A. D. Afton. 2006. Diet shifts of lesser scaup wetland sampling, logistic support, laboratory anal- are consistent with the spring condition hypothesis. Canadian Journal of Zoology 84:779–86. yses, or other help: M. Anderson, A. Anteau, J. Anteau, M. J. and A. D. Afton. 2008a. Amphipod densities and Austin, L. Ball, B. Batt, J. Berdeen, T. Bishop, F. indices of wetland quality across the upper-Midwest, USA. Bolduc, E. Bowers, R. Brady, K. Brennan, R. Wetlands 28:184–96. Anteau, M. J. and A. D. Afton. 2008b. Diets of lesser scaup during Durham, R. Faulkner, J. Fernandez, A. Greer, A. spring migration throughout the upper-Midwest are consistent Hancock, N. Hanke, C. Hansen, R. Helm, D. with the spring condition hypothesis. Waterbirds 31:97–106. Hertel, R. Hier, M. Johnson, R. Johnson, J. Austin, J. E., C. M. Custer, and A. D. Afton. 1998. Lesser scaup (Aythya affinis). The Birds of , Number 338. The Lawrence, A. Leach, C. Loesch, P. Mathews, J. American Ornithologists’ Union, Washington, D. C., USA, and Markl, R. Markl, B. Meixell, E. Moser, T. Mulcahy, the Academy of Natural Sciences, Philadelphia, PA, USA. J. Nyman, M. Ohide, R. Olsen, B. Pardo, S. Badzinski, S. S. and S. A. Petrie. 2006. Lesser scaup nutrient Piepgras, P. Rasmussen, D. Rave, R. Reynolds, J. reserve dynamics during spring on the Lower Great Lakes. Wildlife Society Bulletin 34:395–407. Ringelman, F. Rohwer, T. Rondeau, S. Stephens, Bartonek, J. C. and J. J. Hickey. 1969. Food habits of M. Szymanski, P. Telander, and G. Zenner. We canvasbacks, redheads, and lesser scaup in Manitoba. Condor gratefully acknowledge numerous landowners who 71:280–90. Bellrose, F. C. 1980. Ducks Geese and Swans. Stackpole Books, allowed us to work on their property. We thank the Harrisburg, PA, USA. following agencies or organizations for financial or Besser, J. M., W. G. Brumbaugh, N. E. Kemble, T. W. May, and in-kind support: Ducks Unlimited Inc. USA, Iowa C. G. Ingersoll. 2004. Effects of sediment characteristics on the toxicity of chromium (III) and chromium (IV) to the amphipod Department of Natural Resources, IWWR of Ducks Hyalella azteca. Environmental Science and Technology Unlimited Canada, IWWR of Ducks Unlimited 38:6210–16. 712 WETLANDS, Volume 29, No. 2, 2009

Dahl, T. E. 1990. Wetland loss in the United States, 1780’s to Korschgen, C. E., L. S. George, and W. L. Green. 1985. 1980’s, p. 21. U.S. Fish and Wildlife Service, Washington, Disturbance of diving ducks by boaters on a migrational D.C., USA. staging area. Wildlife Society Bulletin 13:290–96. Ducks Unlimited Inc. 2004. Living Lakes Initiative Plan. Ducks Lindeman, D. H. and R. G. Clark. 1999. Amphipods, land-use Unlimited Inc., Great Plains Regional Office, Bismarck, ND, impacts, and lesser scaup (Aythya affinis) distribution in USA. Saskatchewan wetlands. Wetlands 19:627–38. Euliss, N. H., Jr., J. W. LaBaugh, L. H. Fredrickson, D. M. Low, J. B. 1941. Spring flight of diving ducks through Mushet, M. K. Laubhan, G. A. Swanson, T. C. Winter, D. O. northwestern Iowa. Condor 43:142–51. Rosenberry, and R. D. Nelson. 2004. The wetland continuum: Murkin, H. R. and L. C. M. Ross. 2000. Invertebrates in prairie a conceptual framework for interpreting biological studies. wetlands. p. 201–48. In H. R. Murkin, A. G. van der Valk, and Wetlands 24:448–58. W. R. Clark (eds.) Prairie Wetland Ecology: The Contribution Euliss, H. H., Jr., L. M. Smith, D. A. Wilcox, and B. A. Browne. of the Marsh Ecology Research Program. Iowa State Univer- 2008. Linking ecosystem processes with wetland management sity Press, Ames, IA, USA. goals: charting a course for a sustainable future. Wetlands Rogers, J. P. and L. J. Korschgen. 1966. Foods of lesser scaups on 28:553–62. breeding, migration, and wintering areas. Journal of Wildlife Fast, P. L., R. G. Clark, R. W. Brook, and J. E. Hines. 2004. Management 30:258–64. Patterns of wetland use by brood-rearing lesser scaup in SAS Institute. 2002. SAS/STAT user’s guide, version 9. SAS northern boreal forest of Canada. Waterbirds 27:177–12. Institute, Cary, NC, USA. Gleason, R. A., N. E. Euliss, Jr., D. E. Hubbard, and W. G. Strand, K. A. 2005. Diet and body composition of migrating Duffy. 2003. Effects of sediment load on emergence of aquatic lesser scaup (Aythya affinis) in Eastern South Dakota. M.S. invertebrates and plants from wetland soil egg and seed banks. Thesis. South Dakota State University, Brookings, SD, USA. Wetlands 23:26–34. Strand, K. A., S. R. Chipps, S. N. Kahara, K. F. Higgins, and S. Gleason, R. A., M. K. Laubhan, and N. H. Euliss, Jr. 2008. Vaa. 2008. Pattern of prey use by lesser scaup Aythya affinis Ecosystem services derived from wetland conservation practices (Aves) and diet overlap with fishes during spring migration. in the United States and prairie pothole region with and Hydrobiologia 598:389–98. emphasis on the U.S. Department of Agriculture Conservation Swanson, G. A. and H. F. Duebbert. 1989. Wetland habitats of Reserve and Wetlands Reserve programs. Professional Paper waterfowl in the prairie pothole region. p. 228–67. In A. G. van 1745. U.S. Geological Survey, Reston, VA, USA. der Valk (ed.) Northern Prairie Wetlands. Iowa State Grue, C. E., M. W. Tome, G. A. Swanson, S. M. Borthwick, and University Press, Ames, IA, USA. L. R. DeWeese. 1988. Agricultural chemicals and the quality of Swanson, G. A. and H. Nelson. 1970. Potential influence of fish prairie-pothole wetlands for adult and juvenile waterfowl – rearing programs on waterfowl breeding habitat. p. 65–71. In what are the concerns? p. 55–64. In P. J. Stuber (ed.) E. Schneberger (ed.) Symposium on the Management of Proceedings National Symposium of Protection of Wetlands Midwestern Winterkill Lakes. North Central Division of the from Agricultural Impacts. U.S. Department of Interior, U.S. American Fisheries Society, Winnipeg, MB, Canada. Fish and Wildlife Service, Biological Report 88(16). Tome, M. W., C. E. Grue, and M. G. Henry. 1995. Case studies: Johnson, R. R. and K. F. Higgins. 1997. Wetland resources of effects of agricultural pesticides on waterfowl and prairie In eastern South Dakota. South Dakota State University, Brook- pothole wetlands. p. 565–76. D. J. Hoffman, B. A. Rattner, ings, SD, USA. http://www.npwrc.usgs.gov/resource/wetlands/ G. A. Burton, Jr., and J. Cairns, Jr. (eds.) Handbook of Ecotoxicology. Lewis Publishers, CRC Press, Boca Raton, FL, sdwet/sdwet.htm USA. Kantrud, H. A., G. L. Krapu, and G. A. Swanson. 1989. Prairie Wilkins, K. A., M. C. Otto, G. S. Zimmerman, E. D. Silverman, basin wetlands of North Dakota: a community profile. and M. D. Koneff. 2007. Trends in duck breeding populations, Biological report 85(7.28), U.S. Department of the Interior, 1955–2007. U.S. Department of the Interior, Fish and Wildlife Fish and Wildlife Service, Research and Development, Service, Administrative Report, Laurel, MD, USA. Washington, DC, USA. Zar, J. H. 1996. Biostatisical Analysis. Prentice-Hall, Upper Kahl, R. 1991. Boating disturbance of Canvasbacks during Saddle River, NJ, USA. migration at Poygan, Wisconsin. Wildlife Society Bulletin 19:242–49. Manuscript received 2 July 2008; accepted 5 January 2009.