WILDLIFEMONOGRAPHS (ISSN:0084-0173)

A Publicationof The WildlifeSociety

C t141 ;s

° L @ll I X g r

DETERMINANTSOF BREEDING DISTRIBUTIONSOF DUCKS by

DOUGLASH. JOHNSONAND JAMESW. GRIER

NO. 100 OCTOBER1988 : 0Esu ; e- W$siS

:is

: > r _> .

.

<<-!A./ t-_I'

FRONTISPIECE. Lssser scaup use more permanent wetlands and breed in more northern locations than most of the common North American ducks. (U.S. Fish and Wildlife Service photo) DETERMINANTSOF BREEDING DISTRIBUTIONSOF DUCKS

DOUGLASH. JOHNSON U.S. Fish and Wildlife Service, Northern Prairie Wildlife Research Center, Jamestown, ND 58402 JAMES W. GRIER Zoology Department, North Dakota State University, Fargo, ND 58105

Abstract: Thesettling of breedinghabitat by migratorywaterfowl is a topicof boththeoretical and practical interest.We usethe resultsof surveysconducted annually during 1955-81 in majorbreeding areas to examine the factorsthat affect the distributionsof 10 commonNorth American duck species. Three patterns of settling aredescribed: homing, opportunistic, and flexible. Homing is generallymore pronounced among species that use more stable (more predictable)wetlands, such as the redhead(Aythya americana), canvasback(A. valisineria), lesserscaup (A. affinis), mallard(Anas platyrhynchos), gadwall(Anas strepera), and northern shoveler(Anas clypeata). Opportunisticsettling is more prevalentamong species that use less stable(less predictable)wetlands, such as northernpintail (Anas acuta) and blue-wingedteal (Anas discors). Flexible settlingis exhibitedto variousdegrees by mostspecies. The 10 speciesare shownto fall alonga naturalordination reXecting diSCerent life historycharacteristics. Averagevalues of indicesof r- andK-selection indicated that pintail, mallard, blue-winged teal, and shoveler have the mostfeatures associated with unstableor unpredictableenvironments. Gadwall, American wigeon (Anas americana), andgreen-winged teal (Anas crecca) wereintermediate, and attributes of the divingducks wereassociated with the use of stableor predictableenvironments. Somespecies notablymallard, gadwall, blue-winged teal, redhead, and canvasback tendto fillavailable breedinghabitat first in the centralportions of theirrange, and secondly in peripheralareas. Other species- Americanwigeon, green-winged teal, northern shoveler, northern pintail, and lesser scaup fill theirhabitat in the orderit is encounteredduring spring migration. Age and sex classeswithin species vary in theirsettling pattern. Some of this variationcan be predicted from the matingsystems of ducksin which breedingfemales, especially successful ones, have a greater investmentin habitatresources and are morelikely to returnto the samearea in subsequentyears. WILDL.MONOGR. 100, 1-37 6 WILDLIFEMONOGRAPHS

CONTENTS

INTRODUCTION 6 Breeding Distributions of the Species 13 Acknowledgments 7 Mallard 1 3 A THEORYOF SETTLING 7 Gadwall 16 Homing 8 American Wigeon 16 OpportunisticSettling 8 Green-vinged Teal 21 FlexibleSettling 8 Blue-winged Teal 21 LifeHistory Characteristics 9 Northern Shoveler 24 METHODS 9 Northern Pintail 24 Survey Design 9 Redhead 24 AerialCounts 11 CanvasUck 27 GroundCounts 11 Lesser Scaup 27 Pond Counts 11 General Patterns 30 Computations 12 Relations with Life History Characteristics 30 Analysisof Pond Counts 12 DISCUSSION 31 Breeding-season Distribution,Wintering Areas, Settling Patterns 31 and MigrationCorridors 12 Fillingthe Habitat 32 Relations of Ducks with Local and Continental Relations to Life History Characteristics 32 Ponds 12 Management Applications 34 Relationsto LifeHistory Characteristics 13 LITERATURECITED 35 RESULTS 13 Dynamicsof the WetlandHabitat 13

INTRODUCTION ing estimatesof the breedingpopulation Thesettling of habitatby migratory birds with measuresof productivity(Martin et withintheir geographic range represents al. 1979). Becausereproduction varies habitatselection on a coarsescale (Johnson widelyamong breeding grounds (Hansen 1980),finer than geographical range, but and McKnight1964, Calverley and Boag broaderthan home range and microhab- 1977, Derksenand Eldridge1980), un- itat. Analysisof settlingbehavior on this derstandinghow ducksdistribute them- scaleraises both biological questions and selvesthroughout their range should pro- managementissues. videuseful information for predicting the Onthe biological side, there is a consid- size of fall populations.At a local level, erablebody of theoryconcerning geo- the effectivenessof land management graphicdistribution of highlymobile an- techniquesto increasenumbers and pro- imals such as birds.Both the ultimate ductionof ducksdepends on the extentto factorsthat influence evolutionary fitness whichducks return to the samearea in in the habitatselected and the proximate subsequentyears or seek out new breeding cuesused by animals to selecthabitats have locations(Hochbaum 1946). beenintensively studied. Life history char- Ourobjectives were to determineeco- acteristicsof differentspecies have been logicaland geographiccorrelates of the relatedto habitatsoccupied by thespecies; breeding distributionsof 10 common muchof the discussionon this topic has speciesof NorthAmerican ducks and to been cast in termsof r- and K-selection relatethese to the life historycharacter- (Pianka1974). Distributions and basic bi- isticsof the species.The speciesare mal- ologyare particularly well studied in ducks lard,gadwall, American wigeon, green- (Johnsgard1975, Palmer1976, Bellrose wingedteal, blue-wingedteal, northern 1980),thus providing opportunities for de- shoveler,northern pintail, all of whichare tailedanalysis. dabblers(Tribe Anatini), and canvasback, Thenature of settlingpatterns of ducks redhead,and lesser scaup, all divers (Tribe is importantto waterfowlmanagers at both Aythyini).T}ais analysis is restrictedto the continentaland local levels.To regulate areain NorthAmerica regularly surveyed hunting,for example,managers predict forbreeding ducks (Fig. 1), and data used the sizeof the fall populationby combin- areresults of thosesurveys. BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier 7

1 0

Fig. 1. Strata used in breedingwaterfowl surveys.

We propose3 patternsthat migratory H. Boyd,W. F. Crissey,A. Dzubin,J. L. birdscould use for settlingbreeding hab- Eldridge,R. R. Koford,G. L. Krapu,J. T. itat. We then describethe methodsused Lokemoen,H. W. Miller,J. D. Nichols,T. to gatherand analyzethe data.Distribu- D. Nudds,G. L. Nuechterlein,and A. B. tionsof breedingducks are described and Sargeant.M. R. Milleroffered numerous relatedto winteringareas and migration suggestionsthat improved the manuscript. corridors,local and continentalwetland conditions,and homingand pioneering A THEORYOF SE1TLING tendenciesof each species.We also de- scriberelations among the species and how Migratorybirds, by virtueof theirmo- ourfindings fit with the species' life history bility, have many optionsfor selecting characteristics.We concludewith some habitatin whichto breed(Fretwell and implicationsfor management. Lucas 1969, Gauthreaux1980, Bailey Acknotvledgments.We are grateful 1981).The qualityof a habitathas 2 as- to R. S. Pospahalaand others at the Offlce pects-the ultimatequality and the prox- of MigratoryBird Management, U.S. Fish imatecues by whichthe birdschoose it andWildlife Service, for providing survey (Immelmann1973). data used in the analyses,and to A. D. Ultimatefactors are those that affect Aftonand R. A. Wishartfor use of their long-termsurvival and reproductivesuc- unpublishedinformation on homingof cess.Optimal habitat for survivalof the lesserscaup and Americanwigeon, re- adultmay differfrom the optimumfor spectively.We appreciatethe valuable reproduction;requirements for reproduc- commentson variousdrafts of thisreport tion tend to be morestringent. Not only madeby M. G. Anderson,F. C. Bellrose, mustthe birdbe safefrom predation, but 8 WILI:)LIFEMONOGRAPHS so mustthe eggsand ducklings, life stages OpportunisticSettling thatare highlyvulnerable. Food supplies mustinclude the minimumfor adultsur- Thesecond pattern is fora birdto settle vival,as well as adequate calcium, protein, in the firstsite encounteredalong its mi- andenergy for egg production(Holm and grationpath that appears to offerthe req- Scott1954, Krapu 1979), and nutrients for uisitesfor survival and breeding. This pat- the growthand development of young. ternis akinto the "facultativemigration'> Proximatecues used by ducksto iden- of Pulliamand Parker(1979) and would tify suitablebreeding areas are unknown, be appropriateif habitatconditions are butare thought to be indirectand related unpredictablefrom year to year. More- stronglyto wetlandfeatures. Empirical over,opportunistic settling minimizes any evidencecomes from studies in whichwet- migrationcosts. Dispersal, such as caused landswere created or markedly improved, by opportunisticsettling, is an important oftenleading to large increases in the num- meansof adjustingto majorchanges in ber of breedingducks (e.g., Hochbaum birds'habitats and is especiallyimportant 1946,Hochbaum and Bossenmaier1972, in spatiallyheterogeneous habitats (Gau- McKnight1974). Although ducks typically threaux1980, 1982). One manifestation of usea varietyof wetlandsduring the breed- opportunismis frequentpioneering, the ing period(Dzubin 1969, Dwyer et al. movementof birdsinto habitats that were 1979),many of whichdo not persistfor unavailableor not occupiedduring the the entirebreeding season, the conditions previousyear. Pioneering has been docu- of wetlandsin springoffer good cues to mentedby anecdotesattributing increases pondconditions in thesubsequent summer in ducknumbers in a localarea as a rapid (Leitchand Kaminski 1985). The custom- responseto improvedhabitat and by arymeasure of wetlandquantity in exten- large-scalesurveys that show a redistri- sive waterfowlstudies is the numberof butionof ducksfrom 1 yearto the next wetlandbasins that contain water during when wetlandconditions in part of the somesurvey period, typically May (Cris- areachange markedly. sey 1969). We hypothesize3 patternsof settling FlexibleSettling behaviorand the ecologicaland species characteristicsassociated with each pat- A mixedpattern is for a birdto home tern:(1) homing, (2) opportunistic settling, to the area used the previousyear, but and (3) flexiblesettling. It couldbe ad- moveon (i.e.,become opportunistic) if that vantageous,however, for differentage or habitatis not suitable.Drought displace- sex classeswithin a speciesto followdif- ment (the overflightof normalbreeding ferentpatterns. areas)is includedhere. The pattern is sim- ilarto the <'iElexiblehoming" to wintering Homing areasdiscussed by Bellroseand Crompton (1970)and can be viewedas a compromise Onepattern is alwaysto home,that is, betweenhoming and opportunisticset- for adultsto returnto the breedingarea tling.The habitat chosen after moving on usedthe previousyear and for yearlings mayenhance survival of the adultbut re- to returnto theirnatal area. Such philo- ducereproductive performance. Drought- patryhas been termed "migrational hom- displacedbirds are reportedto havelow ing" (Sowls1955). Homing can be pre- reproductivesuccesss (Hansen and Mc- dicted if the temporal variabilityof Knight1964, Murdy 1966, Calverley and resourcesis low,but would not be advan- Boag1977, Derksen and Eldridge1980), tageousotherwise (Greenwood and Har- and wetlandsin areas frequentedby vey 1982).Simply if the birds'essentials drought-displacednonbreeding birds are areprovided each year, there is no advan- typicallyless productivethan those used tageto switchingbreeding areas by breedingbirds (Murdy 1966). BREEDINGDISTRIBUTIONS OF DUCKS-Johnson and Grier 9

In NorthAmerica most drought-dis-selects. The theoryof settlingEaresented placedducks are thoughtto movenorth- heresuggests that opportunistic responses ward or northwestward(Hansen and to habitatchange will be associatedwith McKnight1964, Crissey 1969 Smith 1970), the use of variablehabitats and withrel- althoughStoudt (1971) suggested that blue- ativelyr-selected species, and that homing wingedteal may move southward.Evi- will be associatedwith the use of stable dencefor drought displacement arises from habitats(e.g., Wiens1976) and with rel- the following2 sources:(1) local studies in ativelyK-selected species. We hypothesize northernareas that show greater densities that a given settlingpattern or selected of ducksSand sometimesthe presenceof breedinghabitat is withina suiteof life speciesnot usuallyfound there, in years historycharacteristics that can be regard- of droughtin the prairieand parkland ed as relativelyr-selected or K-selectedS (Hansen1960} Hansen and McKnight 1964> whilerecognizing thatother kinds of se- Derksenand Eldridge1980); and (2) ex- lectioncould have causedthese suites to tensivesurveys that show greater propor- covary(Stearns 1977). If we are correct, tionsof ducksin northernareas than in then speciesof duckscould be arrayed prairieand parklandduring years of se- alongan axisthat reflects settling pattern vere prairiedrought (Hansen and Mc- andhabitat selection; this axis would cor- Knight1964, Crissey 1969, Smith 1970). respondto an axisof r- and K-selection. Severalauthors (Patt&rson 1979 Bailey Life History Characteristics 1981,Vickery and Nudds 1984) have done relatedcomparisons, mostly attempting to Ecologistslong have recognizedthat placeduck species along a continuumof manylife historycharacteristics are asso- r- andK-selection. ciatedwith one another, e.g., longevity and fecundity.Efforts have been made to re- latesuites of characteristicsto featuresof METHODS the habitatused by the animals,specifi- Survey Design callycontrasting highly variable (and un- predictable)environments with those more Eachyear ducks and ponds are counted stable(and predictable) (e.g. Pianka1974). in the majorNorth American breeding Comparisonsoften have been cast in terms areas.The surveydesign and datagath- of r- andK-selection (MacArthur and Wil- eringtechniques were described in detail son 1967),commemorating the parame- by D. C. Bowden(Review and evaluation tersof the logisticequation of the growth of May waterfowlbreeding ground sur- of populationssubjected to highlevels of vey, unpubl.rep., U.S. Fish and Wlldl. density-independent(r-selected popula- Serv.,Patuxent, Laurel, Md* 1973) Anon- tions) or density-dependent(K-selected ymous(1977) andMartin et al. (1979). populations)mortality (Stearns 1977, Boyce The region from which samplesare 1984).The introductionof theseparame- drawnconsists of about3.03 millionkmS tershas brought much controversy to the (1.17million sq miles)in thenorth-central discussionof covaryinglife historychar- UnitedStates, the prairieprovinces and acteristics.Pending a resolutionof this NorthwestTerritories of Canada and controversywe will use the expressions,Alaska (Fig. 1). About78-84% of North r-selectionand K-selection, to refertQ pOp- Americanmallards breed within the sur- ulationsthat are influenced relatively more veyed area (Pospahalaet al. 1974;R. S. by factorsthat operate in a density-inde- Pospahala,unpubl. memorandum 1985). pendentmanner and populations more in- Becausethe other 9 specieshave breeding flueneedby density-dependentfactors, re- rangesmore restricted to thatarea (Bell- spectively. rose 1980) even greaterpercentages of Thesettling pattern of a speciesmay be theircontinental populations probably nest relatedto the stabilityof the habitatit in the surveyedarea. WILDLIFEMONOGRAPHS 10 Table 1. Strata used in May waterfowl surveys: name, area, number of transects, and area sampled, 195541.

- No. of Area Area tran- sampled Name (sq miles) sectsa (sq miles) Stratum Kenai-Susitna 2,200 6 40 Nelchina 3,900 10 52 2 Tanana-Kuskokwim 9,300 18 132 3 Yukon Flats 0,800 12 80 4 Innoko River 3,400 7 44 s KoyukukRiver 4,100 10 80 6 Copper River Delta 400 1 20 7 Bristol Bay 9,900 1l 92 8 Yukon Delta 26,600 8 260 9 Seward Peninsula 3,850 4 24 10 Kotzebue 5,350 7 48 11 Old Crow Flats 1,970 3 36 12 Mackenzie River Delta 4,935 4 36 13 Northwest TerritoriesForest Tundra 79,217 8 364 14 Northwest TerritoriesForest Tundra 50,462 2 108 15 Slave River Parklands 57,821 4 94.5 16 Northwest TerritoriesClosed Forest and Forest Tundra 69,033 5 297 17 Northwest TerritoriesPre-Cambrian Edge ,733 3 27 18 North Alberta Parklandsand Closed Forest 63,861 8 688.5 19 AthabascaRiver Delta 1,625 4 31.5 20 Northern SaskatchewanMixed Forest 44,922 1l 382.5 21 Northern SaskatchewanMixed Forest 20,278 6 166.5 22 Northern SaskatchewanPre-Cambrian Shield and Mixed 68,320 6 297 23 Forest 24 Manitoba Pre-CambrianShield, Muskeg, and Mixed Forest 83,794 8 490.5 25 SaskatchewanRiver Delta 6,816 6 117 26 Alberta Parklands 26,448 6 265.5 27 Alberta Prairie ,724 3 144 28 Alberta Prairie 2,890 3 135 29 Alberta Prairie 3,235 4 171 30 SouthernSaskatchewan Mixed Forest (west) 18,570 4 162 31 SouthernSaskatchewan Mixed Forest (east) 21,086 5 135 32 SaskatchewanPrairie 37,911 14 571.5 33 Cypress Hills ,345 6 90 34 SaskatchewanParklands 3,164 5 175.5 35 SaskatchewanParklands 9,044 6 126 36 ManitobaInterlake Region 5,500 3 58.5 37 Manitoba Interlake Region 6,485 4 130.5 38 Manitoba Parklands 5,655 3 54 39 Manitoba Parklands 6,552 5 121.5 40 ManitobaParklands 4,536 4 67.5 41 Montana North River 32,902 7 504 42 Montana South River 40,755 7 364.5 43 North Dakota West River 19,835 5 175.5 44 South Dakota West River 27,299 5 216 45 Drift Prairie and MissouriCoteau (ND) 26,625 7 310.5 46 Drift Prairie and MissouriCoteau (ND) 4,238 8 270 47 Red River Valley (ND) 7,821 6 45 48 MissouriCoteau, James River Valley, and Prairie Coteau 24,587 9 315 (SD) 49 South and Eastern South Dakota 15,830 11 171 Total 1,168,624 8,787.5

-

a Fewer trarnsects than indicatedwere surveyedin some years. BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier ll

Theregion is dividedinto 49 strata(Ta- Table 2. Years in which waterfowl breeding-pair surveys and ble1), ranging in sizefrom 1,036 to 424,400 pond surveys were initiated, by stratum. km2(400-163,861 sq miles).Stratification Strata Breeding pairs Ponds is basedboth on politicalboundaries and 1-11 1957 Not done on geographicaldifferences in duckden- 12-20 1956 Not done sities(Stewart et al. 1958).Anderson and 21-25 1955 Not done Henny(1972) briefly described major wa- 26-40 1955 1955 terfowlhabitats in thesurveyed area. Wa- 41-42 1965 1965 terfowlcounts within the variousstrata 43,45 - 47 1958 1958 wereinitiated in yearsranging from 1955 44,48-49 1959 1959 to 1965(Table 2). We useddata through a Data missing in 1963, 1965, and 1966 for stratum 43. 1981. Withineach stratum, a systematicsam- theaerial survey. Efforts are made to cen- ple of transectlines was drawn. The start- susall duckswithin the boundariesof the ing linewas selected randomly (Martin et sampleunit. al. 1979).Transects are 400-m (0.25-mile) Theratio of birdsof eachspecies count- wideand of variablelength. Aerial counts ed on the groundto the numbercounted of ducksand pondsare conductedeach fromthe air,taken over several segments yearalong each transect, and a subsample withina groupof strata,is termedthe "vis- of portionsof transectsis canvassedon the ibilityrate," although its reciprocalmight groundfor ducks. moreaptly bear that name. Visibility rates are calculatedby speciesand year for AerialCounts groupsof strata.For the northernstrata, Fixed-wingaircraft are usedin the ae- whereground surveys are not conducted, rialcounts and are flownalong the tran- visibilityrates are takenas the long-term sectsat a heightof 30-46m (100-150feet) averagefrom strata where they are deter- aboveground and at a groundspeed not mined. to exceed169 km/hr (105 miles/hr). Dates of the survey,in May,are standardized to Pond Counts conformto geographicaldifferences in phenologicalevents. Time of the survey DuringMay aerial surveys, a countof andflight conditions are also prescribed. pondbasins containing water is madein Theaircraft pilot and another observer strata26-49. Included in the tallyare nat- recordall identifiableducks within tran- uralwetlands of typesIII, IV, and V (Shaw sect boundaries(200 m = 0.125 mile on andFredine 1956), which are basinsthat eitherside of the flightline). Unidentified areinundated seasonally, semipermanent- ducksare not recorded.Factors influenc- ly, or permanently,respectively. Streams ing aerialcounts were described by Mar- andrivers that meander across the transect tinsonand Kaczynski (1967). arecounted each time they enter the tran- sect.Also included in the pondcount are GroundCounts man-madewetlands such as stockdams, dugouts,and largeditches that maintain To estimatethe proportionof ducks waterinto summer. Water areas excluded presentthat are seen and identified by the from the countare temporarywetlands aerialsurvey crew, ground counts are con- and sheetwater (expected to persist93 ductedon certainsegments of sometran- weeks),roadside or borrowditches, and sects.Ground counts are not madein the muskegareas. Water bodies in the north- northernareas (strata 1-25) because of in- ernstrata (1-25) are more permanent than accessibility.Crews of 2-4 biologistsmake thosein prairieand parklandstrata and, groundsurveys, preferably the day after therefore,are not counted. Pond counts for 12 WILDLIFEMONOGRAPHS strata26-49 were initiated at various times Breeding-season Distribution, duringthe 1955-65period (Table 2). Wintering Areas, and Migration Corridors Computations Breeding-seasondistributions of ducks Countsof ducks(by species)and areas wereexamined by computingthe average surveyedare summedacross all transects adjusteddensity of each speciesfor each withina stratum,and theirratio, multi- of the 49 strata(averaged across all years pliedby the visibilityrate, is termedthe thatstrata were surveyed). Densities were adjusteddensity. Adjusted density, mul- rankedand plotted on mapsaccording to tipliedby areaof the stratum,yields the thequartile in whichthey fell: highest 12, estimatedpopulation in a stratum.The sum secondhighest 12, thirdhighest 12, and of thesevalues across all stratais the es- lowest13. Plottingaccording to quartiles timatedpopulation within the surveyed facilitatedthe comparisonof distributions area.A similarmethod is usedto deter- amongspecies despite differences in av- minepond count, except that no correction eragedensity. for visibilityis made. The directionfrom which ducksap- proachtheir potential breeding ground was deteminedby theirwintering area and mi- Analysis of Pond Counts grationroute. The pathsthey traverseon Pondcounts provide the only indexto theirspring trips converge within rather breedinghabitat available for mostof the well-definedmigration corridors (Bellrose surveyedarea. Counts in a stratumfor a 1980).Most are generallynorth-south or particularyear are dividedby the area northwest-southeastin orientation,but surveyedwithin that stratumto produce lateralmovements are fairly common. anestimate of ponddensity. Pond densities Patternsof use of migrationcorridors variedby yearand even more so by stra- for the 10 specieswere determined from tum. informationsummarized by Bellrose Becausepond counts in differentstrata (1980).We estimatedthe proportionsof wereinitiated at varioustimes during the membersof a particularspecies that return 1955-65 period,the data were unbal- to theirmajor breeding grounds through anced,as not all stratawere surveyed in (a)southern or southeasterncorridors, en- allyears. As a result,the usual averages by teringthrough the Dakotas,and (b) west- yearor by stratummay be misleading.To ern or southwesterncorridors, entering overcomeproblems resulting from this im- througha northwesternMontana or A1- balance,we calculatedLeast Squares (LS) bertaportal. The importance of migration means,in additionto the usualmeans. LS corridorsto ouranalysis is thatthey affect meansare basedon an assumedlinear theway the population may be distributed model,in thiscase representing pond den- relativeto the habitat;for example, a pin- sity as an additivecombination of effects tail migratingdirectly from California to due to stratumand to year.LS meansby Albertais lesslikely to be affectedby wet- year,as an example, are calculated by add- landconditions in South Dakota than would ing the interceptof the linearequation, one returningfrom a winteringarea in the effect for the desiredyear, and the Texas. averageof allstratum effects. This method is employedby the SASGLM procedure Relations of Ducks with Local (SASInstitute, Inc. 1982). The linear mod- and Continental Ponds el of ponddensities, with 26 year terms and 23 stratumterms, produced a coeffl- To determinehow localwetland con- cientof determinationR2 = 0.75 for the ditionsaffected the distributionof ducks, 601 observations. we calculatedthe correlationcoefficient BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier 13 betweenthe densityof each speciesand aK-selected attribute). The relative extent thedensity of pondsamong years within of opportunismin the responseto habitat eachtransect and stratum. This coeff:icient change was indexedby the mediancor- indicateshow duck numbersand pond relationbetween duck densities and pond numbersvaried together from year to year densities;high values here suggest an ad- withina transect.For analysis,we aver- aptationto unpredictableenvironments agedcorrelation coefflcients over all tran- (hypothetically,an r-selectedattribute). sectswithin a stratum.We calculatedme- We predictthat these 2 measuresare re- dians,means, and weighted means, where latedto eachother. the weightwas the numberof yearson whichthe correlation coefficient was based. RESULTS All 3 measuresperformed similarly, and mediansare used in thefollowing analyses. Dynamics of the Wetland Habitat Duckdistributions may depend not only Averagepond densitiesvaried widely onlocal habitat conditions but also on con- amongthe strata.Highest densities were ditionselsewhere in the breedingrange, observedin southwesternManitoba and especiallyduring drought. To assessthis southeasternSaskatchewan and lowest effect,we calculateda measureof total and western pondsfor the pond-surveyedarea (strata densitiesalong the eastern wasthe edgesof theDakotas and in Montana(Fig. 26-49). Forany year, this measure 2). ponddensity for a stratummultiplied by Variationamong years in pondcounts thearea of thatstratum, and then summed parkland overall strata.We usedonly the datafor wasexpected, as the prairieand whichponds areais wellknown for recurringdroughts 1966-81, the periodduring and interveningwet periods.LS means were countedthroughout strata 26-49. were lower than ordinarymeans, espe- Correlationcoefflcients between the den- Surveys sityof eachspecies and the total pond count ciallyin thefirst few years (Fig. 3). werecalculated for each speciesover all werebegun during a verywet period; pond yearswithin a transectand stratum.The densitiesnever again reached the numbers acrosstran- attainedin 1955.After a declinefrom 1955 medianof thesecoefflcients to 1959,densities of pondshave oscillated sectswithin a stratumwas used as a mea- (10/sq sureof howa speciesin thatstratum varied arounda mean of about4/km2 in relationto totalponds. mile).

Relations to Life History Breeding Distributions Characteristics of the Species We attemptedto measureseveral life Mallard. Mallardsattained highest historycharacteristics for each species. We densitiesin prairieand parklandof the quantifiedthe stability or predictability of southernprairie provinces and in theCop- eachspecies' habitat by developingan in- perRiver and Athabasca River deltas (Fig. dex to the permanenceof basinwetlands 4A).Within the prairie and parkland area, usedby the speciesin an extensivestudy densitiesdecreased with distancefrom by Stewartand Kantrud(1973) in North strataof highestdensities. Dakota.Cropland ponds were scored2, Mostmallards winter in the south-cen- seasonalponds were scored 3, and semi- tralUnited States. We estimatefrom in- permanentponds and reservoirswere formationin Bellrose(1980) that about 70% scored4. Theweighted average was mul- of winteringmallards return in the spring tipliedby -0.1; high negativevalues of throughthe southernentry, or portal(the the resultingindex thus suggest consider- Dakotas),versus about 30% from Califor- ableuse of stablehabitats (hypothetically, nia and the PacificNorthwest through the second quartile

14 WILDLIFEMONOGRAPHS

Ea hi9hest quartiie a3 third quartile ;1 lowest quartile

Fig. 2. Pond density by stratum, 1955-81. westernportal. Thus most mallardsini- of mallardsdisplaced by droughtin the tiallyare exposed to wetlandconditions in southernprairie-parkland breeding range. the southernportion of the surveyedarea. Femalemallards, especially those that Mallarddensities varied with local pond weresuccessful in hatchinga clutchthe counts,the mediancorrelation coefficient being0.417 (Table 3). Highestcorrelations werein Albertaprairie and Saskatchewan 10- | *-*---* CUSTOMARYESTIMATES parkland,but consistently high values were \ LEASTSQUARES calculatedfor the southernportion of the prairieprovinces, where mallardswere verycommon, and in the Dakotas(except the2 easternmoststrata), where they were notas common(Fig. 4B). Mallarddensities covaried positively withtotal ponds in muchof theirprimary .00 t:0.60medi>0.4O,<0 an xo,zo,ca.40r *60l mediLJ c0*20an L n>0*2O,<0 >0*40edi an r 40

MALLARD

ct 3

c

o

ct o

l

g median r r m median r - -median median r ce le _ >-0^20,

cn WILDLIFEMONOGRAPHS

16 Table 3. Medians and interquartile ranges of correlation coef- wallsarriving in springare firstexposed ficients between duck densities and local pond densities, by to wetlandconditions near the southern species, 1955-81. portal. Correlationswith pond densities Numbersof gadwallswere not as closely Interquartile correlatedwith local pond numbers as were S. pecles Median range mallards(Table 3). Correlation coefflcients Mallard 0.417 0.159 werehighest in the centralDakotas (Fig. Gadwall 0.228 0.205 5B),and fairlyhigh in moststrata where American wigeon 0.143 0.208 Green-winged teal 0.199 0.181 gadwallswere common. Blue-winged teal 0.390 0.172 Gadwalldensities were associated pos- Northern shoveler 0.388 0.200 itivelywith total ponds in only5 stratain Northern pintail 0.547 0.157 thesouthern prairie provinces and in east- Redhead 0.154 0.220 (Fig.5C). Negative cor- Canvasback 0.192 0.191 ernNorth Dakota Scaup 0.145 0.182 relationswere found in manystrata in cen- . tral Canadaas well as in westernSouth Dakota,suggesting that these stratare- previousyear, home at fairlyhigh rates ceivedrought-displaced gadwalls. (Table4). Males home at much lower rates, Gadwalls,especially successfully breed- presumablybecause pairing occurs during ingfemales, have fairly strong homing ten- winteror spring migration, and the female dencies(Table 4). Often, however, they do leads the male to the breedingground not homeuntil their second year, the age (McKinney1965). at whichmany gadwalls first breed (J. T. We concludethat mallards tend to re- Lokemoen,H. F. Duebbert,and D. E. turnto breeding areas in the northern states Sharp,Reproductive strategies of prairie andsouthern prairie provinces. When wet- mallards,gadwalls, and blue-winged teal, landconditions in thesestrata are not fa- unpubl.manuscript, U.S. Fish and Wildl. vorable,more mallards than usualshow Serv.,Jamestown, N.D.). up in northernand northeastern strata. If We concludethat most successfulfe- successfullybreeding female mallards malegadwalls return to areasused the pre- homeat a high rate,dispersing females vious year. Younggadwalls and unsuc- likelyare mostlyyearling birds or unsuc- cessfulbreeders likely settle in responseto cessfulbreeders. Because correlations with wetlandconditions in theirprimary range, pondsare higher in the primaryrange of especiallyin the southernpart, judging the mallardthan in the southernportion, fromthe pattern of correlationcoefflcients we suggestthat northward-migrating mal- betweengadwall densities and wetland lardsdo not fill the habitatas they en- counts.When many ponds are dry in the counterit fromthe south,but insteadfill prairies,more gadwalls are counted in cen- the corearea first. traland northern Canada. Gadwall.- Thisspecies had a primary American Wigeon. This species rangecentered southward of themallard's, reachedhigh densitiesin the southern andwas most common in the centralDa- prairieprovinces but wasoften abundant kotasand the southernprairie provinces in thenorth as well (Fig. 6A). Wigeon were (Fig.5A). The gadwallwas less common uncommonsouth of Canada.Within prai- than mallardsin the northernstrata. As rie andparkland habitats, densities stead- with the mallard,highest density strata ily declinedaway from the area of highest weresurrounded by stratawith moderate density,much like the mallardand gad- density,and densitieswere lower farther wall.Wigeon winter throughout the south- away. ernUnited States, along the YucatanPen- Gadwallswinter mostly in Louisianaand insulaof Mexico,and in the Caribbean. Texas. Perhaps90% of spring-arrivingNumbers of springarrivals using the south- gadwallspass through the southernportal erncorridors probably slightly outnumber andonly 10% come through western Mon- thoseentering from the southwest(Bell- tanaor Alberta(Bellrose 1980). Most gad- rose1980). An additional few may migrate BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier 17

Table 4. Reported return rates for 10 species of ducks, by age and sex,a 1955-81.

Adult Yearling Female Male Female Male Species % n % n % n % n Study

Mallard 13+ 15 Sb 20 Ob 13 Sowls (1955) 42 24 Coulter and Miller (1968) 46 113 5+ 140 Doty and Lee (1974) 58 19 14 14 Bishop et al. (1978) 60c 68 31 48 Lokemoen et al. (unpubl. manuscript)d 26 119 Lokemoen et al. (unpubl. manuscript)d Gadwall 37 16 12 8 Ob 9 Sowls (1955) 29 52 Gates (1962) 66C 90 21 164 Lokemoen et al. (unpubl. manuscript)d 40e 101 Lokemoen et al. (unpubl. manuscript)d 41 54 9 236 7 28 2 42 Blohm (1978, 1979) American 38 21 9 11 33 3 0 6 R. A. Wishart (pers. commun.) wigeon Green-winged No information No information teal Ob 1 9 Sowls ( 1955) Blue-winged 14 Ob 30 teal o 58 Ob 200 McHenry (1971) 5 16 1+ 226 Lokemoen et al. (unpubl. 143 manuscript)d Ob 12 SOW1S ( 1955) Northern 42 19 8b 12 shoveler 15 20 11 19 3 116 1 134 Poston (1969) Northern 39 44 13b 1 15 2b 132 Sowls (1955) pintail Redhead 70 23 15 92 Johnson (1978) 70 74 Alliston (1979) Canvasback 75 12 50 Bellrose (1980) 24 76 75 101 1 206 Anderson (1985) 52 27 Lesser 330 10 626 0 557 Trauger (1971) 20 34 scaup 58 3 12 76 4 91 A. D. Afton (pers. commun.) 66 351 9 49 a Reportedrates are minimalestimates of homingrates, as they do not accountfor mortalitybetween seasons or for birdsthat may have homed but were not seen. b Captive-reared. c Successfullynested previousyear. dJ. T. Lokemoen,H. F. Duebbert, and D. E. Sharp, Reproductivestrategies of prairie mallards,gadwalls, and blue-wingedteal, unpubl. manuscript,U.S. Fish and Wildl. Serv.,Jamestown, N.D. e Did not successfullynest previousyear. to breedinggrounds in the parklandsdi- stratawith high densitiesof the species, rectlyfrom wintering locations in theMid- exceptthat wigeonwere commonin 3 dle AtlanticStates (Bellrose 1980). parklandstrata in Albertaand Saskatch- Wigeondensities generally did notcor- ewan,but correlationsthere were low. relateclosely with local pond densities (Ta- Wigeondensities correlated positively ble 3). The highestcorrelations between withtotal ponds in manystrata of thecen- wigeonsand localponds were in western tral portionof its range(Fig. 6C). They SouthDakota and in thesouthwestern part werenegatively associated in manystrata, of theprairie provinces (Fig. 6B). This pat- particularlythose on the easternedge of ternis consistentwith the largemigration its rangeand in the north. of wigeon throughthe westernportal. The few appreciablecorrelations be- Higher correlationcoefficients between tweenwigeon numbers and ponds were in wigeonand pondswere associatedwith the southwesternportion of their range, B h i ghest quart i l e E th i rd quart i l e >o 60 emk >0 medi<0.20 median.20 , anc01:: r r40 I 1}>° g >0*20,<0,40median.40medi 1 1an > r r- medianmedi0.20<-0.20 , an c0 rr* 20

oo

GADWALL

o

o c:

;E

u)

3second quartile 1 lowest quartile >0 40 <0 60

Fig. 5. (A) Average species density by stratum,195541; (B) correlationbetween species density and local pond counts, 1955-81; and (C) correlationbetween species density and total pond counts11955-81. = highest quartile third quarti le }1 0.60 - 1 >0.20t<0.40

5 : lICAN WIGEON

- r -

4

c

o

v: o

- t

ral

Ce median r median r 1 >-0.20,c0.20 >0.40 median r as £ median <-0.20 E >0,40,<0.60 g <0.20 # V V t [2 lowest guartile _ * ea m second quartile 0 between species pond counts, 1955-81; and (C) correlation between species density and local by stratum, 1955-81; (B) correlation Fig. 6. (A) Average species density density and total pond counts, 1955-81. Q . 20 <0 .40 >0 .40

o GREEN-WINGEDTEAL

i r

o

o C)

median r ] median r Omedian r 2 h i ghest quart i le 2 th i rd quart; l e g >o bo - - >-0.20, <0Z20 Smedian r median r median r 3 ssond quarti le C1 lowest warti le m medianXQ40, C0.60 r <0.20 g >0.20,c0.40 <-0.20

Fig. 7. (A) Average species density by stratum, 1955-81; (B) correlationbetween species density and local pond counts, 1955-81; and (C) correlationbetween species density and total pond counts, 1955-81. BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier 21 mostlysouth of the areaof highestden- wingedteal move to northernstrata re- sities.When total pond numbers were high, gardlessof pondconditions to the south. wigeondensities were depressed not only Blue-winged Teal.-Blue-winged teal in manystrata in the north,but also in hada distributionmuch like thatof gad- severalalong the easternedge, suggesting wall,with highest densities mostly in the droughtdisplacement both northward and prairieand parklandof Canadaand the eastward.The limitedevidence suggests Dakotas(Fig. 8A), and was uncommon in fairlystrong homing by thespecies (Table the far-northernstrata. 4). We suggestthat wigeon arrive on their Mostblue-winged teal winter in Central breedinggrounds from the southand the Americaor northern South America. Mod- southwest;many remain in stratainitially est numbersare foundin Mexicoand in encounteredif pondconditions there are the GulfCoast States (Bellrose 1980). A1- satisfactory(thus accounting for the high thougha few mayenter the majorbreed- correlationswith ponds there); otherwise, ing areadirectly from the east,almost all theyoverfly to the northand east. returnin the springthrough the southern Green-upingedTeal. Thisspecies had portaland are first exposed to wetlandsin a distributioncentered farther north than the Dakotas. the otherdabbling ducks. Only 4 of the Densitiesof thisspecies varied markedly stratawith highestdensities were in the withlocal pond counts (Table 3). Consis- southernprairie provinces (Fig. 7A). tentlyhigh correlations were found in most Green-wingedteal winter over a broad stratawhere the species was common (Fig. rangein thesouthern United States and in 8B).Exceptions were in severalstrata along Mexico.About 75So of winteringbirds re- thenorthern and eastern edges of thecore turnto themajor breeding grounds via the areathat had large numbers of blue-winged southerncorridors. Most of therest winter teal, but teal numberswere not strongly in the westernstates or westernMexico associatedwith pond densities. andmove into Montana, Alberta, or north- Blue-wingedteal densitieswere posi- ernbreeding areas through a westernpor- tivelyrelated to totalponds in the eastern tal (Bellrose1980). portionof the southernprairie provinces, Thisspecies exhibited modest but con- northernMontana, all of NorthDakota, sistentcorrelations between its density and andwestern South Dakota (Fig. 8C). Neg- localpond density except in the eastern ative correlationswere detectedin the andnorthern portions of thearea in which northernstrata within its rangeand in the pondswere surveyed (Fig. 7B; Table3). southernAlberta prairie, suggesting these Correlationstended to be high in the strataas ultimatelocations of drought-dis- southwesternstrata (especially eastern placedbirds. Blue-winged teal numbers Montanaand westernSouth Dakota), al- weregreater in stratain Alaskawhen the thoughthe specieswas uncommon there. totalnumber of pondswas low (Hansen Green-wingedteal populationsdis- 1960).Stoudt (1971) suggested drought playedpositive associations with total ponds displacementalso to the south. throughoutmuch of the southernprairie Breedingblue-winged teal are not provinces,in northernMontana, and in philopatric(Table 4). In fact, they are con- easternNorth Dakota (Fig. 7C). Negative sideredstrong pioneers; Bellrose (1980) re- correlationsoccurred in many northern portedseveral instances in whichteal nest- strata. edwell outside their normal breeding range Informationon homingtendencies for in responseto newlyflooded habitat. green-wingedteal is lacking(Table 4). We Althoughnot conclusive,the fact that suggestthat the majorityof birdsarriving correlationswith pond counts were more in springfrom the southsettle in fairly relatedto averagedensity of blue-winged goodnumbers in southernstrata, if pond tealthan to latitudesuggests that the birds conditionsthere are favorable. Otherwise do notsimply fill their breeding habitat as they proceedfarther north. Many green- theyencounter it, butoccupy the primary >0.40,<0.St>0medi 60 anau wnr __an <0.20mediann tn - G?Y->0.20,c0.40r =,Fn . an^_ ;

BLUE-WINGEDTEAL

r . / U 1 r N7 t1 . gm

l O O

t f :) 1 '\ S W-- ) > I 't b

8 2

r OmFian r mmeelan r Emmedian r median V* V >-0w20,<0*20 E i rd quart i l e ZX X B h i ghest quart i l e th , median r , _ <-0<20 - second quartile :1 lowest quartile

1955-81- and (C) correlation between species (B) correlation between species density and local pond counts, Fig. 8. (A) Average species density by stratum, 1955-81; density and total pond counts, 1955-81. t guarti le# =1>0.60lef u X O.4Q,<0.60>0>2o,c0*40- r L -- § c0,20YSq ' >0 3 e 40>0.20,<0.40 I4 I ; >-0*20,<0*20, c-0.20 / e_ C:) a

NORTHERN SHOVELER

>> e B _> C w

9- k Re 1g,90 81>k ,

, /}f:tR't g /) :Qtt Cc

X 7 ' J L

O ' - - I \ / j t -;

CO o r lTedian r > * Mdmedian r /median r medlan 2 hi ghest quart i le E thi rd c

_ r ; E_ median r § -median r median r median loifest ticg sec ^ quar t 1 le -

counts, 1955-81; and (C) correlation between species Fig. 9. (A) Average species density by stratum, 1955-81; (B) correlation between species density and local pond density and total pond counts, 1955-81.

Gd 24 WILDLIFEMONOGRAPHS areafirst. If conditionsthere are not fa- GulfCoast (Bellrose 1980). It appearsthat vorable,they move northward, or possibly onlyabout 40% of springmigrants use the backsouth. southernportal to the breedinggrounds; Northern Shoveler.-Shovelers reached the remainderuse the westernportal or highestdensities mostly in the prairieand proceeddirectly to far-northernbreeding parklandstrata of the southernprairie grounds. provincesand central North Dakota (Fig. Densitiesof this speciescovaried with 9A).Average densities tended to decline localpond densities in the surveyedarea awayfrom the primarybreeding area. moreclosely than any other species (Table Shovelerswinter mostly in California, 3). Correlationswere consistent through- the west coastof Mexico,and the Gulf out thatarea (Fig. 10B), but tended to be Coastof Texasand Louisiana(Bellrose largerwhere pintails were most common. 1980).Nearly 60% of them probablyre- Correlationswere high also in westernand turnin springthrough the southern portal, centralSouth Dakota,although pintail and most of the rest enter throughthe densitiesthere were low. westernportal. The pintailexhibited very strong posi- Densitiesof thisspecies varied markedly tivecorrelations with total ponds through- in responseto localpond numbers (Table out the southernpart of its range(Fig. 3). Correlationswere greatest where shov- 10C).In the northernstrata, associations elerswere common, except in strataalong wereclearly negative. thenortheastern edge of theprimary area, Pintailsmay home at a highrate when whereshovelers were common but did not habitatconditions are stable (Sowls 1955, correlatehighly with ponds (Fig. 9B). Bellrose1980). Henny (1973)suggested Therewas a cleardistinction between thatpintails that overfly to the northin 1 the southernpart of the shoveler'srange, year are likelyto returnthere the next, whereit tendedto correlatepositively with even if wetlandconditions in the south total ponds,and the northernportion, improve.Nonetheless, pintails generally wherethe reverse held (Fig. 9C). Drought- prefertemporary wetlands (Smith 1970, displacedbirds seem to settlenorthwest, Stewartand Kantrud1973); therefore, north,and northeast of the primaryarea. hominglikely occurs at a lowrate. In fact, Althoughshovelers home at a fairlyhigh pintailshave responded opportunistically rate(Table 4), they respondpositively to to newlyavailable wetlands (Stoudt 1971, pondconditions, particularly in thesouth- Hochbaumand Bossenmaier1972) and ernand western part of thesurveyed area. appearto be highlysensitive to drought The patternof correlationsis consistent conditions(Crissey 1969, Smith 1970, with the idea that thereare 2 pathways Henny1973, Derksen and Eldridge 1980). intothe breeding range, one from the south, A reasonablehypothesis is thatpintails en- the otherfrom the west. Birdsarriving teringalong the southern corridor respond from eitherdirection may settle in the initiallyto wetlandconditions in SouthDa- southernor westernstrata, if pondsthere kota,next to conditionsin NorthDakota, areadequate; otherwise, they move on to andeventually proceed northward. Birds the northwest,north, or northeast. arrivingfrom the southwest are influenced Northern Pintail. Pintailsreached initiallyby wetlandconditions in south- very high densitiesin severalnorthern easternAlberta and southwesternSas- strataas well as in the southwesternpart katchewan.If conditionsthere are not suit- of theprairie provinces (Fig. lOA). Within able,they move northward. Many pintails the prairieand parkland region, densities migratedirectly to the far north, bypassing tendedto declineaway from the primary the prairiesand parklands entirely. breedingarea. Redhead.-This divingduck had high- The pintailis the mostnumerous win- estdensities in the parklandof the prairie tering speciesin California.Additional provinces,the mixedforest of southern pintailswinter in Mexicoand alongthe Saskatchewan,and the centralDakotas B h i ghest quart i l e th i r d quar SAwiN)WX, - el21>0,40,<0.60 CE median>0.60XY ,8 W>0*20, r / X LoZc0.20 B<0k median 40> -\ :D >0.20,<0.40 skr >0.40 %C.3 _ median _ Ap > -° _C r208 <-0.20median ,s <° Uz2° nm er a

NORTHERN PINTAIL r r S a. s X a

m : T n ^ t: s --t2/ * -d C ^s e . _k . ^ } ) - O W; . fi { 1 ti [ > ' 4 Z

__, f , W W- e

/ X T J 5 ->

( I t ,

/ I I L __ __ \ / o

> t^l ;median r w/median r ;Uitmedian r _ median

E second quarti le g lowest qux e

Fig. 10. (A) Average species density by stratum, 1955-81; (e3) correlation between species density and local pond counts, 1955-81; and (C) correlation between species density and total pond counts, 1955-81. B highest quarti le thi rd quarti [e gqt wo.40,c0.60median £ 1 mediancQ.20 : r 1g >0.20,<0*40| medlan £ medianc-0.20 r

- S

REDHEAD

- - - ; r

o

o

Ct median r median r ,amedian r I Imedian r E251 >0 60 EJ >0 20 co.40 L-0.20

w . S second quarti le 2 lowest quartfle

Fig. 11. (A) Average species density by stratum, 1955-81; (B) correlation between species density and local pond counts, 1955-81; and (C) correlation between species density and total pond counts, 1955-81. BREEDINGDISTRIBUTIONS OF DUCKS-Johnson and Grier 27

(Fig.11A). Average densities tended to de- mainderare distributedin southernstates cline steadilywith distancefrom these (Bellrose1980). Roughly 75% of returning areas. canvasbacksenter the breedingground via Mostredheads winter along the coast of the southernportal or a southeasternone the Gulfof Mexico(Bellrose 1980). About in the easternDakotas; about 25% use the 90%of themreturn to the majorbreeding westerncorridor or fly directlyto areasin groundalong southern corridors. the north. Correlationsbetween densities of this Correlationswith local pond densities speciesand localponds were mostlylow had a medianof 0.192 (Table3) and were (Table3), includingstrata with highest consistentacross much of the surveyed redheaddensities (Fig. 11B). area,including some stratain which can- Redheadsresponded positively to total vasbackswere not common(Fig. 12B).In pondsin onlya few stratain the southern Manitoba,correlations were low, despite prairie provincesand in south-centralhigh densitiesof the species. In general, NorthDakota (Fig. 11C).Negative cor- there was no relationbetween correlation relationswere calculated for several strata coefficientsand averagecanvasback den- in the northernpart of its rangeand in sities. east-centralMontana. In spiteof low cor- Canvasbackdensities were positively as- relationswith localwetland counts, red- sociatedwith totalponds mostly in central headswere less commonin these strata Alberta, southern Saskatchewan, and whentotal pond numbers were high. southwesternManitoba (Fig. 12C). Cor- Redheadsappear to be stronglyphilo- relationswere negativein severalstrata of patric(Table 4), whichmay also contrib- the northand in centralCanada. uteto theweakness of therelation between Canvasbacksare strongly philopatric local pond countsand numbersof red- (Table4). To someextent, canvasbacks ap- heads.Hochbaum (1946) suggested that pearto occupyhabitat as it is encountered, the redheadas a divingduck, was a poor althoughstrong homing tends to maintain pioneer.McKnight (1974) disagreed, cit- populationsin the primaryportion of their ing a rapidbuildup of redheadsin newly breedingrange. Displacement is primarily impoundedmarshland in Utah. to stratain the northand northwest(Fig. Redheaddensities did notrelate closely 12C). Dry conditionsin the prairie and to localpond counts, possibly because they parklandtended to increase canvasback prefer the more stable semipermanentnumbers in the northernand northwestern wetlands,which lluctuate less in numbers strata.Hansen (1960) also reportedover- than do temporaryand seasonalponds. flightby this species. Moreover,the strength of associationswith Lesser Scaup. The distribution of pondnumbers did not display a latitudinal scaup was very differentfrom the other gradient(Fig. 11B).Thus, some redheads species,with highestdensities attained in seem to respondto pondconditions en- the northand in 2 stratamostly in Mani- counteredon theirnorthward flight, but toba (Fig. 13A). Scaup were uncommon most migrateto their primaryrange. If southof Canada.Although aerial observers wetlandconditions there are not suitable, cannotdistinguish lesser scaup from great- theycontinue northward (see Hansen 1960) er scaup (Aythyamarila), it appearsthat or retreatinto Montana. greaterscaup predominate in coastalstrata Canvasback.-Thecanvasback was most of Alaskaand the NorthwestTerritories commonin the parklandregion and in (strata1, 8, 9, 10, 11, and 13), whereas severalnorthern strata (Fig. 12A).Densi- most of the scaup (an estimated82%) in tiesdeclined with distance away from the the interiorstrata are lesserscaup (Bellrose primarybreeding area. 1980, Boyd 1983). Nonbreeding lesser Abouthalf of thecanvasbacks winter in scaupmay summerfar southof their cus- easternstates and abouta fourthof the tomary breeding areas (Wetmore1920); populationwinters in the west. The re- therefore,the distributionsuggested by the { 6 \ f 5 /--m: )' ' ' 1. \. =

x

CANVASBACK

X _ 0 1 P f , ii | ;D ta

o t ''; a- -

- }' I l'_L _ 8 / ' 8 7 i T j < median r riSr median r r--) rza median r >0.40 L__ >-0.20,<0.20 {sn median r >0.20,c0.40 ti29 r7wA t22: >0.60 223 L23 third quartile r B highest quartile Xmedian r || median lmedian r t _ <-0.20 re=n median r X w iiE3 >0.20,<0.40 r | >0.40,<0.60 L__l <0.20 _] lowest quartile tELj m second quartile L

(C) correlation between species pond counts, 1955-81; and species density and local (B) correlation between density by stratum, 1955-81; Fig. 12. (A) Average species 1955-81 density and total pond counts, j| Xl 8r LFm J 8median J | f ] lr rL -,-l medlan_, \ rj median} > f r j - t-median t r /

-SCAUP

- L;J - t U z Q

Ct

w _ 3 ,_ o

U) o

C l

-

ew C7

2 hi ghest quart i le z thi rd quart i le g median r 3 >0 .20, <0 .40 g medi an r a >-0 "20, <0 . 20

X second quart i le u towest quart i le 1 >0.40, <0 .60 Lg <0. 20 [g3 >0 .20, <0 .40 _ <-0. 20 .

Fig. 13. (A) Average species density by stratum, 1955-81; (B) correlation between species density and local pond countsl 1955-81; and (C) correlation between species density and total pond counts, 1955-81. 30 WILDLIFEMONOGRAPHS aerialsurveys may not be the truebreed- nearlyall gadwalls, blue-winged teal, red- ingdistribution. Also, lesser scaup migrate heads, and scaup enter their primary latein springand may be counted in strata breedingranges from the south. These cor- southof theirultimate destination (A. D. ridorsare takenby about75% of green- Afton,pers. commun.). wingedteal, mallards,and canvasbacks, Lesserscaup winter mostly along the and by about60% of Americanwigeons Atlanticand Gulf coasts(Bellrose 1980). and northernshovelers. Only about 40% The fractionof them returningthrough of returningnorthern pintails use the thesouthern portal to thebreeding grounds southerncorridor. probablyexceeds 90%. Greater scaup win- Overall,species densities tended to cor- ter alongthe coastsof both oceans,and relatepositively with pond densities in the sparinglyin the Gulfof Mexicoand the sametransect. This relation was strongest GreatLakes (Bellrose 1980). Their pri- amongpintails, but also pronouncedin marymigration pathways are either along mallards,blue-winged teal, and shovelers. the PacificCoast or northwestwardacross Threespecies, American wigeon, north- the GreatLakes. ernshoveler, and northern pintail, tended Correlationsbetween scaup densities and to havehigher correlations with ponds in countsof localponds were low (Table3). thosestrata where each species was most The highestvalues were calculated for 9 abundant.These species also were the ones stratain the southernpart of thesurveyed thatmade less use of the southernportal area,none of whichhad large numbers of to the breedingrange. scaup(Fig. 13B).This pattern of correla- Pintaildensities were moreclosely re- tions may reflectthe presenceof non- latedto totalponds than the other species, breedingscaup in the southernpart of the with strongpositive correlations in the area,alluded to above. southernprairie provinces and in the north- Lesserscaup tended to correlateposi- centralstates, and negative correlations in tivelywith totalponds in manystrata in thenorthern strata. The shoveler and mal- central and southernCanada and the lardexhibited similar, but slightly weaker, UnitedStates (Fig. 13C), including many relations.Several species gadwall,wi- stratain whichthe specieswas not com- geon,blue-winged teal, redhead, and less- mon. Negativecorrelations were found er scaup displayednegative correlations onlyin 3 northernstrata and 1 in Mani- in 1 or morestrata in the southernpart of toba,strata that contained large numbers theirprimary range, suggesting that these of scaupin mostyears (Fig. 13A). stratamay be the destinationof drought- Lesserscaup are stronglyphilopatric displacedbirds. (Table4) andonly somewhat affected by Homingrates were highest for the red- localwetland conditions. Nonetheless, the head,canvasback, and lesser scaup; mod- stratashowing highest correlations with erately high for the mallard,gadwall, pondcounts are south of theprimary range Americanwigeon, and northern shoveler; of the species(Fig. 13A),indicating that andlowest for the blue-winged teal. Little somebirds, perhaps nonbreeders, fill the is knownabout homing by thepintail, and habitatas they encounter it. Displacement no informationis availablefor the green- appearsto be to thenorthwest and possibly wingedteal. to the east. General Patterns.-These 10 species Relations with Life History havehighest densities mostly in theprairie Characteristics and parklandof southernCanada. Gad- wall and blue-wingedteal had distribu- Valuesof the 2 indicesof life history tionscentered somewhat more to thesouth; characteristicsand their average are pre- Americanwigeon, green-winged teal, pin- sentedin Table5. As predicted,the neg- tail, and scauphad morenorthern ones; ativeof the firstindex and the other index and the canvasbackwas centerednorth- variedtogether (r = 0.683,P = 0.03).Be- ward and eastward.During migration, causethese indices covary, it is reasonable BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier 31

to combinethem into a single measure Table 5. Indices to life history characteristics: average per- manence values of ponds used by species, median correlation correspondingto the r- andK-continuum. coefficient with pond densities, and average index. Averageindices indicate that the pintail, mallard,blue-winged teal, and shoveler Pondper- Median manence correlation showthe most features associated with un- Species indexa withponds Average predictableenvironments (i.e., are the most Northernpintail -0.296 0.547 0.126 r-selected).Gadwall, American wigeon, Mallard -0.310 0.417 0.053 and green-wingedteal are intermediate,Blue-winged teal -0.325 0.390 0.032 andthe divingducks are mostassociated Northernshoveler -0.340 0.388 0.024 withpredictable environments (i.e., are to- Gadwall -0.336 0.228 -0.054 Green-wingedteal -0.310 0.199 -0.056 wardthe K-selectedend of the contin- Americanwigeon -0.334 0.143 -0.100 uum). Canvasback -0.393 0.192 -0.100 Redhead -0.384 0.154 -0.115 DISCUSSION Lesserscaup -0.389 0.145 -0.122

Settling Patterns a FromStewart and Kantrud (1973); average permanence (cropland ponds= 2, seasonalponds = 3, semipermanentponds and reservoirs = We foundevidence that ducks exhibit 4) times-0.1. all 3 of the hypothesizedpatterns of set- tling.Our literature review suggested that, thewintering grounds or during spring mi- withinspecies, homing was morepreva- gration.Because habitat needs during the lent among adult femalesthan among breedingseason are less critical for males males(for the few speciesin whichmales thanfor females,there is relativelylittle havebeen studied) or yearlings. That adult advantageto malesin homingto familiar femalesare morephilopatric than adult sites.Instead, the male follows the homing malesis consistentwith the seasonalmo- female(McKinney 1965). Greater philo- nogamyof thesespecies and the greater patryby femalesalso is generallyconsis- roleof thefemale in parentalcare (Rohwer tent with the predictionof Greenwood's and Anderson1988); male dispersal may (1980)hypothesis and the socialstructure be largelya consequenceof when and of duckpopulations. The female does not wherethe ducks form pair bonds. Because selecta mateon thebasis of the qualityof of differencesin parentalcare, the repro- his territory.Accordingly, Greenwood's ductivesuccess of malesis limitedby ac- (1980)hypothesis predicts that females are cess to fertilizablefemales, whereas the morelikely to homethan males. reproductivesuccess of femalesis limited Forthe mallard,gadwall, and northern by environmentalresources needed for re- shoveler,at least,there is evidencethat production.Because females spend far hensthat are successful in hatchinga clutch moretime than malesat nestsand with are moreprone to homethan those that young,they are exposedto greaterpre- arenot (Dotyand Lee 1974;Mihelsons et dationrisk (Sargeant 1972, Sargeant et al. al. 1986;J. T. Lokemoen,H. F. Duebbert, 1984),which subsequently results in a pre- and D. E. Sharp,Reproductive strategies ponderanceof males in the population of prairiemallards, gadwalls, and blue- (Johnsonand Sargeant 1977). Thus, all fe- wingedteal, unpubl. manuscript, U.S. Fish malescan generallyfind a mate,and se- andWildl. Serv., Jamestown, N.D.). This lectionshould favor those that are best able phenomenoncould be viewedas a "play- to findquality breeding habitat, which in- the-winner"approach to settlingin a hab- cludessecure nest sites and access to food itat;there is no advantageto be gainedby duringegg laying and incubation. Females changinga habitatthat has been success- maytend to home because familiarity with ful. In fact,some studies have shown in- a breedingarea should facilitate access to creasedreproductive success among birds bothfood and a securenest site. For males, that used nest sites in whichthey were not all of whichcan mate,the selective previouslysuccessful (Dow and Fredga pressureto obtaina femaleconfers an ad- 198S,Blancher and Robertson 1985). vantageto pairingas earlyas possible,on Contrastingthe 10 species, homing seems 2 WILDLIFEMONOGRAPHS most prevalentamong redhead, canvas- to winteringgrounds, which is consistent back, lesser scaup, mallard, gadwall, with predictionsof Fretwelland Lucas Americanwigeon, and shoveler.These (1969).Nudds cited unpublished work of speciesgenerally use more stable wetlands H. A. Kantrud,who foundthat correla- thanthe others(Smith 1971, Stoudt 1971, tionsbetween pond density and density of Stewartand Kantrud1973); their use of breeding-duckpairs decreased with in- thesemore stable habitats would predict creasinglatitude. That result, however, was a greaterhoming tendency (e.g., Wiens basedon countsof all speciescombined. 1976). Incontrast, the 10species examined in this All speciesexhibited opportunistic set- reportfilled the habitatsomewhat differ- tlingto someextent, in thattheir numbers ently. Four of them gadwall,green- fluctuatedwith pond numbers.Such winged teal, canvasback,and lesser movementis especiallyimportant in spa- scaup diddisplay a latitudinaldecline in tiallyheterogeneous habitats (Gauthreaux correlation with pondnumbers, although 1982),such as wetlandsin the prairie, noneof thesespecies was closely associated wheredrought is oftena localphenome- withpond counts. The 3 speciesthat win- non. Mostnotable as opportunistswere terin largenumbers in thewestern states- pintail,blue-winged teal, mallard,and Americanwigeon, northern shoveler, and shoveler.Opportunism is favoredwhen northernpintail- showed high correla- habitatconditions are unpredictable from tionswith pond numbers in southwestern 1 yearto the next.The speciesusing the strata,and the latter 2 didin severalsouth- least predictablewetland habitats is the ern strataas well. Althoughit is possible pintail,which also is the speciesmost that manybirds actually home but then stronglycorrelated with pond counts. Div- backtrackto areasencountered earlier in ing ducks,at the otherextreme, use more migration,it seemsclear that largeseg- predictablewetlands; correspondingly, ments of thepopulations of these7 species thesespecies were less closely tied to pond respondto pond conditionsas they en- counts.More conclusive evidence for op- counterthem. portunismis a patternof strongestcorre- The remainingspecies mallard,blue- lationswith pond counts in stratafirst en- wingedteal, and redhead appearto re- counteredby returningbirds in thespring. spondmore directly to wetlandconditions Thispattern was apparentfor American in the primaryportion of theirbreeding wigeon,green-winged teal, and canvas- ranges,as is evidentfrom highest corre- back and was pronouncedfor gadwall, lationswith pond numbersthere. These shoveler,and pintail. findingsfit the ideal-free distribution mod- Flexiblesettling, indicated by overflight el of Fretwelland Lucas (1969), who pre- in dry years,was shownby mostspecies dictedthat the "mostsuitable" habitat fills whenappropriate. Drought displacement firstand that some individuals are forced was generallyto the northor northwest, to lesssuitable habitats. For several species as suggested earlier by Hansen and of ducks,the mostsuitable habitat may McKnight(1964), Crissey (1969), and not be that closestto winteringareas, as Smith(1970). Some overflight to thenorth- Nudds(1978) conjectured, but rather that east was indicatedfor mallard,gadwall, habitatin theprimary part of thebreeding blue-wingedteal, green-winged teal, red- range. head,and canvasback. For Americanwi- geon,many of whichenter from the west, Relations to Life History droughtdisplacement to the east and southeastwas seen. Characteristics Wefound that 2 independentindices to Fillingthe Habitat suitesof life historycharacteristics (these canbe equallyconsidered as indicesof r- Nudds(1978, 1983) suggested that the andK-selection) varied together. Average habitatof ducksbecomes filled first closest valuesindicated that pintail, mallard, blue- BREEDINGDISTRIBUTIONS OF DUCKS Johnsonand Grier 33 wingedteal, and shovelerhave the most amongduck species is bestmade in terms featuresassociated with unstableor un- of annualsurvival rates, which can be es- predictable environments. Gadwall, timatedfrom long-term banding studies. Americanwigeon, and green-winged teal D. H. Johnson,J. D. Nichols,and M. D. were intermediate,and attributesof the Schwartz(Breeding dynamics of ducks, divingducks were associated with the use unpubl.manuscript presented at Ecol.and of stableor predictableenvironments. Manage.of BreedingWaterfowl Symp., The conclusionsof 3 otherstudies are Winnipeg,Manit., Canada, 18-22 Aug consistentwith ours.For one, Patterson 1987)reviewed studies that used modern (1979)suggested that high correlation coef- statisticalmethods of estimatingsurvival ficientsbetween duck numbers and pond rate.They foundsome variationamong numbersin theCanadian prairie and park- species,but not all of the 10 specieshad landsindicated an opportunisticresponse beenstudied, and estimates available per- to habitatconditions associated with r-se- tainedto differentgeographic areas, time lection.He also determinedcorrelations periods, and banding seasons, making them betweenchanges in ducknumbers and the difficultto compare.Moreover, all studied previouspopulation count, large positive populationshad been subjectedto recre- correlationsinterpreted as evidencefor ationalhunting, so thatestimates may no competitiveself-regulation associated with longerreRect survival rates under condi- K-selection.He concludedthat, among tionsto whichthe specieshad adapted. these10 duckspecies, the mallard,blue- Comparisonsof reproductiveeffort wingedteal, pintail,and shovelerwere amongspecies are often made on thebasis r-selecteddabblers. American wigeon, of age at firstreproduction and clutch (or gadwall,and green-wingedteal wereK- litter)size. For waterfowl, total number of selecteddabblers, and redhead,canvas- eggslaid during a breedingseason is more back,and lesser scaup were further along relevantthan clutch size becauseit takes the spectrumtoward K-selection. differingrenesting probabilities into ac- Bailey (1981)compared mallard and count. canvasbacklife historieswith datagath- We reviewedPalmer (1976) and Bell- eredin Saskatchewanby Stoudt(1971). He rose(1980) for informationon age at first concludedthat the mallardwas moreof reproduction.The 4 specieswe indicated an r-selectedspecies than the canvasback, weremost r-selected (pintail, mallard, blue- in that the mallardresponded more op- wingedteal, and shoveler) typically breed portunisticallyto improvinghabitat and as yearlings,as doesAmerican wigeon, an acocepteda greaterrisk of reproductive intermediatespecies. Of the other2 inter- failure. mediatespecies, the gadwalloften defers Finally,Vickery and Nudds (1984) tested breedinguntil its secondyear, and no in- for density-dependentrelations in duck formationwas presentedon the green- populationson studyareas in Albertaand wingedteal. Large numbers of the 3 most Saskatchewan.They foundthat gadwall, K-selectedspecies canvasback,redhead, northernshoveler, redhead, canvasback, andlesseer scaup do notbreed until they andlesser scaup showed some density de- are 2 yearsold. Thus,age at firstrepro- pendence,characteristic of K-selection. ductionparallels the otherattributes of r- The remaining5 speciesdid not. andK-selection that we examined. It wouldbe worthwhilealso to arraythe Totaleggs laid in a breedingseason is 10 speciesaccording to otherassociates of more difficultto addressbecause it in- r- andK-selection, notably longevity and volvesclutch size and the number of nest- reproductiveeffort. Speciesconsidered ing attemptsmade. Further, clutch size morer-selected tend to haveshorter life- typicallydeclines with later nestingat- spansbut higher fecundity;K-selected tempts,and bothclutch size and proba- speciestypically live longerbut produce bilityof makinga nestingattempt depend feweryoung (e.g., Pianka 1974). on the conditionof the breedingbird and A comparisonof ecologicallongevity its habitat(D. H. Johnson,J. D. Nichols, 4 WILDLIFEMONOGRAPHS and M. D. Schwartz,Breeding dynamics Evenwithout new information, such as of ducks,unpubl. manuscript presented at satelliteimagery providing habitat con- Ecol.and Manage. of BreedingWaterfowl ditions,an understandingof the factors Symp.,Winnipeg, Manit., Canada, 18-22 thatdetermine duck distributions can im- Aug1987). Keith (1961) estimated for an proveestimates of ducknumbers. Johnson Albertastudy area both clutchsize and (1981,1986) showed how auxiliary infor- numbersof nestsper female. We usedthe mationcould be employedto developes- productof those2 valuesas an estimateof timators,known as empiricalBayes esti- totaleggs laid (except for the canvasback, mators,that averaged 20% more accurate whichhad limited sample size, and pintail thancustomary estimators. Auxiliary in- and American wigeon, which Keith formationtook 1 of the following3 forms: thoughtwere not adequately represented). counts of a given speciesin a particular For the 7 remainingspecies, total esti- stratumduring previous years, the pond matedegg productionwas correlated with countin thatstratum during the yearof ourindex of r- andK-selection (r = 0.705, thesurvey, and the total number of ponds P = 0.077;Table 5). withwater within the pond-surveyed area. Theanalyses we present,the otherevi- Althoughthe previous count was generally dencewe review,and the reportsof Pat- aneffective form of auxiliaryinformation, terson(1979), Bailey (1981), and Vickery estimatesfor somespecies in somestrata andNudds (1984) do notjustify any causal werebetter with eitherthe localor con- implications,but suggestan association tinentalpond count. These differences amongcertain life historyfeatures, in- amongspecies and strata corresponded to cludingthe predictabilityof favoredhab- findingsof thisreport. itats. Oneof the mainpurposes of estimating breedingpopulations of ducksis to predict Management Applications the fall flightso thathunting regulations can be establishedconsistent with antici- Effective managementof waterfowl patednumbers of availableducks. For any populationsis enhancedby a basicunder- species,the fall flightdepends on the size standingof the distributionsof the birds of the breedingpopulation and its pro- andthe factorsthat affect those distribu- ductivity.Productivity in turndepends on tions.On a continentalbasis, for example, severalfactors and is variablefrom area mucheffort is expendedin attemptsto to areaand fromyear to year.Again, an estimatethe size of each year'sbreeding understandingof the distribution of breed- populationof key speciesof ducks.Wa- ing ducksis importantin predictingtheir terfowlsurveys to obtainsuch estimates productivity.For example,greater pro- arelabor-intensive and costly. Conversely, ductivitymay be expectedif mostmal- extensivehabitat informationmay be lardsare dispersed across breeding habitats gatheredinexpensively with satelliteim- in the prairiesand parklands, and if pond agery,as a by-productof otheractivities. conditionsthere are good, than if theyare By knowinghow ducksdistribute them- displacedto morenorthern strata. Results selvesin responseto habitatconditions, es- in this reportcan be usedto refinepre- peciallythose of wetlands,surveys could dictionsbeyond such obvious generalities. be designedin a moreoptimal fashion. Ona long-termbasis, there is muchin- Samplingintensity could vary annually to terestin knowingwhether populations of placemore effort in stratawhere the hab- ducks are increasing,fluctuating about itat informationpredicts the presenceof somelong-term mean, or declining.De- more ducks,for example.Results pre- terminingtrends is notas straightforward sentedin thisreport will be usefulin plan- as it wouldseem because of the variable ningextensive waterfowl surveys, but they distributionsof the speciesand the fact alsowarn that the 10 speciesvary, so that thatnumbers of ducksare related to num- an optimaldesign must be a compromise bersof pondswith water.Analyses like amongthe speciesof interest. thoseincluded here can help clarifythe BREEDINGDISTRIBUTIONS OF DUCKS- Johnsonand Grier 35 situation.A studyof mallardtrends was male strategiesof reproduction.Ph.D. Thesis, Univ.Minnesota, Minneapolis. 281pp. presentedby Johnsonand Shaffer (1987). ANONYMOUS.1977. Standardoperating procedures Consideringmore local situations, such foraerial waterfowl breeding ground population as a wetlandmanager might deal with, it and habitatsurveys. U.S. Fish and Wildl.Serv., is equallyimportant to understandhow Off.of Migr.Bird Manage.81pp. themselves.This infor- BAILEY, R. O. 1981. A theoreticalapproach to prob- ducksdistribute lems in waterfowlmanagement. Trans. North mationwill help assessthe ef:fectsof on- Am. Wildl.and Nat. Resour.Conf. 46:58-71. the-groundmanagement practices. For BELLROSE, F. C. 1980. Ducks,geese and swansof example,it maybe feasibleto attemptto NorthAmerica, Third ed. StackpoleBooks, Har- developa large breedingpopulation of risburg,Pa. 540pp. , AND R. D. CROMPTON.1970. Migrational divingducks or someother species with behaviorof mallardsand blackducks as deter- high ratesof homingby attractingpairs minedfrom banding. Ill. Nat. Hist. Surv.Bull. and ensuringhigh production.Surviving 30:167-234. adultfemales and manyof theirfemale BISHOP,R. A., D. D. HUMBURG,AND R. D. ANDREWS. wouldbe likelyto returnto the 1978. Survivaland homing of femalemallards. offspring J. Wildl.Manage. 42:192-196. area,if the breedinghabitat were stable BLANCHER,P. J., ANDR. J. ROBERTSON.1985. Site fromyear to year.This effortwould be consistencyin kingbirdbreeding performance: less likelyto succeedif directedtoward implicationsfor site fidelity.J. Anim.Ecol. 54: northernpintails or blue-winged teal, how- 1017-1027. BLOHM, R. J. 1978. Migrationalhoming of male ever,because of theirreduced homing ten- gadwallsto breedinggrounds. Auk 95:763-766. dencies.The developmentof attractive 1979. Thebreeding ecology of the gadwall wetlandsin southernparts of thebreeding in southernManitoba. Ph.D. Thesis, Univ. Wis- range would more likely attractthese consin,Madison. 177pp. species. BOYCE,M. S. 1984. Restitutionof r- andK-selection as a modelof density-dependentnatural selec- Beyondtheir effortsto improveduck tion. Annu.Rev. Ecol. Syst. 15:427-447. habitatsand populatiorls, managers can fill BOYD, H. 1983. Intensiveregulation of duckhunt- a vitalscientific role. Most of theirman- ing in NorthAmerica: its purposeand achieve- agementactivities can be viewed,not only ments.Can. Wildl. Serv. Occas. Pap. 50. 24pp. CALVERLEY,B. K., AND D. A. BOAG. 1977. Repro- as a positiveaction for the ducks,but as a ductivepotential in parklandand arctic-nesting scientifictest. Theories such as presented populationsof mallardsand pintails(Anatidae). heregenerate predictions of responsesto Can.J. Zool. 55:1242-1251. specificchanges. They can be appliedto COULTER,M. W., AND W. R. MILLER. 1968. Nest- the manager'sactions and results used not ingbiology of blackducks and mallards in north- ernNew England.Vermont Fish and Game Dep. only to evaluatethe impacton the duck Bull. 68-2. 74pp. populations,but alsoto evaluatethe sci- CRISSEY,W. F. 1969. Prairiepotholes from a con- entifictheory. We hope that this report tinentalviewpoint. Pages 161-171 in Saskatoon notonly provides some guidance for man- wetlandsseminar. Can. Wildl. Serv. Rep. Ser. 6. agers,but stimulatesactions that will test DERKSEN, D. V., AND W. D. ELDRIDGE. 1980. its predictions. Drought-displacementof pintailsto the Arctic coastplain, Alaska.J. Wildl. Manage.44:224- 229. DOTY, H. A., AND F. B. LEE. 1974. Homingto nest basketsby wild femalemallards. J. Wildl.Man- LITERATURECITED age. 38:714-719. DOW, H., ANDS. FREDGA. 1983. Breedingand natal ALLISTON,W. G. 1979. The populationecology of dispersalof the goldeneye,Bucephala clangula. an isolatednesting population of redheads(Ay- J. Anim.Ecol. 52:681-695. thya americana). Ph.D. Thesis,Cornell Univ., DWYER,T. J.,G. L. KRAPU,AND D. M.JANKE. 1979. Ithaca,N.Y. 211pp. Use of prairiepothole habitat by breedingmal- ANDERSON,D. R., AND C. J. HENNY. 1972. Popu- lards.J. Wildl.Manage. 43:526-531. lation ecology of the mallard:I. A review of DZUBIN,A. 1969. Commentson carryingcapacity previousstudies and the distributionand migra- of smallponds for ducksand possibleeffects of tion from breedingareas. U.S. Fish and Wildl. densityon mallardproduction. Pages 138-160 Serv.Resour. Publ. 105. 166pp. in Saskatoonwetlands seminar. Can. Wildl. Serv. ANDERSON,M. G. 1985. Socialbehavior of breeding Rep.Ser. 6. canvasbacks(Aythya valisineria): male and fe- FRETWELL,S. D., AND H. L. LUCAS,JR. 1969. On 36 WILDLIFEMONOGRAPHS

territorialbehavior and other factors influencing , AND T. L. SHAFFER. 1987. Are mallards habitatdistribution in birds:I. Theoreticalde- declining in North America?Wildl. Soc. Bull. 15: velopment.Acta Biotheoret. 19:17-36. 340-345. GATES,J. M. 1962. Breedingbiology of the gadwall JOHNSON,D. J. 1978. Age-relatedbreedingbiology in northernUtah. Wilson Bull. 74:43-67. of the redhead duck in southwesternManitoba. GAUTHREAUX,S. A., JR. 1980. The influencesof M.S. Thesis, Texas A&M Univ., College Station. long-termand short-term climatic changes on the 47pp. dispersaland migration of organisms.Pages 103- KEITH, L. B. 1961. A study of waterfowl ecology 174 in S. A. Gauthreaux,Jr., ed. Animalmigra- on small impoundmentsin southeasternAlberta. tion,orientation, and navigation. Academic Press, Wildl. Monogr. 6. 88pp. New York,N.Y. KRAPU,G. L. 1979. Nutrition of female dabbling 1982. The ecologyand evolutionof avian ducks during reproduction.Pages 59-70 in T. A. migrationsystems. Pages 93-168 in D. S. Farner, Bookhout, ed. Proc. 1977 Symp. Midwest Fish J. R. King,and K. C. Parkes,eds. Avianbiology. and Wildl. Conf., Madison, Wis. North Cent. Vol.6. AcademicPress, New York,N.Y. Sect., The Wildl. Soc. GREENWOOD,P. J. 1980. Matingsystems, philo- LEITCH,W. G., ANDR. M. KAMINSKI.1985. Long- patryand dispersal in birdsand mammals. Anim. term wetland-waterfowltrends in Saskatchewan Behav.28:114s1162. grassland.J. Wildl. Manage. 49:212-222. , AND P. H. HARVEY. 1982. The nataland MAcARTHuR,R. H., ANDE. O. WILSON.1967. The breedingdispersal of birds.Annu. Rev. Ecol. Syst. theory of island biogeography. Princeton Univ. 13: 1-21. Press, Princeton, N.J. 203pp. HANSEN, H. A. 1960. Changedstatus of several MARTIN,F. W., R. S. POSPAHALA,ANDJ. D. NICHOLS. speciesof waterfowlin Alaska.Condor 62:136- 1979. Assessmentand population management 137. of North American migratorybirds. Pages 187- , AND D. E. MCKNIGHT. 1964. Emigration 239 in J. Cairns,Jr., G. P. Patil, and W. E. Waters, of drought-displacedducks to the Arctic.Trans. eds. Environmental biomonitoring, assessment, NorthAm. Wildl.Conf. 29:119-127. prediction,and management certain case stud- HENNY, C. J. 1973. Droughtdisplaced movement ies and relatedquantitative issues. Statistical ecol- of NorthAmerican pintails into Siberia. J. Wildl. ogy. Vol. 11. International Cooperative Publ. Manage.37:23-29. House, Fairland, Md. HOCHBAUM,G. S., AND E. F. BOSSENMAIER.1972. MARTINSON,R. K., AND C. F. KACZYNSKI.1967. Responseof pintailsto improvedbreeding hab- Factors influencing waterfowl counts on aerial itat in southernManitoba. Can. Field-Nat.86: surveys,1961-66. U.S. Fish and Wildl. Serv.Spec. 79-81. Sci. Rep. Wildl. 105. 78pp. HOCHBAUM,H. A. 1946. Recoverypotentials in McHENRY,M. G. 1971. Breeding and post-breed- NorthAmerican waterfowl. Trans. North Am. ing movements of blue-winged teal (Anasdis- Wildl.Conf. 11:403-418. cors) in southwestern Manitoba. Ph.D. Thesis, HOLM, E. R., AND M. L. SCOTT. 1954. Studieson Univ. Oklahoma, Norman. 76pp. the nutritionof wild waterfowl.N.Y. Fish and McKINNEY,F. 1965. Spacingand chasing in breed- GameJ. 1:171-187. ing ducks. Wildfowl 16:92-106. IMMELMANN,K. 1973. Roleof the environmentin McKNIGHT,D. E. 1974. Dry-land nesting by red- reproductionas source of "predictive"infor- heads and ruddy ducks. J. Wildl. Manage. 38: mation. Pages 121-147 in D. S. Farner,ed. 112-119. Breedingbiology of birds.Nat. Acad. Sci., Wash- MIHELSONS,H., A. MEDNIS,AND P. BLUMS. 1986. ington,D.C. Populationecology of migratoryducks in Latvia. JOHNSGARD,P.A. 1975. Waterfowlof NorthAmer- ZinatnePubl. House, Riga, Latvia, U.S.S.R.112pp. ica. IndianaUniv. Press, Bloomington. 575PP. MURDY,H. W. 1966. When the prairies go dry. JOHNSON,D. H. 1980. The comparisonof usage Naturalist 17:8-13. andavailability measurements for evaluating re- NUDDS,T. D. 1978. Comments on Calverley and sourcepreference. Ecology 61:65-71. Boag's (1977) hypothesison displaced ducks and 1981. Improved population estimates an evolutionaryalternative. Can. J. Zool.56:2239- throughthe use of auxiliaryinformation. Pages 2241. 436-440in C. J. Ralphand J. M. Scott,eds. Es- 1983. Niche dynamics and organizationof timatingnumbers of terrestrialbirds. Studies in waterfowl guilds in variableenvironments. Ecol- avianbiology No. 6. CooperOrnithological Soc. ogy 64:319-330. and AllenPress, Inc., Lawrence,Ks. PALMER,R. S. 1976. Handbookof North American 1986. EmpiricalBayes estimates of breed- birds. Vols. 2 and 3. Yale Univ. Press, New Ha- ing populationsof NorthAmerican ducks. Ph.D. ven, Conn. 521 and 560pp. Thesis,North Dakota State Univ., Fargo. 187pp. PATTERSON,J. H. 1979. Can ducks be managed by , AND A. B. SARGEANT. 1977. Impactof red regulation?Experiences in Canada.Trans. North foxpredation on the sexratio of prairiemallards. Am. Wildl. and Nat. Resour. Conf. 44:130-139. U.S. Fish and Wildl. Serv.Wildl. Res. Rep. 6. PIANKA,E. R. 1974. Evolutionaryecology. Harper 56pp. & Row, New York, N.Y. 356 pp. Grier BREEDINGDISTRIBUTIONS OF DUCKS- Johnsonand 37 Prairie ducks. Stackpole Co., S., D. R. ANDERSON,AND C. J. HENNY. SOWLS, L. K. 1955. POSPAHALA,R. Pa., and Wildl. Manage.Inst., Wash- 1974. Population ecology of the mallard: II. Harrisburg, conditions,size of the breeding ington, D.C. 193pp. Breeding habitat The evolution of life history and production indices. U.S. Fish STEARNS, S. C. 1977. populations, a critique of the theory and a review of and Wildl. Serv. Resour. Publ. 115. 73pp. traits: Annu. Rev. Ecol. Syst. 8:145-171. H. J. 1969. Home range and breeding the data. POSTON, E., A. D. GEIS,AND C. D. EVANS. 1958. of the shoveler. M.S. Thesis, Utah State STEWART, R. biology Distribution of populations and hunting kill of Univ., Logan. 86pp. 22:333-370. III. 1979. Pop- the canvasback.J. Wildl. Manage. PULLIAM,H. R., AND T. A. PARKER, 1973. Ecological dis- of sparrows.Fortschr. Zool.25: , AND H. A. KANTRUD. ulationregulation of breeding waterfowl populations in 137-147. tribution 1988. Fe- North Dakota. J. Wildl. Manage. 37:39-50. ROHWER,F. C., AND M. G. ANr)ERsoN. factors affecting monogamy,and the tim- STOUDT,J. H. 1971. Ecological male-biasedphilopatry, in the Saskatchewanpark- of pair formation in migratory waterfowl. waterfowl production ing U.S. Fish and Wildl. Serv. Resour. Publ. Current Ornithol. 5:187-221. lands. A. B. 1972. Red fox spatial character- 99. 58pp. SARGEANT, 1971. Populationecology of lesser istics in relationto waterfowl predation.J. Wildl. TRAUGER,D. L. scaup (Aythya affinis) in subarctic taiga. Ph.D. Manage. 36:225-236. 118pp. R. T. EBERHARDT.1984. Thesis, Iowa State Univ., Ames. , S. H. ALLEN,AND D. NUDDS. 1984. Detection on breeding ducks in midcon- VICKERY,W. L., ANDT. Red fox predation effects on annualduck cen- America. Wildl. Monogr.89. 41pp. of density-dependent tinent North 65:96-104. INC. 1982. SAS user s guide: statis- suses. Ecology SAS INSTITUTE, A. 1920. Observationson the habits of tics. SAS Inst., Inc., Cary, N.C. 584pp. WETMORE, birdsat Lake Burford,New Mexico. Auk37:221- SHAW,S. P., ANDC. G. FREDINE.1956. Wetlands United States. U.S. Fish and Wildl. Serv. 247. of the J. A. 1976. Population responsesto patchy Circ. 39. 67pp. WIENS, environments.Annu. Rev. Ecol. Syst. 7:81-120. SMITH,A. G. 1971. Ecological factors affecting wa- productionin the Albertaparklands. U.S. terfowl 18 January 1988. Fish and Wildl. Serv. Resour. Publ. 98. 49pp. Received Accepted 17 June 1988. SMITH, R. I. 1970. Response of pintail breeding populations to drought. J. Wildl. Manage. 34: 943-946.