Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
The Auk 117(3):748-759, 2000 MOVEMENT PATTERNS OF WINTERING IN ARIZONA GRASSLAND SPARROWS CALEB E. GORDON • Department of Ecology andEvolutionary Biology,Universityof Arizona,Tucson, Arizona85721,USA ABSTRACT.--I used mark-recaptureanalysisand radio telemetryto characterizewinter movementpatternsof sixgrasslandsparrowsin southeastern Arizona.Mark-recapturedata weregeneratedby bandingbirdscapturedduringrepeatedflush-nettingsessions conducted on a seriesof 7-ha plotsover three consecutive winters.This resultedin 2,641capturesof 2,006individual sparrowsof the six species.Radiotelemetrywasconductedconcurrently on 20 individualsof four of thesespecies.Recapturedataandradiotelemetryindicatedthat Cassin'sSparrow(Aimophila cassinii)and Grasshopper Sparrow(Ammodramus savannarum) werethe mostsedentary, followedby Baird'sSparrow(Ammodramus bairdii),VesperSparrow (Pooecetes gramineus), SavannahSparrow (Passerculus sandwichensis), and Brewer'sSparrow (Spizellabreweri). Grasshopper, Baird's,Savannah,and Vespersparrowstendedto remain within fixedhomerangesduring winter.With the exceptionof SavannahSparrows,whose movementbehaviorvariedamongstudysites,movementpatternsremainedconstantwithin speciesacrossyearsand studysitesdespiteradicalfluctuations in the absoluteandrelative abundances of all species. Interspecificdifferences in movementpatternsuggestthatspecies in thissystempartitionnichespaceaccordingto theregional-coexistence mechanism. Abundancesof the most sedentaryspecies,Cassin's,Grasshopper,and Baird'ssparrows,were poorlyor negativelycorrelatedwith summerrainfall at the between-yearlandscapescale, whereasabundancesof the more mobile Savannah,Vesper,and Brewer'ssparrowswere stronglypositivelycorrelated.This is consistent with the theoreticalpredictionthat movement constrainslarge-scalehabitatselection,favoringmobilespeciesin fragmentedenvironments.Received 28 April 1999, accepted 10 February2000. where the unpredictabilTHEMOVEMENT BEHAVIOR of an organismde- desertenvironments, terminesthe scaleat whichit perceivesandre- ity of rainfallcreatesseeddistributions thatare spondsto the spatialsubdivisionof its envi- highly patchyin time and space(Davies1984, ronment (Wiens 1976, Levin 1992). Movement Clark 1997, Dean 1997). also is a fundamentalelementof populationIn contrast,an organismwith a fixedhome levelprocesses suchasmetapopulation dynam- rangemovesaroundrepeatedlywithin an area ics, geneflow, habitat selection,and foraging thatis smallrelativeto its abilityto travela givbehavior Understandingthe patterns, con- en distance.Suchhomerangesmay or maynot straints,and adaptabilityof movementbehav- be defendedagainstconspecifics. Thistype of ior is an important step toward understanding movement is suited for exploiting relatively the dynamicsof an organism'secologicalrela- predictableresources. In suchcases,gainingfationships. miliarity with a particularareais morecostefThe range of movementpatternsin nature fectivethan engagingin long-distance explorcan be viewed as a continuum ranging from atory movements(Sinclair1984). random wanderingto movementswithin a fixed home range.Randomwanderingis well suitedfor exploitingunpredictable, patchyresources.In suchcases,the costof exploratory movementis offsetby the benefitof increased encounterswith resourcepatches(Noy-Meir Although random wanderingsand fixed home-rangemovements generallyarelinkedto distincttypes of resourcedistributions,both strategiescanbe viablein many environments. Indeed, the use of multiple movementstrate- 1973, Andersson1980, Sinclair 1984). This rationalehas beenusedto explainthe high degree of nomadismamonggranivorousbirds in lar landscapes hasreceiveda greatdealof attention(e.g.Hutchinson1951,LevinsandCul- gies for exploiting single resourcesin particu- ver 1971, Horn and MacArthur 1972, Tilman 1994). In modelsof regional coexistence, ecologically similar speciescoexistvia a tradeoff E-mail: [email protected] 748 July2000] Sparrow Movement Patterns betweenwidespreadexplorationfor and local exploitationof resources(Brown 1989, Tilman 1994).This tradeoffimpliesthat increasedefficiencyat seekingnew patchescan only be gainedby somesacrificeof an organism's within-patchefficiency, andviceversa.Thisconcept hasbeen'invokedto explainnichedifferences amonggranivorousmammalsand birds in desert ecosystems(Mares and Rosenzweig1978, Brown et al. 1979, Thompsonet al. 1991). In suchcases,granivorousbirds are describedas "cream-skimmers"that explorelarge areasto find the richestpatchesin the landscape.This implies that birds distribute themselvesto matchthe productionof resourcesat the landscape scale.Granivorous rodents, in contrast, are more sedentarybut are more locally efficient, such that they can survive on resource densitiestoo low to supportbirds. This notion of birds as highly mobile cream-skimmersis consistent with patternsof nomadismthathave beendocumentedin many seed-eatingbirds of arid regions of Australia (Davies 1984, Clark 1997)and SouthAfrica (Dean 1997).Movement patternsof granivorousbirds in the American southwest have never been documented. Ratherthan falling on a fixedpoint alongthe nomadism/ home-rangecontinuum,an organism may adjust its movementstrategyin responseto shifting environmentalconditions. Behavioralplasticityis particularlyvaluablein highly fluctuatingand unpredictableenvironmentswherethebenefitsof flexibilityoutweigh the costsof not beingableto specializeon, and therebymaximizeefficiencyfor, oneparticular strategy(Tripp and Colazo 1997).Plasticityof behaviorand/or growthform hasbeenfound in a variety of desert organisms (Fox 1990, Pfennig 1992). Given the variability and heterogeneityof their environment,plasticitywould appear to be advantageousfor sparrowswintering in Arizonagrasslands. Productivityin theseecosystems is largely limited by rainfall (McClaran 1995).The rainfall that occursduringthe summer monsoonseasonlargely determinesthe distributionof habitatconditionsfor wintering sparrows.Summer rainfall controlsthe productionof grassseedsthat comprisethe sparrows' primary food source(PullJam1983, C. Gordonunpubl.data)and the vegetativecover that providesimportantprotectionfrom predators(PullJamand Mills 1977,Lima and Valone 749 1991).Summerrainfall in this regionis patchy at the scale of thunderstorm cells that are 2 to 8 km wide and is highly variableand unpredictable among years (Noy-Meir 1973, McClaran 1995). For this reason,thesesparrows may be faced with radically different spatial distributions of suitable habitat from winter to winter. PullJam and Parker (1979) measured seedproductionat two grasslandstudy plotsin southeastern Arizona over four successive yearsand found that fluctuationsexceedingan order of magnitudewere common.On oneplot, seed productionincreasedmore than two orders of magnitude in responseto a 64% increasein summerrainfall betweenyears. My study addressestwo major questions. First, can variation in movement patterns explain the coexistenceof severalecological(ly similar speciesin this system?Second,does movement behavior constrain habitat selection in this system?To theseends, I developnew field and analyticaltechniques for studyingthe local movement behavior of grasslandsparrows.I usethesein conjunction with radio telemetry to characterizemovementpatternsof six speciesof granivoroussparrowson their winteringgroundsin the grasslandsof southeasternArizonaduringthreeconsecutive winters. In particular, I focuson overall sedentariness,the tendencytoward fixed home-range versus nomadic movements,and plasticity of within-seasonmovementpatterns.I discussthe observedmovementpatternsas strategiesthat are moreor lessadaptedfor the spatiotemporal distributionof resources,patches,and biological interactions in the environment. MATERIALS AND METHODS Studysites.--Fieldwork was conductedat three study sitesin southeastern Arizona. The Research Ranch(TRR) is a 3,200-hapreservelocated8 km southeast of thetownof Elginin SantaCruz County, Arizona. TRR is dominatedby upland grassland vegetation andliesat theedgeof a large(ca.900km2) regionof grasslandin the SonoitaValley.TRR has not been grazed by cattle since 1967. Study plots werelocatedin the northwesternquarterof TRRbetween1,450and 1,500m elevation.The EmpireCienegaResourceConservationArea (EC) is alsolocated within the SonoitaValleygrasslands, roughly10 km from TRR. It lies in Pima and Santa Cruz counties and is leasedto a privaterancherfor cattlegrazing. Within the EC, fieldwork was conducted in the southeastern corner of the Davis Pasture, which is lo- 750 CALEB E. GORDON [Auk, Vol. 117 catedroughly3 km west-northwest of Elginbetween weedyhabitatsaswell asgrassland(Rising1996,C. 1,450 and 1,480 m elevation. The Buenos Aires Na- tional Wildlife Refuge(BANWR) is locatedin the southernAltar Valleyin Pima Countyroughly150 km westof thepreviousstudysites.Thisvalleycontainsa muchsmallerpatchof grasslandhabitat(ca. 100km2) that occursalmostentirelywithin the refuge. BANWR has not been grazedby cattlesince 1985.Studyplotswere locatedin the BorregoGrasslandssectionof the refugebetween1,050and 1,210 m elevation. Vegetation.--All threestudyareaswere locatedin open semidesert or plains grassland vegetation (McClaran1995).Rainfallfluctuateswidely in this habitat,rangingbetween200 and 400 mm per year; 50 to 60%of therainfallusuallyfallsduringthesummer monsoonseasonfrom July to September(McClaran 1995).Thishabitatis dominatedby a variety of perennialbunchgrasses, althoughmanyforbsand several small woody perennials also are common. The mostabundantgrassesincludeseveralspeciesof Bouteloua and Eragrostis, includingthe widespread exoticEragrostis lehmanniana. Theonlywoodyplants taller than 1 m on the studyplotswere a few small mesquitetrees(Prosopis velutina).Thesewere more abundant at the slightly lower and drier BANWR site,wheretheytendedto beconcentrated in arroyos off of study plots. Severalspeciesof oaks (Quercus) occurredin arroyosnear study plots at TRR. Studyspecies.--The studytaxaconsisted of sixspeciesof New World sparrows(Emberizinae):Cassin's Sparrow (Aimophilacassinii),Brewer'sSparrow (Spizella breweri),Vesper Sparrow (Pooecetes gramineus), Savannah Sparrow(Passerculus sandwichensis), Grasshopper Sparrow (Ammodramus savannarum), and Baird'sSparrow(Ammodramus bairdii).Exceptfor the smaller Brewer'sSparrow,these speciesexhibit a greatdealof overlapin bodysizeandbill size,which suggestsa high degreeof overlapin the seedsthat comprisetheirwinterdiet (Pulliam1983,C. Gordon unpubl. data). All of thesespeciesco-occurin midelevationgrasslands of southeastern Arizona. Baird's,Savannah,Vesper,and Brewer'ssparrows are long-distance migrantsthat do notbreedlocally. Theybeginto arrivein southeastern Arizona in late Augustand are absentfrom the areaby early May (Phillipset al. 1964).Cassin'sSparrowsbreedlocally, but it is unknownwhetherthewinteringindividuals I studiedwerelocalbreeders.Grasshopper Sparrows winteringin southeastern Arizonaare composedof locally breeding (C. Gordon unpubl. data) and migrant birds, but the proportionalrepresentationof thesetwo groupsis unknown.Grasshopper, Baird's, and Cassin'ssparrowsare non-flockingspeciesthat are exclusively(Grasshopper and Baird'ssparrows) or largely (Cassin'sSparrow)restrictedto grassland habitat(Rising1996).Savannah,Vesper,and Brewer's sparrowsare flockingspeciesthat use a wider spectrumof vegetationtypesincludingshrubbyand Gordon pers. obs.). Flush-netting.--AtBANWR and TRR, I established six permanent7-ha flush-nettingplots. Each plot consistedof a 96-m net line with a 3.5-haflushing zonefanningout on eachside.Foreachdayof flushnetting,crewsof 13 to 30 people(average22) assembledat 0830MST at the field siteto performthe followingfield protocol.Eight mistnets(2 x 12 m, 36 mm mesh)weresetup alongthe net line at the first of the sixplots.Whennetswereready,thefieldcrew fannedoutalongthe300-mbackedgeof oneflushing zone.With people spreadout evenly alongthe peripheryof the flushingzone,a signalwasgivenand the groupwalkedthroughthe flushingzonetoward the net line, causingsparrowsto flush toward the nets.Thecrewrepeatedtheprocedureforthesecond flushingzoneon the othersideof the net line. After workingbothsidesof a studyplot,eachnetted bird was banded with individuallynumbered aluminumleg bandsor their bandnumberrecorded if a recapture.We thendisassembled thenetsat the firstplot andrepeatedtheprocedureat thenextplot. In this way,we flush-nettedall six plotsat a study sitein randomsequence on eachday of netting. Betweenthe first week of Januaryand the first week of March 1996,flush-nettingcrewsperformed the aboveprocedureon sevenMondaysat TRR. At the sametime of yearin 1997and 1998,flush-netting wasdoneon sevenWednesdays at TRRandon nine Saturdaysat BANWR. Between Januaryand early March1999,threeflushingdayswereconducted on the samestudyplotsat TRR and BANWR. Radio telemetry.--I placed radio transmitterson three Baird'sSparrowsand two GrasshopperSparrowsin lateFebruary1997at theEC andTRRstudy sites.I attached1.0 g radio transmitters(modelBD2, HolohilSystemsLtd.) overtherumpwith leg-loop harnesses (RappoleandTipton1991)madeof cotton candlewick.The radioshave a batterylife of six to eightweeks.Beginningthe day afterradioswereattached,I attemptedto locateeachbird onceper day on 12 days over a seven-weekperiod. Locationsof initial netting and all subsequentrelocationswere recordedusing a global positioningsystem(GPS) unit accurate to within 2 m but not corrected for mil- itary signalscrambling.I conductedthis sameprocedureon sevenVesperSparrowsand eightSavannahSparrowsbetweenlateJanuaryandearlyMarch 1998 at BANWR. Recaptureanalysis.--Goodness-of-fit tests from program RELEASE(Burnhamet al. 1987) showed that the mark-recapturedata setsdid not fit the assumptionsof theJolly-Seber modelsusedin current mark-recaptureanalyses(P < 0.01 for all data sets). Therefore,I createda statisticcalledthe recapture eventrate (RER) that I used as an index of sedentariness,with higherRER valuesreflectingsedentary behaviorand/or lower mortalityduring the study July2000] Sparrow Movement Patterns period. RER is definedsimply as the numberof recaptureeventsdivided by the numberof recapture opportunities.The numberof recaptureopportunities for a given bird is the number of netting days subsequentto the initial captureof that bird during a given season.This statisticmeasuresthe overall tendencyof birds to be recaptured. I testedthe significanceof differencesin RER between data setswith a Monte Carlo simulationprocedure. For each simulation,I compareda pair of datasetsby preservingthe mark-recapture patterns of individual birds but randomly reshufflingindi- 751 the averageof the RER valuesof eachindividual in the data set. This statisticweights all individuals equally and might be called the averageindividual recapture event rate (AI). If individuals are drawn from a homogeneous population,thesetwo statistics shouldbe identical. However,if birds capturedearly in the seasonbehavedifferentlyfrom thosecaptured late in the season, these statistics would differ. The earlybirdshavemoreopportunitiesto be recaptured thanthelatebirdsandare,therefore,heavilyweighted in the RER but not in the AI. I used a simulationprocedureto test for statistically significantdifferencesbetweenthe RERand AI was then recalculatedfor the two sampledata sets. statisticsfor all data setswith at least 20 recapture This procedure was repeated 1,000 times for each events.For each simulation,I preservedthe exact pairwisetestandthentheobserved RERvalueswere structureof initial capturesby datefor all individual comparedwith the distributionsof RERvaluesgen- birds within a given data set.I simulatedrecapture eratedby the simulations. data for eachdata setby settingthe probabilityof reNet learning potentially could confoundRER sta- captureat eachopportunityequalto theRERfor that tisticsby reducingcaptureprobabilityfor previously data set. I repeatedthis procedure1,000 times and capturedbirds. For datasetswith largenumbersof comparedobserveddifferencesbetweenAI and RER recaptures(>20), I testedfor net learningeffectsby with the samplingdistributionsof thesestatistics creating an expected binomial distribution of cap- generatedby the simulations. turesunder the null hypothesisof no net learning (capture probabilitiesare independentof previous RESULTS capture).To do this, I generatedmaximum-likelihood estimatesof binomial parametersas outlined viduals between the two data sets. The RER statistic Flush-netting.--Analysisof movement pat- below. A preliminaryvalueof N (populationsize)wasvisually estimatedfrom the asymptoteof a plot of cumulative individual birds banded with time. The tern was based on the first three field seasons of flush-netting,whichyielded2,641captures of 2,006individualsparrowsamongthe sixfocal speciesand two studysitesaslistedin Table 1. Of the 574 within-seasonrecaptures,only sevenoccurredat a plot otherthan the onein whichthe bird had beeninitially banded.Only two of these(one GrasshopperSparrowand oneSavannahSparrow)requiredmovementof probabilityof capturinga givenbird in a givenflush, p, was calculatedas x (the averagenumberof birds capturedon a samplingdate) divided by N, and 1 p = q was the probability of not capturing a given bird .in a given flush.The probabilityof not capturing a given bird in i samplingdatesis q•,and q' N, therefore,is the numberof birds in the population that were nevercapturedin i samplingdates.This more than 100 m. Therefore, I based subsenumber was then added to the total number of birds that were actuallycapturedat leastoncein the sampling period to producea new estimateof N. This procedurewas then iterated with the new estimate of N until the parameterestimatesconverged onvalues that did not changein subsequentiterations. These parameterswere then used to generateexpected binomial frequency distributions for the numberof capturesfor eachbird in the dataset.Observeddistributionswere then comparedwith the expecteddistributions using chi-squaregoodnessof-fit tests. I calculated RER statisticsfor each data set by lumpingall recaptureeventsand opportunitiesover all individualswithin datasets.This weightseachrecapture opportunity equally, but individual birds with more recapture opportunities (first captured earlierin the season)areweightedmoreheavilythan individualswith fewerrecaptureopportunities(first capturedlaterin theseason).An alternativeis to take quentanalysesof movementpatternsfrom the banding data on same-plotrecapturesas described above. The simulationtestsshowedpronouncedinterspecificdifferencesin RER (Table2). Cassin's Sparrowsand Grasshopper Sparrowswere the most sedentary,followedby Baird'sSparrows, VesperSparrows,and finally SavannahSparrows and Brewer'sSparrows.Within species, RER varied little amongdata sets(Table3). ! found no significant differences in RER betweendatasets(year x studysite)for any spe- ciesexceptSavannahSparrow,which was significantlymoresedentaryat the TRRstudysite than at BANWR. This pattern was consistent across years. Five data sets (four for GrasshopperSparrow, one for Cassin's Sparrow) contained 752 CALEB E. GORDON [Auk,Vol. 117 TABLE1. Number of individualscapturedand within-seasonrecapturesfor six grasslandsparrows.Data are numberof individualbirds, with numberof recaptureeventsin parentheses. TRR = The Research Ranch,BANWR = BuenosAiresNationalWildlife Refuge. Study site 1996 1997 1998 Totala 2 (3) 43 (28) 53 (34) TRR BANWR 0 (0) -- Cassin'sSparrow 0 (0) 9 (3) TRR BANWR 164 (71) -- GrasshopperSparrow 16 (8) 278 (153) 388 (149) 189 (85) 989 (466) TRR BANWR 36 (5) -- Baird's Sparrow 15 (3) 9 (2) 28 (7) 18 (4) 100 (21) TRR BANWR 1 (0) -- Vesper Sparrow 7 (0) 194 (15) 141 (14) 83 (8) 420 (37) TRR BANWR 29 (2) -- Savannah Sparrow 7 (0) 106 (1) 105 (9) 119 (2) 365 (14) TRR BANWR 0 (0) -- Brewer's Sparrow 0 (0) 72 (2) 0 (0) 8 (0) 79 (2) aSpecies totalsforbothstudysites.Thetotalnumberof individualscapturedfor eachspecies is slightlylessthanthesumofsubtotals because individualscapturedin multiple field seasons are countedseparatelyin eachdatasetin whichtheyoccurred,but only oncein the totals. enoughrecaptureeventsto usethe maximum- broken down by weekly intervals to examine likelihoodnet-learningtestsdescribedabove. the probabilityof recapturingindividualsasa In all of these data sets, observed distributions functionof the increasingtime betweenfirst of captures were not significantly different and subsequentcaptures.For a randomlywanfrom thoseexpectedunderthenull hypothesis deringindividual,the probabilityof recapture of independenceof capture events (Table 4). should decrease with time as it "diffuses" This suggests that a strongnet-learningeffect away from its startingpoint. Sparrowswith was not present for Cassin'sSparrows and fixed home rangesmay experiencesomedeGrasshopper Sparrows. creasein recaptureprobability due to mortalEachrecaptureeventand opportunityfor recapturecanbe definedby the amountof time TABLE3. Intraspecificvariation in recaptureevent elapsedbetweenit andtheinitial captureof an rate (RER) amongyears and betweenstudy sites individual bird. Thus, the RER statistic can be for four grasslandsparrows.Only datasetswith at least15 individualsand specieswith at leasttwo such data sets are listed. TABLE2. Interspecificdifferencesin sedentariness of sparrowsas measuredby recaptureeventrate Species 1996 1997 (RER). CASP = Cassin'sSparrow,GRSP = GrassThe Research Ranch hopperSparrow,BAIS= Baird'sSparrow,VESP= VesperSparrow,SAVS= SavannahSparrow,BRSP GrasshopperSparrow 0.1100 0.1270 = Brewer'sSparrow. Baird'sSparrow 0.0382 0.0455 VesperSparrow --Species RER CASPGRSPBAIS VESP SAVSBRSP SavannahSparrowa 0.0220 -- 1998 0.0917 0.0588 0.0234 0.0216 Buenos AiresNational WildlifeRefuge CASP 0.1110 -- GRSP BAIS VESP SAVS 0.0943 0.0446 0.0179 0.0085 ns * ** ** -** ** ** BRSP 0.0044 ** ** ** * ns -- Grasshopper Sparrow -- 0.1000 0.0777 Baird's Sparrow VesperSparrow SavannahSparrow ---- -0.0146 0.0018 0.0404 0.0192 0.0036 aRERwassignificantlyhigherat The Research Ranchthanat Buenns, P :> 0.05; *, P < 0.05; **, P < 0.01 (Monte Carlo simulations). os Aires NWR (P < 0.05; Monte Carlo simulations). July2000] Sparrow Movement Patterns 753 TABLE 4. Estimatedbinomialparametersandchi-squaregoodness-of-fit testsfor net-learningeffectin sparrows.Thetestcompares observeddistributions of individualsin capturefrequencyclasses with expected distributionbasedon the assumption of independence of captureprobabilities (nonet learning). Species Data set Cassin'sSparrow Grasshopper Sparrow Grasshopper Sparrow Grasshopper Sparrow Grasshopper Sparrow 1998BuenosAires NationalWildlife Refuge 1998BuenosAires NationalWildlife Refuge Na pb p 65 315 0.121 0.097 0.67 0.30c 1998 The Research Ranch 686 0.112 0.47 c 1997BuenosAires National Wildlife Refuge 415 0.116 0.25• 1996 The Research Ranch 265 0.129 0.20 Estimatedpopulationsize. Estimatedprobabilityof captureof givenindividual on givennetting day. A singletrap-happyindividualremovedfrom analysis. ity, but otherwisethe probabilityof recapture for GrasshopperSparrowsand Cassin'sSparshould remain relatively constant through rowsto remainwithin fixedhomerangesdurtime. ing this time period.It alsodemonstrates that The probabilityof recapturingGrasshopper mortality was not significantin these cases. Sparrows(all data sets combined)decreased GrasshopperSparrowsand Cassin'sSparrows with time (Fig. 1). However,theRERat thelon- were the only specieswith enoughrecapture gesttimeinterval(eightweeks)wasstillhigher eventsfor this analysisto be performed. than the total RER of the next-mostsedentary Cassin'sSparrows (1998 BANWR data set) species,Baird'sSparrow(Table2). In the 1996 did not show a significantdifferencebetween TRR data set for GrasshopperSparrowsand AI and RER, but GrasshopperSparrowsdid the 1998 BANWR data set for Cassin'sSpar- show this difference.For all four of the large rows, RER remainedalmostconstantthrough datasetsfor Grasshopper Sparrows,A! values time (Fig. 1). In thesecases,individualswere weresignificantlylowerthan the RER(P < 0.05 just aslikely to be recapturedeightweeksfol- for 1998 at BANWR; P < 0.01 for 1998 at TRR, lowing initial captureas in the first week po- 1997at BANWR, and 1996at TRR). Grasshopstcapture.This demonstrates a strongtendency per Sparrowsthat were captured early had higher probabilitiesof recapturethan birds that were not initially captureduntil later in -- GrasshopperSparrow1996 TRR the season. : 0.25 Cassin'sSparrow1998 BANWR Radiotelemetry.--Of the20 birdsthatreceived radio transmitters,three disappearedbefore the battery shouldhave expired,and 10 were found deadbeforethe batteryexpired,having beenkilled by predators.! relocated18 birds at ---A---Total GrasshopperSparrow 0.2 0.15 least twice for a total of 115 relocations. To characterize movementpatterns,I plotted net distancemovedbetweeneachpair of loca- 0.1 tionsas a function of time and performedlinear regression(Fig. 2). The regressionsfor all four specieshad slopesbelow 2.6 m per day and R2 0.05 1 2 3 4 5 6 7 8 Intervalbetween captures(weeks) FIG.1. Recaptureeventrate as a functionof the time interval betweenfirst and subsequentrecapture valuesbelow0.029.This suggeststhat individualsof all four speciestendedto remainwithin fixedhomeranges.Thus,interspecific variation in sedentariness may haveresultedfrom variationin home-rangesize amongspecies. To describe the extent of movement, I also eventsfor total Grasshopper Sparrowsand Grass- comparedthe averagenet distancemovedbehopper Sparrowsin 1996 at The ResearchRanch tween all pairs of locationsfor individuals of (TRR) and Cassin'sSparrows at BuenosAires National Wildlife Refuge(BANWR). eachspecies.Becausenet distancemovedwas not correlated with time, these distances can be 754 CALEB E. GORDON [Auk, Vol. 117 Baird's Sparrow Savannah Sparrow 300 8001 600 200 iI1#1 0 i i 0 10 200 100 i- i i 20 30 40 0 ! 0 10 20 GrasshopperSparrow Vesper Sparrow 200 800 ] , , 150 , * 100 600 400 50 0, el ,I i1,_1teI,P I 4 • * 0 10 20 30 40 , 50 0 0 10 20 Time interval(days) FIG.2. Net distances(distancesbetweeneachpair of radio telemetryrelocations)movedwith time for four grasslandsparrows. thoughtof as the net distancemovedbetween any two locationsof a bird. Thesedata ranked speciesfrom mostto leastsedentaryin exactly the sameorderastheRERfromthenettingdata in Table2. The Grasshopper Sparrowwas the mostsedentarywith an averagenet movement of 84 m, followed by Baird's (113 m), Vesper (142m) andSavannah(186m) sparrows.All interspecificdifferenceswere significant (twotailedt-tests,P < 0.05).Theseobservedaverage ity amongindividualGrasshopper Sparrowsis that two types of individuals are presentin winter, somewith fixedhomerangesand others that move nomadically.This is consistent with manyterritorialsystems which"floaters" represent an excessof individuals relative to the numberof availableterritories(Shutlerand Weatherhead1994, Stutchbury1994,Westcott and Smith 1994).A discrepancyin movement behavioramongindividual Grasshopper Spardistances were overestimates because of GPS rows also may correspondto a differencebesignal scramblingand the error of the GPS tween migrant and residentpopulationsthat unit. GPS locationsfor a fixed point taken on coexiston thesestudy sitesin winter It is also the daysof radio telemetryshowedan average possiblethat differences in recaptureprobabilnet movement of 50.2 m. ities amongindividualswere causedby other It is importantto note that mortality cannot behavioral differences that could lead to differbe distinguished from emigrationassources of ent capture probabilities,such as different low RER. Corroborationof interspecificdifferor flyingheights. encesin RER by the telemetrydata suggests flushingresponses Extent of movement.--Cassin's Sparrowsand that RER does indeed reflect differences in moGrasshopper Sparrows, and to a lesserextent bility among species.Furthermore,the conBaird's Sparrows, were the most sedentary spestancyof RER with weekly intervalssuggests cies. The extent and pattern of movements in thatthe amountof mortalitythatoccurreddurthese sparrows are consistent with those obing the study period was extremelysmall, at least for Cassin'sSparrowsand Grasshopper served for desert rodents (Brown 1989,Brown and Zeng 1989). Thus, the notion that these Sparrows (Fig. 1). birds are highly mobile"cream-skimmers"in this systemmay not be accurate(Mares and DISCUSSION Rosenzweig1978,Brownet al. 1979,Thompson Variation among individuals.--One possibleex- et al. 1991). Somebirds, particularly Cassin's planationfor differencesin recaptureprobabil- Sparrowsand grassland-obligate Baird'sSpar- July2000] Sparrow Movement Patterns 755 rows and GrasshopperSparrows,may behave vannahSparrowsshowedsignificantlydiffermore like the rodents in the above models of ent local movement patterns between study desertgranivorecoexistence, playingthe roles sites, movement patterns of Grasshopper, of highly sedentarybut highly efficientforag- Baird's,and Vespersparrowswere fixed across ers. yearsand study sites(Table3). This behavioral The sparrowsin this study exhibitedsignif- constancy is remarkablegiventhe radicalflucicant interspecificvariation in movementpat- tuations(up to 20-fold) in abundanceof sparterns. Nonetheless,telemetry data indicated rows acrosssites and years (Table 1). Fluctuathat eventhe leastsedentaryspeciestendedto tions near an order of magnitude were obremainwithin fixedhomeranges(Fig.2). This servedfor all focal species.Given the imporpatternmightseemto conflictwith thenomad- tance of densities of interspecific and ic behaviors that have been documented in intraspecificcompetitorson habitatselectionin seed-eating birdsin otherrainfall-limitedeco- various theoretical and empirically demonsystems(Davies 1984, Clark 1997,Dean 1997). strated situations(Fretwell and Lucas1970,RoIn suchsystems, nomadism hasbeensuggested senzweig1981,Pimm et al. 1985),it is surprisas a way for consumersto copewith extreme ing that suchvariationin the densityof potenspatial and temporal variability in resource tial competitorswould haveno effecton movedistribution (Noy-Meir 1973, Davies 1984, ment behavior.One potentialexplanationfor Clark 1997, Dean 1997). Nonetheless,the ob- this pattern is that competitionamongsparservedsedentariness of sparrowsin this study rows may not be an importantecologicalfactor was not inconsistent with nomadic movements in this system.Anotherpossibilityis that the at different scalesof time and space.In fact, movementbehavior of Grasshopper,Baird's sparrowsmay havecombinedbetween-year re- and Vespersparrowsis fixedbecauseof various gional scalemobility with within-winter sed- physiological,morphological or other conentarinessto effectivelyexploit the specific straints,and that thesespeciesare unableto scalesof environmentalheterogeneityand var- adapttheirmovementbehaviorquicklyenough iability in this system.The dramaticfluctua- to trackrapid changes in theirenvironment. A tionsin sparrowabundance acrosssiteswithin third possibilityis thatthedensityadjustments years and acrossyears within sites (Table 1) that birds makein responseto environmental suggestedthat the distributionsof sparrows variability obviate the need for additional bevariedsignificantlybetweenyearsat a regional havioraladjustments. scale.Thispatterncouldresultif sparrowsunMovement patternsandcommunitystructure.derwent a brief period of exploratorymove- FollowingHutchinson's (1951)conceptof a fumentupon arriving on the winteringgrounds. gitive species,severaltheoreticalstudieshave This would allow them to match their distrisuggestedthat variationin movementpatterns bution to the variableand patchydistribution functionsas a nicheaxis,allowingthe coexisof resourcesin any given year, driven by the tence of otherwiseecologicallyidenticalspesummer rains. cies in competitively structured communities In contrastto the between-yearvariability of (e.g. Levins and Culver 1971, Horn and Macis the summer rainfall pattern, the pattern of Arthur 1972,Tilman 1994).This coexistence summer rainfall remains fixed over the course permitted by a tradeoff between efficiencyat of any givenwinter. For this reason,the distri- exploringfor new patchesand efficiencyat exbutionsof seedsand grasscoverremain rela- ploiting a particular patch (Brown 1989, Tiltively constantduring the courseof a winter.It man 1994). In other words, the best colonizer is unlikelythat sparrowsalter thisdistribution of new patchescannotalsobe the bestcompetsignificantlyduringthe courseof mostwinters, itor within patches. becausethey consumeonly 20 to 30% of the In recent decades,the validity of equilibriseed resourcesin many winters (PullJamand um-based, competition-structuredcoexistence Dunning 1987). Fixed home ranges during theorieshas been questioned(Connell 1975, midwintermay be an adaptationto exploitthis Connorand Simberloff1979,Strong1983),parwithin-winter constancyby remainingseden- ticularly in highly fluctuating environments tary and conservingenergy. such as grasslandand shrubsteppe(Wiens Plasticity ofmovement patterns.--Although Sa- 1974, Rotenberryand Wiens 1980). The ratio- 756 CALEB E. GOl•OON [Auk,Vol. 117 nale for this argumentis that in communities that they shouldbe able to make a living on wherebiologicalinteractions neverreachequi- patcheswith lower resourcedensities. librium and consumers do not saturate re- sources,competitivepressureshould be low. Withoutcompetition,no reasonexistsfor niches to be competitively structured. Resource partitioning is not required for speciesto coexist. The wide fluctuationsin sparrow abundanceacrosstime and space(Table1) suggest thatwintergrassland sparrowassemblages in- Significanceof movementstrategiesin a fragmentedlandscape.--Recent modelshave shown that sedentaryspecies,includingcompetitive dominants, are more adversely affected by landscapefragmentationthan are moremobile species(Dytham 1994;Tilman et al. 1994,1997; Moilanenand Hanski 1995).If suitablepatches becomewidely separatedin space,sedentary speciesare unableto find and colonizethem. deed are far from being in equilibrium. Furthermore,sparrowsare not believedto saturate Danielson (1991) and Pulliam and Danielson seed resourcesduring most winters (Pulliam (1991)modeledthe effectof sedentarybehavior on habitat selectionby varying the amountof and Brand 1975, Pulliam and Parker 1979, Pulliam and Dunning 1987).Therefore,it may not patchesthat individualssampledbeforethey be appropriateto invokethe notionof resource chosea patchto inhabit.In their simulations, partitioningto explainthe significantdiffer- individuals that sampled a high number of patchesoccupieda higher percentageof the betterhabitatthan individualswith low patch competitive coexistencemay not apply to co- sampling.In landscapeswhere high-quality habitat patcheswere rare, the ability to find existingspeciesof grasslandsparrows. Nonetheless, competitivestructuringshould goodpatcheslimitedpopulationgrowth,anda not be ruled out completely.Competitiondoes high level of exploratorypatchsamplingwas encesin movement patterns among speciesre- cordedin this study. In short, the notion of not need to be intense or constant to affect the distributionof nichesin a community(Connell 1980). PullJam(1985) suggestedthat competition and resourcepartitioningamongsparrows winteringin Arizona grasslands are important only in rare yearswhen seedproductionis extremelylow and competitionfor seedsmay be especiallyintense.Furthermore,evenif predatorslimit sparrowpopulationsbelow their ability to saturatefood resources,nichepartitioning may be required for coexistence(Holt 1977).Therefore,differences in movementpat- necessaryfor speciesto survive. This has an important implicationfor the managementof highly sedentaryspeciessuch as GrasshopperSparrows, Baird's Sparrows, and Cassin'sSparrows.In additionto beingaffectedby pure increasesor decreases in their preferredhabitattypes,sedentaryspecies may suffer increased adverse effects relative to more mobile speciesif the distributionof suitable habitat becomesfragmented.One important caveatis that in this study,the mobility of sparrows was measured at the scale of several hect- ares.This small scalemay not accuratelyreflect terns may function as an important coexistence regionalmovements, which perhapsare more mechanismfor ecologicallysimilar grassland appropriatelymatchedto the spatiotemporal sparrowswintering in Arizona grasslands. scaleof environmentalvariabilityin this sysEvenin the absenceof nichepartitioningin tem.Figure3 suggests thattheselocal,within- responseto competition,the tradeoff between year mobility differencesindeedreflectmovewithin-patch efficiency and among-patch ment constraintsat the regionalscaleas well. searchingmay posesignificantconstraintsfor At the between-yearand betweenstudy site sparrows. Speciesthat sacrificedsome ability scales,abundances of the more locallysedento explorethe landscapemay have done so to tary specieswere more poorly correlatedwith gain efficiencyat the local scale.Indeed, the summerrainfall (a negativecorrelation in the three mostsedentaryspeciesin my study,Cas- caseof Baird'sSparrow)than were abundances sin's Sparrow, Grasshopper Sparrow, and of the moremobilespecies.AlternativehypothBaird'sSparrow,are short-wingedrelativeto esescannotbe ruled out, but a likely explanaothersparrowsand areknownto be weak fliers tion for this patternis that sedentarybehavior (Rising1996).Thisleadsto the predictionthat constrainsthe ability of GrasshopperSparthese sedentaryspeciesare more locally effi- rows, Baird'sSparrows,and Cassin'sSparrows cient than are the more mobile sparrows,and to matchtheir winter spatial distributionsto July2000] Sparrow Movement Patterns Baird's Sparrow 40R2 =0.18 30 Vesper Sparrow 2ooR2= 0.64 ø • 150 20 •• 100 10 50 0 03 757 0 100 0 200 300 0 GrasshopperSparrow 100 200 Savannah Sparrow 500R2 420R2 4oo=0.21, 400 =0.38• 300 80••e ,o 20 200 100 o 0 300 40 , 100 *, , 200 o 300 0 100 200 , 300 o Cassin's Sparrow o Brewer'sSparrow 40 70 30 50 ,o 40 20 30 20 10 10 0 i , , 100 200 300 • 0 . . 100 200 300 Total July-Septemberrainfall (mm) FIc. 3. Relationship betweensummerrainfalland sparrowcapturesin subsequent winter fromJanuary to March1996to 1999.The threemostsedentaryspeciesat left are muchmorepoorly(or negatively)correlatedwith rainfallthanthe threemoremobilespeciesat right,suggesting thatmovementconstraints play an importantrolein large-scalehabitatselection.Eachdatapoint represents the totalnumberof individuals capturedin sevenflush-nettingsessions at a singlestudysiteper winter.The 1997and 1998BANWRdata setswere standardizedto sevenflush-nettingdatesby discardingdatafrom the eighthand ninth netting sessions. I extrapolatedthe 1999totalsfrom threeto sevendatesby dividingthe three-weektotalsby the averageproportionof birdsin a seven-week datasetcapturedby thethird week.Tocalculatetheseaverages, I divided the total numberof birdscaughtby the third weekby the total numberof birds that had been caughtby the seventhweekin all datasetswith at leastsevennettingdates.Julyto Septemberrainfalldata arefrom rain gaugesat the headquarters of TRR andBANWR,1 to 5 km from all studyplots.R2valuesare givenonly for specieswith sevendatapoints.Brewer'sSparrowsand Cassin's Sparrowsdo not occurregularly at the higher-elevation TRRstudysite;hence,onlythreedatapointsrepresenting threeflush-netting seasonsat BANWR are presented. the distributionof preferredconditions(high rainfall areas)in the landscape. The grassland-obligate GrasshopperSparrows and Baird'sSparrowsundoubtedlyhave sufferedfrom widespreadconversionof grass- lands to shrublands in southeastern Arizona overthe past120years(Bahre1991).The sedentarybehaviorof thesespeciessuggeststhat managersneedto considerthe distributionof habitatpatchesin the landscape,aswell asthe 758 CALEB E. GORDON [Auk,Vol. 117 structuralcomposition of the patches.In other CLARK, M. E 1997. A review of studies of the breeding biology of Australianbirds from 1986-95:Biwords, it's not just a questionof creatingthe asesand consequences. Emu 97:283-289. right conditionsin a particularpatch:it also CONNELL, J. H. 1975. Some mechanismsproducing may matterwherethe patchis relativeto other structure in natural communities: A model and patcheswithin the range of existingpopulaevidencefrom field experiments.Pages461-490 tions of sparrows.Managing for relativelysedin Ecologyand evolution of communities(M. L. entaryspecies suchasCassin's Sparrow,Baird's Cody and J. M. Diamond, Eds.). Harvard UniSparrow,and GrasshopperSparrowmay reversityPress,Cambridge,Massachusetts. quire maintaininga landscapewith aggregat- CONNELL, J. n. 1980.Diversity and the coevolution ed patchesof suitablegrassland,or coreareas, of competitors, or the ghostof competition past. Oikos 35:131-138. as opposedto a landscapewith an equal area of grasslandspreadout amongwidely separat- CONNOR, E. E, AND D. SIMBERLOFF.1979. The assembly of speciescommunities:Chanceor compeed smallpatches. tition? Ecology60:1132-1140. DANIELSON, B. J. 1991. Communitiesin a landscape: The influenceof habitatheterogeneityon the interactionsbetweenspecies.AmericanNaturalist This work wassupportedby The ResearchRanch ACKNOWLEDGMENTS Foundation, the United States Fish and Wildlife Ser- 138:1105-1120. DAVIES,S. J. J. E 1984. Nomadism as a responseto vice (BuenosAires NWR), the researchtraining desert conditions in Australia. Journal of Arid group in the analysisof biologicaldiversificationat Environments 7:183-195. the Universityof Arizona, the departmentof EcoloDEAN, W. R. J. 1997. The distributionandbiologyof gy and EvolutionaryBiologyat theUniversityof Arnomadic birds in the Karoo, SouthAfrica. Jourizona, the American Ornithologists'Union, the nal of Biogeography24:769-779. JamesR. Silliman Foundation,and a predoctoralfellowshipfrom NSE I thank R. Robichaux,M. Rosen- DYTHAM,C. 1994. Habitat destructionand competizweig, S. Russell,R. W. Mannan, H. R. Pulliam, and M. L. Morrisonfor guidancein designingthisproject and for commentson prior versionsof the manu- tive coexistence:A cellular model. Journalof An- imal Ecology63: 490-491. Fox, G. A. 1990.Componentsof floweringtime var- iation in a desert annual. Evolution 44:1404script. I also thank the roughly 300 people who 1423. helpedflushsparrows,especiallyC. Clark,N. Crook, E. Davis,B. Flack,C. Hanus,R. Hummel,M. Kaplan, FRETWELL,S. D., AND H. L. LUCAS,Jr. 1970. On terA. Lehovich, W. Leitner, T. Locascio,J. McCabe, L. ritorial behaviorand other factorsinfluencing habitat distribution in birds. Acta Biotheoretica McMurray,A. and D. Mecikalski,C. Pope,B. Shank, 14:16-36. K. Short,J.Taylor,and B. Whitcomb. LITERATURE CITED ANDERSSON, M. 1980.Nomadism and site tenacityas alternatereproductivetacticsin birds.Journalof Animal Ecology49:175-184. HOLT, R. D. 1977. Predation, apparent competition, and the structureof prey communities. Theoretical PopulationBiology 12:197-229. HORN,H. S., ANDR. H. MACARTHUR.1972.Competition amongfugitive speciesin a harlequinenvironment. Ecology53:749-752. HUTCHINSON, G. E. 1951.Copepodologyfor the or- BAHRE, C. J.1991.A legacyof change,historichuman nithologist.Ecology32:571-577. impact on vegetationin the Arizona borderLEVIN, S.A. 1992.Theproblem of patternandscale lands. Universityof Arizona Press,Tucson. BROWN,J. H., O. J. REICHMAN, AND D. W. DAVIDSON. 1979. Granivory in desert ecosystems.Annual Review of Ecologyand Systematics10:201-227. BROWN,J. H., ANDZ. ZENG.1989.Comparativepopulation ecologyof eleven speciesof rodents in the Chihuahuandesert.Ecology70:1507-1525. BROWN,J. S. 1989. Desert rodent community structure: A test of four mechanisms of coexistence. EcologicalMonographs59:1-20. BURNHAM, K. P., D. R. ANDERSON, G. C. WHITE, C. in ecology.Ecology73:1943-1967. LEVINS,R., AND D. CULVER.1971. Regional coexistence of speciesand competitionbetweenrare species.Proceedings of theNationalAcademyof Sciences USA 68:1246-1248. LIMA, S. L., AND T. J. VALONE. 1991. Predators and avian community organization: An experiment in a semi-desertgrassland.Oecologia86:105112. MARES, M. A., AND M. k. ROSENZWEIG.1978. Grani- vory in North and SouthAmericandeserts:Roanalysismethodsfor fish survivalexperiments dents,birds, and ants.Ecology59:235-241. basedon release-recapture. AmericanFisheries MCCLARAN, M.P. 1995.Desertgrasslandsand grassSocietyMonographsNo. 5. es. Pages 1-30 in The desert grassland(M.P. BROWNIE,ANDK. H. POLLOCK. 1987. Design and July2000] Sparrow Movement Patterns McClaran and T. V. Van Devender, Eds.). Uni- versity of Arizona Press,Tucson. MOILANEN, g., AND I. HANSKI. 1995. Habitat destruc- tion and coexistenceof competitorsin a spatially-realistic metapopulation model. Journal of Animal Ecology64:141-144. NoY-MEIR, I. 1973. Desert ecosystems: Environment and producers.Annual Review of Ecologyand Systematics4:25-41. PFENNIG, D. W. 1992.Polyphenism in spadefoot toad tadpolesasa locallyadjustedevolutionarilystable strategy.Evolution46:1408-1420. PHILLIPS, g., J. MARSHALL, AND G. MONSON. 1964. The birds of Arizona. Universityof Arizona Press, Tucson. PIMM, S. L., M. L. ROSENZWEIG, AND W. MITCHELL. 1985.Competitionand food selection:Field tests of a theory.Ecology66:798-807. PULLIAM,H. R. 1983. Ecologicalcommunitytheory and the coexistence of sparrows.Ecology64:45- 759 ROSENZWEIG, M. L. 1981.A theory of habitat selection. Ecology62:327-335. ROTENBERRY,J. t., AND J. g. WIENS. 1980. Habitat structure,patchiness, and avian communitiesin North Americansteppevegetation:A multivariate analysis.Ecology61:1228-1250. SHUTLER,D., AND P. J. WEATHERHEAD.1994. Move- ment patternsand territory acquisitionby male Red-winged Blackbirds. Canadian Journal of Zoology 72:712-720. SINCLAIR, g. R. E. 1984. The function of distance movementsin vertebrates.Pages240-258 in The ecologyof animal movement(I. R. Swingland and E J. Greenwood, Eds.). Clarendon Press,Oxford. STRONG, D. R., Jr. 1983.Natural variability and the manifold mechanismsof ecologicalcommunities. American Naturalist 122:636-660. STUTCHBURY, B. J. 1994. Competitionfor winter territoriesin a Neotropicalmigrant:Theroleof age, sex and color. Auk 111:63-69. 52. PULLIAM,H. R. 1985. Foragingefficiency,resource THOMPSON, D. B., J. H. BROWN, AND W. D. SPENCER. 1991. Indirect facilitation of granivorous birds partitioning, and the coexistenceof sparrow by desertrodents:Experimentalevidencefrom species.Ecology66:1829-1836. foragingpatterns.Ecology72:852-863. PULLIAM,H. R., AND M. R. BRAND.1975. The productionand utilization of seedsin plainsgrass- TILMAN, D. 1994. Competition and biodiversity in spatiallystructured habitats.Ecology75:2-16. land of southeastern Arizona.Ecology56:1158TILMAN, D., C. L. LEHMAN, AND C. YIN. 1997. Habitat 1166. PULLIAM, H. R., AND B. J. DANIELSON. 1991. Sources, sinks,and habitat selection:A landscapeperspective on population dynamics..American Naturalist 137:S50-S66. PULLIAM,H. R., AND J. B. DUNNING. 1987. The influ- enceof food supplyon local densityand diversity of sparrows.Ecology68:1009-1014. PULLIAM,H. R., AND G. S. MILLS. 1977. The use of spaceby wintering sparrows.Ecology58:13931399. PULLIAM,H. R., ANDt. g. PARKER IIE 1979. Popula- destruction,dispersal,and deterministicextinction in competitive communities.American Naturalist 149:407-435. TILMAN, D., g. M. MAY, C. L. LEHMAN, AND M. A. NOWACK. 1994. Habitat tinction debt. Nature destruction and the ex- 371:65-66. TRIPP,K. J., ANDJ. g. COLAZO.1997. Non-breeding territoriality of SemipalmatedSandpipers.Wilson Bulletin 109:630-642. WESTCOTT, D. g., AND J. N.M. SMITH. 1994. Behavior and socialorganizationduring the breedingsea- son in Mionectes oleagineus, a lekking flycatcher. tion regulationof sparrows.Fortschritteder Condor 96:672-683. Zoologie 25:137-147. RAPPOLE, J. H., AND g. R. TIPTON.1991. New harness WIENS,J. g. 1974.Climaticinstabilityand the "ecologicalsaturation"of bird communitiesin North designfor attachmentof radio transmittersto Americangrasslands.Condor76:385-400. small passerines.Journalof Field Ornithology WIENS,J.g. 1976.Populationresponsesto patchyen62:335-337. vironments.AnnualReviewof Ecologyand SysRISING,J.D. 1996.The sparrowsof the United States and Canada. AcademicPress,San Diego, California. tematics 7:81-120. AssociateEditor: J. D. Brawn