Habitat and Nesting of Le Conte's Sparrows in the Northern Tallgrass Prairie

University of Nebraska - Lincoln DigitalCommons@University of Nebraska - Lincoln

USGS Northern Prairie Wildlife Research Center US Geological Survey

2005

Habitat and nesting of Le Conte's Sparrows in the northern tallgrass prairie

Maiken Winter State University of New York

Jill A. Shaffer USGS Northern Prvirie Wildlife Research Center, [email protected]

Douglas H. Johnson USGS Northern Prairie Wildlife Research Center, [email protected]

Therese M. Donovan University of Vermont, [email protected]

W. Daniel Svedarsky University of Minnesota

See next page for additional authors

Follow this and additional works at: https://digitalcommons.unl.edu/usgsnpwrc

Part of the Other International and Area Studies Commons

Winter, Maiken; Shaffer, Jill A.; Johnson, Douglas H.; Donovan, Therese M.; Svedarsky, W. Daniel; Jones, Peter W.; and Euliss, Betty R., "Habitat and nesting of Le Conte's Sparrows in the northern tallgrass prairie" (2005). USGS Northern Prairie Wildlife Research Center. 204. https://digitalcommons.unl.edu/usgsnpwrc/204

This Article is brought to you for free and open access by the US Geological Survey at DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in USGS Northern Prairie Wildlife Research Center by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln. Authors Maiken Winter, Jill A. Shaffer, Douglas H. Johnson, Therese M. Donovan, W. Daniel Svedarsky, Peter W. Jones, and Betty R. Euliss

This article is available at DigitalCommons@University of Nebraska - Lincoln: https://digitalcommons.unl.edu/ usgsnpwrc/204 Journal of Field Ornithology, Vol. 76, No. 1 (Winter, 2005), pp. 61-71

J. Field Ornithol.76(1):61-71, 2005 Habitatand nesting of Le Conte's Sparrows in the northern tallgrass prairie

MaikenWinter,l 6 Jill A. Shaffer,2Douglas H. Johnson,2Therese M. Donovan,3 W. Daniel Svedarsky,4Peter W. Jones,5and Betty R. Euliss2

State Universityof New York,College of EnvironmentalSciences and ForestrynSyracuse) New York13210 USA 2 U,S, GeologicalSurvey) Northern Prvirie WildlifeResearch Center Jamestown) North Dakota 58401 USA 3 U.S. GeologicalSurvey, Vermont Cooperative Fish and TMildlifeResearch Unit) Universityof Vermont) Burlingron,Vermont 05405 USA 4 NorthwestResearch and OutreachCenter, University of Minnesota, Crookston,Minnesota 56716 USA 5 Natural ResourcesCommittee, South Burlington, Vermont05403 USA Received5 April2004; accepted 24 May2004

ABSTRACT.Little is knownabout the breedingbiology of the Le Conte'sSparrow (Ammodramus leconteil), probablybecause of its secretivenature. We providenew informationon severalaspects of Le Conte'sSparrow breedingbiology, including rates of nestparasitism by Brown-headedCowbirds (Molothrus ater) and potential factors affectingbreeding densities and nesting success of the species.Our study was conducted in the tallgrassprairie of northwesternMinnesota and southeastern North Dakota during 1998-2002. Breeding densities varied among years, but this variationwas not clearlylinked to climaticpatterns. Vegetation had someinfluence on densitiesof Le Conte'sSparrows; densities were highest in grasslandswith moderateamounts of bareground. Prairie patch size andthe percentageof shrubsand trees in the landscapehad no recognizableinfluence on density.Nesting success washighly variable among sites and years and increased slightly with distancefrom trees. Rates of nestparasitism werelow (1 of 50 nestsparasitized), and clutch sizes were similar to thoseof otherstudies of LeConte's Sparrows. SINOPSIS. Habitat y anidamiento deAmmodramus leconteii en las praderas de yerbas altas del norte Se concocepoco sobre el anidamientodel pinzonAmmodramus leconteii, probablemente por su naturalezasecre- tiva.Proveemos informacion de variosaspectos de la biologiareproductiva, incluyendo la tasade parasitismode partede Molothrusater, y los factorespotenciales que afectanla densidadreproductiva y el exitode anidamiento de la especie.El estudiose llevoa cabode 1998-2002,en las praderasde yerbasaltas del noroestede Minesotay el surestede Dakotadel Norte.La densidadreproductiva vario entre los diferentesahos de estudio,aunque la variacionno estuvovinculada a patronesclimaticos. La vegetacion influyo en la densidaddel ave:se encontraron las mayoresdensidades en yerbasalescon areasmoderadas de suelodesnudo. El tamahodel parchoy el porcentaje de arbustosy arbolesen el paisaje,no influyoen la densidad.El exito de anidamientoresulto variable entre localidadesy ahos, y aumentoen algocon la distanciaa los arboles.La tasa de parasitismofue baja (1 de 50 nidos) y el tamahode la camadafue similaral de otrosestudios del ave. Keywords: Ammodramusleconteii, bird density, cowbird parasitism, habitat ecology, Le Conte's Sparrow, nesting biology,patch size

BreedingLe Conte'sSparrows (Ammodramus fluctuations that may be causedby changesin leconteis)occupy mesic grasslandsand marshes soil moistureand consequentchanges in vege- of centraland southernCanada and the north- tation structure(Igl and Johnson 1995, 1999; ernmostparts of the centralUnited States,but Igl 1999). Most publishedinformation on this also nest in otherhabitats such as fallowfields, specieshas been descriptiveand lacks quanti- hayfields,or drainageditches (Lowther1996; tativeevaluation of factorsthat influencebreed- Igl and Johnson 1999). Little is known about ing densities and nesting success (Lowther the speciesbecause of its crypticplumage and 1996)- elusivebehavior (Lowther 1996). In addition, Here we providesome detail on severalas- this species exhibits large annual population pects of the breedingbiology of the Le Conte's Sparrow.In particular,we reportinformation on inter-annualand inter-regionalvariation in 6 Correspondingauthor. Current address: Labora- tory of Ornithology,Cornell University, 159 Sap- breedingdensities; relationships between breed- suckerWoods Rd., Ithaca,New York14850 USA. ing density and nesting success,and habitat, Email:[email protected] landscape,and climaticvariables; and various

61 J. Field Ornithol. 62 M. Winteret al. Winter 2005

Crooksto with a preference for areas of little shrub or tree cover. In addition, we selected study plots based on size of the grassland patch (small, <65 ha; large, >100 ha) and on the percentage of shrubs and trees within a 1-km radius (treed, >10%; open, The mean size of the contiguous grassland-10%). patches in which study plots were located was 226.1 ha (SD = 341.7; range, 2.4-1245.6 ha), and the mean percent- of within the •50kd;'16 O• age shrubs and trees landscape was 12.4% (SD = 10.9; range, 1-46%). Most Glyndon of our study sites were relatively level and prone a 0m to flooding after heavy rains, thus providing habitat suitable for the Le Conte's Sparrow Sheyenne (Lowther 1996). We defined a grassland patch as a contiguous stretch of native or reseeded grassland that was not interrupted across 75% of its width. Inter- could be two- or four-lane roads 1. area in the northern ruptions by Fig. Study tallgrassprairie, their or wooded 1998-2002. The was conducted in three re- (and rights-of-way), by strips study or wetlands that were at least 20 m wide. The gions near Crookston and Glyndon, Minnesota sizes of a (MN), and in Sheyenne National Grassland,North patches containing study plot ("patch were determined from Dakota (ND), with each region containing 11-18 size") digital orthophoto study plots. Bird surveys were conducted in each quads (DOQ's) in a geographic information study plot, whereas nest searchingwas conducted in system (Map and Image Processing System 6.6 a subset of the study plots. and 6.7; MicroImages 2002). The DOQ's also were used to delineate and calculate areas of trees or shrubs within and to a distance of 1 aspects of breeding ecology including philo- km from the edge of each study plot ("land- patry, nesting success, rates of nest parasitism scape"). A 1-km radius was large enough to in- by Brown-headed Cowbirds (Molothrus ater), clude landscape data from outside even the and clutch size. largest grassland patches, but small enough not to overlap with the buffer zone of a neighboring METHODS AND ANALYSIS study plot. Because the size of the buffer zone varied with the size of the study plot, we used Study sites. We examined variation in the percentage of the total area that was covered breeding densities of Le Conte's Sparrows in 44 by shrubs and trees within the 1-km buffer in study plots that were established in three re- all analyses. We were not able to distinguish gions of the northern tallgrass prairie (Fig. 1): other types of landscape cover, such as agricul- southeast of Crookston ("Crookston," 15 plots) tural fields, with the available GIS data. and east of Moothead ("Glyndon," 18 plots) in Field methods. On each plot we con- northwestern Minnesota (1998-2001), and at ducted bird surveys along 100-m wide transects Sheyenne National Grassland ("Sheyenne," 11 between 05:00 and 10:00 CST twice during plots) in southeastern North Dakota (1999- each field season between early June and early 2001). A total of 160 plot-years of density data July. We mapped all birds (heard or seen) and were available because not all plots were cen- their flight paths on outlines of the study plots sused during each year. Nesting success was es- to minimize double-counting. For analyses we timated on 11 plots at Crookston, 10 plots at used the maximum number of males counted Glyndon, and nine plots at Sheyenne, totaling in each plot and standardized this number to 108 plot-years. Plots ranged from 1.3 to 20.6 the number of males per 100 ha ("density"). ha in size (x = 10.6, SD = 5.5 ha, N = 44 We mist-rietted Le Conte's Sparrows from plots), and were selected based on uniformity late May to late July of each year on four study of vegetation structure within and among plots, plots in the Crookston area, using 36-mm mesh Vol. 76, No. I Le Conte's Sparrow BreedingBiology 63 nets (Winter et al. 2001). Most Le Conte's region and year. Palmer Drought Severity (PDS) Sparrowswere captured incidentally to our re- indices for May of each year were obtained for searchthat focused on SavannahSparrows (Pas- northwestern Minnesota and southeastern North serculussandwichensis), Clay-colored Sparrows Dakota from the National Climatic Data Cen- (Spizella pallida), and Bobolinks (Dolichonyx ter. Both CSM and PDS indices have previously oryzivorus). been shown to be useful predictors of abun- We quantified plot vegetation annually in dances of certain bird species (Johnson 1996; Igl early July by randomly sampling plots in such and Johnson 1999). a way that the number of samples (10 to 32) Statistical analysis. Statistical analyses varied directly with plot size. At each sampling were done using SAS software (SAS Institute site we estimated ground cover (bare ground, 1995). The Tukey-Kramer multiple comparison litter, grass, forb, and woody vegetation) within test and the maximum likelihood approach a 20 x 50 cm Daubenmire frame (Daubenmire were used to compare mean breeding densities 1959). At each corner of the Daubenmire and the variability in mean breeding densities frame, we measured maximum vegetation among study plots, years, and regions. We used height and litter depth to the nearest 0.5 cm. a mixed-model analysis to determine which var- We obtained visual obstruction readingsin each iables influenced Le Conte's Sparrow density, cardinal direction to the nearest 5 cm by plac- with study plot nested within region. Year was ing a Robel pole (Robel et al. 1970) in the included as a repeated effect, because bird center of the Daubenmire frame. For analyses, counts were conducted on the same study plots S.... L ,,,e meanvalues for each plot because for several years. We used GLIMMIX, a SAS vegetation variablesmeasured in the same plot Macro for generalized Linear Mixed Models were not independent of each other. (Wolfinger and O'Connell 1993) to determine Between 15 May and 31 July each year, we which variables influenced nesting success (as found nests by flushing adults from the nest described below), with study plot nested within vicinity or by observing adults returning to region, and year as random effect. their nests (Winter et al. 2003). Nests were In order to reduce the number of potential monitored every three to four days to deter- models used in the analyses, we identified veg- mine clutch size, incidence of cowbird parasit- etation variables that were most promising. We ism, and nest failureor success. A nest was con- randomly split the data sets into two halves and sideredparasitized by a Brown-headedCowbird ran analyses on each half, separately for each if it contained at least one cowbird egg or nes- vegetation variable (as linear or squared values); tling, and it was considered successful if it this was done 10 times, each time using differ- fledged at least one host offspring. After a nest ent, randomly drawn data subsets. We used this became inactive, we measured the same vege- procedure in lieu of procedures such as stepwise tation variablesas described for plot vegetation regression because such analyses yield biased at four sampling points within 0.5 m of the standard errors and P-values. Variables were in- nest site and at one sampling point directly at cluded in all further analyses on breeding den- the nest site (Winter 1999a). For analyses we sity if they had a P < 0.1 more frequently than used the mean values derived from the five sam- in a quarter of all analyses. Because of the low pling points at one nest. In addition, we mea- number of significant effects found for nesting sured the height of each nest from the ground success, we used any variable that had a P < to the bottom of the nest cup to the nearest 0.3 for the final analysis. Then we ran all pos- 0.5 cm. We visually estimated the percentageof sible combinations of models that included any the nest concealed by vegetation when viewed of the selected vegetation variables and their in- from above ("nest cover") and the distance of teractive effects with region. Because of the each nest to the nearest tree. small number of nests, we minimized the num- We obtained monthly precipitation data for ber of models on nesting success by excluding 1996-2001 for each region from the National interactive terms. We used Akaike's Information Climatic Data Center (http://lwf.ncdc.noaa.gov/ Criterion for model selection adjusted for small oa/ncdc.html). From the precipitationdata, we sample sizes (AICc; Anderson and Burnham calculatedthe Conserved Soil Moisture (CSM) 2002) to determine vegetation models that best index (Williams and Robertson 1965) for each described the data. J. FieldOrnithol. 4 M. Winter et v1. Winter2005 For all furtheranalyses of densityor nesting partialdepredation. Only those time periods success,we consideredthe modelwith the low- were includedthat providedclutch-size infor- est Akaikevalue (either the null model or a mationfor morethan one nest. Clutchsize and model that had a vegetationvariable added) as nest and plot vegetationvariables are presented base model. To determinewhether distance to as means+ SD. trees(for nestingsuccess only), patchsize, percentageof shrubsand treesin the landscape,or RESULTS AND DISCUSSION climaticvariables improved the fit of the base model,we sequentiallyadded those variables to Density and return rates. Meandensities the base model.All variablesthat had a AAICC of Le Conte'sSparrows on our studyareas (42.5 value <4 in the previousset of models were males + 3.7/100 ha) weresimilar to or higher kept for subsequentmodels becausewe could than densitiesreported in other studiesin the not excludethe possibilitythat those variables northernGreat Plains. Densities were compa- had some influenceon nestingsuccess. After all rableto thoserecorded in ConservationReserve variableshad been added (yieldingthe "final Programfields in EddyCounty, North Dakota, model"), we calculatedmodel-averaged esti- duringthe peaksof irruption(Igl andJohnson matesand their 90% confidenceintervals (An- 1995, 1999). Densitieswere muchhigher than dersonet al. 2000). However,if the finalmodel those describedby other studiesin North Da- had a lSAIC:value >4 comparedto the base kota (Renken and Dinsmore 1987; Madden model,we consideredthe basemodel to be the 1996), which reportedfewer than 5 males/100 best model and did not performmodel-aver- ha. Averagedensities of other speciesin our aging. study were 91.1 males/100 ha for Savannah We computedtWO estimatesof daily proba- Sparrows,48.6 males/100 ha for Bobolinks, bility of nest survival:the Mayfieldestimates and 20.4 males/100ha for GrasshopperSpar- (Mayfield1975; Johnson 1979) and estimates rows. from logistic-exposuremodels (Shaffer2004). Density of Le Conte's Sparrowsvaried Whereasthe Mayfieldmethod requires nests to among regionsand years(Table 1) . Eachyear, be groupedinto specificcategories, the newly densityof Le Conte'sSparrows was highestin developedlogistic-exposure method allows each Crookstonand lowestat Sheyenne,despite the nest to have uniquevalues of covariates.This specles, c ramatlc range expanslon ln recent methodis thereforemore appropriate when in- years(Igl and Johnson 1995, 1999; Leddyet vestigatingcertain factors that might influence al. 1999). The Sheyennesite, locatedin south- nestingsuccess. For the logistic-exposureanal- easternNorth Dakota, is near the southern ysis,we split the datainto two nestingintervals edge of the species' typical breedingrange (beforeand afterthe penultimatenest check), (Lowther1996; Igl andJohnson 1999), where- such that the numberof observationintervals as the Crookstonarea is closer to the core. used in the analysisis greaterthan the number Within each region, mean densitywas much of nests (observationintervals should not be lower in 2001 than in the other yearsof the confused with the intervals between nest study (Table1). checks).To estimatethe probabilityof nestsur- In spite of the regionaland annualdifferenc- vival throughoutthe incubationand nestling es in densitiesof Le Conte'sSparrows, the spe- stages,we raisedeither daily nest survivalrate cies'densities were more variable among study to the powerof 21 (assuming13 d for the in- plots within a region(26% of all variationin cubationperiod and 8 d for the nestlingperiod; densitywas explainedby variationamong study Ehrlichet al. 1988).We also usedthese interval plots) than among regions (13%) or among lengthsto determinedates of nest initiationby yearswithin a region(10%). The errorterm (as back-datingnests with known hatching or derivedby maximumlikelihood) accounted for fledgingdates. 51% of all variation,indicating that much of We calculatedmean clutch size within 10-d the variationin densitywas the resultof year- intervals to determine whether clutch size to-yearvariation within individualstudy plots. changedduring the breedingseason, using only This variationwas almosttwice as high as for nests found duringincubation, to decreasethe the Clay-coloredand Savannahsparrows (Win- likelihoodthat clutchsize had been reducedby ter et al., in press). Vol. 76, No. 1 Le Conte's Sparrow Breeding Biology 65

Table 1. Average density of Le Conte's Sparrows(males per 100 ha) in study plots in northwesternMin- nesota (Crookston and Glyndon) and southeasternNorth Dakota (Sheyenne), USA. Differences in density yearswithin and differences regions significant at = 0.05, as derived from the Tukey- among region among o- Kramertest, are indicated by different superscripts(N = number of study plots surveyedper year). Crookston Glyndon Sheyenne Year x SD N k SD N i SD N 1998 72.7a 67.5 15 53.9a 53.2 18 0 1999 62.2a 36.3 15 44.0a 34.6 18 18.2a 27.2 11 2000 77.0a 57.4 15 40.8a 42.3 18 18.7a 20.2 11 2001 41.0b 33.7 14 9.3b 17.4 14 1.7b 2.9 11 Total 63.6a 51.6 59 38.6b 42.3 68 12.8c 20.6 33

Large annual fluctuations in Le Conte's Spar- dence that the return rates were as low as they row density have also been described by Igl and appeared. In the one other study that docu- Johnson (1995, 1999). Erratic population fluc- mented return rates in Le Conte's Sparrows tuations are typical for most grassland passer- (Murray 1969), one of seven males and none ines because they are highly dependent on spe- of six female birds were resighted. cific vegetation features that vary dramatically Low return rates also have been documented in response to local moisture conditions (Igl for other Ammodramus species, such as the and Johnson 1999). This is especially true for Baird's Sparrow (A. bairdii; Green et al. 2002), Le Conte's Sparrows, which require relatively the Henslow's Sparrow (A. henslowii; Skipper mesic habitats (Dechant et al. 2003). High var- 1998; Herkert et al. 2002), and the Grasshop- iability in population sizes reflects the species' per Sparrow (A. savannarum; Vickery 1996; ability to respond quickly to changing habitat Skipper 1998). Low return rates are assumed to quality and is essential for birds inhabiting be typical of grassland sparrows in the northern highly variable environments (Wiens 1973). A Great Plains of North America because of the species that is not flexible in its choice of breed- area's highly variable climatic conditions (Igl ing location is much more prone to experience and Johnson 1999). In contrast, high return the negative effects of habitat deterioration, as rates are found among Ammodramus species in is the case for the sedentary Cape Sable Seaside eastern North America, such as the Grasshop- Sparrow (A. maritimus mirabilis; Jenkins et al. per Sparrow (Vickery 1996), the Saltmarsh 2003). Sharp-tailed Sparrow (A. caudacutus, Greenlaw The large annual variation in density within and Rising 1994; DiQuinzio et al. 2001), and our study plots is consistent with the low return the sedentary Florida Grasshopper Sparrow (A. rates that we documented in our study. Of the savannarum floridanus, Delany et al. 1995). 93 birds that were banded, only one was de- It seems likely that annual variation in cli- tected on our study area in subsequent years. A matic conditions played some role in the an- female returned to the same site, a small (15- nual variation in Le Conte's Sparrow density. ha) prairie surrounded by trees, a year after she Igl and Johnson (1999) found a strong rela- had been banded. Le Conte's Sparrows have tionship between density in this species and cli- been reported to exhibit a "colonial" tendency matic factors. However, we found no effect of when breeding (Lowther 1996), and breeding either the PDS index or the CSM index on densities on our study plots at which Le Conte's breeding density (Tables 2, 3), possibly because Sparrows were banded were relatively high (58 climatic variability during our study was too to 130 males/100 ha). Therefore, the likelihood low to detect an effect. of recapturing or resighting a banded bird in tew of the vegetation variables that we mea- the years following banding should have been sured affected Le Conte's Sparrow density. Two relatively high, if it returned. However, the Le models that described Le Conte's Sparrow den- Conte's Sparrow was not a focal species, and sity had a AAICc value <4. The first model therefore effort and capture rates of this species included year, the squared value of bare ground, across years were unequal, reducing our confi- and the interactive effect of region and vegeta- J. FieldOrnithol. 66 M. Winter et al. Winter2005

Table 2. The final model for Le Conte's Sparrowdensity was determined by sequentiallyadding variables to the best vegetation model, which included year, bare ground, and region X vegetation height. Data were collected in study plots in three regions of the northern tallgrassprairie, USA, 1998-2001 (N = 160 plot- years). Model variablesa AAICc Akaike weight Patch size added Vegetation 0.00 0.55 Vegetation, patch size 0.65 0.40 Vegetation, patch size x region 5.08 0.04 Landscapeadded Vegetation, patch size 0.00 0.59 Vegetation, patch size, landscape 1.03 0.35 Vegetation, patch size, landscape x region 4.50 0.06 Climate addedb Vegetation, patch size, landscape 0.00 0.55 Vegetation, patch size, landscape, CSMc 1.67 0.24 Vegetation, patch size, landscape, PDSd 1.86 0.22 a At each step, all variableswere retained in the model that had a AAICc value < 4. All parameterswere estimable, and all models included two random effects (year,and plot nested within region). The number of estimable parametersis thereforethe number of variablesin the model plus 2. b All models that had a AAICc value < 4 after adding climate variableswere used for model-averaging (Table 3). c Conserved Soil Moisture index. d Palmer Drought Severity index. tion height (AAIC,:= 0; Akaike weight = 0.76, The first model therefore was used as a base N = 160 plot-years). Relative to the models model for all subsequent analyses (Table 2). considered, this model had three times greater Although the models that included patch size support by the data than the second-best model or the percentage of shrubs and trees in the (which also included the squaredvalue of grass landscape had slightly higher Akaike values cover; AAIC.l = 2.28; Akaike weight = 0.24). than the null model, there was strong support

Table 3. Model-averagedresults with 90% lower and upper confidence limits (LC and UC) on the effect of vegetation variables, patch size, the percentage of shrubs and trees within 1 km surrounding the study plots ("Landscape"),and climate (ConservedSoil Moisture index [CSM], and PalmerDrought Severityindex [PDS]) on Le Conte's Sparrowdensity in the northern tallgrassprairie, USA, 1998-2001 (N = 160 plot- years).

Parameter Year Region Estimate LC UC Intercept 1.95 -47.61 51.51 (Bare ground cover)2 -0.05 -0.07 -0.03 Vegetation height x region Crookston 1.29 0.71 1.86 Glyndon 0.49 -0.54 1.52 Sheyenne 0.16 -0.55 0.88 Patch size 0.01 -0.01 0.03 Landscape -0.44 -1.06 0.18 PDS 1.14 -3.35 5.63 CSM -0.04 -0.19 0.11 Year 1998 26.53 7.40 45.66 1999 24.92 8.86 40.99 2000 22.45 9.97 34.94 Vol. 76, No. 1 Le Contes SparrowBreeding Biology 67 fromthe data (40%and 35%, respectively)for severalmeters away from the nest (Winteret the model that includedboth vegetationvari- al. 2003). In our study,we found 42 nests in ablesand patchsize or the percentageof shrubs the Crookston region, eight nests in the and trees(Table 2). AAICCvalues for the veg- Glyndonregion, and none at SheyenneNation- etation models that includedpatch size, per- al Grassland.These numbersreflect the varia- centageshrubs and treesin the landscape,and tion in density of Le Conte'sSparrows across climaticvariables were similarto that of the the threeregions (Lowther 1996; this study). best vegetation model that did not include In our study area,the species'breeding sea- thoseadditional factors (Table 2). We therefore son appearedto be relativelyshort: the nestswe averagedthe estimates and standard errors found were initiatedbetween 17 May and 8 acrossthe best threemodels (Burnham and An- July.Between 24 May and 3 June, 53% of all derson2002; Table3). nests were initiated,with a mediandate of 29 Model-averagingshowed only threevariables May. Becausewe concentratednest-searching that had confidenceintervals that did not in- effortson other species,we may have missed cludezero (Table3), thus indicatinga close re- potentialre-nesting or double-broodingevents. lationshipbetween those variablesand density These seem likely given that the length of the in Le Conte's Sparrows.Le Conte's Sparrow Le Conte'sSparrow breeding season can extend density decreasedwith either high or low to the end of Augustor even earlySeptember amountsof bareground, increased with higher (Murray1969; Faanes1981), and that some vegetationat Crookston, and varied among Ammodramusspecies have been reportedto years. As a typical Ammodramusspecies, Le havetwo broodsper season(Greenlaw and Ris- Conte'sSparrows forage mainly on or nearthe ing 1994; Vickery1996; Winter 1999a;Green ground(Lowther 1996). Moderateground cov- et al. 2002; Herkertet al. 2002). er is probablybeneficial for ground-foraging Nests containedon average4.51 (+0.56, N speciesbecause it providesa balancebetween = 33) eggs,which is similarto the clutchsizes foragingefficiency and coverfrom predators. A reportedfrom othernesting studies of Le Con- similarrelationship has previouslybeen shown te's Sparrows(Lowther 1996). The clutch size for Baird'sSparrows (Winter 1999b). also is similarto otherAmmodrarmus species in- These resultsindicate that predictionsof Le habitingtallgrass or mixed-grassprairie in the Conte'sSparrow density based on patch size Midwest(Vickery 1996; Davisand Sealy1998; and percentageof shrubsand treesin the land- Greenet al. 2002; Herkertet al. 2002; Davis scapewill be uselessif the models do not in- 2003b). As shownfor otherspecies (Dhondt et cludeinformation about local vegetation struc- al. 2002; Davis 2003b), clutch size decreased ture.Similarly, Johnson and Igl (2001) did not later in the breedingseason (rate: - 0.18 + find an overalleffect of patchsize on Le Conte's 0.03 eggs per 10 d, X = 0.92, F= 34.8, P= Sparrowdensity in the northernGreat Plains. 0.010, N= 5 10-d intervals).An averageof This patterndiffers from that in most other 3.32 (+1.61, N= 34) eggs hatchedand 2.24 Ammodramusspecies, which have been de- (+2.01, N= 45) young fledged. From suc- scribedas areasensitive (Vickery 1996; Winter cessful nests, 3.88 young (+0.71, N= 26) and Faaborg1999; Benoit and Askins 2002; fledged.The numberof young fledgedfrom all Greenet al. 2002; Herkertet al. 2002; Davis nests combinedwas higherthan that reported 2003a).In our study,mean patch size may have for the closelyrelated Henslow's Sparrow (Win- been too large, and the percentageof shrubs ter 1999a) and Baird'sSparrow (Davis 2003b). and treesin the landscapetoo low, to demon- Of 45 nestswith knownfates, 26 nestswere strateinfluences on distributionsin Le Conte's successful,18 nests were depredated,and one Sparrows. nest was abandonedfor unknownreasons. The Nesting biology. We found 50 Le Conte's estimatedMayfield daily nest survival rate based Sparrownests during five years. Few Le Conte's on 432 exposuredays and 19 unsuccessfulnests Sparrownests have been found, probablybe- was 0.956 + 0.009 (N= 45). This estimate cause of the species'highly secretivebehavior was almostidentical to that of the logistic-ex- and well-concealednests (Lowther1996). Like posuremodel (0.957, 95% CL 0.905-0.981). other Ammodramusspecies, this specieswalks Basedon these nesting successestimates, nest ratherthan flies to its nest and usuallyflushes survivalduring the incubationand nestlingpe- J. FieldOrnithol. 68 M. Winteret al. Winter2005

Table 4. Vegetation characteristicsof Le Conte's Sparrownests and of plots in which Le Conte's Sparrows were found in the northern tallgrassprairie, USA, 1998-2002.

Nest (N = 43)a Plot (N = 109)b i SD x SD t P Bare ground cover (%) 0.5 1.5 3.3 7.8 -3.70 < 0.001 Grass cover (%) 37.6 15.8 35.0 9.9 -0.07 0.948 Forb cover (%) 15.6 11.6 20.7 9.3 -1.30 0.202 Litter cover (%) 45.3 17.3 38.7 13.3 2.23 0.033 Woody cover (%) 0.2 0.7 1.4 1.7 -6.71 < 0.001 Litter depth (cm) 6.1 2.5 5.2 2.8 1.33 0.185 Vegetation height (cm) 37.3 9.6 46.8 9.1 -3.16 0.002 Visual obstruction (dm) 1.9 0.9 2.6 0.8 -3.24 0.001 a Number of nests for which vegetation was characterized. b Number of plot-years. For analysiswe used only those study plots in which nests were found (N = 27 plots). riods was 38.9% and 39.7%, respectively.This SD = 21.5, N = 46). Well-hidden nests are rate of nesting success was higher than the rates typical for most Ammodramus species (Lowther for three other grassland species that we have 1996; Vickery 1996; Winter 1999a; Green et investigated so far (Savannah Sparrow, Clay- al. 2002; Herkert et al. 2002). Although nests colored Sparrow,and Bobolink; M. Winter et were on average placed far from trees (k = al., unpubl. data). Potential factors influencing 123.1 m, N = 46), variation in nest placement high rates of nesting success (and low rates of among nests was considerable (SD = 124.6). brood parasitism, as described below) are well- We therefore have no clear evidence that Le hidden nests, secretive behavior, distance of Conte's Sparrows avoid trees for nest place- nests from potential perch sites for avian nest ment, as has been documented for several grass- predatorsand cowbirds, and the location of our land passerines (O'Leary and Nyberg 2000; study areas outside the area of highest cowbird Johnson et al. 2003). Compared to the vege- densities (based on Breeding Bird Survey data; tation at random sites in the study plots in Price et al. 1995). which Le Conte's Sparrows were found, nest Only one of 50 Le Conte's Sparrownests was sites were in areas with less bare ground and parasitized by a Brown-headed Cowbird. This woody vegetation and with higher litter cover nest contained two host young and two cow- (Table 4). In addition, Le Conte's Sparrows bird eggs; all young were depredatedduring the chose nest sites that were surrounded by shorter nestling stage. This cowbird parasitismrate was and less dense vegetation. lower than that of any focal grasslandpasserine None of the nest vegetation variables could that we studied (M. Winter et al., unpubl. explain variation in nesting success because data). Although Lowther (1996) noted that none of the vegetation models had an Akaike cowbird parasitismoccurred relatively frequent- value lower than the null model. Vegetation ly in Le Conte's Sparrows,that view was based may be a poor indicator of the likelihood of a largely on studies with sample sizes of fewer nest being successful (Winter 1999a; Hughes et than six nests. Among other Ammodramusspe- al. 2000; Winter et al., in press). This apparent cies, low parasitismrates have been reportedfor lack of influence of vegetation structure on Henslow's Sparrows(Winter 1999a; Herkert et nesting success might be due to the diverse spe- al. 2002; Shaffer et al. 2003), but rates for cies composition and distribution of nest pred- Baird'sSparrows and GrasshopperSparrows are ators in grassland systems (Pietz and Granfors moderate to high (Shafferet al. 2003). 2000; M. Winter et al., unpubl. data). Nesting Nests typically were placed on or close to the success of Le Conte's Sparrows was most influ- ground (X = 2.6 cm high, SD = 3.6, N = 46 enced by the distance to trees (Table 5); nesting nests) within dead plant material (96% of 44 success increased slightly with increasing dis- nests were built in litter), and were largely cov- tance to trees (slope = 0.02 per 100 m, SD = ered by surrounding vegetation (X = 86.9%, 0.02, N = 39). Such a positive effect was ex- Vol. 76, No. 1 Le Conte's Sparrow Breeding Biology 69

Table 5. The final model for Le Conte's Sparrow nesting success was determined by sequentially adding variablesto the null model, using Shaffer'slogistic-exposure model (Shaffer 2004). Data were collected in study plots situated in two regions of the northern tallgrassprairie, USA, 1998-2002 (N = 39 nests). See Table 2 for analysis procedureand definitions of abbreviations.

Model variables AAICc Akaike weight Distance to tree added Distance 0.00 0.96 Null 6.36 0.04 Patch size added Distance 0.00 0.84 Distance, patch size 3.37 0.16 Landscapeadded Distance, patch size 0.00 0.79 Distance, patch size, landscape 2.63 0.21 Climate added Distance, patch size, landscape 0.00 0.64 Distance, patch size, landscape, PDS 2.05 0.23 Distance, patch size, landscape, CSM 3.25 0.13

M. Hochachka for statistical because nest and cowbird par- on the manuscript, and W. pected predation This research would not have asitism are close to and programming advice. frequently higher woody been without the enormous effort of all the field and Winter et al. possible edges (Johnson Temple 1990; assistants and volunteers who participated in the project. 2000). This study was funded by the U.S. Geological Survey and Neither patch size, percentage of shrubs and the U.S. Fish and Wildlife Service Regions 3 and 6. trees in the landscape, nor climate improved the fit of the distance model on nesting success (Ta- LITERATURE CITED ble 5). We expected lower nesting success in small grassland patches and in areas of relatively ANDERSON, D. R., ANDK. P. BURNHAM.2002. Avoiding amounts of shrubs and trees because of pitfalls when using information-theoretic methods. high of Wildlife 66: 912-918. the amount of habitat in Journal Management larger woody edge 1-_- , AND W. L. THOMPSON. 2000. Null such areas. In forested systems, small patch sizes hypothesis testing: problems, prevalence, and an al- and low forest cover in the surroundings can ternative. Journal of Wildlife Management 64: reduce nesting success of forest-breeding birds 912-923. K. ANDC. 2003. (Robinson et al. 1995; Donovan et al. 1997). BALENT, L., J. NORMENT. Demograph- ic characteristics of a Grasshopper popu- In few studies have demon- Sparrow grassland systems, lation in a highly fragmented landscape of western strated a negative effect of patch size on nesting New York State. Journal of Field Ornithology 74: success (Winter and Faaborg 1999; Balent and 341-348. R. ASKINS.2002. Norment 2003; Davis 2003a), and effects of BENOIT, L. K., AND A. Relationship between habitat area and the distribution of tidal landscape variables on nesting success have not marsh birds. Wilson Bulletin 114: 314-323. yet been clearly demonstrated. Patch sizes in BuRNHAM,K. P., AND D. R. ANDERSON.2002. Model our study may have been too large, and the selection and inference: a practical information-the- 2nd ed. New York, percentage of shrubs and trees in the landscape oretic approach, Springer, NY. DAUBENMIRE, R. 1959. A canopy-coverage method of too few, to influence nesting success of Le Con- vegetational analysis. Northwest Science 33: 43-64. te's Sparrows. DAVIs, S. K. 2003a. Habitat selection and demography of in a ACKNOWLEDGMENTS mixed-grassprairie songbirds fragmented landscape. Ph.D. dissertation. University of Regina, We appreciate the efforts of T. L. Shaffer, who de- SK. veloped the methodology that we used for our analyses . 2003b. Nesting ecology of mixed-grassprairie of nesting success. We thank L. D. Igl for thoughtful songbirdsin southern Saskatchewan.Wilson Bul- comments on earlier drafts of the manuscript, S. K. Da- letin 115: 119-130. vis and an anonymous reviewer for helpful comments , ANDS. G. SEALY.1998. Nesting biology of the J. FieldOrnithol. 70 M. Winteret al. Winter2005

Baird's Sparrow in southwestern Manitoba. Wilson PIMM. 2003. Why sparrow distributions do not Bulletin 110: 262-270. match model predictions. Animal Conservation 6: DECHANT, J. A., M. L. SONDREAL, D. H. JOHNSON, L. 39-46. D. IGL, C. M. GOLDADE, A. L. ZIMMERMAN, AND JOHNSON,D. H. 1979. Estimating nest success: the B. R. EULISS.2003. Effects of management practic- Mayfield method and an alternative. Auk 96: 651- es on grassland birds: Le Conte's Sparrow. Northern 661. Prairie Wildlife Research Center, Jamestown, ND. 1996. Waterfowl communities in the northern http://www.npwrc.usgs.gov/resource/literatr/ Plains. In: Long term studies of vertebrate com- grasbird/lcsp/lcsp.htm (12 December 2003). munities (M. L. Cody, and J. A. Smallwood, eds.), DELANY,M. E, C. T. MOORE,AND D. R. PROGULSKE. pp. 391-418. Academic Press, San Diego, CA. 1995. Territory size and movements of Florida ,, AND L. D. IGL. 2001. Area requirements of Grasshopper Sparrows. Journal of Field Ornithol- grassland birds: a regional perspective. Auk 118: ogy 66: 305-309. 24-34. DHONDT, A. A., T. L. KAST,AND P. E. ALLEN.2002. , AND J. A. DECHANTSHAFFER. 2003. Geographical differences in seasonal clutch size var- Effects-- of management practices on grassland birds. iation in multi-brooded bird species. Ibis 144: 646- Northern Prairie Wildlife Research Center, U.S. 651. Geological Survey, Jamestown, ND. http://www. D. P. W. C. AND W. R. EDD- DIQUINZIo, A., PATON, npwrc.usgs.gov/resource/literatr/grasbird/grasbird. LEMAN.2001. Site fidelity, philopatry, and survival htm (4 September 2003). of in promiscuous Saltmarsh Sharp-tailed Sparrows JOHNSON,R. G., AND S. A. TEMPLE.1990. Nest pre- Rhode Island Auk 118: 888-899. dation and brood parasitism of tallgrass prairie T. P. E. M. AND E DONOVAN, M., W. JONES, ANNAND, birds. Journal of Wildlife Management 54: 106- R. THOMPSON.1997. Variation in local-scale edge 111. effects: mechanisms and landscape context. Ecology LEDDY,K. L., K. E HIGGINS,AND D. E. NAUGLE.1999. 78: 2064-2075. Effects of wind turbines on upland nesting birds in EHRLICH,P. D. S. AND D. WHEYE.1988. R., DOBKIN, Conservation Reserve Program grasslands. Wilson The birder's handbook. Simon and Schuster, New Bulletin 111: 100-104. York, NY. LOWTHER,P. E. 1996. Le Conte's Sparrow (Ammodra- C. A. 1981. Birds of the Croix River FAANES, St. Valley: mus leconteit). In: The birds of North America (A. Minnesota and Wisconsin. North American Fauna Poole, and E Gill, eds.), no. 224. of Nat- U.S. Fish and Wildlife Academy 73, Service, Washington, ural Sciences, PA, and American Or- D.C. Philadelphia, nithologists' Union, Washington, D.C. GREEN, M. P. E. S. L. S. K. DAVIS, T., LOWHER, JONES, MADDEN, E. M. 1996. Passerine communities and bird- AND B. C. DALE. 2002. Baird's Sparrow (Ammodra- habitat relationships on prescribe-burned, mixed- mus In: The birds of North America (A. bairdiz). in North Dakota. thesis. Mon- Poole, and E Gill, eds.), no. 638. The Birds of grass prairie M.S. tana State University, Bozeman, MT. North America, Inc., Philadelphia, PA. MAYFIELD,H. F 1975. for nest GREENLAW, S., AND D. 1994. Suggestions calculating J. J. RISING. Sharp-tailed success. Wilson Bulletin 87: 456-466. Sparrow (Ammodramus caudacutus). In: The birds MICROIMAGES, INC. 2002. Lincoln, NE. of North America (A. Poole, and E Gill, eds.), no. http://www. 112. of Natural Sciences, microimages.com Academy Philadelphia, G. 1969. A of the Le PA, and American Union, MURRAY,B. comparative study Ornithologists' Washing- Conte's Auk 86: ton, D.C. and Sharp-tailed sparrows. 199- 231. HERKERT, J. R., P. D. VICKERY, AND D. E. KROODSMA. C. AND D. W 2000. Treelines 2002. Henslow's Sparrow (Ammodramushenslowiz). O'LEARY, H., NYBERG. In: The birds of North America (A. Poole, and E between fields reduce the density of grassland birds. Gill, eds.), no. 672. The Birds of North America, Natural Areas Journal 20: 243-249. P. AND D. A. GRANFORS. 2000. Inc., Philadelphia, PA. PIETZ, J., Identifying HUGHES, J. P., R. J. ROBEL, AND K. E. KEMP. 2000. predators and fates of grassland passerine nests us- miniature video cameras. of Wildlife Factors influencing Mourning Dove nesting success ing Journal in CRP fields. Journal of Wildlife Management 64: Management 64: 71-87. 1004-1008. PRICE, J., S. DROEGE, AND A. PRICE. 1995. The summer IGL, L. D. 1999. Le Conte's Sparrow: ephemeral jewel atlas of North American birds. Academic Press, of the northern Great Plains. North Dakota Out- London, UK. doors 61: 10-13. RENKEN, R. B., AND J. J. DINSMORE. 1987. Nongame ,-, AND D. H. JOHNSON.1995. Dramatic increase bird communities on managed grasslands in North of Le Conte's Sparrow in Conservation Reserve Pro- Dakota. Canadian Field-Naturalist 101: 551-557. gram fields in the northern Great Plains. Prairie ROBEL,R. J., J. N. BRIGGS,A. D. DAYTON,AND L. C. Naturalist 27: 89-94. HULBERT.1970. Relationship between visual ob- , AND -. 1999. Le Conte's Sparrows breed- struction measurements and weight of grassland ing in Conservation Reserve Program fields: precip- vegetation. Journal of Range Management 23: 295- itation and patterns of population change. Studies 297. in Avian Biology 19: 178-186. ROBINSON, S. K., E R. THOMPSON, III, T. M. DONO- JENKINS, C. N., R. D. POWELL, O. L. BASS, AND S. L. VAN, D. R. WHITEHEAD, AND J. FAABORG. 1995. Vol.76, No. 1 Le Conte's Sparrow Breeding Biology 71

Regional forest fragmentation and the nesting suc- western Missouri prairie fragments. Wilson Bulletin cess of migratory birds. Science 267: 1987-1990. 111: 515-527. . of fire on Baird's hab- SAS INSTITUTE, INC. 1995. SAS/STAT user's guide, re- 1999b. Impacts Sparrow itat in northern Studies in Avi- lease 6.03. SAS Institute, Inc., Cary, NC. mixed-grass prairie. an 19: 171-177. SHAFFER, J. A., C. M. GOLDADE, M. E DINKINS, D. H. Biology ,, AND J. FAABORG. 1999. Patterns of area-sensi- JOHNSON, L. D. IGL, AND B. R. EULISS. 2003. tivity in birds. Conservation Bi- Brown-headed Cowbirds in grasslands: their habi- grassland-nesting 13: 1324-1434. tats, hosts, and to Prairie ology response management. , S. E. HAWKS, J. A. SHAFFER, AND D. H. JOHN- Naturalist 35: 146-186. SON.2003. Guidelines for finding nests of passerine T. L. 2004. A unified to SHAFFER, approach analyzing birds in tallgrass prairie. Prairie Naturalist 35: 197- nest success. Auk 121: 526-540. 211. return to SKIPPER,C. S. 1998. Henslow's Sparrows pre- -- , D. H. JOHNSON, J. A. DECHANT, T. M. DON- vious nest site in western Maryland. North Amer- OVAN, AND W. D. SVEDARSKY. 2001. Evaluation of ican Bird Bander 23: 36-41. the bird conservation area concept in the northern VICKERY, P. D. 1996. Grasshopper Sparrow (Ammodra- tallgrass prairie. Annual report: 2001. Northern mus savannarum). In: The birds of North America Prairie Wildlife Research Center, U.S. Geological (A. Poole, and E Gill, eds.), no. 239. Academy of Survey, Jamestown, ND. http://www.npwrc.usgs. March Natural Sciences Philadelphia, PA, and American gov/resource/2002/bca2001/bca2001.htm (4 Union, Washington, D.C. 2002). Ornithologists' FAABORG.2000. Evidence for WIENS, A. 1973. Patterns and in ,AND J. J. processes grassland on levels bird communities. 43: edge effects multiple in tallgrass prairie. Ecological Monographs Condor 102: 256-266. 237-270. ANDJ. A. SHAFFER.In press.Variability WILLIAMS,G. D., AND G. W. ROBERTSON. 1965. Esti- , in vegetation-, effects on density and nesting success most wheat mating probable prairie production of grassland birds. Journal of Wildlife Management. from precipitation data. Canadian Journal of Plant WOLFINGER, R., AND M. O'CONNELL. 1993. General- Science 45: 34-47. ized linear mixed models: a pseudo-likelihood ap- WINTER, M. 1999a. Nesting behavior and nest success proach. Journal of Statistical Computation and of Henslow's Sparrows and Dickcissels in south- Simulation 48: 233-243.