Purple Martin Population Status, Nesting Habitat Characteristics, and Management in Sacramento, California

PURPLE MARTIN POPULATION STATUS, NESTING HABITAT CHARACTERISTICS, AND MANAGEMENT IN SACRAMENTO, CALIFORNIA

DANIEL A. AIROLA, Jones & Stokes, 2600 V Street, Sacramento, California 95818 JESSE GRANTHAM, National Audubon Society, 205 N. Carrizo Street, Corpus Christi, Texas 78401 (currentaddress: U.S. Fish and Wildlife Service, Hopper MountainNational Wildlife Refuge Complex, 2493 PortolaRoad, Suite A, Ventura, California 93003)

ABSTRACT:The PurpleMartin (Prognesubis) has been eliminated from mostof California'sCentral Valley. The last known populationnests in elevatedroadways ("bridges")in the city of Sacramento. We reviewed bird records (1949-1990), conductedsurveys to assesspopulation size and trend (1991-2002), and evaluated managementactions to protectand increasepopulations. Martins ceased nesting in buildingsin Sacramentoin the 1960s and 1970s as the EuropeanStarling (Sturnus vulgaris)population exploded, but adoption of verticalweep holes for nesting beneathbridges in the 1960s hasallowed martins to persist.The knownSacramento populationincreased from fourto sevencolonies and from approximately105 to 135 nestingpairs from 1992 to 2002. Martincolonies are locatedin longer(>85 m) bridge sectionsthat provideat least 6 m of unobstructedairspace beneath the colonies. Managementhas included a pilot projectto insertwire "nestguards" in weep holesto increasesurvival of young and protectingnesting coloniesduring construction activitiesnear nests.Initial resultsindicate that nestguards reduce mortality of young fallingfrom nests and do not resultin increasedcompetition for nestsites by starlings and other species.Exclusion of martinsfrom nestingat a large colonyto prevent disturbancefrom a constructionproject reduced nesting use of the site;such exclusion is not considerednecessary to protect nestingbirds. Martin recoverymay require substantialmanagement, including protecting and enhancingexisting colonies, en- couragingcolonization of otherelevated road sites, establishing martin use of artificial neststructures in unoccupiedregions, and enlistingpublic adoption of management responsibility.

The PurpleMartin (?rognesubis) was once considered"fairly common" in California(Grinnell and Miller 1944) but recentlyhas been considered rare to very uncommonin the state (Garrettand Dunn 1981, Zeiner et al. 1990, Williams 1998). The state of California identifiedthe martin as a species of special concern becauseof a well-documentedand drastic population decline and substantialreduction in the species' geographic range(Remsen 1978). A reviewof the statusof the martinin Californiaalso confirmedthat Sacramentoremains the only locationin the CentralValley where the speciesis known to breed(Williams 1998). The major suggestedcause of martin declinein Californiais competition for nest holes with the nonnativeEuropean Starling (Sturnus vulgaris) (Remsen1978, Garrett and Dunn 1981, Williams 1998, 2002). Detrimen- tal effectsof starlingand HouseSparrow (Passer domesticus) competition are well documentedelsewhere in the U.S. (Brown 1981, 1997) but not specificallyin California.Williams (1998) has also suggestedthat lossof riparianhabitat and snags that representedsuitable nesting habitat may have

Western Birds 34:235-251, 2003 235 PURPLE MARTIN NESTING IN SACRAMENTO contributed to martin decline in California. Pesticide effects have not been documented but cannot be dismissed. In the easternU.S., PurpleMartins have relied on birdhousesand gourds as nestingsites for many years(Jackson and Tate 1974, Brown 1997). In the west, Purple Martinshave only recentlymade substantialuse of nest boxesin Oregon,Washington, and BritishColumbia (Copley et al. 1999, Fouts 1996, Horvath 1999). In California,martins formerly nested com- monly in buildingsbut have seldomused nest boxes (Grinnelland Miller 1944, Garrett and Dunn 1981, Williams 1998). In the 1930s and 1940s, useof buildingsin citiesand towns in Californiawas thought to have increasedthe state'smartin populationand range (Willett 1933, Grinnell and Miller 1944). Today, however,no urban areasexcept Sacramentoare known to supportnesting martins (Williams 1998). As summarizedby Williams(1998), PurpleMartins were reported nesting in treesin Sacramentoas earlyas 1853 (Baird1858) and were documented nestingin buildingsas early as 1924 (Bryant 1924). Nestingin buildings occurredmainly in holesformed by hemicylindricalroof tiles(i.e., Spanish tiles).No recentnests have been reported in buildingsin Sacramento(B. and H. Kimball pers. comm.); all recent nestinghas occurredin weep holes beneathbridges (including elevated freeways). We initiatedthis study in 1991 and 1992 to evaluatethe recenthistorical and current populationstatus and nest site characteristicsof the Purple Martin in Sacramentoand to assessthe effectivenessof management actionsfor the species.Grantham continued monitoring through 1997 at one site(see Study Area), and Airola reinitiated monitoring in 2002 to assess the populationtrend. This assessmentalso incorporatesinformation col- lectedat Sacramentocolonies by Williams(1998). The study'sobjectives were to: ß identifyrecent historical and currentmartin colonies in the Sacramento area. ß evaluatechanges in localdistribution and habitatuse at martin colony locations. ß estimatecurrent populations and recenttrends at colonies. ß identifyfactors that may influenceselection of colonysites. ß evaluateeffectiveness of recent managementactions to protect and enhancemartin nestinghabitat. ß identifythreats to coloniesand additionalresearch and management needs.

STUDY AREA

We surveyedfor Purple Martins at bridgesand buildingsin the Sacra- mentoarea wherethey had previouslybeen observed nesting, as well as at other bridgeswith similarcharacteristics. All sitesused by martinswere in bridgesbuilt of steel and concretebox girders(Tonias 1995). All these bridgesspan urban areas and railroad tracks; none crosseswater. The bridgessupport an enclosedchamber beneath the roadsurface. Weep holes on the undersideof the chamberrelieve air pressureduring heating and

236 PURPLE MARTIN NESTING IN SACRAMENTO coolingand draincondensation, but theydo not drainwater from the road surface. The downtownarea contains the greatestnumber of bridgesin the region. We surveyedsuitable bridge sections to locatecolonies in 1991, 1992, and 2002. Colonieswere definedas occupiedsites separatedby areas of unsuitableor unoccupiedhabitat that exceedthe typicaldistances between nestswithin occupiedsites. We studiedfour occupiedcolony sites in detailin 1991 and 1992: ß I Street--Interstate 5 at I Street in Old Sacramento. ß 20th Street--U.S. Highway50 between19th and 21st streets. ß Broadway--thejunction of State Route 99 and U.S. 50, aboveand adjacentto Broadway. ß 35th Street--U.S. 50 between 34th Street and Stockton Boulevard. Granthamcontinued monitoring the 35th Streetcolony through 1997. We studiedthree additionaloccupied colonies in 2002: ß Sutterville--SuttervilleRoad overpassof the Union Pacific Railroad (UPRR)yard. ß S Street--CapitalCity Freeway(Business Route 80) at S Street. ß RosevilleRoad--I-80 and the adjacentlight rail parking-accessramp above Roseville Road and the UPRR tracks.

METHODS

Review of Recent Historical Records Becauseavailable information on the earlyhistorical status of the Purple Martin is limitedand hasbeen summarized(Williams 1998), we confinedthe statusreview to the period since the mid-1900s, for which a substantial numberof recordswere available.We reviewedapproximately 970 martin records(1949-1990) from the Sacramentoarea that Betty and Harold Kimballcompiled from their personalfield notes, as wellas recordsof other membersof the SacramentoAudubon Society (unpubl. data). Other local recordsfrom CaliforniaDepartment of Fishand Game fileswere provided by J. Estepand R. Schlorff(pers. comm.). Many recent historicalrecords noted locationsof birds near known nestingcolonies but did not confirm breedingactivity. We considered breedingat a site to be possibleif recordsnoted that birdswere present during the peak breedingseason (mid-April through July) at a specific buildingor site,probable if birdswere reportedentering holes or carrying nestingmaterial, and confirmed if adultswere observedcarrying food to a nest hole or feedingdependent young, if juvenileswere observedwithin holes,or if recentlyfledged juveniles were reportednear a knownsite. We used1965-1990 recordsonly to documenttrends in colonylocations and nestinghabitats. We could not determineabundance and detailed occupationpatterns because of limitedinformation on nestingactivity. Also, the reportedincidental counts of martinsare not reliableindicators of colony size(see Results). The firstyear of siteoccupation was especially difficult to establishbecause all potentialsites were not surveyedsystematically. Site

237 PURPLE MARTIN NESTING IN SACRAMENTO abandonmentdates are more reliablebecause once the siteswere known, observerstended to visitthem annually(B. and H. Kimballpers. comm.). Surveysfor BreedingColonies We surveyedpreviously reported martin coloniesin 1991, 1992, and 2002. We also surveyedmany other sites in Sacramentothat could be nestinghabitat; these includedlonger sectionsof elevatedfreeway, over- passeswith weep holes,and buildingswith Spanishtile roofs.We surveyed 25 potentialnesting areas from April to July by searchingfor flyingand perchedbirds. We conductedmost surveysbefore 11:00 or after 17:00 becausebirds tended to be more activeduring these periods, especially on hot days(Airola unpubl. data).

PopulationEstimation at Colonies Determiningpopulation sizes of nestingcolonies in bridgesis difficult becausemartins can be countedonly as they enter or leaveholes or when they are perchedor foragingnear the colonies.At any time, manybirds in a colonycannot be countedbecause they are insidenest holes, foraging or perchingaway from colonies,or obstructedfrom observationby the over- head structure.Also, most chamberswithin bridgescannot be enteredto observenests. Accordingly, we usedtwo primary approachesto evaluating populationsizes: observation of nestingbehaviors and hole use,and direct counts. Evaluation of NestingBehaviors and Hole Use. With the assistanceof a group of trained volunteers,we estimatedpopulations at coloniesby observingnesting behavior and mappingbird useof individualweep holes throughoutthe nestingseason. We developedand testedthis method in 1991, but becauseeffort and methodswere inconsistentthat year,we have not reportedthose results. We preparedmaps of availableweep holesfor each colonysite and assignedeach hole a uniquealphanumeric code. We visitedeach colonysite repeatedly during the nestingseason to countbirds and map holesentered by martins.During the two yearsof mostthorough surveys(1992 and 2002), we monitored each occupiedcolony for an averageof 8 hourson an averageof 13 daysat 3- to 9-day intervals(Table 1). We documentedmore than 2000 individualhole entriesby martins annuallyduring 1992 and 2002. We estimatednumbers of nestingpairs on the basisof diagnosticnesting behaviorand the frequencyof hole use. Diagnosticbehaviors that demon- stratednesting use of a weep hole by a pair includedadults entering holes with food, adultscarrying fecal sacsfrom holes, vocalizationsof begging young,visible presence of youngin nestholes, and presence of deadyoung beneath nest holes. Diagnosticnesting behaviors were difficultto observeat some sites because of limited access and limited time to observe arrivals at and departuresfrom holes(i.e., birdswere often visiblefor only a few seconds). Accordingly,we also consideredholes to be occupiedby nestingpairs if adultsmade repeatedvisits to specificholes over the courseof the nesting period.We consideredholes entered two or more timeson eachof two or

238 PURPLE MARTIN NESTING IN SACRAMENTO

Table 1 Survey Effort AssessingPurple Martin Breeding Populationsin Sacramentoduring Two Yearsof ComprehensiveSurveys

1992 2002

Colony Days Hours Days Hours

I Street 5 3.9 34 20.8 20th Street 17 5.3 12 6.7 Broadway 7 4.7 10 4.4 35th Street 20 15.0 13 12.0 Suttervillea 3 1.2 12 4.3 S Street• 3 1.3 10 5.0 RosevilleRoad b 0 0.0 4 6.1 Total 31.4 59.3

aUnoccupiedin 1992. bNotsurveyed in 1992. moredays separated by morethan one weekduring the nestingperiod to be occupiedby a nestingpair. In general,the numbersof individualvisits and dayswith visitsrecorded for mostpairs considered to be nestingexceeded this minimum standard.In addition,we combinedhole-use counts with less diagnosticbehaviors (especially carrying nesting material) to indicateoccu- pancy. Estimationof populationsby behavioralobservation and hole mappingis affectedby the amount of observationtime. Levelsof effort at three sites occupiedin both 1992 and 2002 were comparable,but we spent substan- tiallymore time monitoringthe I Streetsite in 2002 than in 1992 (Table1). Low surveyeffort for the S Streetand Suttervillesites in 2002 reflectedthe absenceof the speciesfrom these sites that year. We have included populationdata collectedby Williams(1998), who usedmethods generally consistent with ours. Direct Counts. Becausedetermining population sizes on the basisof nestingbehaviors and hole use was so labor intensive,we evaluatedthe simplermethod of directlycounting numbers of individualsobserved at colonies(direct counts). We evaluatedthe effectivenessof directcounting as a censustechnique or population index by comparingdirect counts to numbers determined on the basis of behavioral observation and hole-use mapping.Direct counts were madeduring each visit by repeatedlycounting the numberof adultsobserved flying or perchingnear coloniesor knownto be in weepholes simultaneously, then recordingthe highestof thesecounts for each day. The proportionof the total populationat each site observed duringdirect-count surveys was calculated by dividingthe maximumnumber of birds observedduring direct countsover the seasonby the number of nestingindividuals determined to be at colonieson the basisof holeuse and nestingbehavior.

239 PURPLE MARTIN NESTING IN SACRAMENTO

Habitat Characteristicsat NestingColonies Nestingsites and bridgesections not usedfor nestingwere evaluatedfor a varietyof characteristics,including ß lengthand width of the bridgesection. ß heightof holesfrom the groundor other flightobstructions. ß pedestrianactivity, lighting, and other humanuses (including construction activity). ß availabilityof nestingmaterial and perch sites. ß relative levels of vehicular traffic beneath sites. ß distanceto non-urbanforaging areas.

NestlingMortality Discoveryof deadnestlings below weep holesduring population surveys promptedus to investigatethe lossof nestlingsfrom nest holes as a mortality sourcefurther. On the assumptionthat young more than 7 days old that werefound beneath nest holes had fallenaccidentally, we hypothesizedthe causeto be the lackof any physicalbarrier at the upperedge of the vertical nest hole. We systematicallymonitored mortality beneath nests at several colonieswhere vegetationor debris did not obscureour ability to locate fallenyoung. We conductedsurveys for fallennestlings every one to three daysfrom mid-Juneto mid-July,the peak periodof susceptibility,and left a sample of dead young in the field to assesslosses due to scavengingby predatorsand pulverizationby car tires.We determinedages of youngfrom referencephotographs (Rogillio 1992, Hill 1999). Effectivenessof Nest Guardsin ReducingNestling and FledglingMortality As a pilot project,Grantham devised wire nestguards and installedthem in occupiedand suitableweep holesat the 35th Streetsite from 1992 to 1996 (Figure1). The guards,made of 0.5-inchhardware cloth (wire screen), were designedto providean internalbarrier (i.e., a "fence")to prevent youngfrom fallingfrom nests(see Kostka et al. 2003 for nest-guarddesign and installation).Effectiveness of nestguards in reducingnestling mortality was evaluatedby comparingresults of surveysfor fallenyoung before and after installation(1992 and 2002). Effectsof Nest Guardson Roostingby Fledglings Followingfiedging, Purple Martin youngtypically roost at nightin the nest cavityfor 1 to 12 days(Brown 1997). Becausewe observedrecenfiy fledged youngexperiencing difficulty reentering weep holesto roostin the evening (seeResults), nest guardswere also designedto providea wire ladderto enhance the ability of recent fledglingsto reenter holes to roost. We monitoredthe returnof fledglingsat the 35th Streetsite during the post- fiedgingperiod in 1992 and 1993 to assesstheir abilities to enterroost holes with and without nest guards.

240 PURPLE MARTIN NESTING IN SACRAMENTO

Figure 1. Female Purple Martin with food perchingat a wire '•nestguard" inserted into a weep hole beneathan elevatedfreeway in Sacramento,California. Photo by SacramentoBee/Chris Crewell

CompetitorUse of Weep Holeswith Nest Guards Weep holesappear to representthe only nestingsubstrates in the Central Valleywhere starlings have not beenable to displacemartins. Because of the concern that the insertionof nest guardscould make nest holes more accessibleto starlingsand other potentialcompetitors, Airola evaluateduse by competitorsof holeswith and withoutnest guards at the 35th Street site in 2002. Potentialeffects of wire sleeveson competitiveinteractions were analyzedby evaluatingpatterns of useby the species(i.e., useof holeswith and withoutnest guards and changesin holesused over time by martinsand competitors). Locationsof HolesUsed by Martinsand Starlingswithin Nest Colonies In 2002, Airola initiated a studyof hole-uselocations by martins and starlingsat the 35th Street colony.He mapped hole useby starlingsduring martin surveysat this site• Hole locationswere characterizedby their position(i.e., outermost,next to outermost,or interior)relative to the edge of the bridge.To assessfor selection(non-random use) of nestinglocations by eachspecies at thiscolony. the numberof holesused in eachposition was comparedagainst an expectedvalue. based on the proportionalavailability of holesin eachcategory. Differences were testedfor significanceby the ½hi- squaregoodness-of-fit test (Sokaland Rohlf 1995).

241 PURPLE MARTIN NESTING IN SACRAMENTO

RESULTS

Recent Historical Status

Martinswere recordedduring the breedingseason at 16 sitesin Sacra- mento from 1949 to 1990. Many identifiedsites supported suitable nesting habitat,but most records consisted of singleor severalisolated sightings, and breedingwas not confirmedduring most years (Figure2). Sites regularly usedby martinsprior to 1991 were assignedto sevenmajor groupsfor an assessmentof breedingstatus over the 42-year period (Figure 2). Two groups,downtown buildings and the CapitalCity Freewaysite betweenL and Q streets,are no longerused, while the other five siteshave been used since1991. Before 1964, all martin observationsand nestingrecords were at buildingsin downtownSacramento. No confirmednesting at or associa- tion with buildingswas noted after 1974. Martins were first noted near bridgesin 1964, and breedingin bridges(nest building) was first observed in 1967 (Figure2). Martin use was reported once historicallyat both the Sutterville(the site wherebridge use was first documented in 1964) and Broadwaycolonies but then went unreportedfor extendedperiods (Figure 2, Table2). Thesesmall colonies may have been overlookedor used intermittentlyduring the interveningyears. With the exceptionof the Capital City Freewaybetween L and Q streets,the three other freewaysites have been occupied regularly sincethey were firstdiscovered (Figure 2, Table 2). Number of Coloniesand PopulationStatus Four bridgessupported nesting colonies in 1991 and 1992 (Table 2). Fifteen bridgessimilar in designand characterto occupiedsites were not occupied.No martinswere found during 1991 and 1992 surveysat nine buildingswith tile roofsthat the birdshad occupiedprior to 1972, nor were

195o 1960 197o 1980 1990 i I

, DowntownBuildings CapitolCity Freeway betweenL & Q Streets

20th Street

I Street

35th Street

Sutterville Broadway ! Figure2. Known occupancyand breedingstatus of Purple Martin nestingareas in Sacramento,California, 1949-1990.

242 PURPLE MARTIN NESTING IN SACRAMENTO

Table 2 EstimatedBreeding Pairs of Purple Martins at Sacramento Colonies 1992-2002 a

Colony 1992 1993 1994 1995 1996 1997 2002

I Street 23 25 21 37 20th Street 38 48 39 40 14 Broadway 14 10 4 3 8 35th Street 30 34 36 41 34 35 29 Sutterville 0 b 4 S Street 0 b 14 Roseville Road 29

Total 105 104 105 135

øSurveysconducted in 1992 by Airola and Grantham, 1993-95 by Williams(1998), except by Grantham at 35th Street, 1996-97 by Grantham, and in 2002 by Airola. Lack of a number indicatesthat no surveywas conducted. bSitewas surveyed but no martinswere found.

they found at other suitablebuildings. Also, no martin use of buildingshas been observedor reported in subsequentyears. Starlings and House Sparrowswere observednesting at many of the formerbuilding nest sites. We found sevenactive colonies in 2002, includingthe four sitesactive 1991-1995 (Table2). Two other sites(S Streetand Sutterville)that were not activein 1991 or 1992 were activein 2002. RosevilleRoad, which was not surveyeduntil 2002, was reportedoccupied as early as 1996 but was not monitored. Distancesbetween nearest coloniesaveraged 2550 m and rangedfrom 600 to 9600 m. We documenteddiagnostic nesting cues for 62% of pairs that we characterizedas nestingin 1992 and for 70% of pairsin 2002; the balance of nestingpairs were so designatedon the basisof hole use. In 1992, 105 pairs were estimatednesting at the four colonies(Table 2); three colonies eachsupported 23-38 pairs,and the Broadwaycolony supported 14 pairs. From 1993 to 1997, numbersremained stableat most monitored sitesand for the populationas a whole. They appearto havedeclined, however, at the Broadwaysite and increasedslightly at the 35th Street site. In 2002, after a 5-year gap in monitoring,numbers of recordedpairs had increasedslightly at 35th Street and substantiallyat I Street. Numbers decreasedsubstantially at 20th Street followingblockage from 2000 to 2002 of 68% of nestingholes used in 1992; thisblockage was undertaken to preventdisturbance during construction of a rail line near the colony(G. Moir and J. Mentzer pers. comm.). The two new colonies(Sutterville, not reportedused since 1964, and S Street)together supported 18 nesting pairs, and the previouslyunsurveyed Roseville Road colony supported approximately29 pairs(Table 2). In total, between1992 and 2002 (thetwo yearsof mostconsistent and intensivesurveys) the known populationat all Sacramentocolonies increased by 32% (from 105 to 135 pairs). Despite concernsabout overestimationof populationson the basisof countsof nesthole use alone (Williams 1998), we believethat our integrated

243 PURPLE MARTIN NESTING IN SACRAMENTO approach to assessingnumbers of nesting pairs produced reasonable estimatesof actualnesting populations for most sitesin 1992 and 2002. Moreover,the consistencyin effort at most sitesprovides a reasonably reliablebasis for evaluatingpopulation trends over this period. Similarity of populationsizes at sites during different years (Table 2) suggeststhat numbersdid not fluctuatesubstantially on an annualbasis. The highestpopulation estimates generated by the direct-countmethod at three coloniesin 1992 represented12% (I Street), 26% (20th Street), and 28% (Broadway)of the populationsestimated by nest-holemapping at each site.Thus, the direct-countmethod clearly underestimated populations, and the proportionof the known breedingpopulation counted by this method was inconsistent from site to site. Habitat Conditionsat Nestingand Non-NestingAreas Bridgeshosting martin colonieshad an abundanceof weep holesappar- entlysuitable as nestsites. There are more than 1800 holeswithin 100 m of the sevennesting colonies. Habitat conditionsat occupiedsites varied substantially and were similar to those at many unoccupiedsites (Airola unpubl. data). On the basisof qualitativeevaluation, martins did not appearto selector rejectsites because of the widthof bridgesections, level of humanuse beneath sites, presence or absenceof lighting,availability of nestingmaterial, or distanceto non-urban foragingareas. Certain minimumconditions, however, were presentat all colonysites. Occupiedareas were alwayswithin a spanof bridgeat least85 m long(mean 301 m, standarddeviation 184 m). There was at least6 m of verticalspace beneathall occupiedbridge sections (mean 8.3 m, standarddeviation 1.5 m); martinswere absentfrom all surveyedareas with lessspace. Martins abandonedone formerbreeding site (Capital City Freewaybetween L and Q streets;Figure 2) in the mid-1980s when flightspace was reducedto less than 3 m by constructionof a two-storyparking lot. Also, all occupiedsites had unobstructedflight access and low to moderatetraffic levels. NestlingMortality and Effectivenessof Nest Guards In 1992, prior to installationof nestguards, we found32 nestlingsthat had fallenfrom nest holes(1.07 per nest)at 35th Street and 12 (0.32 per nest)at 20th Street;in 2002, alsoprior to nestguard installation, we found threefallen young (0.21 per nest)at S StreetIn 2002, sevenyears following installationof nestguards at 35th Street,23 (79%) of 29 nestingpairs used weepholes with nestguards, and onlyone deadnestling was found (0.03 per nest).Most nestlingsfell from nests8-14 daysafter hatching.Scavenging ratesof fallenyoung appeared to be low; all sixdead nestlings that we left in place at various colonies were still on the ground during subsequent monitoringperiods 7-17 days later. Even young that fell onto active roadwayswere visiblefor more than a week;consequently, fallen nestlings did not likelyescape detection during surveys. Fledglings'Use of Weep Holeswith Nest Guardsfor Roosting On eveningreturns to the 35th Streetcolony prior to installationof nest guards,we observedthat many fledglingshad difficultyreentering the

244 PURPLE MARTIN NESTING IN SACRAMENTO verticalweep holesto roost.Young birds would misjudge the hole'slocation and not achievesufficient vertical velocity to ascendthrough the hole. Fledglingsshowed extreme fatigue, which put them at riskof collisionwith passingvehicles. They eventuallygave up anddrifted off, apparentlyto roost outsideholes. Some young began roostingin nest boxes that had been erectedthere. After nest guardswere installed,fledglings regularly flew to holes,grasped the wire-meshliner, and readilyclimbed into holes. CompetitorUse of Weep Holes with Nest Guards Monitoringat 35th Street in 2002 showedthat no potentialnest-hole competitorsregularly used holes with nest guards. We observedstarlings, the mostcommon competitors, entering 24 weep holes.Nesting behavior was confirmedat 10 holes, and we observeduse on multipledays, which we consideredto suggestnesting, at four other holes.Only two (8%) of the 24 holesthat starlingsentered had nestguards, and none of the siteswhere we observedeither breedingbehavior or multiple-dayuse had nestguards. Of the holes where we observeduse by White-throatedSwifts (Aeronautes saxatalis) (5 holes), House Sparrows(2), and Rough-wingedSwallows (Stelgidopteryxserripennis) (1), nonecontained nest guards. Some of this differencein use probablyreflects both site fidelity by martins and our purposefulinstallation of nest guardsat activemartin nest sites. Use of Hole Locationswithin Coloniesby Martinsand Starlings Starlingsselected the outermostholes of the bridgeat 35th Street more often than would be expectedif selectionwere random; 13 of 24 (54%) holesused were the outermostholes at the bridgeedge, while only 86 (19%) of 444 availableholes at 35th Streetwere at the edge(• = 20.09, P < 0.001, d.f. -- 1). Starlingsused the next•to-the-outermostholes in propor- tion to their availability(17% usedversus 19% available)and usedinterior holesat lowerthan expected levels (7 of 268 interiorholes, •2 __10.36, P < 0.005, d.f. -- 1). Martinsdid not nest in the outermostholes at thissite, but thisuse pattern did not differ significantly from that expected by chance (•2 -- 2.3, P > 0.10, d.f. -- 1). Martins did use outer holesat other colonies.

DISCUSSION

PopulationStatus and Long-TermTrends The 1949-1990 records indicate that in Sacramento martins nested solelyin buildingsuntil they begancolonizing bridges in the late 1960s. The effect on the martin populationof lossof buildingsas an availablenesting substratethus appears to havebeen offsetat leastpartially by the adoption of bridges.Whether the abandonmentof the colony at the Capital City Freewaybetween L and Q streets(coinciding with constructionof a two- storyparking lot beneaththe freewaythat restrictedflight space) caused a populationdecline cannot be determined;individuals could have movedto new sites,or other establishedcolonies may have grown independently. Comparisonof populationsizes in 2002 with numbersin previousyears showsthe populationin Sacramentoto be stableor possiblyincreasing.

245 PURPLE MARTIN NESTING IN SACRAMENTO

Most of the apparent increase,however, is attributableto the RosevilleRoad colony,which was not surveyedprior to 2002, makingthe determinationof an increaseuncertain. In the larger context, the Sacramentocolonies are remarkablefor being one of the largestconcentrations of nestingmartins in NorthAmerica that do not relyon nestboxes (Williams 1998). Nonetheless, the Sacramentopopulation is stillsmall and restrictedto sevensites that are widelyseparated from other nestingareas; consequently, it is vulnerableto disruption.

Reasonsfor Persistenceof the Purple Martin in Sacramento This studyhelps address the questionof why PurpleMartin colonies have persistedin the Central Valley only in elevatedbridges in Sacramentoand why martinshave not been reportedat bridgeselsewhere in the Central Valley. Nest sitesare superabundantin girderand concrete-boxbridges, while availabilityof naturalnest sitesin riparianhabitats has declinedwith the degradationand lossof riparian and oak woodlandhabitats in the Central Valley (Katibah 1984). Reducednest-site availability, however, appears insufficientto explain regionalmartin decline;martins have used many cavitytypes (including buildings) that are stillreadily available but unoccupied throughouta large portionof the species'former range. Abandonmentof coloniesin tile roofsin Sacramentowas complete by the mid-1970s. This decline,and the adoptionof newly constructedelevated freewaybridges, coincided with the periodwhen the starlingpopulation in lowlandareas of Californiaexploded (DeHaven 1973, Small1974, Garrett and Dunn 1981, Robbinset al. 1986). This timing suggeststhat starlings may have excludedmartins from tile roofsbut couldnot excludethem from bridges.The abandonmentof buildingsalso coincides with the lossof many coloniesin naturalhabitats as starlingsincreased (Williams 1998). The persistenceof martinsin Sacramentomay reflectseveral interrelated factors.First, nest sitesare superabundantin bridges.Second, the vertical weep holes(especially those in the interiorportions of bridges)may not be optimalfor useby starlings,although starlings regularly nest in weepholes in and near martin colonies in low to moderate numbers. Third, low food abundancein the highlyurbanized area of nestingcolonies may limit starling populationsfrom reachinglevels that wouldresult in nestsite competition. Urban-nestingmartins in Sacramentodo not appear to be limited by food. Martinsfrequently foraged directly above urban colonies.During the periodof peak fooddemand for young,martins regularly fed them dragon- flies and damselflies(order Odonata), which were superabundantin the immediate vicinity of the nest colonies.The prevailingsoutherly winds ("Delta breeze") may carry odonatesto the coloniesfrom surrounding ricefields,where they reproduceabundantly, or the insectsmay disperseto urban areas on their own. The lack of reportsof martinsat the many other steeland concretebox i girderbridges in the CentralValley and elsewhere in Californiaindicates that use of thesestructures is rare. If suchwere not the case,we wouldexpect birdersin urbanareas to havereported martins in manyareas, as they have donein Sacramento.We didnot findmartins, for example,in an apparently

246 PURPLE MARTIN NESTING IN SACRAMENTO idealbridge in Stockton,50 milessouth of Sacramento;moreover, we have seenno martinsin manylarger bridges throughout the CentralValley that appearsuitable. OutsideSacramento, martins have been reported using elevated highway structuresonly at bridgeson State Route 1 along the Californiacoast (Williams1998). Martinshave also continued nesting in otherareas, mostly within forestswhere lack of foraging opportunitieshas limited starling populations(Small 1974, Williams1998). The currentpattern of useof elevatedstructures may be explainedby the coincidenceof the constructionof suitablebridges with the invasionand subsequentpopulation increase of EuropeanStarlings. Martins use bridges in Sacramento,where bridgeswere availablebefore starlingpopulations exploded,as well as in coastalregions, where starlings have not yet become abundant.Martins are not foundwhere suitablebridges were built subse- quentto starlingestablishment (e.g., Stockton, many Central Valley sites). In theseareas, starlings presumably eliminated martin populations in buildings and natural habitatsbefore bridgeswere available,and no local source populationis availablenow to colonizethe new habitat. MonitoringMethods BecauseSacramento may support as much as a third of the known nestingPurple Martins in California,and becausethe species'status is so precariouselsewhere in the state(Williams 1998), the Sacramentopopula- tion warrantsconsistent monitoring. Development of techniquesto monitor martin populationsand trends in inaccessiblebridge sites, however,is problematic.Numbers observed vary with time of day,temperature, season, and time spentobserving. In particular,the direct-countsurvey method is unreliablein assessingpopulations or trends;this method not onlyunderes- timatedpopulations at bridgecolonies but did so inconsistentlybetween sites,despite substantial time spentat sitesattempting to obtainthe highest possiblecount. Such resultscast doubt on other populationestimates derived from direct counts (e.g., historicalrecords in Williams 1998); however,it is likelythat the degreeof inaccuracyis greater at largercolonies, suchas those in Sacramento, than at smallerones. Estimatingmartin populationsby evaluatingnesting behavior and hole useprovides a basisfor assessingpopulation trends. However, the potential for inaccuracystemming from variationsin observationtime, observerskill, and visibilitycannot be fully avoidedwithout accessto nest chambers. Accordingly,the population sizes we report should be viewed only as estimates,although they were determinedin a consistentmanner and are thereforeappropriate for assessingpopulation trends. Assumptionsused in populationestimation by mappingof hole use(i.e., the accuracyof usinga certainnumber of visitsto concludenesting use) can be evaluatedby more intensivestudy. Where accessis possible,direct observationof potentialnest siteswith camerasor remote viewingequip- ment (e.g., Kostkaet al. 2003) can be usedboth for monitoringand to test and refine assumptionsused in interpretingpopulation estimates derived from mappinghole useand evaluatingbehavioral cues.

247 PURPLE MARTIN NESTING IN SACRAMENTO

Habitat Characteristics and Protection Needs The lack of obviousdifference between the characteristicsof occupied bridgesites and a largenumber of unoccupiedsites, as well as the abundance of unoccupiedweep holes within occupied colonies, suggests that availability of nestinghabitat is not limitingmartin populationsin Sacramento.The apparentlysuitable unoccupied habitat appears capable of supportinga populationat leastfive timeslarger than currentlyexists. Why martinshave not occupiedthe availablenesting habitat in Sacra- mentomore fully, even within existing colonies, is perplexing.One possibil- ity is that nestlingmortality resulting from falling from weep holes has precludedor dampenedany populationincrease. Moreover, the inabilityof many fledglingsto roost in weep holes without nest guardsmay have reducedproductivity, presumably because excluded young suffer from exposure,injury, predation, or starvationas a resultof separationfrom their parents. An alternativeexplanation for the lack of occupancyof apparently suitablehabitat is that someimportant but subtlehabitat characteristics may remainundetected. The importanceof availableperch sites adjacent to nest sitesthat facilitatenest defensefrom other pairs is one factor that may warrant further examinationand is readilyamenable to experimentalma- nipulation. A third possibilityis that unknown factors affectingother California martinsmay alsobe affectingthe Sacramentopopulation. Starling competition, however,does not appear to be affectingmartin populationsat Sacramentobridge colonies. Potential population limitations that couldbe imposedon winteringgrounds or migratoryhabitat remain unexplored, but recent population increasesin responseto adoption of nest boxes in Oregon, Washington,and British Columbia(Fouts 1996, Copely et al. 1999) suggestthat factors on wintering and migratory areas are not significantlong-term limiting factors. Even though an abundanceof apparentlysuitable unoccupied bridges existsin Sacramento,the martinsthere have strongfidelity to traditional colonysites. Consequently, maintaining the suitabilityof occupiedsites is an important conservationpriority. Land usesthat reduce vertical airspace beneath coloniesto less than 6 m pose the primary threat to current Sacramentomartin colonies.Such usesinclude construction of two-story parkingstructures and storagebuildings, which eliminated the colonyin the Capital City Freeway betweenL and O streets(Figure 2). Constructionof buildingsand growthof landscapetrees immediately adjacent to bridgeshas also restricted martin access to suitable habitat. Existinguses protect four colonysites (I Street, Sutterville,35th Street, and RosevilleRoad) from land-usechanges. Portions of the I Streetcolony receiveheavy pedestrian use, but martins appear unaffected. Sutterville and RosevilleRoad are in overpassesabove railroads. Much of the 35th Street colonyis over a Caltransequipment yard. In contrast,20th Street,Broad- way, and S Streetare at leastpartially above vacant land owned by Caltrans, which is activelysoliciting lessees for thesesites (L. Vieira pers. comm.). While martinshave readilyaccepted highly urbanized sites and tolerateof manyforms of humanactivity, protection of airspaceand access to colonies

248 PURPLE MARTIN NESTING IN SACRAMENTO in Caltransleasing decisions is necessaryto maintainthe long-termoccu- pancyof theseexisting colonies. Treatmentof Coloniesduring Construction Projects The relocationof martinsat 20th Streetby installingplugs in nestholes from 2000 to 2002 apparentlycaused a substantialdecline in this colony (Table 2). The effectson the entire Sacramentopopulation cannot be determinedwith certainty,but suchexclusion threatens the traditionaluse of colonies.Because colonies are sofew, actionsthat excludebirds from a large numberof occupiednest holesat any givencolony, as was done at 20th Street, shouldbe avoided.Since 1991, the area beneath the 20th Street colonyhas been usedduring the nestingseason to stockpiletopsoil and to store trucksand constructionmaterials, but this activitydoes not appear have to disturbedthe breedingbirds (Airola pers. obs.). Given the Sacra- mento martins' past record of tolerance of substantialhuman activity beneath nest sites, we believe it is unwise to excludebirds from nest sites during constructionprojects, except where the nestingareas would be affecteddirectly. Scheduling activities from Septemberto mid-March,however,would ensure that disturbanceof nestingPurple Martins is avoided. NestlingMortality and Effectivenessof Nest Guards Lack of direct accessto nest sitesprecluded the countingof young at fledglingage and hencethe assessmentof the nestguards' effectiveness in enhancingcolony productivity.The installationof nest guardsand their extensiveuse by martinsat 35th Streethas not resultedin an increasein the sizeof that colony,but populationsize at individualcolonies is not a strong indicatorof colonyproductivity because first-year breeders seldom return to their natal colonies (Miller et al. 2001). Many other factors could be influencingthe sizesof individualcolonies. Nestlingmortality due to fallingfrom weep holes is highenough to be of concern as a factor inhibitingproductivity. Reasons for the differencein nestlingmortality rates between the two coloniesstudied (20th Streetand 35th Street) are not readilyapparent because the bridgesare of nearly identicaldesign. Pre- and post-installationdata from the pilotstudy indicate that nestguards are effectivein reducingnestling mortality and that martins readilyaccept nest guards and continue to reusetreated sites for manyyears afterinstallation. The nestguards do not appearto fosterincreased compe- titionwith starlingsor otherspecies; hence, they have not demonstratedany negativeconsequences. Therefore, we are proceeding,with the help of volunteers,to installadditional nest guardsat other colonies(Kostka et al. 2003).

Use of Nest Boxesin Management Managednest boxes and gourdsare the PurpleMartin's primary nest sites in the eastern United States (Brown 1997) and the Pacific Northwest (Horvath1999, S. Kostkapers. comm.). Grantham initiated a programto encouragemartins to use nest boxesat the 35th Street site. While some fledgedjuveniles used boxes for roosting,in subsequentyears no adults

249 PURPLE MARTIN NESTING IN SACRAMENTO nested or roosted in boxes becausethey were taken over by House Sparrows.While encouragingmartins at bridge sitesto use nest boxes appears possible,we and other researchersand managers(Kostka et al. 2003) do not considerthis a high priority. Developinga box-dependent population requiresintensive human management,posing risks for the establishedcolonies. Unless threats to bridgecolonies (e.g., incompatible land uses,starling competition) increase substantially, we believeit is better to allow the existingcolonies to breed in bridges,with provisionof less intrusivemanagement assistance (i.e., installingnest guards, protecting airspace,and avoidingexclusion during construction). A carefullymanaged programto attractmartins to nestboxes, by playingtaped dawn songsand mountingdecoys (Kostka 2000), mightplay an importantrole in restoring martinsto former areasof their rangeand increasingthe total population. Successin sucha programwill depend on developinga dedicatedgroup of professionalsand volunteersto implementlong-term management.

ACKNOWLEDGMENTS

We wish to thank many collaboratorsfor their assistance.Mark Hada, Mike Rushton,Sam Garcia, and Bud and MargaretWiddowson assisted with population surveys.Stan Kostkaprovided valuable advice and review.James Hill, III, provided adviceon interpretingresults. Ted Beedy,Tim Manolis,and Ron LeValleyreviewed the manuscript,and Larry Goral editedit. We thank Betty and Harold Kimballfor recordingand providing their and the SacramentoAudubon Society's 42 yearsof bird records.Julie Mentzer, DebbieMartin, and Gene Moir providedkey informationon statusand managementactivities at severalcolony sites. The CaliforniaDepartment of Parksand Recreation,California State Railroad Museum, supported Mark Hada's monitoringefforts. Caltrans provided access to sitesand assistancein installingnest guards.The NationalAudubon Society supported Grantham's efforts. Jones & Stokes providededitorial, graphics, logistical, and financialsupport.

LITERATURE CITED

Baird,S. F. 1858. Birds,in Reportsof explorationsand surveysto ascertainthe most practicableand economicalroute for a railroadfrom the MississippiRiver to the PacificOcean, vol. IX, pp. 314-315. War Dept., Washington,D.C. Brown, C. R. 1981. The impact of starlingson Purple Martin populationsin unmanagedcolonies. Am. Birds35:266-268. Brown, C. R. 1997. PurpleMartin (?rognesubis), in The Birdsof North America(A. Pooleand E Gill, eds.),no. 287. Acad. Nat. Sci., Philadelphia. Bryant, H. C. 1924. Sacramento'swestern Purple Martin colony.Condor 26:195. Copely, D., Fraser,D., and Finlay, J. C. 1999. Purple Martins:?rogne subis:A BritishColumbia success stow. Can. Field-Nat.113:226-229. DeHaven, R. W. 1973. Winter populationtrends of the starlingin California.Am. Birds 27:836-838. Fouts,D. R. 1996. Ten yearshelping Purple Martins in the PacificNorthwest. Purple Martin Update 7:7-11. Garrett, K., and Dunn, J. 1981. Birdsof SouthernCalifornia: Status and Distribution. Los AngelesAudubon Soc., Los Angeles. Grinnell,J., and Miller, A. H. 1944. The distributionof the birdsof California.Pac. Coast Avifauna 27.

250 PURPLE MARTIN NESTING IN SACRAMENTO

Hill, J. R., III. 1999. The growthof nestlingPurple Martins. Purple Martin Update 8:18-24. Horvath, E. 1999. Distribution,abundance, and nest site characteristicsof Purple Martinsin Oregon. Ore. Dept. Fishand WildlifeTech. Rep. 99-1-01. Jackson,J. A., andTate, J., Jr. 1974. An analysisof nestbox use by PurpleMartins, HouseSparrows, and starlingsin easternNorth America.Wilson Bull. 86:435- 445. Katibah,E. F. 1984. A brief historyof riparian forestsin the Central Valley of California,in CaliforniaRiparian Systems: Ecology, Conservation, and Produc- tiveManagement (R. E. Warnerand E. M. Hendrix,eds.), pp. 23-29. Univ.Calif. Press,Berkeley. Kostka,K. 2000. Socialattraction: A new techniquefor establishinga Purple Martin colonysite. Purple Martin Update9:26-29. Kostka,S., Airola,D. A., andHada, M. 2003. Conservationof bridge-nestingPurple Martinsin Sacramento,California. Purple Martin Update 12:2-6. Miller, H. D., Miller,J. M., and Hill, J. R., III. 2001. Post-fiedgingwandering by hatchingyear Purple Martins:A color-bandingstudy. Purple Martin Update 10:2-4, 13. Remsen,J. V., Jr. 1978. Birdspecies of specialconcern in California:An annotated list of decliningor vulnerablebird species.Calif. Dept. Fish and Game Wildlife Mgmt. BranchAdmin. Rep. 78-1. Rogillio,C. 1992. Purple Martin RehabilitationManual. The Nature Societyin conjunctionwith He]pingHands, Inc., Metairie,LA. Robbins,C. S., Bystrak,D., and Geissler,P. H. 1986. The breedingbird survey:Its firstfifteen years, 1965-1979. U. S. Fishand Wildlife Serv. Resource Publ. 157. Small, A. 1974. The Birdsof California.Collier Books, New York. Sokal,R. R., and Rohlf, E J. 1995. Biometry(3rd ed.). W. H. Freeman,New York. Tonias,D. E. 1995. BridgeEngineering: Design, Rehabilitation, and Maintenanceof ModernHighway Bridges. McGraw-Hill, New York. Willett, G. 1933. Revisedlist of the birds of southwesternCalifornia. Pac. Coast Avifauna 21. Williams, B. D.C. 1998. Distribution, habitat associations,and conservationof Purple Martins breedingin California.M.S. thesis,Calif. State Univ., Sacra- mento. Williams,B. D. C. 2002. Purplemartins in oak woodlands,in Proceedingsof fifth symposiumon oak woodlands:Oaks in California'schanging landscape (R. B. Standiford,D. McCreary,and K. L. Purcell,tech. coords.), pp. 323-334. USDA ForestServ. Gen. Tech. Rep. PSW-GTR-184. Zeiner,D.C., Laudenslayer,W. F., Jr., Mayer,K. E., andWhite, M. 1990. California's Wildlife,vol. 2: Birds.Calif. StatewideWildlife Habitat RelationshipsSystem, Calif. Dept. Fishand Game, Sacramento.

Accepted 12 December 2003

251