SHORT COMMUNICATIONS 549 The Condor 99549-553 0 The Cooper Omithologlcal Society 1997 COMPARISON OF PELLETS VERSUS COLLECTED BIRDS FOR SAMPLING DIETS OF DOUBLE-CRESTED CORMORANTS CLAYTONE. DERBY AND JAMES R. LOVVORN~ Department of Zoology and Physiology, University of Wyoming, Laramie, WY 82071 Abstract. For many fish-eating birds, two types of sample sizes of colonial diving birds in given months samplescan be usedto determine diets of adultsduring or years are usually < 25 and often < 10 (e.g., Baltz both breeding and nonbreedingperiods: esophagus and and Morejohn 1977, Kennedy and Greer 1988, Otten- gizzard contentsof birds collected at feeding sites,and bather et al. 1994), althoughgreater samples are some- pellets (indigestible residue) cast up at roostsor breed- times possibleat wintering concentrations(Glahn et al. ing colonies. We compared these two methods for 1995). Collecting birds also requiresthat all important Double-crested Cormorants (Phalucrocorux auritus) feeding sites be located and sampled. collected on the North Platte River, Wyoming and Gathering pellets at colonies or roosts does not re- nesting at a nearby colony both before and after fin- quire killing birds, can provide integrative samplesof gerling trout were stockedin the river. Before stocking, all feeding sites, and can result in larger sample sizes there were no significant differences between pellets with less time and effort (Veldkamp 1995, Warke and and collected birds in percent massesof different fish Day 1995). However, this method cannot be used if types in different length classes. After stocking, the colonies or roosts are inaccessible(e.g., on cliffs), are two methodsyielded similar resultsfor different length directly over water where pellets cannot be retrieved classes,but differed in relative proportionsof suckers (on pilings, etc.), or have too few birds to yield enough and trout. Patternsin percent occurrenceof crayfish in pellets in a given period. Also, disruption of colonies the diet suggestthat the time of day adults are col- while gatheringpellets, causingincreased predation on lected might affect comparability of the two methods. eggs and young and nest abandonment,might often Key words: diet sampling, Double-crested Cor- have greater impactson bird populationsthan shooting morant, Phalacrocoraxauritus, fish-eating birds, oto- adults away from colonies (Bunnell et al. 1981). For liths, pellets, trout, sucker. cormorants, the latter problem is especially great in areas such as the Intermountain region of western Two methodscan be used to sample diets of adult fish- North America, where colonies often contain only 20- eating birds during both breeding and nonbreedingpe- 200 cormorant nests (Findholt 1988) and many dep- riods. The first is analysisof esophagusor gizzard con- redating gulls. tents from birds that are collected, usually by shooting Otoliths (hard inner-ear bones of fish) found in pel- (Baltz and Morejohn 1977, Ottenbacher et al. 1994, lets or collected birds often are used to identify fish Glahn et al. 1995). The second method is analysis of species and estimate lengths of fish remains (Craven pellets (indigestible residue) cast up by adults at col- and Lev 1987, Martucci et al. 1993, Veldkamp 1995). onies or roosting sites (Keller 1995, Brown and Ewins However, in some studies of captive cormorants, di- 1996). A third type of sample, sometimes combined gestive acids partly or completely dissolvedsome oto- with pellets (e.g., Craven and Lev 1987), is more re- liths. These changescaused underestimates of size for cently ingestedmaterial regurgitatedwhen birds at col- larger fish and of total numbers of smaller fish, and onies are disturbed (Blackwell et al. 1995. Findholt distorted the apparent species composition of prey and Anderson 1995). However, when searchinga dis- (Johnstoneet al. 1990, Harris and Wanless 1993, Veld- turbed colony for regurgitationsit is often not possible kamp 1995). Other workers have concludedthat these to distinguish regurgitationsof adults versus chicks, effects are negligible (Martucci et al. 1993), do not which might be fed different foods (Harris and Wan- apply to field situations (Zijlstra and Van Eerden less 1993, Veldkamp 1995). In contrast,pellets usually 1995), and can be largely eliminated by excluding oto- are not producedby cormorant chicks < 7 weeks old liths that are clearly eroded (Suter and Morel 1996). (Russell et al. 1995, Trauttsmansdorffand Wassermann Regardlessof these concerns, otoliths continue to be 1995, Zijlstra and Van Eerden 1995). Collecting birds away from colonies or roosts, es- used (Suter 1995, Veldkamp 1995, Warke and Day pecially pelagic divers, can be difficult and time-con- 1995) because they greatly increase the number of suming. Also, sensitive population status of the birds food items identified. or proximity to people can make shooting birds bio- Despite variationsin desirability and possiblyresults logically or politically undesirable.For these reasons, of pellets versuscollected birds, the two methodshave not been comparedat the same location and time dur- ing known changesin prey availability. We performed r Received 1 June 1996. Accepted 14 January 1997. such an analysisfor Double-crestedCormorants (Phal- z Correspondingauthor. e-mail: [email protected] acrocorax auritus) feeding on an inland coldwater riv- 550 SHORT COMMUNICATIONS er before and after the river was stocked with finger- 1974) and the edge of the thick end of each otolith ling trout. was measuredwith an ocular micrometer.This method was intended to reduce error in measurementsof total METHODS otolith length caused by variable erosion of the op- Our study was conducted on the North Platte River posite, thinner end of the otolith. We also excluded near Casper, Wyoming, and on two nesting islands in from analysis otoliths that were clearly eroded (Suter Soda Lake, 2.5 km north of Casper. Nesters included and Morel 1996). Total fish length was estimatedfrom both Double-crested Cormorants (maximum of 243 separateregression equations relating otolith radius to and 77 pairs on the two islands, respectively, in 1994) total fish length for trout, suckers (Cutustomus com- and several hundred pairs of California Gulls (Larus mersoni and C. cutustomus), and minnows (mainly caZz@zicus). The 83-km section of river upstreamof Rhinichthys cutaractue and Pimephales promelus) Casper to Gray Reef Dam was stocked on 27 June (Derby and Lovvorn 1997). Fresh massesof fish in 1994 with 54,100 rainbow trout (Oncorhynchus my- each length class were calculated from allometric kiss) and 38,900 cutthroat trout (0. clarki), both spe- equations of mass versus total length for samples of cies 11.4-12.6 cm long. whole fish of the different species. From this section of river, we shot and collected 12 In both pellets and collected birds, we matchedpairs cormorantsfrom 2 to 23 June, and 11 cormorantsfrom of otoliths (two occur in each fish) whenever possible, 30 June to 14 July. In the previous year, we found that so numbers of fish analyzed equaled the number of cormorantscollected while actively foraging often had matched pairs plus the number of unmatchedotoliths. empty esophagi and gizzards, whereas birds collected These sucker species do not eat fish, nor do rainbow while drying their wings (typically on rocks in the riv- trout in all studies to date in Wyoming (e.g., Hubert er) usually containedfood. et al. 1994); electroshockingon 9-10 May 1994 re- Fresh pellets (moist, mucus-coatedcapsules cast up vealed almost no cutthroattrout of piscivoroussize in the study area. Thus, otoliths in diet samplesprobably bv. adult birds: Harris and Wanless 1993) were eath- ered from both nesting islandsbefore trout stockingon did not result from otoliths in the stomachsof prey 24 June (3 1 pellets) and after stocking on 1 July (15 eaten by cormorants.Percent massesof each fish type pellets). Lack of vegetation, high densities of cormo- (trout, sucker, minnow) in each of five length classes rant nests, and many breeding gulls made these colo- (8-cm intervals from 0 to 40 cm) were calculated for nies very sensitive to human disturbance;visits every each individual sample (pellet or collected bird) (ag- other week caused complete abandonmentof one of gregate percentage method of Swanson et al. 1974). the islandsthe previous year. Thus, we restrictedpellet Crayfish (Cambaridae) and tiger salamanders(Ambys- collection to single visits before and after stocking. tomu tigrinum) could not be counted based on miscel- The colony was searchedbetween 06:30 and 07:30 to laneousfragments, so for thesefoods we recordedonly minimize heat stress on nestlings not brooded during percent occurrence (percentageof individual samples our presence. containing the food item). We gathered pellets only in the area of cormorant To test for differences between samplingmethods in nests, which were generally < 1 m apart. In over 200 percent massesof fish in different length classes,we hr of dawn-to-duskobservations during incubationand performed analysesof variance (ANOVA) with PROC early brood-rearing the previous year, the dense nest- GLM (SAS Institute Inc. 1987). We used arcsine ing areas were seldom if ever entered by adults that squareroot transformationsto stabilize variances.For were not obviously paired. In studies of Great Cor- percent occurrence of crayfish and salamanders,we morants (Phalucrocorux turbo) and European Shags used percentagetests (Sokal and Rohlf 1981). (P. uristotelis), chicks < 7 weeks old did not produce RESULTS pellets,