The Marbled Teal Marmaronettaangustirostrisis a globally threatened species (BirdLife International
2000)considered tobethe mostprimitive memberof thepochardsAythyini (Livezey1996). Recently ithas been shown to be welladapted to exploit the high spatial and temporal variation in habitat availability found in natural Mediterranean wetlands (Green
However,dataondiethasso farbeenlimitedtoashort study inTurkey,which found Scirpusseedsto bethe majorcomponent, combinedwitharangeofinverte brates(Green &Selva2000).
Herewecompare MarbledTealdietinthebreeding and post-breeding periods in Donana in southwest SpainandSidiMoussa-Oualidia inMorocco,two wetlands used by a common migratory population (Green 1993, Navarro &Robledano 1995). Wecon sider how diet changed across the annual cycle in eachwetland.Wealso investigateiftheinvertebrate component of the diet was more important during thebreedingseason, andtheplantcomponentmore important duringthenon-breedingseason,aswith northern-temperateduckspecies (Krapu& Reinecke
biased towards harder, lessdigestible componentsof thediet(Swanson&Bartonek1970; Sedinger1986). Thus,wealsousegutcontentsfromsmallernumbersof birdsasareference.
MATERIALSANDMETHODS Studyarea SamplesinSpainwerecollectedfromVetaIa Palma (VLP), an extensive brackish fish farm of 3125 ha within Donana Natural Park(36°57’N, 6°14’W), one ofthetwomostimportant wetlandsforMarbledTealin Spain(Green &Navarro1997,Green 2000),Thissite is used to culture estuarine fish such as European Seabass Dicentrarchuslabrax, Flathead Mullet Mugil cephahdsandGiltheadSeabream Sparusauratusaswell asAtlanticDitch Shrimp Palaemonetesvarians. Submergedvegetationis dominatedbyWigeongrass Ruppiamaritima with small amountsoffennel pondweed Potamogeton pectinatus,and the shoresand islandsarecoveredwithsaltmarshvegetation,especially ArthrocnemummacrostachyumandSuaedaspp.
Hotba-WIadSalem (32°55’N 08°49’W;‘ElHotba’ or EH) in the Sidi Moussa-Oualidia complex on the Atlanticcoast. Submerged vegetation at this site is morevaried than atVLP,beingdominatedbyRuppia butmixedwithlargeareasofcharophytes andP.pecti natus.There isabundantemergent vegetation (Juncus, Scirpus, Phragmitesand Typha). The shorelines are dominated by Salicorniaand other saltmarsh plants (seeElHamoumietat.(2000)andGreen etat.(2002a) formoredetailsofER).
Faecalsamples Freshfaecalsampleswerecollected fromislandswhere monospecificgroupsof MarbledTealwereobserved resting.Before collection,birdswereobservedwitha telescope,andasketchwas madeusingclumpsof vegetation as reference points before wading to the islandstocollect freshfaeces.We areconfidentthat eachsamplecamefroma differentindividual, except samplescollected frombroods.At VLP, fivefaecal sampleswerecollected fromMarbledTealbroods,four on26July1996wheretwobroodswereresting(females with 7and 8classlic ducklings) and one on 18July
1997 from afemale with 16 classIla ducklings (see Green 1998bforage classes).Eachsamplemayhave contained faeces from several ducklings and/or the
11May1999(the height ofthenesting periodatthis latitude, Green 1998b).
Faecalsampleswerestoredindividually intesttubes andair-dried.Priortoanalysistheywererehydratedin water(or24hoursand then shaken usingaHeidoiph vortex to loosen them. They werethen washedina
0.04 mmsieve,preserved in 70%ethanolandexam inedwitha10—25xbinocular microscope.Animaland plant food items were sorted and identified to the lowestpossibletaxonomiclevelusingreferencematerial ofpotentialfooditemscollectedatthestudysiteswith sweepnets, together withsuitable keys(seeGreen & Selva 2000,Sanchezetal.2000).The volumeofthe faecalsamplerepresented by eachfooditemwas estimated usingfivecategories ofabundance:absent,
<10%,10—50%,51—90%and>90%oftotalvolume. The minimum number of individual invertebrates consumed wasestablished usinghead parts, elytraor otherfeatures(seeVeitmanetat.1995,e.g. chironomid larvae heads survived digestion and were readily counted).Whole invertebratesfromreferencematerial were used to measure the average volume of each taxon. For organisms with volume >0.01 ml, the volume of ten individuals selected at random was measured by displacement. Linear measurements of smaller organisms were taken, and their volume was estimated based on similar geometric forms (e.g. a sphere oracylinder). Owing totheir similarsizeand form,unidentifieddipteranlarvaeorpupaewereassigned thevolumeofchironomids. Thevolumeofunidentified insectswasestimated usingthemeanvolumeofinsects inreferencesamples.Thesedatawerethenusedtoesti mate theoriginal volumeat ingestion ofthe inverte bratesrepresented inthefaeces(e.g.eachchironomid
Faecal samples with more than 10% green plant material wereexamined morecloselyusinga40—250x optical microscope.Epidermalstructure wascompared withmicrophotographsofvariousaquaticplantspecies inSpain(Amat&Soriguer1982,Green&Selva2000). Intotal, 22sampleswereanalysedinthisway(atER, five collected in October, 13 in May; at VLP,one collectedfromaJulybrood,threecollected inAugust).
Gutsamples SixbirdswerefounddeadatVLPfrom23September to
8October 1997.Fooditemsfoundinthegizzardandin oesophagus plus proventriculus were identified and stored in 70%ethanol. The actual volume of the contentsofthegulletwasmeasuredasabove.
The percentage ofindividual samplesinwhicheach fooditemwasrecorded(i.e.thepercentageoccurrence) wascalculated for faecal and gut samples for each siteandseason.Estimatedvolumesof invertebrates represented inthefaeces,andactual volumetric measurements ofallfooditemsintheoesophagusplus proventriculus were expressed as the mean of indi vidualvolumetric percentages foreachsample(aggre gatepercentage) and/orpercentageof totalvolume combined forallsamples(aggregatevolume,Swanson etal.1974).
50% (at VLP) ofsamples inone ofthe twoseasons were tested statistically, and Bonferroni corrected. Thesedifferent thresholds wereappliedbecauseofthe differences in sample sizes,and none of the compo nents with lower percentage occurrences could have shownstatistically significantdifferences.