The Auk 111(4):853-862, 1994

HIGH SURVWAL ESTIMATES OF GRIFFON VULTURES (GYPS FULVUS FULVUS) IN A REINTRODUCED POPULATION

FRANgOISSARRAZIN? 2 CONSTANTBAGNOLINI, 2 JEAN LOUIS PINNA? ETIENNE DANCHIN, 1 AND JEAN CLOBERT• •EcoleNormale Supgrieure, Universitg Pierre et Marie Curie,Laboratoire d'Ecologie, CNRS, URA 258, 46 rue d'Ulm, 75230, Paris Cedex 05, France; 2Fondsd'Intervention pour les Rapaces--Grands Causses, Mairie, 12720Peyreleau, France; and 3parc National des Cgvennes,Chdteau de Florac, BP 15, 48400 Florac, France

ABSTRACT.--Littleis known of the life history of vultures.The reintroductionprogram of GriffonVultures (Gyps fulvus fulvus) in the Causses(south of the MassifCentral, France) and extensivemonitoring by capture-mark-resightingof the releasedbirds allowed us to obtain the first estimatesof their survival. Adult survival rates are high (œ= 0.987 + SE of 0.006). A releaseeffect on adult survival was detected(only 0.743 + 0.006 survival during the first year after release).Young born in the wild (lessthan three yearsold) had an annual survival rate of 0.858 + 0.039.Mortality causesand erraticbehavior of immaturebirds are considered in order to assessthe effectivenessof this reintroduction program. Our resultsindicate that reintroductionsof vultures and similar speciesshould use adults that have bred in captivity within the target area rather than juvenilesor immatures.Received 9 February1993, accepted 19 August1993.

OLD WORLD VULTURES such as Griffon Vultures have suitableadapted and standardizedanaly- (Gypsfulvus fulvus) are long-lived birds that are sis methods been available to achieve such es- highly sensitive to environmental changes timates (for review, see Pollock 1991, Lebreton (Houston 1987). Being exclusivelyscavengers, et al. 1992, Nichols 1992). Crude estimates of they depend on the availability of large car- survival have been made for only one Gypsspe- casses.Loss of this food resourceand direct per- cies, the Cape Vulture (G. coprotheres;Piper et secutionof birds by hunting and indirect poi- al. 1981,Robertson 1984). In our study,we mon- soningled to the decline of southernEuropean itored every individual in the population from vulture populations over the last few hundred 1981 to 1991 and used these data to calculate years (Houston 1982). Griffon Vultures in the survival rates.Our secondobjective was to com- Causses(south of the Massif Central, France) pare survival rates of wild-born and released were extirpated in 1945 (Berthet 1946). In 1968, captive birds to evaluatetechniques for the re- a successfulreintroduction project was insti- introductionprogram. Overall, our purposewas gated and more than 100 Griffon Vultures now to use capture-resighting data to model and occupy the Caussescolony composedof both compare survival estimatesin three categories captive-bred birds and their wild-born off- of individuals of this population: (1) wild-born spring (Bonnet et al. 1990, Terrasse 1990). We individuals, (2) birds releasedas adults, (3) birds report on two aspectsof the recently established released as immatures. population. Initially, we analyzed our three data setssep- Our first aim was to gain a better understand- arately. First, the wild-born individuals were ing of the biology and especiallyof the de- studiedsince they were not likely to be affected mographic parametersof this poorly studied by any reintroduction effect. These resultswere species(Donazar 1987, Elosegui 1989). In par- then compared with the survival estimatesfor ticular, survival is the demographic parameter released adults in order to test the release effect. to which population growth is most sensitive We then estimated the survival rates of birds for such long-lived animals (Lebreton and releasedas young. Finally, the three classeswere Clobert 1991). Few survival estimateshave been grouped together to compareparameters of the made becausethey require long-term studies different data sets and to obtain the best-pos- and extensive field observation;only recently sible estimates.

853 854 $ARRAZINET AL. [Auk,Vol. 111

Releese period problemsof legibility from distancesover 300 m led [] Releasedimmatures us to use a codeof more resistantfour-layer DARVIC [] Wild-bornbirds 20- , Releasedadultscolorbands. Banding occurred on differentoccasions: (1) before release (n = 59); (2) at nest for wild-born chickswhen 60 to 80 days old (n = 59); (3) at nest for tame adults (n = 3); (4) during recapturesessions to lO compensatefor band lossesor to changethe identi- fication system(in 1988 n = 29; in 1991 n = 31). Ob- servationswere essentiallymade by threepeople. Birds were observedwith a 20-60x telescopeyear-round to obtain information on their overall activities o t 980 1981 t 9821983 t 984 1985 19861987 1988 1989 t 990 (breeding,feeding, etc.). All resightingswere entered Year into a dbase III+ database,and individual capture- resighting historieswere constructedfrom the ob- Fig. 1. Number of marked Griffon Vultures en- servationsmade during the breedingperiod (between tering Caussespopulation per year.Numbers of year- 1 Januaryand 30 Septemberof each year). ly releasedbirds differ from thosegiven by Bonnet et aL (1990) and Terrasse (1990) becausevultures re- Analyses.--Althoughbanding recoveries have been leased after mid-November were considered here as used to estimatesurvival of the Cape Vulture (Hous- releasedduring followingyear. One bird escapedfrom ton 1974, Piper et al. 1981, Piper 1990), we did not aviary in 1980 and came back three years later. The use them becauseof their limited importancein the 17 individuals born in 1991 were not taken into ac- contextof our study (Lakhani and Newton 1983,An- count in survival analyses. dersonet al. 1985,Lebreton 1985 unpubl. report). As our population was consideredto be an open popu- lation (Pollock et al. 1990, Pollock 1991, Nichols 1992), METHODS survival and resighting rates (respectivelynoted ß and P) were computedwith captureresightings mod- Reintroductionand management technique.--The study els (Clobert et al. 1985, Clobert et aL 1987, Lebreton area covers all the Grands Causses of south-central et aL 1992).Models combining survival and resight- France where the Tarn, Jonte, and Dourbie rivers flow ing parameterswere noted modelsß P. Theseparam- into deep canyons.The releasesite was locatedon eterswere estimatedusing SURGE 4 (Pradel and Le- the southern rim of the CausseM(•jean (44ø12'N, breton 1991, Lebretonet al. 1992), a programthat 3ø15'E),where aviariesfor captivebreeding were built allows use of user defined models.The constancyof at the top of a cliff overlookingthe gorgesof the Jonte ß and P, as well as the effectsof release(r), age (a), River. In 1970, a first attempt to release four juveniles and time (t) on • and P, could be tested as in standard failed becausethe youngbirds rapidly left the area. analysis of variance or of covariance. Models with A stockof captiveadults was then acquiredwith the interactions (a + t + a,t) were noted a,t, whereas intention of establishingbreeding pairs in the avi- models without interactions were noted a + t (see aries (Bonnet et al. 1990, Terrasse 1990). The release Table 1 for model notations). Details of models struc- of theseadults began in December1981 and was car- ture are available in Pradel et al. (1990), Pradel and ried on with youngerbirds until 1986(Fig. 1). During Lebreton (1991), Lebreton et al. (1992). these five years, 61 Griffon Vultures were released Because we were led to consider numerous models, (one banded and two unbanded birds escaped un- we usedAkaike's procedure,which allows compari- intentionally in 1980 and 1985, respectively).Food son of non-necessarilynested models.This new sta- resourcesin the area mainly consistedof sheepand tistical method is recommended in multivariate anal- cattle carcassesresulting from natural mortality. In ysis(Lebreton et al. 1992).It givesa penaltyto models addition, there was an active educationprogram for with too many parameters.Thus, it is useful to retain local farmers and hunters, sites for supplementary the most-parsimoniousmodels. The lower the Akaike's feeding were established,nest cliffs were protected information criterion value (AIC), the more appro- from climbers, and power lines were converted to priate is the model for the data. The similarity of ß prevent birds being electrocuted(Bonnet et al. 1990, and P parametersacross groups and the dependence Terrasse 1990). of theseparameters on capture-resightinghistory were Monitoring.--Weconducted detailed monitoring of tested with TEST1 and TEST3, respectively,of the the population.Each marked individual wore a metal computer program RELEASE(Burnham et al. 1987, band on one tarsus and an identification system on Lebreton et al. 1992). Because Griffon Vultures are the other. This systemchanged during the monitor- monomorphic(Newton 1979,Fry 1983,Elosegui 1989), ing period.The firstbirds to be releasedwere equipped sexwas not taken into accountin our analyses. with a combination of two plastic color bands. Be- Some Griffon Vultures sCartedto breed when they causeof the loss of some of these bands,engraved were four years old, so we considered birds to be DARVIC white bands were used from 1988. However, adultsfrom that age onwards.These age classescor- October1994] SurvivalofReintroduced Vultures 855

T^!•I•E1. Numberof parameters(np) and Akaike's information criterion values (AIC) for differentmodels of survival(•b) and resighting (P) parameterstested on capture-resighting histories of wild-bornindividuals. Lettersindicate the effect:(a) age;(c) constantover time; (t) time (i.e. one parameterper year).Subscript numberscorrespond to rangeof differentage classes described in models(e.g. •b[a3.6] is model of survival ratewith two ageclasses three and six yearslong, respectively). When no subscriptincluded, each age is fully identified.The lower the AIC value,the more appropriate is themodel for thedata. Selected model for •band P parametersis in bold.

Survivalmodels Resightingmodel (P) (•) C t t/laa.6 t/xa.7 a•.8 aze t/3.6 a•.5 t/3.6*t a3.6+ t az, + t c np 2 10 5 4 3 3 3 3 16 11 11 AIC 248.58 248.03 251.46 250.07 250.30 248.28 248.03 249.95 248.91 246.56 247.74 t np 10 17 13 12 11 11 11 11 23 18 18 AIC 258.98 257.20 261.33 259.85 260.81 258.34 258.34 260.01 258.70 256.07 251.06 a np 10 18 13 12 11 11 11 11 24 19 19 AIC 258.98 257.86 263.00 261.13 260.25 259.14 259.58 260.66 261.10 258.05 258.69 ai.x•.6 np 5 13 8 7 6 6 6 6 19 14 14 AIC 249.22 247.86 254.01 252.01 250.71 250.02 250.58 251.22 253.08 248.71 248.69 a•.•.7 np 4 12 7 6 5 5 5 5 18 13 13 AIC 248.20 246.91 252.77 250.96 249.60 248.96 249.12 250.16 251.10 247.32 247.77 a•.8 np 3 11 6 5 4 4 4 4 17 12 12 AIC 250.13 249.47 253.37 252.00 252.07 250.27 249.95 265.68 250.82 248.37 249.64 a2.7 np 3 11 6 5 4 4 4 4 17 12 12 AIC 246.23 245.53 251.64 249.91 248.73 248.00 247.88 248.80 249.60 246.02 246.71 a•.• np 3 11 6 5 4 4 4 4 17 12 12 AIC 246.11 245.04 264.75 248.75 248.01 246.94 247.49 248.11 249.89 245.66 245.72 a4.5 np 3 11 6 5 4 4 4 4 17 12 12 AIC 248.20 247.09 252.08 250.29 250.02 248.50 248.67 250.11 249.23 246.25 247.25

respondedto the calendaryear, exceptthe first one, erogeneity in the histories of birds releasedas which was shorter(from fledging in July and August adults and birds released as young (TEST1 = to 31 Decemberof the sameyear, as in Houston1974). 27.01, df = 9, P = 0.0014). Thus, capture-resight- All wild-born individuals were banded as nestlings. ing historiesof individuals depended on their Therefore,their agewas known preciselyand cohorts origin and thesethree groupswere studiedsep- includedonly birds of the sameage. Other vultures arately. that were born in our aviarywere alsoaged precisely. This was not the casefor some captive birds of un- Modelselection for wild-bornindividuals.--This known origin. A minimal age for such individuals group contained59 individuals whose capture- was estimatedfrom plumagecharacteristics. We rec- resightinghistories stretched over nine cohorts ognized five age classes(zero, one, two, three, and (1982-1990; Fig. 1). TEST3 did not reveal any four or more years old). Cohorts of releasedbirds heterogeneity among individual capture-re- were constituted of birds released at the same time, sighting histories (TEST3 = 5,946, df = 9, P = but probably not born in the same year. 0.745). We testedfor constancyand for age and/or RESULTS time effect in both survival and capture rate. For the age effecton survival rate, different age Group splitting.--The first successfulrepro- classeswere evaluated, looking for maximal duction in the wild occurred in 1982 (Fig. 1). parsimonyof parameters.In the sameway, we By 1990, 60 Griffon Vultures had been born in tested an age effect and a time effect on the the wild (59 of whom were marked); these were different age classeson resightingsrates. Mod- in addition to the 59 marked birds that had been els with interactions (Pa-t) and without inter- released. Therefore, 118 individual histories action (Pa + t) were evaluated. Table 1 de- were analyzed. scribes the models and their AIC values. TEST1 showed significant differences be- The model [•a3.6,Pt] was accepted (AIC = tween capture-resightinghistories of wild-born 245.04, number of parameters np = 11), al- and released birds (TEST1 = 67.38, df = 14, P though closestmodels had similar AIC values = 0.0001). Similarly, there was significant het- (Table 1). Therefore, the resighting rate de- 856 S^RP,•ZXNET AL. [Auk, Vol.

TASLE2. Number of parameters(np) and Akaike's TABLE3. Number of parametersand Akaike's infor- information criterion (AIC) values for different mation criterion values for different models ß P modelsß P testedon capture-resightinghistories tested on capture-resightinghistories of birds re- of birds releasedas adults (seetext and Table 1 for leased as iramatures (see text and Table 1 for no- notations).The lower the AIC value, the more ap- tations). The lower the AIC value, the more appro- propriateis the model for thedata. Selected model priate is the model for the data. Selectedmodel for for ß and P parametersis in bold. ß and P parameters is in bold.

Survival Resighting model (P) Survivalmodel Resightingmodel (P) model (•) r t c (•) c t a2.6+ t t np 22 21 12 c np 2 9 10 AIC 174.06 153.96 165.58 AIC 998.92 858.66 768.19 c np 12 12 2 t np 9 15 16 AIC 181.36 163.55 173.02 AIC 854.71 737.21 653.04 r np 21 22 12 r•.6 np 4 11 12 AIC 167.43 151.17 160.97 AIC 906.97 777.58 690.13 •ao np 13 13 3 r2.6 np 3 10 11 AIC 156.16 137.56 147.83 AIC 905.98 779.58 698.65 r•.19 np 14 14 4 r•7 np 3 10 11 AIC 156.80 138.43 148.46 AIC 973.46 844.19 762.46 r2.9 np 13 13 3 ß • np 10 17 18 AIC 161.31 143.56 152.93 AIC 828.94 712.83 642.14 ß • np 7 14 15 AIC 852.68 731.87 648.16 ß • np 6 13 14 AIC 928.81 802.99 725.57 pended on time and the survival rate ß only • Age since releaseand time effect on first two years followed by constant survival. dependedon age (i.e. ß wasconstant from zero • Time effecton first two yearsfollowed by a constantsurvival. to threeyears old and constantover the follow- ½Time effecton first year followedby a constantsurvival. ing ages).Parameter values will be discussed after the last analysiswith the three groupsto- gether. on resighting rate. Moreover, survival rate was Modelselection for birdsreleased as adults.--We different during the first year after release. released 39 adult Griffon Vultures from 1980 to Modelselection for birds released as immatures.• 1985(Fig. 1). TEST3revealed heterogeneity in From 1983 to 1986, 20 Griffon Vultures were capture-resightinghistories (TEST3 = 9.33, df releasedwhen lessthan four yearsold (Fig. 1). = 3, P = 0.009). Nevertheless, because this test Accordingto the survival estimatesfor the two compares,for eachresighting occasion, the cap- previousgroups, the survivalrate of thesebirds ture-resightinghistories of birdsnever seen be- releasedas young should have been subjectto fore and those of birds seen before this occasion a release and age effect. However, becauseof (accordingto a Jolly-Sebertime effect;Burnham the low number of birds of each age classre- et al. 1987),a strongage effectcould causethis leased,we could not define a group for each apparent heterogeneity.Unfortunately, the age classand we used only one group. Hetero- smallnumber of individualsprevented us from geneityshould have been important in thisdata making a batch-by-batchanalysis (Burnham et set, although it was not detected by TEST3 al. 1987). (TEST3 = 9.365 df = 5, P = 0.095). The models As these birds were released when at least with release effect (•r) should be considered as four yearsold, their survivalshould have been rough approximationsbecause they described constantaccording to the model for wild-born histories of birds releasedat different ages(e.g. individuals. Cohorts contained birds of differ- when fledgling or when two years old). Ac- ent agesso it was not possibleto test any age cordingto the modelstested for wild-born vul- effect. However, the releaseeffect was estimated tures,the hypothesisof differencein resighting using modelsthat describedan "age-since-re- ratesduring immaturity and maturity (Pa•.• + lease"effect (•r). Differenttypes of delayedef- t) was tested (seeTable 3 for tested modelsand fect were tested (Table 2). AIC values). The selected model was [•r•.•0, Pt] (AIC = The model [•, Paz• + t] (Table 3) was selected 137.56,np = 13). Thus, there was a time effect (AIC = 642.14, np= 18). In resighting rate, a October1994] SurvivalofReintroduced Vultures 857

Possiblebreeding 1,0-

1.0-

O,g ÷

0.8

0.7

0.6 lg83 lg84 lg85 lg86 1987 lg88 lg89 lggo lggl

Year

0.5 1 2 3 4 5 Fig. 3. Annual resightingrates (wisker indicates Time since release or birth (years) SE) of Griffon Vultures according to their origin: re- leased adults and released immatures after the first Fig. 2. Increaseof survival rates(wisker indicates two yearsfollowing releases(dark bars);wild-born SE) with time in releasedadults (full bars) and wild- birds and released immatures during the first two born birds (openbars). Released adults can breed im- yearsfollowing releases(light bars). mediately,whereas fledglings cannot breed before four years of age. fidence interval [CI] = 0.965-0.995; Fig. 2). An- time effect with a parallelism over two "age" nual survival was lower during the first three classes(the first two yearsfollowing releaseand years of life for wild-born individuals (0.858, the rest of life) was detected. Survival rate de- CI = 0.761-0.919). The effect of release on adult pended on the interactionof time and "age since survival rates was strong during the first year release" during the first two years following after release (adult survival rate after release is release and was constant over age and time 0.743, CI = 0.594-0.851). In the same way, an- thereafter. This result was due to the impossi- nual survival rates of released immatures varied bility of selectinga more-parsimoniousmodel, during the first two years following releases probably becauseof the influence of age at re- and had a mean value of 0.755. lease. Resighting rateschanged over time, but were Model selectionfor the threedata sets.--To de- the same for adults of releasedbirds and gen- velop the most-parsimoniousmodel, the three erally were significantly lower for wild-born previous groups were studied together. Four birds and released immatures during the two models were compared.Model A was the sum firstyears following releases(Fig. 3). Before1986, of the three previously acceptedmodels with- high observationpressure on a small number out parameterequality between groups(AIC = of individual explainsthat resightingrates reach 479.23, np = 42). Model B was the same model 1.0. with equality of adult resighting rates of re- Causesof mortalityand removalfrom popula- leased birds (released adults and released ira- tion.--From 1981 to 1991, 27 marked Griffon matures after the first two years following re- Vultures were recovered dead (n = 18) or de- leases;AIC = 470.18,np = 26). Model C equalized liberately removed (n = 9; Table 4). Electrocu- wild-born birds resighting rates and those es- tion was the most important known cause of timated during the first two yearsafter releases mortality. From December 1981 to May 1992, of immatures,with a parallelismover time with 10 banded and 1 unbanded juvenile and im- adult resighting rates of releasedbirds (AIC = matureGriffon Vultures(unknown origin) were 455.80,np = 25). Finally, model D that equalized killed in the Grands Causses. Most of these birds adult survival rates over the three groups was were found electrocuted at the bottom of trans- accepted(AIC = 452.88, np = 23) and gave the former pylons. In the Grands Causses,no case final estimatesof resighting and survival pa- of shootingor poisoninghas been proved since rameters. the beginning of the reintroduction program. Survivaland resightingrates.--The values ob- However, 1 of 17 fledglings born in 1991 was tained were exceptionallyhigh, particularly for found poisoned in Senegal (Kaffrine depart- adult survival (0.987 with 95% asymmetriccon- ment, 14øN, 16øW) in February 1992. This re- 858 S^RRAZlhlET ^L. [Auk, Vol. 111

T^I•I,E4. Summaryof deathsand reasonswhy birdswere deliberatelyremoved from wild from 1981to 1991. Samplesize (n) indicatesinitial number of birds.

No. immatures No. adults

Released Wild-born Released Wild-born (n = 20) (n = 59) (n = 39) (n = 25) Total Death Electrocution 3 7 2 0 12 Exhaustion 0 1 1 0 2 Unknown 1 2 1 0 4 Removal Exhaustion 0 0 4 0 4 Starvation 0 4 0 0 4 Injury 0 0 1 0 1 Total 4 14 9 0 27

covery gives evidenceof long-distancemigra- the bias on survival estimatesarising from un- tion and the risks run by Griffon Vultures equal resighting probabilities of individuals outside our study locality. could be neglected (Carothers 1973). However, Human intervention to help birds in diffi- ignoring definitive tag loss (i.e. loss of both cultiesoften occurredjust after releasesand af- plasticand metal bands,or lossof plasticbands ter fledging each year. The large number of only, if individuals are not recaptured during birdwatchers in the area during the summer the study period) can entail a severe underes- increasesthe chancethat fledgingbirds that get timate of survival rates (Kremers 1988). Nev- into difficultiesalong roadsor rivers can be re- ertheless,because of their high values,survival captured,fed, and released. However, this effort estimateswere obviously not affectedin a sub- only led to saving 2 releasedadults of 11 re- stantial way by band lossin our study. capturedbirds, and no wild-born fledgling, so Also, we cannot consider that the duration of it did not have any artificial effect on the sur- sampling periods was instantaneous,because it vival estimates. was three times longer than the interperiod, and mortality and emigration could have oc- DISCUSSION curred during this time. For example, most of the removals or observed death of released Methodologicalaspects.--The present method adults occurred less than two months after re- of samplingdata violated two of the five main lease.Thus, the temporal location of mortality assumptionsof the capture-recapturemodels in had no biological significanceaccording to the open populations(Lebreton 1985 unpubl. re- intersampling period. Since most of annual first port, Clobert et al. 1987,Pollock 1991,Lebreton resightingsof individuals (89.46% _+SE of 1.26) et al. 1992, Nichols 1992): (1) existence of band were made during the first half of the sampling losses;(2) relatively long duration of the sam- period, we assumedthat the short duration of pling period (nine months) in comparisonto the intersampling period did not change the the intersamplingperiod (three months).How- statisticalbackground of this study, but only ever, different elements allowed us to give ro- causedus to moderate the interpretation of our bustness to our results. results. Some plastic identification bands were lost, Finally, the time effect on resighting rates but none of the metal bands were. Five of the was due essentiallyto variationsin observation birds that lost plasticidentification were iden- activity. It has declined after 1984 becausethe tified when recaptured. This could have in- main observerbegan to spend time collecting ducedsome heterogeneity, although moderate carcasses.The lower values of resighting rates heterogeneityin capture rates has little effect in wild-born birds and released immatures dur- on survival estimates(Nichols et al. 1982, Clob- ing the first two yearsfollowing releasescould ert et al. 1993). Moreover, as resighting rates have different methodological or biological were higher than 0.5, except in 1987 (Fig. 3), causes(see below). Resighting pressure could October1994] Survivalof Reintroduced Vultures 859 have been higher on adult breedersbecause of first year could be due to someconsequences of the intensive monitoring of nests.Nonbreeders captivity: the energetic cost of learning to fly; were occasionallyresighted at roost or feeding the necessityto learn how to feed; and behav- placesin the proximity of nesting cliffs, but ioral problemsfor the most imprinted birds (Ta- were not subject to the intensive observations ble 4). However, the birds that successfully of the breeding birds. withstood this period had no more problems, Biologicalaspects.--Overall, the survival rates as shown by the steep increasein their survival observedin this study are high. Our estimate rate. of adult survival rate from one year after release The annual survival rate of wild-born birds onwards,or after threeyears for wild-born birds is significantlylower during the firstthree years (0.987, CI = 0.965-0.995), is higher than those of life (0.858 + 0.039 vs. 0.987 + 0.006), which obtainedby Piper et al. (1981) for CapeVultures includesthe period up until one year beforethe using banding recoveries.High adult survival earliest first breeding (Fig. 2). This age effect is confirmedby the absenceof any recoveries observed on survival rates of wild-born birds of dead adultsamong wild-born birds (Table 4). can have variousexplanations concerning post- Different biological factors can explain this fledging dependenceperiod and natal dispers- very high adult survival rate in the Griffon Vul- al. tures. First, high survival rate can be related to In fact,the periodof learning to fly following the other life-history characteristicsof this spe- fledgingseems to be a critical period for a Grif- cies, which produces a maximum of one off- fon Vulture. Despite the observationsof two spring per year after the four to five yearsneed- fledglingsfeeding on sheepcarcasses a few days ed to reach maturity (Simmons 1986). Our after fledging, juveniles still depend on adult estimatesare similar to those obtained by cap- feeding for severalweeks after fledging, as in ture-recapture methods in other large soaring Cape Vultures (Robertson 1985). Perhaps com- birds that forage extensivelyfor spatiotempo- petition with adults at feeding sites prevents rally unpredictable resources:Wandering Al- them obtaining sufficient food. In the Causses, batrosses(Diomedea exulans; mean adult survival juvenile Griffon Vultures were fed at the nest 0.968); and Light-mantled Albatrosses(Phoebe- by their parents up to 90 days after fledging. tria palpebrata;0.973; Weimerskirch et al. 1987). Newly fledgedGriffon Vultures often use flap- Second,adult breeding birds are not known to ping flight, which is of high energetic costfor have dispersedonce they startreproduction and these soaring birds (Pennycuick 1972). Fur- breeding dispersal has not been described in thermore, somejuveniles face considerabledif- other Gypsspecies. Therefore, emigration could ficulties returning to the nest after their first not affect adult survival estimates. Third, Grif- flight. Intraspecificcompetition at feeding sites fon Vultures have no predatorsor competitors and inefficientflight probablyleads some new- in this studyregion, and their low densitymin- ly fledgedGriffon Vultures to exhaustion(Table imizes intraspecificcompetition. Furthermore, 4). As only a small number of fledglingswas the number of sheepcarcasses constituting the released (n = 4), we could not estimate the im- main food resource was substantial and could pact of release on their survival. support a population of about 300 Griffon Vul- Some unbanded immature Griffon Vultures tures in summer and 1,200 in winter (Briquet immigratedto our colony(two unbandedbirds 1987). Finally, the life span of the Griffon Vul- were captured in 1988 and, during the study ture (37 years in captivity; Newton 1979) ex- period, five to nine unbanded birds were seen; ceedsthe total durationof this study.Thus, no one of them bred in 1990 and 1991). Thus, al- senescenceis to be expected.However, it is like- though no bird born in the Causseswas re- ly that such survival rates are not peculiar to corded in the nearest Griffon Vulture colonies our population.Long-lived speciesare extreme- in Spain or in the Pyrenees,they may belong ly sensitive to small variations in survival rates to the samemetapopulation. Natal dispersalfa- (Lebreton and Clobert 1991). Thus, the rapid voring gene flow between populations(for re- increasein SpanishGriffon Vulture populations view, seeJohnson and Gaines 1990) could occur (Donazar 1987, Donazar and Fernandez 1990) in this population. The erratic behavior of im- may be due to a slight increasein survival rates maturesthat cameback someyears later to the reaching values similar to thosereported here. colony may be responsible for the lower re- The effect of release on adult survival in the sighting ratesof wild-born birds (Fig. 3). Erratic 860 $ARRAZINErAL. [Auk,Vol. 111

behaviorof young birds hasbeen shown to take The presentreintroduction schedulediffered them to North and West Africa in autumn (Elo- from previousreintroduction programs by re- segui 1989, Berthold et al. 1991). The juvenile leasingadult pairsin order to establisha breed- recoveredin Senegalshows that birds born in ing group.In former reintroductionprojects in- the Causseshave restoredthis migration pat- volving raptors such as the Bearded Vulture tern. However, we could not detectany effect (Gypaetusbarbatus; Coton and Esteve 1990), the of age on resighting rates of wild-born birds White-tailed SeaEagle (Haliaeetusalbicilla; Love despite the fact that the model [45a3.6,Pa•.6 + t] 1983),and the PeregrineFalcon (Peregrinus an- was very closeto the selectedone [45a3.6,P,] (Ta- atum;Holroyd and Banash1990), only imma- ble 1). This could be due to variations in the ture birds were released.Taking into account age at first breeding together with few tem- the biologicalcost of our method (during the porary band lossesin wild-born adults. firstyear after releasethe apparentmortality of Nevertheless,the juvenile survival rate esti- adults was higher [0.257] than later on [0.013], mated in the Caussesis much higher than that the survival rate from releaseto first breeding estimatedby Piper et al. (1981),or by Robertson is higher with adults(0.743) than with juveniles (1984) for the Cape Vulture. Even though the (0.8583x 0.987 = 0.623; Fig. 2), even ignoring methodsused by theseauthors had manybiases a possiblerelease effect on juveniles. In addi- that could explain sucha discrepancy,the im- tion, immatures are much more likely to emi- mature survival rate observed in the Causses is grate, especiallywhen there is no established probablyhigher than in otherpopulations. Both naturalpopulation, which is alwaysthe casefor an increaseof the populationdensity and a de- reintroduction projects.Thus, we proposethat creaseof human effort to make resourcespre- reintroduction protocols for Griffon Vultures dictablein the future are likely to affect im- and similar speciesshould involve adults that mature survival rates. This encouragesus to have bred in captivity within the target area study natural populationsin order to discrim- rather than fledglings or even immatures.Two inate between intrinsic characteristics of sur- new reintroduction projectsof Griffon Vulture vival rates and those due to the management in Friuli (northeasternItaly) and the southern of reintroducedGriffon Vulture populations. French Alps will useour method. Becausenatal Conservationaspects.--The reintroduction dispersalrapidly developedfrom Spain to the program strictly followed the recommenda- Caussescolonies, these projects are expectedto tions of the International Union for Conser- reinforce the exchange possibilitiesbetween vation of Nature concerningpreparatory mea- subpopulationsin order to restore a southern sures,introduction schedule,and monitoring. EuropeanGriffon Vulture metapopulation. From a conservationpoint of view, the educa- tion programwas successful:no human perse- ACKNOWLEDGMENTS cution (shooting,poisoning) was recorded(Ta- We are very grateful to M. Terrasseof the Fonds ble 4). Only one caseof disturbanceat the nest d'Intervention pour les Rapacesand J. Bonnet of the by climberswas reported. Until now, the effi- Parc National des Cevennes,who designedthe field cient collaborationof local farmershas provid- reintroduction and monitoring. Data were used in the ed an abundant food resourceat feeding sta- framework of the agreement between the Parc Na- tions (Bonnet et al. 1990, Terrasse 1990). tional des Cevennes,the Fonds d'Intervention pour Moreover, farmers now tend to leave carcasses les Rapaces,and the Laboratoired'Ecologie of the in the field more often so that Griffon Vultures EcoleNormale Sup6rieure.The Centre de Recherche canforage by themselves.As for CapeVultures sur la Biologie des Populationsd'Oiseaux supported banding. We alsothank M. S. Gaines,D.C. Houston, in South Africa (Markus 1972) and other large and J. D. Nichols for their reviews and J. M. Thiollay raptorssuch as SpanishImperial Eagles(Aquila and I. Swingland for improving an earlier draft of adalberti;Ferrer and Hiraldo 1991), electrocu- the manuscript.T. Boulinier and G. Sorci provided tion is the main cause of death (Table 4). It fruitful discussions.This study was supportedby a increasesjuvenile mortality of Griffon Vultures SRETIEgrant (nø90088)of the FrenchMinistry of the in Caussesas in Israel (Leshem 1985). A collab- Environment. orationwith Elecricit•De Francefrom 1988per- LITERATURE CITED mitted us to modify about 300 pylons near the releasecenter to reducethe possibilityof elec- ANDERSON, D. R., K. P. BURNHAM, AND G. C. WHrrE. trocution. 1985. Problems in estimating age-specificsur- October1994] SurvivalofReintroduced Vultures 861

vival rates from recovery data of birds ringed as FERRER,M., AND F. HIRALDO. 1991. Evaluation of young. J. Anita. Ecol. 54:89-98. managementtechniques for the SpanishImperial BERTHET, G. 1946. Les derniers Vautours fauves du Eagle. Wildl. Soc. Bull. 19:436-442. Massif Central. Rev. Soc. For. Franche-Comte, 1-6. FRY,D. M. 1983. Techniques for sexing monomor- BERTHOLD, P., J. GREISINGER,E. NOWAK, AND U. phic vultures. Pages 185-196 in Vulture biology QUERNER.1991. Satellite tracking of a migrating and management (S. R. Wilbur and J. A. Jackson, Griffon Vulture in Spain. J. Ornithol. 132:327- Eds.).Univ. California Press,Berkeley. 329. HOLROYD, G., AND U. BANASH. 1990. The reintro- BONNET,J., M. TERRASSE,C. BAGNOLINI,AND J. L. PINNA. duction of the Peregrine Falcon,Falco peregrinus 1990. Installation et extension d'une colonie de anatum, into southern Canada. Can. Field Nat. Vautoursfauves (Gyps fulvus fulvus) r•introduite 104:203-208. dans les Grands Causses du Massif Central. L'O- HOUSTON,D.C. 1974. Mortality of the CapeVulture. iseau et la R.F.O. 60:181-206. Ostrich 45:57-62. BRIQUET, R. 1987. Les ressources alimentaires des HOUSTON,D.C. 1982. The statusof European vul- Vautoursfauves dans la r•gion desGrands Causs- tures and their rangesin the Soviet Union. Vul- es. Paris, Rapport stageenseignement optionnel ture News 8:4-6. de 4•me ann•e-E.N.V. Maisons-Alfort. HOUSTON,D.C. 1987. Management techniques for BURNHAM, K. P., D. R. ANDERSON, G. C. WHITE, C. vultures feeding and releases.Pages 15-29 in BROWNIE,AND K. H. POLLOCK.1987. Design and Breeding and managementin birds of prey (D. analysis methods for fish survival experiments J. Hill, Ed.). Univ. Bristol, Bristol. basedon release-capture.Am. Fish. Soc. Mono- JOHNSON,M. L., AND M. S. GAINES. 1990. Evolution gr. 5. of dispersal:Theoretical models and empirical CAROTHERS,A.D. 1973. The effectsof unequal catch- testsusing birds and mammals.Annu. Rev. Ecol. ability on Jolly-Seber estimates. Biometrics 29: Syst. 21:449-480. 79-100. KREMERS,W. K. 1988. Estimation of survival rates CLOSERT,J., R. JULIARD,AND R. H. MCCLEERY. 1993. from a mark-recapturestudy with tag loss. Bio- The componentsof local recruitment.Pages 281- metrics 44:117-130. 293 in Marked individuals in the study of bird LAKHANI,K. H., AND I. NEWTON. 1983. Estimating population (J.D. Lebreton and P.M. North, Eds.). age-specificbird survival rates from ring recov- BirkhauserVerlag, Basel. eries-Can it be done? J. Anita. Ecol. 52:83-91. CLOBERT,J., J. D. LEBRETON,AND D. ALLAINE. 1987. A LESRETON,J. D., AND J. CLOSERT.1991. Bird popu- general approach to survival estimation by re- lation dynamics,management and conservation: captures or resightings of marked birds. Ardea The role of mathematicalmodelling. Pages105- 75:133-142. 125 in Bird population studies.Relevance to con- CLOBERT,J., J. D. LEBRETON,M. CLOBERT-GILLET,AND servationand management(C. M. Perrins, J. D. H. COQUILLARD. 1985. The estimation of surviv- Lebreton, and G. J. M. Hirons, Eds.).Oxford Univ. al in bird populationsby recapturesor sightings Press, Oxford. of marked individuals.Pages 197-213 in Statistics LEBRETON,J. D., K. P. BURNHAM, J. CLOBERT,AND D. in ornithology (B. J. T. Morgan and P.M. North, R. ANDERSON.1992. Modeling survivaland test- Eds.).Springer, Berlin. ing biological hypothesisusing marked animals: COTON, C., AND R. ESTEVE. 1990. La r•introduction A unified approachwith casestudies. Ecol. Mon- du Gypa•te barbudans les Alpes.In Reintroduc- ogr. 62:67-118. tions et renforcementsde populationsanimales LESHEM, Y. 1985. Griffon Vultures in Israel: Electro- en France. Compte rendu du colloque de Saint cution and other reasonsfor a declining popu- Jeandu Gard Supl. 5, Revue d'Ecologie,La Terre lation. Vulture News 13:14-20. et la Vie:227-241. LOVE,J. 1983. The return of the Sea-eagle.Univ. DONAZAR,J.A. 1987. Apparent increasein a Griffon Press,Cambridge. Vulture (Gypsfulvus) population in Spain.J. Rap- MARKUS,M. 1972. Mortality of vultures causedby tor Res. 21:75-77. electrocution. Nature 238:228. DONAZAR,J. A.,AND C. FERNANDEZ.1990. Population NEWTON,I. 1979. Population ecologyof raptors.T. trends of the Griffon Vultures Gypsfulvus in & A.D. Poyser,Berkhamsted. northern Spain between 1969 and 1989 in rela- NICHOLS,J.D. 1992. Capture-recapturemodels using tion to conservation measures. Biol. Conserr. 53: marked animalsto study population dynamics. 83-91. BioScience 42:94-102. ELOSEGUI,I. 1989. Vautourfauve (Gyps fulvus), Gy- NICHOLS, J. D., S. L. STOKES,J. E. HINES, AND M. J. paetebarbu (Gypaetus barbatus), Percnopt•re d'E- CONROY. 1982. Additional comments on the as- gypte (Neophronpercnopterus): Synth•se biblio- sumptionof homogeneoussurvival ratesin mod- graphiqueet recherches.Acta Biologica Montana ern bird banding estimation models. J. Wildl. 3. S•rie Document de Travail. Manage. 46:953-962. 862 S^P,R•ZXNETAL. [Auk, Vol. 111

PENNYCUICK,C.J. 1972. Soaringbehaviour and per- document. C.E.F.E./C.N.R.S. BP 5051, 34033 formance of some East African birds, observed Montpellier, CEDEX, France. from a motor-glider. Ibis 114:178-218. ROBERTSON,A. S. 1984. Aspectsof the population PIPER,$. E. 1990. Dispersal, nomadism and conser- dynamicsof Cape Vultures in the Capeprovince. vation in the Cape Vulture. Ring 13:217-218. Ostrich 55:196-206. PIPER,$. E., P. J. MUNDY, AND J. A. LEDGER. 1981. ROBERTSON,A. S. 1985. Observationson the post Estimatesof survival in the Cape Vulture Gyps fledging dependenceperiod of Cape Vultures. coprotheres.J. Anim. Ecol. 50:815-825. Ostrich 56:58-66. POLLOCK,K.H. 1991. Modeling capture,recapture, SIMMONS,R. 1986. Delayed breeding and the non- and removal statistics for estimation of demo- adaptivesignificance of delayedmaturity in vul- graphic parametersfor fish and wildlife popu- tures:A fruit fly perspective.Vulture News 15: lations: Past,present, and future. J. Am. Stat. As- 13-18. soc. 86:225-237. TERRASSE,g. 1990. R•introduction du Vautour fauve POLLOCK,K. H., J. D. NICHOLS,C. BROWNIE,AND J. E. dans les grands Causseset renforcement de la HINES. 1990. Statistical inference for capture- population de Vautour percnopt•re. Pages213- recaptureexperiments. Wildlife Monogr. 107. 226 in Reintroductionset renforcementsde pop- PRADEL,R., J. CLOBERT,AND J. D. LEBRETON.1990. ulations animales en France. Compte rendu du Recentdevelopments for the analysisof capture- colloque de Saint Jean du Gard Supl. 5, Revue recapture multiple data set. An example con- d'Ecologie, La Terre et la Vie. cerning two Blue Tit populations.Ring 13:193- W•MERSKmCH,H., J. CLOBERT,AND P. JOUVF2,rnN.1987. 204. Survival in five southern albatrosses and its re- PRADEL,R., AND J. D. LEBRETON.1991. User's manual lationshipwith their life history. J. Anim. Ecol. for program SURGE version 4.1 Mimeographed 56:1043-1055.