Nesting habitat selection by the Spanish imperial eagle Aquila adalberti Luis M. Gonzalez ICONA, Servicio de Vida Silvestre, Gran Via San Francisco, 435, Madrid 28005, Spain Javier Bustamante & Fernando Hiraldo Estación Biológica de Doñana, Consejo Superior de Investigaciones Cient(ficas, Pabellón del Peru, Avda. Maria Luisa s/n, Sevilla 41013, Spain () Nesting habitat selection by the Spanish imperial eagle Aquila adalberti was quantitatively assessed. Nest sites chosen did not differ from the available habitat with regard to physiography and vegetation, but the eagles tended to avoid areas affected by human disturbance. Nest sites used by subadults and those of adults that were recently established tended to be in more disturbed areas than those used by adults or in traditional nesting localities. Management recommendations for more effective conservation of the eagle’s habitat are discussed. ‘endangered’ in the IUCN Red Data Book (King, 1981; Wilcox, 1988). The current world population is estimated at just over a hundred pairs, all in the Iberian Peninsula (Gonzalez et al., 1987). Factors influencing the present distribution of the Spanish imperial eagle have recently been studied (Gonzalez et a!., 1990), but nesting habitat has been described only qualitatively (Valverde, 1960; Garzón, 1974; Meyburg, 1975). In this paper we provide a quantitative description of their nesting habitat and test for differences between available and selected habitat, between pairs of adult and subadult breeders, and between traditional and new nest sites. INTRODUCTION One of the most important causes of species decline is the loss or alteration of habitat (Greenway, 1967; Temple, 1978; King, 1981). Snyder and Snyder (1975) drew attention in the United States to a progressive loss of suitable habitat for many species of birds of prey, and suggested that habitat preservation was of prime importance in maintaining raptor populations at an acceptable size. This view coincides with that of various authors in other areas (Newton, 1979; Steyn, 1983). Since then, numerous studies on the conservation and management of birds of prey threatened by local or global extinction have quantified data on nest sites (Grubb, 1976; Morris, 1980; Bednarz & Dinsmore, 1981; Newton et a!., 1981; Andrew & Mosher, 1982; Reynolds et a!., 1982; Gilmer & Stewart, 1984; Rich, 1986; Kost rzewa, 1987). The Spanish imperial eagle Aquila adalberti Brehm 1861, currently considered as a separate species from the Eastern imperial eagle Aquila heliaca (Hiraldo et a!., 1976; Collar & Andrews, 1988; Gonzalez eta!., 1989), is one of the rarest birds of prey at a worldwide level and is considered METHODS Two national censuses of the Spanish imperial eagle have been conducted, in 1971—74 (Garzón, 1972, 1974) and in 1981—86 (Gonzalez et a!., 1987). The nests recorded during the second census were used in the present study. A total of 108 nest sites, belonging to at least 104 different breeding pairs, were marked on 1:50000 topographic maps prepared by the Spanish Army Cartographic Service and on land use maps prepared by the Agriculture Ministry. We then evaluated the habitat within a radius of 325 km 45 from each nest, i.e. half the average distance between nests of neighbouring pairs, following the methods of Grubb (1976), Howell et a!. (1978), Bednarz and Dinsmore (1981), Gilmer and Stewart (1984) and Rich (1986). In addition 108 random sites were sampled to evaluate habitat available to the species, using the 53 sheets of the ‘L’ series 1:50 000 topographic map of Spain (covering an area of 26712km2) in which the species had been located (Gonzalez et al., 1987). To avoid bias due to differences in habitat and level of human influence among the various breeding areas, random sampling was stratified, the number of random sites included on each map sheet being equal to that of previously marked nest sites. Random sites were located on the maps using random generation of coordinates with a calculator, and were at least 65 km from any nest site. Since the eagle nests in trees, random sites which fell in unwooded areas (farmland, uncultivated land, meadows, rocky ground, urban centres, reservoirs, etc.) were rejected (Bednarz & Dinsmore, 1981; Gilmer & Stewart, 1984; Rich, 1986; Speiser & Bosakowski, 1987). For every nest site and random site we measured 19 variables on the maps (Table 1). As land use maps were unavailable for some areas, some variables could not be measured for 20 nest sites and 18 random sites. Land use and habitat were verified by field visits, averaging five visits per nest site to each of the 53 map sheets used. Breeding subadults (3—4 years old) are readily distinguished by their light brown plumage spotted with black compared with the black plumage of adults (Valverde, 1960). Pairs with at least one subadult breeder were analysed separately from pairs in which both were adult, in order to test possible differences in habitat selection. In pairs where both members were adults we also analysed separately those breeding at ‘traditional’, i.e. those occupied during both censuses, and at ‘new’ nest sites, occupied only since 1981. Mean values for nest site and random site variables were compared using t-tests. A stepwise discriminant function analysis was conducted using Table 1. Variables used to characterize the nest sites of the Spanish imperial eagles Mnemonic code 1. 2. 3. 4. 5. 6. DINBUIL” DVILL” DPAVRO DUNPAVRO’ HEIGHT’ TOPIND 7. KPAVRO’ 8. KEL° 9. INHAW 10. DTW’ 11. TCOV” 12. PATINDb 13. QUEFORb 14. SABFORb 15. CONFORb 16. OTHFORb 17. SCRUBb 18. OPLANb 19. INACC Meaning Distance from nest to nearest inhabited building. Distance from nest to nearest urban centre. Distance from nest to nearest paved road. Distance from nest to nearest unpaved road passable by vehicle. Height of nest above sea-level (m). Topographic irregularity index. Total number of 20 m contour lines, cut by two lines diametric to the sampling circle in the directions N—S and E—W. Kilometres of paved roads in the circular sampling area, to the nearest 05 km. Kilometres of electric power lines in the circular sampling area, to the nearest 05 km. Number of inhabitants in a radius of 325 km around the nest, calculated from the number of inhabited buildings and the proportion of urban centres included in this radius, measured on the topographic map (20 inhabitants precision). Distance from nest to the border of wooded area. Percentage of tree cover in wooded area housing the nest. Habitat patchiness index, number of times two lines diametric to the sampling circle in the directions N—S and E—W cut the limit of zones with different agricultural use or different vegetation. Percentage of surface covered by dense oak forest Quercus spp. (cover >70%). Percentage of the surface covered by savannah-like forest (pasture and scattered trees): Quercus spp., Olea europaea and occasionally Pinus spp., with canopy cover <70% and ground cover with pasture and cultivated farmland. Percentage of the surface covered by coniferous forest. Percentage of surface covered by other species of trees: predominantly Eucalyptus, Fraxinus and Populus. Percentage of surface covered by mediterranean scrubland. Percentage of surface occupied by open land, herbaceous formations or ground without vegetation (pasture, non-forested cultivated lands, meadows and rocky ground).C Inaccessibility index, an estimate of difficulty of access by foot in the area calculated as a function of the steepness of the relief and the surface of scrubland according to the formula INACC = SCRUB + 2 x TOPIND. All horizontal distances measured in kilometres to the nearest 005 km unless stated otherwise; percentage values to 1%. From topographic maps. 1’From farming and land use maps. To avoid redundancy in the discriminant analysis the habitat unsuitable to the eagle (urban centres, reservoirs) is not considered as an independent variable, being the amount lacking to 100% of the surface once variables 13 to 18 are added. C BMDP (Dixon & Brown, 1983) to identify the set of variables which best separated nest sites and random sites. RESULTS Table 3. Comparison of 19 nest site variables (means and standard deviations) between (a) adult pairs (n = 89) and pairs with at least one subadult (n = 19); (b) traditional nest sites (known since 1971) (a = 55) and new nest sites (only since 1981) (a = 34), all nests with both adults (a) Variable0 Nesting habitat did not differ from random habitat in any of the variables related to vegetation structure (Table 2). The only statistically significant differences between the two groups could be found in those variables relating to the physical environment and degree of human influence. Nests were in more rugged terrain (TOPIND), and in more inaccessible areas (INACC), with less paved roads (KPAVRO) and power lines (KEL), than random sites. Nests were also located further away from paths (DUNPAVRO), roads (DPAVRO), inhabited buildings (DINBUIL), and villages (DVILL). In the stepwise discriminant analysis nest sites and random sites were best distinguished by the following relationships: Nest sites = 1806 78 + 0042 90 KPAVRO + 0153 68 TOPIND Adult pairs DINBUIL* DVILL DPAVRO DuNpAvRO***+ HEIGHT TOPIND KPAVRO+ KEL* + + + INHAB + + + *DTB *TCOB+ + *PATIND* *QUEFOR+ + + *SABFOR+ *CONFOR**+ + + *OTHFOR *SCRUB**+ + *OPLAN+ + *INACC* Subadult pairs K +SD 18 69 39 [0 679•8 0•9 41 28 09 4367 91 153 40 06 29oo 04 4[0 126 76 23•9 201 24 257 167 54Q — K 13 55 36 05 5547 114 55 57 22 9290 07 350 72 151 280 42 1 6920 06 450 173 206 179 255 75 220 186 307 59 29 150 320 370 +SD — 07 3.3 28 03 3252 90 106 59 5 76[0 07 2&0 84 246 170 146 59 1F9 303 24I Random sites = 1761 09 + 0122 64 KPAVRO +011503 TOPIND Table 2. Comparison of 19 habitat variables (means and standard deviations) between 108 nest sites and 108 random sites (b) Variable VariabIes Nest sites X DINBUIL** DVILL DPAVRO*** DUNPAVRO* HEIGHT TOPIND* KPAVRO***+ + KEL***+ + + INHAB DTB * *TCDB *PATIND *QUEFOR *SABFOR *CONFOR *OTHFOR+ + SCRUB * *OPLAN INACC* * ±SD 174 091 &65 398 387 279 087 081 65780 42070 1460 920 430 660 F20 340 53700 2565{)0 046 069 4200 3300 760 1350 990 1780 2280 2640 1750 2440 720 250 2370 2090 1940 2F80 5090 3020 2 **=p<O.O1, *** =p<o.ool. New nests K ±SD X ±SD ±SD 138 102 132 386 246 254 044 075 59970 38000 1)00 980 980 860 300 600 98100 3076* 040 079 4300 3000 1410 140 740 2700 1170 130 1130 1970 3980 1620 3140 22* 1600 2010 1930 3160 variables marked *, n = 88 for nest sites and n =90 for random sites. Significance of Levene F-test for difference between variances: + + =p <01)1, + + + =p <01)01; and of Student 1-test for difference between the means, for equal or different variances according to each case: * =p <01)5, ° Traditional nests Random sites DINBILL DVILL DPAVRW+ + + DUNPAVRO HEIGHT++ TOMND+ + KPAVRO+ + KBL++ INnAr+++ +DTB +TCDB +PATIND+ tQUBPOR +SABFOR +CQNFOR+ +U1HPOR +SCRUB 4’CPLAN++ •INACC+ a 19 78 48 1-0 6104 143 25 0-3 92-0 0-4 36-0 112 7-6 26-6 13-5 3-1 25-7 19-0 50-6 10 [7 38 30 09 4808 101 4-2 1-4 4800 07 35-0 6-0 14-9 27-6 21-9 8-5 22-0 22-1 33-4 55 25 0-9 792-1 16-9 6-4 1-0 611-0 0-4 49-0 14-8 7-5 19-9 30-0 1-4 25-5 13-1 58-9 09 41 17 0-8 330-6 7-1 6-5 3-0 13840 0-7 340 8-3 15-6 28-7 27-5 8-5 22-4 11-2 260 For variables marked *,the samples are n = 72 and n = 16 for adults—subadults, respectively, and n =43 and n = 29 for traditional—new, respectively. Significance of Levene F-test for difference between variances: + =p <05, + + =p <01)1, + + + = p < 01)01; and Student 1-test for difference between the means, for equal or different variances according to each case:*=p<01)5,**=p<01)1,***=p<0.001. 48 Luis M. Gonzalez, Javier Bustamanie, Fernando Hiraldo We found that 614% of the nest sites and 633% of the random sites were correctly classified. A jack-knife classification reduced the correct classification of nest sites to 6O2%, but maintained that of random sites at 63-3%; the kappa statistic (Titus et a!., 1984) showed a classification rate 24% better than chance (kappa = O2358, Z = 3-15; p <0-01). Comparison of the nesting habitat of adult pairs and pairs with at least one subadult member (Table 3) showed that subadult pairs occurred in areas with greater human influence, their nest sites being closer to villages and roads, in more accessible areas and with more kilometres of power lines. Confining the analysis to adult pairs only, a similar tendency was obtained when comparing the habitats of ‘tra­ ditional’ and ‘new’ nests (Table 3). New nest sites were in areas with significantly more human inhabitants and kilometres of roads, and tended to be located closer to villages and roads than were traditional ones. DISCUSSION Unlike other species of raptors described by McGowan (1975), Reynolds et at. (1982), Bednarz and Dinsmore (1982), Janes (1985) and Rich (1986), the Spanish imperial eagle does not seem to prefer one habitat rather than another in terms of cover and vegetation: nest selection in this respect reflects the proportions of habitat found in the area, and the eagle does not appear to seek out areas more forested than the average. This may be partly due to the fact that the population we studied is located in an area influenced by centuries of human activity (see Bauer, 1981). Given the eagle’s size and mobility in relation to the size of the patch (PATIND values much greater than 1, Table 2), a tendency towards a particular habitat specialization would appear difficult (see MacArthur & Pianka, 1966), and the variables most associated with the choice of nesting habitat indicate attempts to avoid human disturbance, as in other birds of prey (Grubb, 1976; Andrew & Mosher, 1982; Kostrzewa, 1987). The 61-4% correct classification inthe discriminant function was only 24% better than expected by chance. We believe that the lack of a higher discrimination was due to the heterogeneity of the sample studied, which included breeding areas with very different landscapes, and to individual variation in the birds’ general intolerance to the presence of humans, as occurs with other large birds of prey (Grubb, 1976). The fact that pairs with at least one subadult were found in more disturbed habitats can be attributed to several possible causes: (1) These pairs were less selective in choosing their nesting habitats. This cannot be totally excluded as there were no significant differences for such pairs between nest sites and nearby random sites (n 31) for any of the 19 variables examined. We do not, however, consider that subadults are less selective; in 90% of the cases one member of the pair was an adult and pairs formed only by adults at ‘new’ nest sites also occurred in more disturbed habitats. (2) They were mainly new pairs, choosing marginal habitat because less disturbed habitat was saturated by traditional adult pairs. This was suggested by our field observations, which showed that the population is growing in marginal habitats around traditional nesting areas (Gonzalez ci al., 1987). This expansion strategy seems to be motivated by the species’ philopatric behaviour; of 10 eagle chicks ringed and recovered as breeding adults, nine subsequently nested <10km from their birthplace and only one 80km away (ICONA Ringing Center and Doñana Biological Station ringing files). We do not think these data are biased as ringing has been undertaken across the whole distribution area for the last ten years. This behaviour, also shown in other species of birds of prey (Newton, 1979), would favour the saturation of existing nesting nuclei (Diamond, 1976, 1984) and the occupation of surrounding areas, even if the habitat is disturbed, before recolonization of more suitable habitats at greater distances. In fact, new pairs of adults are also settling in such areas that are more disturbed than average (Table 3). (3) Turnover rates of pair members in these more disturbed habitats were higher, and this increases the frequency of a subadult being incorporated into the breeding population. This has been suggested for the golden eagle Aquila chrysaetos by Steenhof ci at. (1983), and is supported here by the fact that there are significantly more kilometres of power lines (KEL) in the breeding territories of subadults. Electrocution is a frequent Spanish imperial eagle cause of death in birds of prey (Olendorif, 1981) and was the most common cause of non-natural death in the Spanish imperial eagle (501% of known causes, n = 42), followed by shooting (287%) in the period 1981—1988 (Gonzalez, 1989). The predicted increase in the mortality of breeding birds in habitats with a higher level of human disturbance increases the probability that subadults will breed in these areas (Newton, 1979). We believe that the saturation of safer nest sites in the traditional breeding areas, and the increased mortality of breeders in more disturbed habitats, are responsible for the subadults’ choice of nesting habitat rather than simply poor judgement on the part of these birds. Management implications Management of Spanish imperial eagle nesting habitat should include an inventory of existing and potential habitats. This can be accomplished in part by using the information in this paper and in a previous one on the factors conditioning the eagle’s present distribution (Gonzalez et a!., 1990). Construction of new forest paths and power lines in the actual nesting habitat should be avoided. Power lines already crossing these should be modified to reduce electrocution risks, following the recommendations made by Olendorif (1981). Since this species appears to be sensitive to disturbance, we recommend that all public use, forestry works, etc. should be reduced within a radius of 500 m around nests from the initiation of breeding until young fledge—from February to August. Planting should be carried out along some of the existing forest paths to reduce or eliminate the eagle’s visual contact with possible disturbance (Andrew & Mosher, 1982). As the population seems to be expanding around existing nesting nuclei where the remaining habitat is affected by high levels of human disturbance, we suggest that chicks produced in captivity, or taken from nests with large broods (3—4 chicks) to reduce mortality caused by cainism (Meyburg & Garzón, 1973), should be released in other less disturbed areas formerly occupied by the species. ACKNOWLEDGEMENTS Special thanks go to J. Garzón, José L. Gonzalez, B. Heredia, F. Palacios and the staff of Unidad de 49 Zoologia Aplicada INIA (Madrid), the Estación Biológica de Doñana, CSIC (Sevilla) and Parque Nacional de Doñana (ICONA) for their help during the field work. We received helpful review comments from K. Titus and two anonymous referees. The study was financially supported by ICONA and the DGICYT Projects No. 2007 and PB87-0405. REFERENCES Andrew, J. M. & Mosher, J. A. (1982). Bald eagle nest site selection and nesting habitat in Maryland. J. WildL Manage., 46, 383—90. Bauer, E. (1981). Los Monies en Ia Historia de España. Ministerio de Agricultura, Madrid. Bednarz, J. C. & Dinsmore, J. J. (1981). Status, habitat use and management of red-shouldered hawks in Iowa. J. Wildi. Manage., 45, 236—41. Bednarz, J. C. & Dinsmore, J. J. (1982). Nest-sites and habitat of red-shouldered and red-tailed hawks in Iowa. Wilson Bull., 94, 31—45. Collar, N. J. & Andrews, P. (1988). Birds to Watch: the ICBP World Check-list of Threatened Birds. ICBP Techn. Pubis, No.8. Cambridge. Diamond, J. M. (1976). Island biogeography and conservation: strategy and limitations. Science, N.Y., 193, 1027—29. Diamond, J. M. (1984). Normal extinctions of isolated populations. In Extinctions, ed. M. H. Nitachki. University of Chicago Press, Chicago, pp. 191—246. Dixon, W. J. & Brown, M. J.(1983). BMDP, Statistical Software. University of California Press, Los Angeles. Garzón, J. (1972). Especies en peligro: el aguila imperial. Adena, 4, 8—12. Garzón, J. (1974). Contribución al estudio del status, alimentación y protección de las Falconiformes en Espafla Central. Ardeola, 19, 270—330. Gilmer, D. S. & Stewart, R. E. (1984). Swainson’s hawk nesting ecology in North Dakota. Condor, 86, 12—18. Gonzalez, L. M. (1989). Historia Natural del aguila imperial ibérica (Aquila adalberti) en Espafla. Tesis Doctoral, Universidad Autónoma Madrid. Gonzalez, L. M., Gonzalez, J. L., Garzón, J. & Heredia, B. (1987). Censo y distribución del aguila imperial ibérica (Aquila heliaca adalberti) en España durante el periodo 1981—1986. Bol. Est. Cenir. Ecol., 16, 99—109. Gonzalez, L. M., Hiraldo, F., Delibes, M. & Calderén, J. (1989). Zoogeographic support to consider the Spanish imperial eagle as a distinct species. Bull. Brit. Ornith. Club, 109, 86—93. Gonzalez, L. M., Bustamante, J. & Hiraldo, F. (1990). Factors influencing present distribution of the Spanish imperial eagle (Aquila adalberti). Biol. Conserv., 51, 311—19. Greenway, J. C. (1967). Extinct and Vanishing Birds of the World. Dover Publishers, New York. Grubb, T. 0. (1976). A survey and analysis of bald eagle nesting in Western Washington. MSc thesis, University of Washington. Hiraldo, F., Delibes, M. & Calderón, J. (1976). Sobre el status taxonómico del aguila imperial ibérica. Doñana Ada Vert., 3, 171—82. Howell, J., Smith, B., Holt, J. B. & Osborne, D. R. (1978). Habitat structure and productivity in red-tailed hawks. Bird Banding, 49, 162—71. King, W. (1981). Endangered Birds of the World. The Red Data Book. Vol. Ayes. JUCN/ICBP, London. 50 Luis M. Gonzalez, Javier Bustamante, Fernando Hiraldo Kostrzewa, V. A. (1987). Quantitative Untersuchungen zur Habitattrennung vom Mausebussard (Buteo buteo), Habitch (Accipiter gentilis) und Wespenbussard (Pernis apivorus). .1. Orn., 128, 209—29. Janes, S. W. (1985). Habitat selection in raptorial birds. In Habitat Selection in Birds, ed. M. L. Cody. Academic Press, New York, pp. 159—87. MacArthur, R. A. & Pianka, F. R. (1966). On optimal use of a patchy environment. Amer. Nat., 190, 603—9. McGowan, J. D. (1975). Nesting habits and reproductive success of goshawks in interior Alaska. Population status of raptors. Proc. Conf Raptor Conservation Techniques, Fort Collins, Colorado, 22—24 March 1973, Part 6. Raptor Res. Rep., 3, 147—55. Meyburg, B.-U. (1975). On the biology of the Spanish imperial eagle (Aquila heliaca adalberti). Ardeola, 21, 245—83. Meyburg, B.-U. & Garzón, J. (1973). Sobre Ia protección del aguila imperial ibérica (Aquila heliaca adalberti) aminorando artificialmente Ia mortalidad juvenil. Ardeola, 19, 107—28. Morris, M. M. (1980). Nest site selection by the red-shouldered hawk (Buteo lineatus) in southwestern Quebec. MS thesis, McGill University. Newton, I. (1979). Population Ecology of Raptors. T. & A. D. Poyser, Berkhamsted. Newton, I., Davis, P. E. & Moss, D. (1981). Distribution and breeding of red kites in relation to land-use in Wales. J. App!. Ecol., 18, 173—86. Olendorif, R. R. (1981). Suggested practices for raptor protection on power lines. The state of the art in 1981. Raptor Res. Rep., 4, 1—111. Reynolds, R. T., Meslow, E. C. & Wright, H. M. (1982). Nesting habitat of coexisting Accipiter in Oregon. J. Wild!. Manage., 46, 124—38. Rich, E. (1986). Habitat and nest site selection by burrowing owls in the sage brush steppe of Idaho. J. Wild!. Manage., 50, 548—55. Snyder, N. T. R. & Snyder, H. A. (1975). Raptors in range management. In Proc. Symposium on Management of Forest and Range 1-Jabitats for Nongame Birds (D. R. Smith, Techn. Coordinator). USDA Forest Service Gen. Tech. Rep., No. WO-l, pp. 190—209. Speiser, R. & Bosakowski, T. (1987). Nest site selection by northern goshawk in Northern New Jersey and Southwestern New York. Condor, 89, 387—94. Steenhof, K., Kochert, M. N. & Doramus, J. H. (1983). Nesting of subadult golden eagles in southwestern Idaho. Auk, 100, 743—6. Steyn, P. (1983). Birds of Prey of Southern Africa. Croom Helm, London. Temple, S. A. (1978). The concept of managing endangered birds. In Endangered Birds, ed. S. A. Temple. University of Wisconsin Press, Madison, Wisconsin, pp. 3—8. Titus, K., Mosher, J. A. & Williams, B. K. (1984). Chancecorrected classification for use in discrimination analysis: ecological applications. Amer. MidL Nat., 111, 1—7. Valverde, J. A. (1960). La population d’aigles imperiaux (Aquila heliaca adalberti) des marismas du Guadalquivir; son evolution depuis un siècle. A!auda, 28, 20—26. Wilcox, B. (1988). IUCN Red List of Threatened Animals. IUCN Conservation Monitoring Center, Cambridge.