Conservation in Conflict: the Tale of Two Endangered Species

GARY W. ROEMER* AND ROBERT K. WAYNE† *Department of Fishery and Wildlife Sciences, New Mexico State University, P.O. Box 30003, MSC 4901 Las Cruces, NM 88003-8003, U.S.A., email [email protected] †Department of Organismic Biology, Ecology, & Evolution (OBEE), University of , 621 Charles E. Young Drive South, Los Angeles, CA 90095–1606, U.S.A.

Abstract: With its umbrella of provisions, the U.S. Endangered Species Act ( ESA) provides critical protection to threatened or endangered wildlife. It provides minimal guidance, however, on identifying taxa worthy of conservation, lacks guidelines for resolving endangered species conflicts, and subsequent recovery programs often focus on the species rather than the ecosystem. These deficiencies are exemplified by the recovery pro- gram for the San Clemente Loggerhead ( ludovicianus mearnsi) and the recent proposed rule to grant federal protection to one of its predators, the island fox ( Urocyon littoralis). Recovery actions that have included euthanasia of foxes have likely contributed to a 40–60% decline in the population size of the San Clemente Island fox ( U. l. clementae), a subspecies listed by the state of California as threatened. The U.S. Fish and Wildlife Service proposes to list four other subspecies of the island fox as endangered but excludes the San Clemente Island fox and the sixth and last subspecies, the San Nicolas Island fox ( U. l. dickeyi), ignoring their evolutionary distinctness and the recent decline in population size of U. l. clementae. Using published mor- phological and genetic information, we show that the shrike’s current taxonomic and legal status should be reevaluated. We also reexamine the current recovery program for the shrike and conclude that the implemen- tation of the ESA’s provisions to protect the shrike was species-centric. The shrike recovery program is primar- ily centered on two approaches: the release of captive-bred and control of native and non-native pred- ators. The predator control program has contributed to the endangerment of the distinct San Clemente Island fox. Given that five of the six fox subspecies face extirpation, the proposed rule to list only four of the six as en- dangered is inadequate. This endangered-species conflict might have been avoided through a more balanced ecosystem approach that considers the ecological role of all native taxa and strives to enhance habitats critical to both the shrike and the fox.

Conservación en Conflicto: el Cuento de Dos Especies en Peligro Resumen: Con su normatividad general, el Acta de Especies en Peligro de E. U. A. (1973) (AEP) proporciona protección decisiva a vida silvestre amenazada o en peligro. Sin embargo, proporciona una orientación mínima para identificar taxones merecedores de conservación y aumenta las directrices para resolver conflic- tos con especies en peligro, y los subsecuentes programas de recuperación a menudo enfocan en la especie y no en el ecosistema. Estas deficiencias están ejemplificadas por el programa de recuperación para el alcaudón ( Lanius ludovicianus mearnsi) y la recientemente propuesta ley para otorgar protección federal a uno de sus depredadores, el zorro ( Urocyon littoralis). Las acciones de recuperación, que han incluido la eutanasia de zorros, probablemente han contribuido a la declinación del 40–60% de la población del zorro de San Cle- mente ( Urocyon l. clementae), una subespecie listada como amenazada por el estado de California. El Servicio de Pesca y Vida Silvestre de E. U. A. propone listar otras cuatro subespecies de Urocyon littoralis como en peli- gro pero excluye a Urocyon l. clementae y la sexta y última subespecie, U. l. dickeyi, ignorando su singular- idad evolutiva y la reciente declinación de la población de Urocyon l. clementae. Utilizando información morfológica y genética publicada, mostramos que el estatus taxonómico y legal de Lanius ludovicianus mearnsi deben ser reevaluados. También reexaminamos el actual programa de recuperación para Lanius lu-

Paper submitted May 13, 2002; revised manuscript accepted January 15, 2003. 1251

Conservation Biology, Pages 1251–1260 Volume 17, No. 5, October 2003

1252 Conservation in Conflict Roemer & Wayne dovicianus mearnsi y concluimos que la instrumentación de las disposiciones de AEP para proteger a la espe- cie son especie-céntricas. El programa de recuperación de Lanius ludovicianus mearnsi se centra principal- mente en dos tratamientos: la liberación de alcaudones criados en cautiverio y el control de depredadores nativos y no nativos. El programa de control de depredadores ha contribuido a poner en peligro a Urocyon l. clementae. Dado que cinco de las seis subespecies de zorro encaran la extirpación, la propuesta ley de listar solo cuatro de las seis como en peligro es inadecuada. Este conflicto de especies en peligro pudo haberse evi- tado con un enfoque ecosistémico más balanceado que considere el papel ecológico de todos los taxones nati- vos y se esfuerce por incrementar los hábitats críticos tanto para el alcaudón como para los zorros.

Introduction cies and the sixth and final subspecies, the San Nicolas Island fox (U. l. dickeyi), from potential federal protec- The U.S. Endangered Species Act (1973) is widely con- tion. Both of the islands harboring the excluded subspe- sidered the most significant wildlife protection legisla- cies are the jurisdiction of the U.S. Navy. We (1) com- tion ever enacted ( National Research Council 1995). pare the evolutionary distinctiveness of the San Clemente The ESA mandates that the Secretary of the Interior de- Loggerhead Shrike to that of the island fox, with special termine whether a species is endangered or threatened reference to the San Clemente and San Nicolas island with extinction “solely on the basis of the best scientific foxes, (2) contrast the current species-centric recovery and commercial data available . . . ” (U.S. Code. 1999. program for the shrike with the potential benefits of an 16 U.S.C. Section 1533). Once a species is listed, it is ecosystem approach, and (3) examine the proposed rule protected by an umbrella of prohibitions, including Sec- to list only four of six subspecies of the island fox and tion 7, which “prohibits any federal action that will jeop- comment on its efficacy. We posit that the momentum ardize the future of any endangered species, including of legal, institutional, and public support for conserva- any threat to designated critical habitat” ( National Re- tion of the shrike has led to an endangered-species con- search Council 1995). However, the act and subsequent flict that has contributed to the endangerment of the en- legal opinions provide little guidance on how to identify demic island fox. taxa worthy of conservation or how to resolve endangered- species conflicts, such as when one endangered species preys on another ( National Research Council 1995). The actions taken to recover the critically endangered San Cle- Evolutionary Distinctiveness of the San Clemente mente Loggerhead Shrike (Lanius ludovicianus mearnsi) Loggerhead Shrike on San Clemente Island, California, and the subsequent in- adequacy of a proposed rule to list one of its predators, the First described by Ridgway (1903), the San Clemente Log- island fox (Urocyon littoralis), as endangered are exam- gerhead Shrike was later considered an endemic form ples of how biased implementation of the provisions of found only on San Clemente Island, California (Miller the ESA can lead to an endangered-species conflict. 1931). In his taxonomic study of the Loggerhead Shrike, In the case of the shrike recovery program, a species- Miller (1931) described 11 different subspecies. In addi- centered philosophy was adopted that resulted in the tion to L. l. mearnsi, he also described another island implementation of a comprehensive predator-control pro- form, L. l. anthonyi, a subspecies found on three other gram that contributed to a decline in the San Clemente Is- Channel Islands (Santa Catalina, Santa Cruz, and Santa land fox (U. l. clementae), a subspecies listed as threat- Rosa), and two mainland forms, L. l. gambeli and L. l. ened by the state of California (California Department of grinnelli, both found in California. In his classification, Fish and Game 1987). Four other subspecies of the island Miller (1931) used a method that relied heavily on the fox (U. l. littoralis, U. l. santarosae, U. l. santacruzae, wing cord–to–tail length ratio ( WC:TL). However, the and U. l. catalinae) are already critically endangered ( Roe- WC:TL is a poor character for classifying shrike subspe- mer 1999; Timm et al. 2000; Roemer et al. 2003), are the cies; its use led to the misclassification of up to 39% of focus of an ecosystem-centered conservation strategy shrike specimens (Collister & Wicklum 1996). Recently, (Coonan 2001; Roemer et al. 2001b, 2002, 2003), and Patten and Campbell (2000) applied a more rigorous ap- were recently recommended to be listed as federally en- proach. They used nine plumage and mensural charac- dangered by the U.S. Fish and Wildlife Service ( USFWS; ters coupled with a discriminant function analysis to as- U.S. Department of the Interior 2001). However, the sess the subspecific status of L. l. mearnsi. Historic samples proposed rule ignored the recent decline in the San of L. l. mearnsi collected between 1897 and 1939 were dis- Clemente Island fox and excluded both this subspe- tinguishable from other shrike subspecies and were most

Conservation Biology Volume 17, No. 5, October 2003

Roemer & Wayne Conservation in Conflict 1253 closely allied with L. l. anthonyi, the only other island shrike population, most likely L. l. anthonyi, in the past form. However, shrike specimens obtained from the re- 80 years. cent San Clemente Loggerhead Shrike recovery program Thus, both morphological and genetic evidence were indistinguishable from L. l. anthonyi, prompting suggest that gene flow likely occurred between L. l. Patten and Campbell (2000) to conclude that “Our mor- mearnsi and L. l. anthonyi ( Mundy et al. 1997b; Patten phological analyses indicate that Loggerhead Shrikes on & Campbell 2000). Although Mundy et al. (1997a, 1997b) San Clemente Island are no longer assignable to the concluded that L. l. mearnsi was sufficiently distinct from mearnsi subspecies. Instead, they appear to be hybrids neighboring subspecies to warrant current conservation or intergrades between mearnsi and anthonyi . . . .” efforts, Patten and Campbell (2000) suggest that a more Molecular genetic analyses of current and historic popu- prudent conservation strategy is to protect all the Channel lations of southern California Loggerhead Shrikes also have Island shrike populations rather than focus conservation been used to assess the distinctiveness of L. l. mearnsi efforts only on San Clemente Island. ( Mundy et al. 1997a, 1997b). Mundy et al. (1997a) found Mainland shrikes migrate to San Clemente Island, and frequency differences in mtDNA haplotypes among popu- captive-reared L. l. mearnsi are found at mainland locales lations of the two island subspecies (L. l. mearnsi and L. l. ( Juola et al. 1997; Mundy et al. 1997a, 1997b). These anthonyi) and two mainland populations of L. l. gambeli. movements, along with the evidence for introgression, Four different mtDNA haplotypes are recognized, and, al- are troubling because they strongly suggest that L. l. though maximum pairwise sequence divergence among mearnsi is hybridizing with other island and mainland populations is low (1.25%), strong genetic substructure is shrikes and may no longer be a distinct form. The ESA does evident. The FST values are high for island-versus-mainland not extend protection to hybrids except in special cases comparisons (0.54–1.0) and between L. l. mearnsi and L. l. (e.g., Florida panther [Puma concolor coryi]; National anthonyi (0.61), indicating a low level of female-biased Research Council 1995), and, although a draft intercross gene flow among shrike populations (Mundy et al. 1997a). policy for the ESA has been written, it has not been L. l. mearnsi has two haplotypes, one identical to that in adopted (U.S. Department of the Interior & Department mainland and the other the sole haplotype among the of Commerce 1996). If the San Clemente Loggerhead nine birds assayed from L. l. anthonyi. This suggests the ex- Shrike is interbreeding with other shrike subspecies, istence of an island-specific haplotype, but there are no there is little precedence for continuing an intensive re- haplotypes unique to L. l. mearnsi. covery program that affects other native species, includ- Mundy et al. (1997b) completed their investigation of ing clearly defined and critically endangered endemic the phylogeographic relationships among southern Cali- taxa. fornia shrike populations by comparing contemporary (1995) and historic (1915) samples of L. l. mearnsi with contemporary samples from mainland populations Evolutionary Distinctiveness of the Island Fox of L. l. gambeli. The contemporary population of L. l. mearnsi is depauperate in genetic variation compared Island foxes are a unique dwarf form endemic to the six with the mainland populations, having lower allelic di- largest of the eight California Channel Islands ( Moore & versity and an expected heterozygosity (He ) averaging Collins 1995). Several researchers have explored the col- 60% that of mainland populations. Significant differ- onization history, evolution, and phylogeography of is- ences in allele frequency among microsatellite loci also land foxes by using morphological and molecular genetic occur between the historic and contemporary island evidence (Gilbert et al. 1990; Collins 1991a, 1991b, 1993; samples. The authors concluded that genetic variation Wayne et al. 1991; Goldstein et al. 1999; Roemer et al. was low in L. l. mearnsi prior to the most recent decline 2001a). and that genetic drift is probably the main evolutionary Collins (1993) used 29 cranial and mandibular charac- mechanism responsible for the low levels of variation in ters measured from 2207 island and gray fox ( U. both the historic and contemporary samples of L. l. cinereoargenteus) specimens to correctly classify 91% mearnsi. The contemporary L. l. mearnsi sample has of all island fox specimens to their island of origin. Wayne two microsatellite alleles that are not in the historic sam- et al. (1991) used seven restriction enzymes to explore ple, but, as with the mtDNA haplotypes, there are no mi- mtDNA variation in both island and gray foxes. These two crosatellite alleles unique to L. l. mearnsi. Interestingly, species do not share mtDNA haplotypes (Table 1), all the historic samples of L. l. mearnsi have a single whereas island foxes have a unique restriction site for mtDNA haplotype (haplotype A), whereas the contem- restriction enzyme Hha, a synapomorphy that clusters porary sample has two haplotypes ( haplotypes A and B) all subspecies into a monophyletic group (Wayne et al. ( Mundy et al. 1997b). The authors concluded that these 1991). The San Clemente Island fox has a mtDNA haplo- differences are likely a result of introgression, or gene type shared only with San Miguel Island foxes, suggesting flow, that occurred between L. l. mearnsi and another that the San Clemente population was derived from them

Conservation Biology Volume 17, No. 5, October 2003

1254 Conservation in Conflict Roemer & Wayne

Table 1. The total number and number of unique mtDNA 1981; Moore & Collins 1995). Finally, island foxes from haplotypes and microsatellite alleles found in gray foxes (Urocyon different subspecific populations would be unable to hy- cinereoargenteus) and island foxes (U. littoralis).* bridize unless transfer among islands was facilitated by hu- mtDNA Microsatellite mans (Collins 1991a, 1991b). haplotypes loci no. no. total unique total unique Conservation of the San Clemente Loggerhead Gray fox 7 7 120 53 Island fox 5 5 86 19 Shrike and Decline of the San Clemente Island Fox Island fox subspecies San Clemente Island fox, The San Clemente Loggerhead Shrike is one of the most U. l. clementae 1 0 39 8 endangered birds in the world. L. l. mearnsi was re- Santa Catalina Island fox, U. l. catalinae 3 1 49 2 duced to an estimated 17 individuals in the wild in 1988 San Nicolas Island fox, before a recovery program was initiated in 1991 ( U.S. U. l. dickeyi 1 1 19 2 Department of the Interior 1977; Juola et al. 1997). The Santa Cruz Island fox, program is centered on the island, funded by the U.S. U. l. santacruzae 2 0 44 2 Navy, and supported by both public and private organi- Santa Rosa Island fox, U. l. santarosae 2 0 47 7 zations. San Miguel Island fox, The shrike recovery program has been successful at U. l. littoralis 2 0 31 2 raising and releasing shrikes and establishing breeding *Mitochondrial DNA data are from Wayne et al. (1991), and micro- pairs, but the population has remained extremely small. satellite data are from G.W.R. and R.K.W. (unpublished data). In July 2000, the wild population was estimated to be 25 individuals and has never exceeded 35 individuals in the history of the program ( Mader et al. 2000; Blackford et ( Wayne et al. 1991). The San Nicolas Island fox has a al. 2001). Breeding failures attributed to nest predation unique mtDNA haplotype not found in any other island have plagued the recovery program and led to a plan for fox subspecies (Table 1). mitigating threats to shrike nestlings and fledglings ( U.S. The genetic difference between the island fox and gray Department of Agriculture 1998). Although diet studies fox revealed by the mtDNA analyses is mirrored by analy- suggest that birds typically form a very small portion ses of nuclear DNA (Gilbert et al. 1990; Wayne et al. 1991; ( 3–6%) of the diet of the island fox (Laughrin 1977; Crooks Goldstein et al. 1999). Of 19 microsatellite loci that are & Van Vuren 1995), San Clemente Island foxes have been polymorphic across the two species, 120 alleles are found observed preying on nestling shrikes (Cooper et al. 2001). in the gray fox, and 53 (44%) of these are unique to gray Mitigation included live-capture and removal or euthana- foxes (Table 1). Similarly, of 86 alleles in island foxes, 19 sia of foxes and euthanasia of native species of raptors (22%) are unique to them ( Table 1). Nei’s (1978) aver- and non-native nest predators. This plan was endorsed age unbiased genetic distance between the gray fox and by the USFWS and initiated by the U.S. Department of Ag- island fox is 0.76. Phylogenetic analyses of these data con- riculture’s division responsible for damage con- sistently cluster all island fox subspecies together (only 2 trol. As a result, in 1999 at least 32 foxes were euthanized of 183 fox genotypes were misassigned), with the gray or permanently removed from the island to zoological in- fox as the ancestral population, supporting the finding stitutions. Other native species were also killed, includ- from the mtDNA analysis that island foxes are mono- ing 105 Common Ravens (Corvus corax), 48 American phyletic. Importantly, the San Clemente Island fox con- Kestrels (Falco sparverius), 27 Red-tailed Hawks (Buteo tains both unique minisatellite and microsatellite alleles jamaicensis), and nine Barn Owls (Tyto alba) (Elliot & not found in other island fox populations, and the San Popper 1999). After an article appeared in the Los Ange- Nicolas Island fox also contains unique microsatellite al- les Times describing the control program (Schoch 1999), leles despite its low genetic variability (Gilbert et al. 1990; the U.S. Navy stopped killing foxes. Goldstein et al. 1999). The San Clemente Island fox has During the late 1990s, the U.S. Navy experimented with eight unique microsatellite alleles, and the San Nicolas Is- other means of predator control. They used a commer- land fox has two (Table 1). From the standpoint of ge- cially available shock collar system to exclude foxes from a netic diversity, the San Clemente Island fox and the San limited area surrounding shrike nest sites (Cooper et al. Nicolas Island fox are well-defined and important sub- 2001). A wire broadcasts a signal to a battery-powered species contributing to the overall genetic diversity of collar that then delivers a shock to a collared fox. The the species. Further, the morphological and genetic evi- system worked well at nest sites because foxes were dence support both the current classification of the is- continually shocked until they left the vicinity of the land fox as a separate species ( Wilson & Reeder 1993) and nest. To exclude foxes from a larger area, where shrike the subspecific status of each island fox population ( Hall fledglings might land, the USFWS required that release

Conservation Biology Volume 17, No. 5, October 2003 Roemer & Wayne Conservation in Conflict 1255 sites and canyons containing shrike nests also be wired. Foxes readily transgressed these wired areas because they would no longer experience the shock deterrent after moving a short distance away from the perimeter wire. Further, a “three strikes” rule was established: foxes that transgressed a wire-protected area three times would be held in captivity or euthanized. In 1999, 49 foxes were held temporarily in small pens (0.55 m2 ) in addition to the 32 foxes that were permanently removed or eutha- nized. Of the foxes held captive, 20 females were sus- pected of having dependent pups in the wild (Cooper et al. 2001). Adult female San Clemente Island foxes wean an average of 1.25 (SE 0.015) pups per reproductive event ( Roemer 1999). Thus, an estimated 25 pups are Figure 1. Estimated trend in island fox population expected to have starved to death as a result of confine- size on San Clemente Island. Average population size ment of their mothers. In 2000 no foxes were euthanized, (upper and lower 95% confidence interval, UCI and but 71 foxes, about 12% of the estimated population, LCI, respectively) between 1988 and 1997 was esti- were held in captivity for an average of 35 days (range mated based on densities determined from a capture- 6–114 days) during the fox reproductive season (Cooper recapture approach (Roemer et al. 1994, 2001b). et al. 2001). Estimates from 1998 to 2002 (Est N) are based on a Shock-wired canyons where shrikes nest are prime fox deterministic population model with vital rates de- habitat, and the consistent removal of foxes from these rived from the previous 10 years of data (Roemer et areas created population sinks into which foxes would al. 2000). In 1992, predator management to protect persistently disperse (Cooper et al. 2001). Resident and the shrike was initiated. dispersing foxes were essentially removed from the breed- ing pool, either by being placed in captivity or by repeated capture and confinement in traps. In 2000 alone 13,347 program has significantly affected the San Clemente Is- box traps were set to capture foxes and non-native preda- land fox and contributed to the endangerment of the is- tors in an area encompassing approximately two-thirds of land fox. the island, or about 100 km2 (Cooper et al. 2001). These traps were set over 279 nights (76% of the year) and re- sulted in captures of 180 foxes 1460 times; one fox was Status and Conservation of the Island Fox captured 66 times (Cooper et al. 2001). Capture success (defined as the number of foxes captured divided by the Recently, the USFWS was petitioned by two nonprofit or- number of traps set) of foxes on San Clemente Island in ganizations and subsequently proposed to list four sub- 2000 (10.9%) was considerably lower than capture suc- species of the island fox as endangered ( U.S. Department cess in 1998 (36.3%) (Cooper et al. 2001; Roemer et al. of the Interior 2001). These four subspecies, U. l. littora- 2001b). lis, U. l. santarosae, U. l. santacruzae, and U. l. catali- To assess the potential impact of the removal and con- nae, have declined drastically owing to hyperpredation finement strategies on the fox population, we constructed by Golden Eagles (Aquila chrysaetos; Roemer et al. 2001b, a Lefkovitch matrix based on 10 years of demographic 2002) and infection with canine distemper virus (Timm et data (1988–1997) collected at three sampling sites on San al. 2000). U. l. clementae and U. l. dickeyi were not in- Clemente Island (Roemer et al. 2000, 2003). This matrix yielded a population growth rate () of 0.9566, or ap- proximately a 4.4% decline in annual population size. Table 2. Mean density (SE), estimated population size, and mean We used this rate of decline to estimate the trend in pop- number of island fox pups captured on three demographic grids on San Clemente Island, California, between 1988 and 2000. ulation size from 1998 to 2002. We estimated that the San Clemente Island fox population will be at approxi- Density Population mately 60% of its pre-1992 population size by 2002 ( Fig. Perioda ( foxes/km2) sizeb No. of pups 1). A comparison of fox demographic data collected on 1988–1992 6.4 (0.41) 929 6.7 (0.68) San Clemente Island suggests the decline may be even 1993–1997 5.3 (0.28) 771 4.3 (1.24) more significant. Average density, population size, and 1999–2000 2.8 (0.71) 410 2.8 (1.01) the mean number of pups captured have declined by aData from 1988 to 1997 are condensed from Roemer et al. (2000, nearly 60% between 1988 and 2000 (Table 2). We sug- 2001b ), and data from 1999 and 2000 are from Schmidt et al. (2002). Trapping was not conducted in 1998. gest that the single-species philosophy that has charac- bPopulation size was simply estimated as the mean density multi- terized the San Clemente Loggerhead Shrike recovery plied by island area (145 km2).

Conservation Biology Volume 17, No. 5, October 2003 1256 Conservation in Conflict Roemer & Wayne cluded in the proposed listing, despite a substantial popula- tributes to the overall genetic diversity of the species. Fur- tion decline in U. l. clementae and evidence of morpho- ther, Section 3(6) of the ESA states that “The term ‘endan- logic and genetic distinction in both subspecies (Gilbert gered species’ means any species which is in danger of et al. 1990; Wayne et al 1991; Goldstein et al. 1999; Roe- extinction throughout all or a significant portion of its mer et al. 2000, 2001b; U.S. Department of the Interior range . . . ” (U. S. Code. 1999. 16 U.S.C. Section 1533). Five 2001). of six island fox subspecies are now in danger of extinc- An independent assessment of the status of the island tion, representing a combined land area equaling 93.5% of fox argues that the entire species, not just select subspe- the species’ range. Clearly, the entire species should re- cies, be listed as endangered ( Roemer et al. 2003). Us- ceive federal protection. ing the World Conservation Union (IUCN) Red List cate- gories ( World Conservation Union 2001), Roemer et al (2003) recommended that the subspecies on San Miguel and Santa Rosa islands, U. l. littoralis and U. l. santaro- Single-Species Approach versus an Ecosystem sae, be listed as extinct in the wild. Extensive surveys Approach to Conservation have found no evidence of wild foxes present on either San Miguel or Santa Rosa islands, and there are, respec- Rational arbitration of the conservation conflict between tively, only 28 and 45 foxes currently in captivity (Coonan the fox and the shrike should not only rest upon a 2001, 2002). The subspecies on Santa Cruz Island, U. l. thoughtful evaluation of the evolutionary history and adap- santacruzae, and the subspecies on Santa Catalina Island, tive diversity of both taxa ( Moritz 1994; Crandall et al. U. l. catalinae, are recommended to be listed as critically 2000; Fraser & Bernatchez 2001) but should also consider endangered. Each has been reduced in number by approx- the ecological significance of both species (Carroll et al. imately 90% in the past 10 years, and each subspecies has 1996). For example, the loss of fox populations on the a population size of 250 adults, with all individuals in a northern Channel Islands has had immediate community- single subpopulation ( IUCN criteria A.1 & C.1) ( Roemer level impacts. The decline in foxes has resulted in an 1999; Timm et al. 2000; Roemer et al. 2001b, 2002). The increase in the island spotted skunk (Spilogale gracilis estimated probability of extinction in the wild for U. l. amphiala) on Santa Cruz Island, owing to competitive santacruzae is 50% within 10 years (IUCN criterion E) release (Roemer et al. 2002), and an increase in deer ( Roemer et al. 2001b ). The San Clemente Island fox, mouse (Peromyscus maniculatus) densities on San Miguel U. l. clementae, is recommended to be listed as endangered Island, owing to relaxed predation (Coonan et al. 2000). because it has a population estimated at 2500 mature The shrike’s effect on the community is unknown. individuals, has been reduced by 50% over the past 10 The current conservation conflict might have been years, and is declining, and all individuals are in a single avoided if the ecological importance of a variety of spe- subpopulation that occupies 500 km2 (IUCN criteria cies had been considered and if an ecosystem approach A.4, B.2b,c & C.2b) (Roemer 1999; Roemer et al. 2001b; to conservation had been adopted. A primary problem Schmidt et al. 2002). The San Nicolas Island subspecies, in shrike conservation is the absence of appropriate hab- U. l. dickeyi, is recommended to be listed as vulnerable itats. Habitat degradation that reduces foraging perches because the population is estimated to number 1000 and increases the density and cover of herbaceous vege- individuals ( IUCN criterion D) ( Roemer et al. 1994; Roe- tation, thereby decreasing the availability of prey, has been mer 2001). Furthermore, U. l. dickeyi is one of the most identified as a key cause of declining shrike populations genetically uniform wild populations ever described (Gil- worldwide (Temple 1995). Habitats on San Clemente Is- bert et al. 1990; Wayne et al. 1991; Goldstein et al. 1999), land were degraded through historic ranching activities and past demographic studies suggest that it went and overgrazing by feral herbivores (Scott & Morrison through a population crash in the 1970s, recovering in 1990), and, although the U.S. Navy made considerable the early to mid-1980s ( Laughrin 1980; Kovach & Dow progress toward arresting habitat loss by eradicating feral 1981, 1985). herbivores, habitat restoration following the eradication ef- In sum, the biological determination made in the pro- forts has been limited. Furthermore, captive-reared shrikes posed rule is inadequate because it excludes both U. l. were also released without the necessary refugia from clementae and U. l. dickeyi from potential federal pro- predators (appropriate nesting substrate and escape cover) tection. These two subspecies are small populations that that existed previously when foxes and shrikes actually are either declining or have been through an historic coexisted. Habitat enhancement is needed in the form of population decline and thus are vulnerable to future fluc- shrub establishment to increase the number of perches tuations in population size. Both subspecies may be sig- and improve escape cover and nesting substrate for nificantly affected by the introduction of a lethal canine shrikes. disease, such as the canine distemper epizootic that deci- The proliferation of exotic annual grasses following fe- mated the Santa Catalina Island fox (Timm et al. 2000). ral herbivore removal (Klinger et al. 1994; Laughrin et Both subspecies have unique genetic variation that con- al. 1994) may also have exacerbated the problem by re-

Conservation Biology Volume 17, No. 5, October 2003 Roemer & Wayne Conservation in Conflict 1257 ducing prey availability for both foxes and shrikes. For example, foxes on Santa Cruz Island utilize native grass- lands (Nassella spp.) and canyons and creekbeds more than expected (2 47.0, p 0.001) and avoid exotic annual grasslands (Avena, Lolium, and Bromus) (Fig. 2). Exotic annual grassland harbors abundant insect prey but is significantly taller and denser and has significantly greater litter duff (analysis of variance, p 0.001 in all cases) than other habitats (Fig. 3). We hypothesize that the greater cover afforded by exotic annual grasses re- duces the availability of insect prey to foraging foxes, in a sense creating insect refugia. Also, exotic annual grasses may hinder shrikes from obtaining insect prey (Lynn et al. 2000). Thus, in addition to shrub establishment, habitat enhancement that controls the spread of exotic annual grassland and maintains native perennial grassland is needed. Such efforts may actually allow for species co- existence. Finally, an ecosystem approach is often money better spent. On the northern Channel Islands, the U.S. Na- tional Park Service and other cooperators have adopted an ecosystem approach to conserving three endangered subspecies of the island fox (Roemer 1999; Coonan 2001; Roemer et al. 2001b, 2002; U.S. Department of the Interior 2002). This program’s actions include the live- trapping and translocation of Golden Eagles, the main source of fox mortality; the reintroduction of the extir- pated Bald Eagle (Haliaeetus leucocephalus), to restore this component of the fauna and as a potential hindrance to future colonization of the islands by Golden Eagles; the eradication of feral pigs (Sus scrofa), to prevent fur- ther loss of native flora and to remove the food supply Figure 3. (a) Abundance of invertebrate prey of foxes Golden Eagles require; the removal of exotic plants, such and shrikes ( Jerusalem crickets [JC, Stenopelmatus as fennel (Foeniculum vulgare), to increase the distribu- fuscus] and ground-dwelling beetles [TB, Coleoptera: tion of native plants; and the captive propagation, re- Carabidae and Tenebrionidae]) summed for 12 pit- fall traps located in each of two habitats, exotic an- nual ( EAG) and native perennial ( NPG) grassland, on Santa Cruz Island (G. Roemer, unpublished data). ( b) Physical structure of the four most prevalent habi- tats within a coastal grassland community on the west end of Santa Cruz Island, California. Metrics include mean height of vegetation, mean percent cover, and mean litter depth. Data were derived from 325 1-m2 vegetation plots randomly placed across the study area. Within each category, like numbers indicate habitats that are not significantly different from one another.

lease, and monitoring of island foxes, including disease and parasite surveillance. This comprehensive program will cost several million dollars and will span several years; it seeks to enhance the entire ecosystem while restoring Figure 2. Patterns of habitat availability and use in is- the fox. land foxes (n 14) inhabiting a coastal grassland com- In contrast, the shrike program has hinged on preda- munity on the west end of Santa Cruz Island, California. tor control and captive propagation and release at a sub-

Conservation Biology Volume 17, No. 5, October 2003 1258 Conservation in Conflict Roemer & Wayne stantial cost, $2.3 million spent in 1999 alone ( Morrison lution to conserving our natural heritage that might be et al. 1995; Schoch 1999; Cooper et al. 2001). Although improved by a proactive ecosystem approach (Carroll et predator control is a necessary tool with which to con- al. 1996). We have a better chance of conserving imper- trol or eradicate harmful non-native predators such as fe- iled species and the habitats on which they depend if en- ral cats (Felis catus), its negative effects on native taxa dangered-species recovery programs weigh the value of cannot be ignored. Allocating funds toward habitat res- all populations to the ecosystem, consider the evolution- toration that would benefit the island ecosystem may ary legacy and adaptive uniqueness of vulnerable taxa, be one approach that could help alleviate the current and take into account the likelihood of a program’s suc- endangered-species conflict. cess relative to its financial cost.

Conclusions and Recommendations Acknowledgments

The U.S. ESA and its implementation have been criticized We thank the U.S. Navy for supplying some of the re- on many grounds (Clark et al. 1994), some of which are ports requested, and we thank the following individuals exemplified by the current conflict between the shrike for thoughtful comments on the manuscript: D. Cowley, and the fox. Biased implementation of the ESA has re- C. J. Donlan, B. LaHaye, E. Main, C. Moritz, T. Smith, B. Van sulted in a species-centric approach to protecting the San Valkenburgh, M. Willson, and two anonymous reviewers. Clemente Loggerhead Shrike, a listed subspecies of equiv- This work was supported by grants from the National Geo- ocal evolutionary distinction and ecological importance. graphic Society, the National Park Service, the University The subsequent recovery program has hinged upon ag- of California, Los Angeles, and the New Mexico Agricul- gressive control of native and non-native predators, which tural Experiment Station. is contributing to the decline of the San Clemente Island fox, a distinct evolutionarily significant unit that plays an important ecological role. The USFWS ignored the decline Literature Cited in the San Clemente Island fox and the importance of this subspecies and the San Nicolas Island fox to the Blackford, A. V., C. Couroux, H. Carlisle, J. Plissner, and N. Warnock. overall genetic diversity of the species, whereas the four 2001. Breeding report: 2000 population monitoring of the San Clemente Loggerhead Shrike on NALF, San Clemente Island, Cali- remaining island fox subspecies were being considered fornia. U.S. Navy, Natural Resources Management Branch, South- for federal protection. This lack of consideration may in- west Division, Naval Facilities Engineering Command, San Diego. crease the risk of extinction for these subspecies and the California Department of Fish and Game (CDFG). 1987. Five-year status species as a whole, especially if current practices to save report on the island fox (Urocyon littoralis). CDFG, Sacramento. the San Clemente Loggerhead Shrike continue and the Carroll, R., C. Augspurger, A. Dobson, J. Franklin, G. Orians, W. Reid, R. Tracy, D. Wilcove, and J. Wilson. 1996. Strengthening the use of U.S. Navy is not subject to future Section 7 consultations. science in achieving the goals of the Endangered Species Act: an as- Efforts to preserve the San Clemente Loggerhead Shrike sessment by the Ecological Society of America. Ecological Applica- need reconsideration. As has been suggested for an en- tions 6:1–11. dangered subspecies of the California Gnatcatcher (Poliop- Clark, T. W., R. P. Reading, and A. L. Clarke. 1994. Endangered species tilla californica californica; Cronin 1997), we recom- recovery, finding the lessons improving the process. Island Press, Washington, D.C. mend that both the San Clemente Loggerhead Shrike and Collins, P. W. 1991a. Interaction between island foxes (Urocyon the other Channel Island form, L. l. anthonyi, be treated as littoralis) and Indians on islands off the coast of Southern Califor- management units by instituting a cooperative agreement nia. I. Morphologic and archaeological evidence of human assisted among the various resource agencies and stakeholders. dispersal. Journal of Ethnobiology 11(1):51–81. We recommend that all actions that negatively affect the Collins, P. W. 1991b. Interaction between island foxes (Urocyon littoralis) and Native Americans on islands off the coast of South- San Clemente Island fox be halted and that some of the ern California. II. Ethnographic, archaeological and historical evi- funds used for predator control be allocated toward hab- dence. Journal of Ethnobiology 11(2):205–229. itat restoration. Finally, we recommend that all subspe- Collins, P. W. 1993. Taxonomic and biogeographic relationships of the cies of the island fox receive federal protection and that island fox (Urocyon littoralis) and gray fox (U. cinereoargenteus) any action that may affect an island fox subspecies be from western . Pages 351–390 in F. G. Hochberg, editor. Third California islands symposium: recent advances in re- brought to the attention of all concerned parties, espe- search on the California Islands. Santa Barbara Museum of Natural cially to independent organizations that have no finan- History, Santa Barbara, California. cial or political interests on the islands ( Reading & Miller Collister, D. M., and D. Wicklum. 1996. Intraspecific variation in Log- 1994). gerhead Shrikes: and implication for subspecies Given that the number of listed species or species that classification. The Auk 113:221–223. Coonan, T. J. 2001. Recovery plan for island foxes (Urocyon littoralis) are candidates for listing exceeds 1500 ( Hoekstra et al. on the northern Channel Islands. Park headquarters, Channel Is- 2002), endangered-species conflicts will become com- lands National Park, Ventura, California. monplace. Single-species management is a traditional so- Coonan, T. J. 2002. Findings of the island fox conservation working

Conservation Biology Volume 17, No. 5, October 2003 Roemer & Wayne Conservation in Conflict 1259

group, Ventura, California. Park headquarters, Channel Islands Na- update on the status of resources. Santa Barbara Museum of Natural tional Park, Ventura, California. History, Santa Barbara, California. Coonan, T. J., C. A. Schwemm, G. W. Roemer, and G. Austin. 2000. Lynn, S., J. A. Martin, D. M. Cooper, K. M. Wakelee, G. A. Schmidt, and Population decline of island foxes (Urocyon littoralis) on San D. K. Garcelon. 2000. Research efforts to aid in the recovery of the Miguel Island. Pages 289–297 in D. R. Browne, K. L. Mitchell, and San Clemente Loggerhead Shrike—1999. U. S. Navy, Natural Re- H. W. Chaney, editors. Proceedings of the fifth channel islands sources Management Branch, Southwest Division, Naval Facilities symposium. U.S. Department of the Interior, Minerals Management Engineering Command, San Diego. Service, Pacific Outer Continental Shelf Region, Camarillo, California. Mader, T., N. Warnock, and W. Ostheimer. 2000. Final report: 1999 popu- Cooper, D. M., E. L. Kershmer, G. A. Schmidt, and D. K. Garcelon. lation monitoring of the Loggerhead Shrike on NALF, San Clemente Is- 2001. San Clemente Loggerhead Shrike predator research and man- land, California. U. S. Navy, Natural Resources Mangement Branch, agement program—2000. Institute for Wildlife Studies, Arcata, Cali- Southwest Division Naval, Facilities Engineering Command, San Diego. fornia. Final report. U. S. Navy, Natural Resources Management Miller, A. H. 1931. Systematic revision and natural history of the Amer- Branch, Southwest Division, Naval Facilities Engineering Command, ican shrikes (Lanius). University of California Publications in Zool- San Diego, California. ogy 38:11–242. Crandall, K. A., O. R. P. Bininda-Emonds, G. Mace, and R. K. Wayne. Moore, C. M., and P. W. Collins. 1995. Urocyon littoralis. No. 489. Mam- 2000. Considering evolutionary processes in conservation biology. malian species. The American Society of Mammalogists, Brigham Trends in Ecology & Evolution 15:290–295. Young University, Provo, Utah. Cronin, M. A. 1997. Systematics, , and the Endangered Spe- Moritz, C. 1994. Defining ‘evolutionarily significant units’ for conserva- cies Act: the example of the California gnatcatcher. Wildlife Soci- tion. Trends in Ecology & Evolution 9:373–375. ety Bulletin 25:661–666. Morrison, M. L., C. M. Kuehler, T. A. Scott, A. A. Lieberman, W. T. Crooks, K. R., and D. Van Vuren. 1995. Resource utilization by two in- Everett, R. B. Phillips, C. E. Koehler, P. A. Aigner, C. Winchell, and sular endemic mammalian carnivores, the island fox and island T. Burr. 1995. San Clemente Loggerhead Shrike: recovery plan for spotted skunk. Oecologia 104:301–307. an endangered species. Proceedings of the Western Foundation of Elliot, T., and B. Popper. 1999. Predator management report for the Vertebrate Zoology 6:293–295. protection of the San Clemente Loggerhead Shrike. U. S. Navy, Nat- Mundy, N. I., C. S. Winchell, and D. S. Woodruff. 1997a. Genetic dif- ural Resources Management Branch, Southwest Division Naval Fa- ferences between the endangered San Clemente Loggerhead Shrike cilities Engineering Command, San Diego. Lanius ludovicianus mearnsi and two neighboring subspecies Fraser, D. J., and L. Bernatchez. 2001. Adaptive evolutionary conserva- demonstrated by mtDNA control region and cytochrome b se- tion: towards a unified concept for defining conservation units. quence variation. Molecular Ecology 6:29–37. Molecular Ecology 10:2741–2752. Mundy, N. I., C. S. Winchell, T. Burr, and D. S. Woodruff. 1997b. Gilbert, D. A., N. Lehman, S. J. O’Brien, and R. K. Wayne. 1990. Ge- Microsatellite variation and microevolution in the critically endan- netic fingerprinting reflects population differentiation in the Cali- gered San Clemente Island Loggerhead Shrike (Lanius ludovi- fornia channel island fox. Nature 344:764–767. cianus mearnsi). Proceedings of the Royal Society London Series B Goldstein, D. B., G. W. Roemer, D. A. Smith, D. E. Reich, A. Bergman, and 264:869–875. R. K. Wayne. 1999. The use of microsatellite variation to infer patterns National Research Council. 1995. Science and the Endangered Species of migration, population structure and demographic history: an evalu- Act. National Academy Press, Washington, D.C. ation of methods in a natural model system. Genetics 151:797–801. Nei, M. 1978. Estimation of average heterozygosity and genetic dis- Hall, E. R. 1981. The mammals of North America. 2nd edition. John Wiley tance from a small number of individuals. Genetics 89:583–590. & Sons, New York. Patten, M. L., and K. F. Campbell. 2000. Typological thinking and the Hoekstra, J. M., J. A. Clark, W. F. Fagan, and P. D. Boersma. 2002. A conservation of subspecies: the case of the San Clemente Island comprehensive view of Endangered Species Act recovery plans. Loggerhead Shrike. Diversity and Distributions 6:177–188. Ecological Applications 12:630–640. Reading, R. P., and B. J. Miller. 1994. The black-footed ferret recovery Juola, F. A., W. T. Everett, and C. E. Koehler. 1997. Final report: 1996 program: unmasking professional and organizational weaknesses. population and habitat survey of the Loggerhead Shrike on NALF Pages 73–100 in T. W. Clark, R. P. Reading, and A. L. Clarke, edi- San Clemente Island, California. U.S. Navy, Southwest Division, Na- tors. Endangered species recovery: finding the lessons, improving val Facilities Engineering Command, San Diego. the process. Island Press, Washington, D.C. Klinger, R. C., P. T. Schuyler, and J. D. Sterner. 1994. Vegetation re- Ridgway, R. 1903. Descriptions of new genera of species and subspe- sponse to the removal of feral sheep from Santa Cruz Island. Pages cies of American birds. Proceedings of the Biological Society of 341–350 in W. L. Halvorsen and G. J. Maender, editors. The fourth Washington 16:105–112. California islands symposium: update on the status of resources. Roemer, G. W. 1999. The ecology and conservation of the island fox. Santa Barbara Museum of Natural History, Santa Barbara, California. Ph.D. dissertation. University of California, Los Angeles. Kovach, S. D., and R. J. Dow. 1981. Status and ecology of the island Roemer, G. W. 2001. Annual progress report 2000: summary of the de- fox on San Nicolas Island, 1980. Technical memorandum TM–81– mography of the San Nicolas Island fox (Urocyon littoralis dickeyi). 28. Pacific Missile Defense Center, Point Mugu, California. Final report. U. S. Navy, Point Mugu Naval Air Station, Camarillo, Cal- Kovach, S. D., and R. J. Dow. 1985. Island fox research on San Nicolas ifornia. Island. Annual report. Department of the Navy, Pacific Missile Test Roemer, G. W., D. K. Garcelon, T. J. Coonan, and C. Schwemm. 1994. Center, San Diego. The use of capture-recapture methods for estimating, monitoring, Laughrin, L. 1977. The island fox: a field study of its behaviour and and conserving island fox populations. Pages 387–400 in W. L. Hal- ecology. Ph.D. dissertation. University of California, Santa Barbara. vorsen, and G. J. Maender, editors. The fourth California islands Laughrin, L. 1980. Populations and status of the island fox. Pages 745– symposium: update on the status of resources. Santa Barbara Mu- 749 in D. M. Power, editor. The California islands: proceedings of a seum of Natural History, Santa Barbara, California. multidisciplinary symposium. Santa Barbara Museum of Natural Roemer, G. W., P. S. Miller, J. Laake, C. Wilcox, and T. J. Coonan. History, Santa Barbara, California. 2000. Island fox demographic workshop report. Final report. Laughrin, L., M. Carroll, A. Bromfield, and J. Carroll. 1994. Trends in Channel Islands National Park, Ventura, California. vegetation changes with removal of feral animal grazing pressures Roemer, G. W., D. A. Smith, D. K. Garcelon, and R. K. Wayne. 2001a. on Santa Catalina Island. Pages 523–530 in W. L. Halvorsen and The behavioural ecology of the island fox. Journal of Zoology 255: G. J. Maender, editors. The fourth California islands symposium: 1–14.

Conservation Biology Volume 17, No. 5, October 2003 1260 Conservation in Conflict Roemer & Wayne

Roemer, G. W., T. J. Coonan, D. K. Garcelon, J. Bascompte, and L. 2000. Investigation into the decline of island foxes on Santa Cat- Laughrin. 2001b. Feral pigs facilitate hyperpredation by Golden Ea- alina Island. Institute for Wildlife Studies, Arcata, California. gles and indirectly cause the decline of the island fox. Animal Con- U.S. Department of Agriculture Animal and Plant Inspection Service–Wild- servation 4:307–318. life Services. 1998. Final environmental assessment: predator damage Roemer, G. W., C. J. Donlan, and F. Courchamp. 2002. Golden eagles, management to protect the federally endangered San Clemente Log- feral pigs and insular carnivores: how exotic species turn native gerhead Shrike on San Clemente Island. U.S. Navy, Southwest Divi- predators into prey. Proceedings of the National Academy of Sci- sion, Naval Facilities Engineering Command, San Diego, California. ences of the United States of America 99:791–796. U.S. Department of the Interior. 1977. Determination that seven Chan- Roemer, G. W., T. J. Coonan, L. Munson, and R. K. Wayne. 2003. Canid nel Island and plants are either endangered species or action plan for the island fox. In press in C. Sillero-Zubiri, J. R. threatened species. Federal Register 42(155):40682–40685. Ginsberg, and D. W. Macdonald, editors. Foxes, wolves, jackals and U.S. Department of the Interior. 2001. Endangered and threatened dogs: status survey and conservation action plan. 2nd edition. wildlife and plants: listing the San Miguel Island fox, Santa Rosa Is- World Conservation Union, Gland, Switzerland. land fox, Santa Cruz Island fox and Santa Catalina Island fox as en- Schmidt, G. A., D. K. Garcelon, and J. Sloan. 2002. Fox monitoring and dangered (50 CFR Part 17). Federal Register 66(237):63654– research in support of the San Clemente Loggerhead Shrike preda- 63665. tor control program on Naval Auxiliary Landing Field, San Clem- U.S. Department of the Interior. 2002. Santa Cruz Island primary resto- ente Island, California. Report of the Institute for Wildlife Studies, ration plan. Final environmental impact statement. Park Headquar- Arcata, California. U.S. Navy, Natural Resources Mangement ters, Channel Islands National Park, Ventura, California. Branch, Southwest Division, Naval Facilities Engineering Com- U.S. Department of the Interior and Department of Commerce. 1996. mand, San Diego. Proposed policy on the treatment of intercrosses and intercross Schoch, D. 1999. Hard lessons in the tale of the fox and the shrike spe- progeny (the issue of “hybridization”); Proposed rule (50 CFR Part cies on San Clemente Island, the Navy has been killing a rare preda- 424). Federal Register 61(26):4709–4713. tor in a bid to save its nearly extinct prey. Los Angeles Times 17 Wayne, R. K., S. B. George, D. Gilbert, P. W. Collins, S. D. Kovach, D. May:1. Girman, and N. Lehman. 1991. A morphologic and genetic study of Scott, T. A., and M. L. Morrison. 1990. Natural history and manage- the island fox, Urocyon littoralis. Evolution 45:1849–1868. ment of the San Clemente Loggerhead Shrike. Proceedings of the Wilson, D. E., and D. M. Reeder. 1993. Mammal species of the world: a Western Foundation of Vertebrate Zoology 4:23–60. taxonomic and geographic reference. 2nd edition. Smithsonian In- Temple, S. A. 1995. Priorities for shrike research and conservation. stitution Press, Washington, D.C. Proceedings of the Western Foundation of Vertebrate Zoology 6: World Conservation Union (IUCN). 2001. IUCN red list categories and 296–298. criteria. Version 3.1. Species Survival Commission, IUCN, Gland, Swit- Timm, S. F., J. M. Stokely, T. B. Gehr, R. L. Peebles, and D. K. Garcelon. zerland.

Conservation Biology Volume 17, No. 5, October 2003