Biological Conservation 116 (2004) 35–47 www.elsevier.com/locate/biocon The conservation status of Townsend’s shearwater Puffinus auricularis auricularis Juan E. Martı́nez-Gómeza,*, Jeff K. Jacobsenb a Department of Biology, Villanova University, Villanova, PA 19085, USA Department of Biological Sciences, Humboldt State University, Arcata, CA 95521, USA b Received 8 March 2002; received in revised form 18 March 2003; accepted 20 March 2003 Abstract Townsend’s shearwater (Puffinus auricularis auricularis) is an endangered seabird endemic to the Revillagigedo Archipelago. It nested on Socorro, Clarion, and San Benedicto Islands. It was extirpated by the Barcena volcano on San Benedicto in 1952, and there are no recent indications of nesting. Introduced mammals—pigs and rabbits—preyed on them and destroyed habitat at Clarion; shearwaters were extirpated by 1988, and no breeding attempts have been reported since. Our results confirm that Socorro holds the last breeding grounds. We found breeding colonies above 800 m and a minimum population of 1100 individuals. This represents a significant reduction in distribution and population size. Intensive cat predation at Socorro could potentially kill ca. 350 females per season, and sheep progressively destroy nesting areas. Population projections suggest that demographic instability could occur in less than 100 years under severe predation and habitat degradation. Only low predation rates would allow population persistence for more than 150 years in spite of a declining population. Thus, the immediate eradication of all introduced mammals is necessary to prevent the extinction of this seabird. # 2003 Elsevier Ltd. All rights reserved. Keywords: Townsend’s shearwater; Puffinus auricularis; Revillagigedo Archipelago; Socorro Island; Matrix models 1. Introduction Townsend’s shearwater (Puffinus auricularis auricularis; AOU, 1998) is an endangered seabird endemic to the Revillagigedo Archipelago (Ainley et al., 1997). Three of the four islands in that region were suitable for nesting by Townsend’s shearwater: Socorro, Clarion, and San Benedicto. These islands are among the least impacted in the Pacific, because limited human colonization began only recently and damage caused by introduced animals can be reversed almost entirely. The archipelago is noteworthy for its high levels of endemism in flora and fauna (Levin and Moran, 1989; Brattstrom, 1990). Townsend’s shearwater was first discovered at Clarion in 1889 (Townsend, 1890). Subsequently, breeding grounds were detected there and at San Benedicto in 1897 (Anthony, 1898), and at Socorro in 1925 (McLellan, * Corresponding author. Present address: Department of Biology, University of Missouri—St. Louis, St. Louis, MO 63121, USA. E-mail address: mimodes@jinx.umsl.edu (J.E. Martı́nez-Gómez). 0006-3207/03/$ - see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0006-3207(03)00171-X 1926). The colonies at San Benedicto were destroyed by the eruption of the Barcena volcano in 1952 (Brattstrom and Howell, 1956), and introduced mammals extirpated those at Clarion by the late 1980s (Everett, 1988; Howell and Webb, 1989). Thus, it would seem that the last remaining breeding grounds for this subspecies are on Socorro. Although Townsend’s shearwater is considered at risk, lack of information has made it impossible to determine its precise status (Collar et al., 1992). Some authors have considered this shearwater to be the most endangered seabird of the offshore islands of the Mexican Pacific (Everett and Anderson, 1991). Mounting evidence suggests that it is critically endangered (BirdLife, 2000). Population size has not been assessed adequately, but there are reports ranging from several hundred to several thousand birds. Jehl (1982) estimated a Socorro population of at least 1000 breeding pairs by counting birds returning to land. Spear et al. (1995) used at-sea density data from 15 years (early 1980s–mid 1990s) of surveys to estimate total pelagic populations of 46,400 36 J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 individuals, of which 10,600 were breeding pairs. During the breeding season, individuals in reproductive condition are at Socorro, but most non-breeding individuals are found in the Middle American Trench offshore Mexico’s Pacific coast from Jalisco to Guerrero (Howell and Engel, 1993). Habitat destruction and predation caused by introduced mammals are the most serious threats to Townsend’s shearwater. On Clarion, breeding colonies were destroyed by the late 1980s. In January 1986, only unoccupied burrows were observed (Everett, 1988). In February 1988, the damage to nesting areas caused by introduced pigs and rabbits was extensive and shearwater carcasses were observed frequently on the ground (Howell and Webb, 1989). On Socorro, the impact of cat predation has been contested. Jehl and Parkes (1982) suggested that cat predation was an imminent threat after observing several shearwater carcasses, presumably killed by cats. Rodrı́guez-Estrella et al. (1991) and Wehtje et al. (1993) did not agree with that assessment because they failed to find cat sign in the shearwater’s breeding areas. More recently, Martı́nez-Gómez and Curry (1995; 1996) found shearwater remains in cat scats and stomach contents, as well as carcasses eaten in a way indicative of cat predation (c.f. Veitch, 1985). Such observations confirm that cats pose a serious threat to this species. In this paper, we provide an assessment of the current status of Townsend’s shearwater as well as information about its population size and distribution. We focused our efforts on the extant breeding colony at Socorro, but surveyed Clarion to determine if shearwaters had returned. The effect of habitat destruction, cat predation, and demographic stochasticity on this endangered subspecies was evaluated using matrix population projections. By combining information on population size and estimated levels of habitat loss and predation, we were able to estimate the time frame in which Townsend’s shearwater may reach demographic instability. We also discuss several conservation steps necessary to avert this species’ extinction. 1987, 1 February–7 March 1988, 15 January–7 March 1989, 18 January–18 April 1991, 5 March–4 April 1995, 18 January–4 April 1996, 18 January–18 April 1997, 18 January–18 April 1998, 18 January–18 April 1999, 4 February–18 April 2000, and Clarion on 5 April 1995, 19 January 1996, 9–14 April 1997, 4–18 March 1998, 5 February–6 April 1999, 5 February–6 April 2000. Most data were gathered during 1993–1997. 2.1. Distribution and population size Socorro and Clarion were surveyed to estimate shearwater numbers and determine the extent of current breeding areas. On Socorro, 40 campsites served as nocturnal sampling stations (Fig. 1). Nine were located below 500 m, and 31 above 500 m around Mount Evermann. Most of these sites were placed in five different regions around Mount Evermann (north, northwest, east and west forests, and the central plateau). A coarse estimate of breeding population size was obtained by summing the maximum count of birds returning to each of the five regions throughout our study. Birds were detected visually or by listening to their calls during moonless nights. We did not count birds at their burrows because of the risk of facilitating cat predation. Our population estimate was compared with that of Spear et al. (1995) to help us choose a realistic initial population vector for matrix projections. On Clarion, the entire island was surveyed, but especially those areas where shearwaters had been previously reported (Fig. 2; Everett, 1988; Howell and Webb, 1989) and areas with suitable habitat. A night station was established near the military garrison to listen for incoming shearwaters. We also searched for shearwaters resting or flying at sea during daily near-shore (5 km) surveys around Socorro and Clarion. We looked carefully for shearwaters flying offshore of Clarion as these birds could be prospecting for burrows or attempting to breed. In addition, marine surveys were conducted from naval transport vessels while approaching or departing these two islands. 2.2. Predation 2. Methods Opportunistic terrestrial and maritime samplings were conducted on Socorro and Clarion at different periods during 1986–2000 while conducting studies on Socorro Mockingbirds (Mimodes graysoni; JEMG) and Humpback whales (Megaptera novaeangliae; JKJ). JEMG visited Socorro from 20 May to 5 June 1988, 6–19 January and 6 June–5 August 1993, 18 February–4 June 1994, 20 March–19 May 1995, 20 January–3 April 1996, and 19 June–10 August 1997, and Clarion on 6 June 1988, 19 March 1995, and 19 January 1996. JKJ visited Socorro on 15 January–21 February 1986, 15 January–7 March On Socorro, we searched for cat sign in multiple areas of the island (Fig. 1). Trails to the different sampling stations were used as sampling transects, and cat scats were collected every 4–6 days. Cats were trapped with leg-hold, Tomahawk, and Havahart traps. Additionally, military personnel hunted cats with firearms. Scats and stomach contents were examined to determine cat diet. Bird remains were identified from feathers, beaks, bones and/or claws; mouse and lizard remains from bones, fur, skin and/or scales; and invertebrate remains by their shape and texture. Our sampling scheme allowed us to obtain an idea of cat distribution and diet, but it J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 37 Fig 1. Map showing the locations of trails used to gather cat sign and the location of 40 nocturnal sampling stations. Squares with dots show those stations where shearwaters were observed landing. was not designed to obtain a population estimate. In addition to cats, Socorro red-tailed hawks (Buteo jamaicensis socorroensis) are a natural predator of Townsend’s shearwater (Martı́nez-Gómez and Curry, 1995). We assessed qualitatively predation by red-tailed hawks by inspecting nest contents and by searching under hawk perches for shearwater remains. Previously published information on cat distribution and diet (Martı́nez-Gómez and Curry, 1996) is supplemented with subsequent findings. We estimated the number of shearwaters (s) killed per cat per breeding season using the proportion (p) of cat scats that had shearwater remains following the methods of Apps (1983):
s ¼ ðpÞðdÞ r1 ðkÞðbÞ of an ‘‘average’’ cat. We assumed that cats defecate one scat per day (d=1), that shearwater remains in a scat came from a single bird (r=1) and that cats killed 90% (k=0.9) of the shearwaters while the other 10% was scavenged from birds killed by hawks. We also assumed a shearwater breeding season of 8 months (b=240 days). Thus, the total number of female shearwaters (T) killed by cats during the breeding season would vary proportionally to the size (N) of the predator population [T=(s/2)(N)]. Although these estimates might not be accurate in the presence of a prey-specialist predator, they are still helpful in evaluating the impact of different hypothetical predation levels. Thus, if all cats have similar prey preferences, the proportions of the different prey items represent the diet Matrix models have been used as a heuristic tool to evaluate the contribution of different factors to the 2.3. Population projections 38 J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 Fig. 2. Previously reported locations of Townsend’s Shearwater breeding grounds at Isla Clarion. Of these, two were found on hilltops and two in a central plateau between two hills. demography of endangered species (Caswell, 2001). The impact of cat predation and/or various anthropogenic factors on the populations of Newell’s (Puffinus a. newelli) and black-vented (P. opisthomelas) shearwaters has been explored using deterministic matrix models (Ainley et al., 2001; Keith et al., 2002). Encouraged by the versatility of these models, we estimated a possible time frame in which Townsend’s shearwater may reach demographic instability by projecting the change of a hypothetical population affected by habitat degradation, cat predation and demographic stochasticity. Only females and their daughters are considered in our projections; thus, all demographic parameters are adjusted accordingly. We explored six possible scenarios: (1) the fate of a slow growing population, (2) the impact of severe habitat degradation, (3) low and (4) high predation scenarios and severe habitat degradation combined with (5) low and (6) high predation. For each model, we determined the finite rate of increase (l), the annual rate of decrease, population size at t=150, time to extinction (arbitrarily set as the point where population size becomes less than 2), and the time in which the last 100 females occur. Matrix analyses and projections were carried out using PopTools 2.5.3 for Windows (http:// www.cse.csiro.au/poptools); projection files are available from the authors upon request. 3. Life cycle reconstruction Prior to demographic modeling, a reliable reconstruction of this species’ life cycle is required. The breeding biology for this seabird can be reconstructed from the reports of expeditions to the archipelago in 1897 (Anthony, 1898, 1900; Kaeding, 1905) and 1925 (Hanna, 1926; McLellan, 1926). Young as large as adults (half grown,  35 days old if growth patterns are similar to those of Manx shearwater; Brooke 1990) and one egg in advanced incubation ( 50 days if it was about to hatch) were reported on San Benedicto and Clarion by both parties during 30 April–1 May. Wellgrown young (perhaps 60–70 days old) were reported during 20–27 May on Clarion (Anthony, 1898; Kaeding, 1905). However, only young in pteroptyle down from both islands were archived in museum collections. Those specimens could not be older than 2 weeks considering their wing chords (1 May: USNM 162003=29 mm, CAS 28164=25 mm, CAS 28165=36 mm; 25 May: USNM 162004=41 mm). Thus, it appears that birds begin gathering at the islands by mid-November to prospect for burrows; a pre-laying period of one or two months should follow (Jehl, 1982). They should begin laying from late January to mid-March and fledging young from late May to mid-July as previously conjectured (Jehl, 1982), with little to no fledging occurring after. Ainley et al. (2001) suggested that the islands in the archipelago might have a different laying phenology based on a report of very small young on San Benedicto by late May (Anthony, 1900). However, closer inspection of that account suggests a printing error because the other reports of the 1897 expedition only mention young on that island by late April (Anthony, 1898; Kaeding, 1905). J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 It is also conceivable that if Townsend’s shearwater has a similar life history to that of Newell’s shearwater (e.g. Ainley et al., 1997), young birds would not return to the breeding grounds, subadults would return after adults have finished laying their eggs, and pair bonding would begin during the fourth and fifth years. Most birds would begin breeding by ages 5 or 6, if age of first reproduction parallels that of the Manx shearwater (e.g. Brooke, 1990). 4. Stable population The fate of a stable population was studied by assembling a matrix using demographic parameters from similar sized shearwaters. We followed the criteria for a post-breeding census (e.g. McDonald and Caswell, 1993; Caswell, 2001); therefore, the first survival rate (P21) represents the transition from fledglings to 1-yearold birds (that probability would be incorporated into fertility values under a pre-breeding census). Survival rates (Pij) increased with age; for the first three years of life they were 0.654, 0.780, 0.890, and 0.905 thereafter (e.g. Perrins et al., 1973; Ainley et al., 2001). Our matrix has two fertility values (F15, F16); these values were modeled as the product of adult survival rates, breeding probability and breeding success of first-time and experienced breeders. We used breeding success values from an undisturbed colony of the Manx shearwater; females in their first breeding year produce fewer female chicks than experienced females. Breeding probability was adjusted to produce a finite rate of increase (l) slightly greater than one; such a value represents the minimum proportion of breeding females required to maintain a stable population. In doing so, our model resembles the small growth rates observed in other pelagic birds (Russell, 1999). Our work on Socorro Island prompted us to begin our projections using the worst-case demographic scenario suggested for this seabird (Spear et al., 1995). Thus, our initial population vector consisted of a total of 8750 females of which 2625 (30%) were sexually mature; the remaining birds were distributed evenly among classes 1–5 (1225 per class). Population projections were run for 150 years. Average population trajectories and 95% confidence intervals for these projections were obtained through Monte Carlo simulations incorporating demographic stochasticity; we ran 10,000 simulations for each projected matrix. We allowed the number of survivors at time t+1 to be a binomial random number based on the survival rate and number of individuals in the previous age class at time t. Fertility components varied according to published deviation estimates for similar sized shearwaters. Matrix parameters for this and subsequent models are shown in Table 1. 39 5. Habitat degradation Presumably, habitat degradation and loss could impact demographic parameters of a life cycle in various manners. In the absence of data, most parameters were left the same as in a stable population, but breeding probability was reduced to model the loss of nesting burrows. We derived a value for breeding probability from our colony-based census on Socorro and the number of mature females in the initial population vector. We assumed that habitat destruction proceeds at a rate that maintains breeding probability constant despite a declining population. Confidence intervals were obtained through Monte Carlo simulations. 6. Cat predation The impact of cat predation was estimated by recalculating survival rates and fertility values of age classes affected by predation. The number of survivors under low and high predation scenarios results from subtracting the number of female shearwaters taken by 1 or 10 ‘‘average’’ cats (sensu Apps, 1983) and those dying of natural causes from the initial population vector. We assumed that only the three oldest age classes were taken by cats. The total number of killed shearwaters represents the sum of all deaths happening during the entire breeding season. Our model takes into consideration the fact that predation of birds 4–5 years old occurs predominantly in the first months of the breeding season, whereas that of birds 6 years old or older happens at the end. For practical purposes, we considered predation on nestlings to be nil because they remain in their burrows until fledging; upon departure to the sea they will not return to the breeding grounds until they reach sexual maturity. Thus, the number of 6-year-old females or older eaten by cats was determined by multiplying the estimated total number of breeding females killed under low and high predation scenarios by the proportion of that age class in the stable age distribution of the stable population. The remaining number of killed females was divided equally between ages four and five. Although arbitrary, this assignation describes a plausible pattern of mortality where mature females would suffer the greatest impact due to their longer stay on the island (but see Ainley et al., 2001; Keith et al., 2002). After the impact of predation was incorporated, survival rates in the age classes affected by predation varied randomly following a normal distribution with an average and standard deviation determined from annual variation in scat contents (Table 1). Then, we estimated the number of survivors in affected age classes as a binomial random number as in previous models. By keeping average survival rates constant, we assumed 40 J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 Table 1 Demographic parameters used in population projections refer to females in the populationa Non-zero matrix elements Stable population Habitat degradation (HD) Low predation (LP) High predation (HP) HD & LP HD & HP Variation around parameter Initial population vector F15 F16 P21 P32 P43 P54 P65 P66 Fertility components Breeding Probability 5th Age Class Breeding Probability 6th Age Class Breeding Success 5th Age Class Breeding Success 6th Age Class 0.193777 0.240314 0.654 0.780 0.890 0.905 0.905 0.905 0.059843 0.074215 0.654 0.780 0.890 0.905 0.905 0.905 0.190951 0.236454 0.654 0.780 0.890 0.898595 0.898595 0.89736 0.178701 0.218336 0.654 0.780 0.890 0.840945 0.840945 0.828601 0.058970 0.073023 0.654 0.780 0.890 0.898595 0.898595 0.897360 0.055187 0.067427 0.654 0.780 0.890 0.840945 0.840945 0.828601 b 0.011741c 0.011741c 0.014004c – – 1225 1225 1225 1225 1225 2625 0.68 0.68 0.31488 0.3905 0.21 0.21 0.31488 0.3905 0.68 0.68 0.31488 0.3905 0.68 0.68 0.31488 0.3905 0.21 0.21 0.31488 0.3905 0.21 0.21 0.31488 0.3905 0.07054d 0.059235d 0.07054 0.059235 – – – – b c c c a Survival rates for the stable population come from Ainley et al. (2001). Fertility values are the product of survival rates from the 5th and 6th age classes, breeding success (Brooke, 1990), and breeding probability (adjusted to obtain a stable population). Fertility values and survival rates were adjusted further to incorporate habitat degradation and predation. The initial population vector corresponds to the worst demographic scenario suggested by Spear et al. (1995). b Variation around fertility values resulted from the variation in its three components. c Variation in survival rates followed a binomial distribution. Survival rates for the last three age classes first varied according to differences in prey proportions observed in scats; then as a binomial variate. d Variation around breeding probabilities was arbitrarily set equal to that of breeding success (from Brooke, 1990) for the same age class. that cats inflict the same proportional impact as the population declines. This assumption might not hold in the presence of prey-specific predators or if densitydependent factors occur. Confidence intervals for the remaining parameters were established as before. 7. Habitat degradation and cat predation The combined effect of these two risk factors was evaluated by incorporating the impact of habitat degradation into the two previous predation models. This exercise allows one to evaluate the relative impact of each threatening factor and to identify those age classes that require protection when habitat loss and predation act concurrently. 8. Results 8.1. Distribution and population size Our surveys confirmed that Socorro holds the last breeding colonies of Townsend’s shearwater. At Clarion, our only marine detection was of five shearwaters 37 km east of the island on 19 January 1996 while in transit in a navy vessel. No shearwaters were seen in the vicinity of Clarion during daily offshore surveys during 1997–2000. On Socorro, shearwaters were found frequently in groups ranging from 3 to 10 individuals and less frequently in groups > 50 individuals. We saw > 100 shearwaters resting on the waters 23 km west of Socorro on 13 January 1996 and > 20 foraging individuals 35 km southwest of the island on 20 January 1996. These results suggest that most shearwaters are found at Socorro. Terrestrial surveys corroborated that pattern. At Clarion Island in 1998–2000, no active burrows were detected. The areas where shearwaters had previously been observed nesting (Fig. 2) no longer had suitable habitat and were infested by rabbits. On Socorro, breeding areas were found on Mount Evermann in dense shrubs and forests above 800 m (Fig. 1). Colonies apparently remained active until mid-summer. We detected shearwaters coming to land during June and July 1993 and 1997. However, during the same months in 1997, we detected only a few individuals returning to montane areas at night. We also found fledglings landing on the Naval Base in June 1993, perhaps attracted by streetlights. Our findings suggest that the breeding phenology of Townsend’s shearwaters coincides with our reconstruction and that breeding colonies are not active beyond July. However, both of these years were influenced by the El Niño Southern Oscillation (ENSO) and perhaps are not representative of other years. Breeding and roosting areas of Townsend’s shearwater coincide with the core set of territories of Socorro Mockingbirds around Mount Evermann. This area covers about 6 km2 and represents < 10% of the island’s surface (Martı́nez-Gómez and Curry, 1996). The dominant trees are Ilex socorroensis and Guettarda insularis, while Triumfetta socorrensis is the dominant herbaceous J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 species (Martı́nez-Gómez et al., 2001). The dense and intricate herbaceous layer may provide protection from predators. We detected shearwaters landing in 22 of the 31 sampling stations above 500 m; most sightings were concentrated in forested areas north and northwest of Mount Evermann (Figs. 1 and 3). Below 500 m, we observed few individuals flying above the stations; at lower elevations, landings happened only at the naval base, perhaps as a result of light attraction. Based on these sightings, we obtained a rough estimate of a minimum breeding population of ca. 1100 birds. 8.2. Predation On Socorro, cats and red-tailed hawks preyed on the endemic shearwater during the breeding season. During 41 our study we trapped 92 cats of which 40 (43%) were found above 500 m. Overall, the majority of cat sign was found south of the mountain summit where habitat is highly degraded by sheep (Fig. 3). However, cats are present in pristine areas to the north as indicated by cat tracks detected at Playa Blanca. In 37 scats and 46 stomachs, the percentage of bird remains was significantly higher in scats (2=11.33, df=1, P < 0.001) and stomachs ( 2=13.56, df=1, P < 0.001) collected above 500 m than in those below (Table 2). The high incidence of bird remains indicates that cats pose a significant threat to Townsend’s shearwater and Socorro’s avifauna (Table 3). On the basis of shearwater remains in scat samples, one ‘‘average’’ cat (sensu Apps, 1983) could potentially kill ca. 36 female shearwaters if only scats collected above 500 m were tallied (Table 4). Using Fig. 3. Approximate location and number of shearwaters detected during this study. Five different breeding areas were found around Mount Evermann above 800 m. The largest breeding grounds were found North and Northwest of the island’s summit where the vegetation has not been damaged severely by introduced sheep. Black circles indicate the location of cat sign (scats, tracks, prey remains, captures) observed along the main road, trails, and walking paths on the island. J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 42 only scats above 500 m for predation estimates on this seabird is justified because of the elevation bias abovementioned. Table 2 Incidence of bird remains in cat scats and stomach contents below and above 500 m of elevationa Scats (m) Bird remains Present Absent Stomachs (m) 0–500 N=12 501–1000 N=25 0–500 N=35 501–1000 m N=11 4 8 23 (8) 2 1 34 6 (4) 5 a Numbers in parentheses indicate instances in which Townsend’s shearwater remains were found. Table 3 Percentage of different prey items found in cat scats and stomachs collected at all elevations of the island during 1993–1996a Scats N= 100% Stomachs 37 N= 100% 46 Birds Puffinus auricularis Thryomanes sissonii Parula pitiayumi Mimodes graysoni Pipilo maculatus Mimus polyglottos Reptiles and mammals Urosaurus auriculatus Mus musculus Invertebrates Schistocerca spp. Scolopendra sp. Beetles 27 8 7 3 2 2 1 7 6 1 31 26 10 1 7 4 2 1 15 14 1 20 19 6 2 73% 22% 19% 8% 5% 5% 3% 19% 16% 3 84% 70% 27% 3% Birds Puffinus auricularis Pipilo maculatus Parula pitiayumi Reptiles and mammals Urosaurus auriculatus Mus musculus Invertebrates Schistocerca spp. Scolopendra sp. Gecarcinus planatus 15% 9% 4% 2% 33% 30% 2% 43% 41% 13% 4% a Notice the high incidence of bird remains in scats. Townsend’s shearwater was the most common prey item found in scats and stomachs. Because some scats and stomachs contained several prey types, their addition does not add up necessarily to the total or partial counts for each category. Data from 1993 to 1994 (Martı́nez-Gómez and Curry, 1996) were supplemented with findings from 1995 to 1996. No scats were collected in 1997 because intense rain due to El Niño conditions washed away cat sign. During the summer of 1993, many clusters of shearwater feathers were found on the forest floor in a manner suggestive of red-tailed hawk predation (c.f. Veitch 1985). Cats probably killed some of these birds first and hawks scavenged their carcasses later. Shearwaters seem to be a common item offered by red-tailed hawks to their nestlings; remains of this seabird were found in two inactive hawk nests and later in an active nest with a nestling on 25 April 1994. 8.3. Population projections Projections of the six scenarios modeled in this paper and their demographic benchmarks are summarized in Table 5 and Fig. 4. Besides the stable population, only the projection under low predation (1 ‘‘average’’ cat) had a population that persisted for more than 150 years. Time to extincion was reached first (t=63) in the model that incorporated high predation and habitat degradation. In the other models, population extinction took more than 100 years. Under habitat degradation and high predation, it took only 30 years for the number of breeding females to drop below 100; a similar situation was reached before 100 years in the models for habitat degradation, high predation, and habitat degradation with low predation (Table 5). Thus, severe habitat loss combined with intensive predation could have catastrophic consequences for Townsend’s Shearwater in less than 100 years. Although habitat degradation, high predation, and habitat degradation with low predation would allow population persistence for more than 100 years, this may not occur because of the small number of breeding females present at that time. Only very small predation rates would allow population persistence and a large number of breeding females for more than 150 years in spite of declining conditions. Annual rates of decrease when low and high predation levels were combined with habitat degradation were only slightly higher than the addition of the individually modeled factors; thus, it cannot be concluded that either Table 4 Estimated number of female Townsend’s shearwaters killed by cats during a breeding season following an approach similar to that of Apps (1983)a Year 1994 1995 1996 Mean SD All years a Scats found >500 m 13 7 5 – – 25 Proportion of scats with shearwater remains (p) 0.307 0.286 0.400 0.331 0.061 0.32 Shearwaters killed per cat per breeding season s=(p)(d)(r1)(k)(b) 66.312 61.776 86.400 71.496 13.105 – Female shearwaters killed per cat per breeding season (s/2) 33.156 30.888 43.200 35.748 6.552 – Total number of females killed T=(s/2) (N) Low Predation N=1 average cat High Predation N=10 average cats 33.156 30.888 43.200 35.748 6.552 – 331.560 308.880 432.000 357.480 65.524 – Only cat scats collected above 500 m were used in these calculations because of a bias in prey contents due to elevation. J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 43 Table 5 Demographic benchmarks for the six models discussed in the text Model l Annual rate of decrease (%) N t150 95% CI range Time to extinction 95% CI range Time to 100 last females 95% CI range Stable Habitat Degradation (HD) Low Predation (LP) High Predation (HP) HD & LP HD & HP 1.00137 0.94322 0.99432 0.93625 0.93592 0.87311 – 5.68 0.56 6.37 6.40 12.68 10257 1 3863 1 0 0 7614–13521 0–8 2831–5144 0–8 0–2 0–0 – 147 – 142 130 63 – 118–177 – 106–172 107–152 53–75 – 71 – 58 62 30 – 66–67 – 53–63 58–67 28–31 predation or habitat degradation had a predominant role in population decline. In all cases, sensitivity and elasticity analyses obtained from the deterministic models indicate that fertility and survival of the 6th age class are the most critical elements in the demography of this species (Table 6). Another pattern derived from our projections is that by modeling the number of survivors as a random binomial number, confidence intervals are rather narrow because under a binomial distribution small populations will show less variance. Also, under low predation conditions average population trajectories show slightly larger numbers than these derived from a deterministic run; in the remaining models deterministic and average trajectories overlap. 9. Discussion All the evidence gathered indicates that Townsend’s shearwater is critically endangered. A decline in its population can be inferred from its shrinking colony and high predation rates. Our results confirm that previously reported colonies on Socorro below 800 m are no longer occupied (e.g. Jehl and Parkes, 1982; Castellanos and Rodrı́guez-Estrella, 1992; Wehtje et al., 1993; Llinas-Gutiérrez, 1994), and that cat predation could affect seriously the demography of this species. Available estimates of population size, including ours, are less than ideal because none followed a demographic methodology based on meticulous monitoring of breeding colonies. Jehl’s (1982) estimate of 1000 breeding pairs was, after all, an educated guess. He used the number of birds seen at one location to infer the number of individuals across the island. Implicitly, he assumed a homogeneous distribution of shearwaters at a certain elevation. Population estimates by Spear et al. (1995), although based on a regression model that accounts for spatial heterogeneity, could be inaccurate because data for their regression were gathered over several years. Their approach probably introduced a temporal bias to their analyses if the shearwaters were more abundant during the initial years but declined subsequently. Interestingly, their estimate for P.a. newelli was within the confidence limits of colony-based estimates obtained independently (cf. Ainley et al., 2001). The estimate of 1100 individuals presented in this paper is certainly an underestimate because birds were counted cautiously to avoid tallying individuals more than once. Nevertheless, the dramatic change in the amount of occupied habitat in recent decades (Jehl and Parkes, 1982; Castellanos and Rodrı́guez-Estrella, 1992; Wehtje et al., 1993; Llinas-Gutiérrez, 1994), and the difference in order of magnitude between the estimates of Spear et al. (1995) and ours very likely indicate a real population decline. The trend is consistent, too, with the results of our models. Previous researchers probably failed to recognize the high incidence of cat predation on the island’s avifauna because of short surveys and sampling schemes concentrated below 500 m (Veitch, 1989, Rodrı́guezEstrella et al., 1991). Additionally, cat sign may be difficult to detect when relative humidity is high (June to January approximately) because land crabs (Gecarcinus planatus) effectively erase all tracks and destroy cat scats overnight. Nonetheless, cat predation appears to be pervasive on Townsend’s shearwater (Table 3), and the number of shearwaters potentially killed by cats during a breeding season is alarming (Table 4). Furthermore, simple estimates for a species that produces one young per year and reaches maturity at 5–6 years must maintain adult survival rates of at least 85%, and all mature birds should breed successfully to uphold a viable population (Perrins, 1991). Thus, even limited predation could bring adult survival rates below that threshold (e.g. Tomkins, 1985; Berruti, 1986; Cruz and Cruz, 1987; Seabrook, 1990; Ainley et al., 2001; Keith et al., 2002). The archipelago does not have large human settlements. Therefore, required conservation actions are simply (1) the removal of introduced mammals and (2) habitat restoration. The impact of severe predation and habitat degradation will cease immediately once introduced mammals are removed. Losses due to the native hawks should have a minor effect because of their small population size (Walter, 1990). Restoration of degraded habitats can be accomplished because preserved regions on Socorro have the necessary seed 44 J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 banks from which plants can recolonize. On Clarion, however, the impact of rabbits is pervasive and the seed stock for many plants might be greatly reduced. Enclosures of several sizes could be placed in degraded regions; native plants can be seeded or transplanted to these areas to constitute nuclei from which native vegetation will spread. On Socorro, Dodonaea viscosa, and Triumfetta socorroensis are good candidates for Fig. 4. Demographic projections for the total population of female shearwaters incorporating demographic stochasticity, the impact of habitat degradation, and predation. Confidence intervals (95%) were obtained through Monte Carlo simulations (10,000 runs). Under low predation the deterministic run was slightly lower than the average trajectory obtained from simulations; in the remaining models these two trajectories are practically the same. Numbers in parentheses in the middle projections represent survival rates of the last three age classes. J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 45 Table 6 Sensitivities (sij) and elasticities (eij) for the non-zero matrix elements in the six models discussed in the texta Non-zero elements F15 F16 P21 P32 P43 P54 P65 P66 Stable Population Habitat Degradation (HD) Low Predation (LP) High Predation (HP) HD & LP HD & HP sij eij sij eij sij eij sij eij sij eij sij eij 0.2805 0.2634 0.1051 0.0881 0.0772 0.0759 0.0699 0.6568 0.0054 0.0632 0.0686 0.0686 0.0686 0.0686 0.0632 0.5936 0.0180 0.4258 0.0500 0.0419 0.0367 0.0361 0.0349 0.8268 0.0011 0.0335 0.0346 0.0346 0.0346 0.0346 0.0335 0.7933 0.0289 0.2680 0.1053 0.0883 0.0774 0.0766 0.0705 0.6536 0.0055 0.0637 0.0693 0.0693 0.0693 0.0693 0.0637 0.5898 0.0386 0.3015 0.1112 0.0932 0.0817 0.0865 0.0783 0.6116 0.0074 0.0703 0.0777 0.0777 0.0777 0.0777 0.0703 0.5412 0.0187 0.4353 0.0503 0.0422 0.0369 0.0366 0.0354 0.8243 0.0012 0.0340 0.0351 0.0351 0.0351 0.0351 0.0340 0.7903 0.0276 0.5215 0.0561 0.0470 0.0412 0.0436 0.0418 0.7899 0.0017 0.0403 0.0420 0.0420 0.0420 0.0420 0.0403 0.7496 a Numerical values in the table were rounded for presentation. Notice that in all cases the highest values are associated with the survival of birds 6 years old or older. In a declining population, the second largest sensitivity is the fertility value associated with the 6th age class; only in the stable population is the second largest fertility value that of the 5th age class. such a purpose. Introduced plants should be removed at the same time. Hopefully Townsend’s shearwaters will prospect for burrows on Clarion once introduced mammals have been eradicated (e.g. P. tenuirostris; Bradley et al., 1999). Nonetheless, strategies to reintroduce the shearwaters once mammals have been removed should be considered. Some of the strategies proposed for the recovery of Newell’s shearwater could be implemented on Clarion (see Telfer, 1983). Playbacks, olfactory baits, and decoys should be attempted first. If these efforts fail, a translocation program must follow. Chicks that have not left their burrows, and thus have not yet imprinted on their natal island’s skies, are required (Telfer, 1983; Brooke, 1990). Although our matrix projections are not based on demographic parameters derived from Townsend’s Shearwater and simplify complex demographic components of real life cycles, they show key demographic elements that should be considered in conservation actions. For instance, they demonstrate clearly the importance of adult survival and reproduction and thus the necessity to protect and preserve current breeding areas. They also show that habitat degradation and cat predation must be confronted at the same time. From a conservation standpoint, attention to one threatening factor would only be advisable if the impact of the other is negligible; a situation that very likely does not occur on Socorro. Townsend’s shearwater demography will experience a minimum change only if few cats prey on them. Based on our trapping experience, we believe that no more than a dozen cats may roam breeding grounds on Socorro. In montane areas, the dense and intricate understorey where the burrows are located provides refuge from predators; most shearwaters are taken by cats when they land in parts of the forest where the understorey is practically absent. Under these conditions, a low but steady decline should be expected. The pattern observed at Socorro differs from predation scenarios where target species lack protection from vegetation and nesting areas are readily accessible to predators (e.g. Black-vented shearwater; Keith et al., 2002). In the latter conditions, even few predators may cause significant damage to the population. The situation may be complicated for Townsend’s shearwater because as sheep continue destroying vegetative cover the impact of predation will increase dramatically. Control efforts during the breeding season should be implemented to minimize cat predation on shearwaters and secure remaining breeding areas on Socorro from further destruction by sheep. Special care should be taken to avoid the risk of demographic instability. Ultimately, the risk and timing of extinction will depend on actual predation rates on adults and fledglings, the rate at which nesting habitat is lost, and on how quickly the pool of mature birds found at sea diminishes as times goes by. Intensive demographic monitoring of Townsend’s shearwater will yield accurate estimates of survival rates, fecundity, productivity, breeding phenology and movements that will allow more realistic population projections. More importantly, such a program will aid in evaluating the performance of the different demotypes (sensu Ricklefs, 1991) in the population and the efficiency of eradication, restoration, or translocation programs. Acknowledgements We would like to express our gratitude to the Mexican Navy for their continuous and generous support throughout all years of this study. Financial, logistical support, and/or equipment, were provided by the American Museum of Natural History, the Cooper Ornithological Society, the Association of Field Ornithologists, the American Ornithologist’s Union, 46 J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 Sigma—Xi the Scientific Research Society, the Island Endemics Institute, the California Academy of Sciences, Villanova University, Humboldt State University, and Idea Wild—Biodiversity Research Assistance Organization, the R.V. Odyssey of the Whale Conservation Institute, Cetacean Society International, and the National Geographic Society. A. Jesse and J. Dean kindly offered measurements from nestlings in the California Academy of Sciences and the U.S. National Museum. This manuscript benefited from the comments of D. G. Ainley, R. Ricklefs, E. Hansen, A. Scheuerlein, J. Sánchez, B. Tershy, B. Keitt, S. Fallon, J. Uribe, J. Bollmer and two anonymous reviewers. References Ainley, D.G., Telfer, T.C., Reynolds, M.H., 1997. Newell’s and Townsend’s shearwater (Puffinus auricularis). No. 297. In: Poole, A., Gill, F. (Eds.), The Birds of North America. The Academy of Natural Sciences and The American Ornithologists’ Union, Philadelphia, US, Washington, DC. Ainley, D.G., Podolsky, R., Deforest, L., Spencer, G., Nur, N., 2001. The status and population trends of the Newell’s shearwater of Kaua’i: insights from modeling. Studies in Avian Biology 22, 108–123. American Ornithologists’ Union, 1998. Check-list of North American Birds, seventh ed. American Ornithologists’ Union, Washington, DC. Anthony, A.W., 1898. Avifauna of the Revillagigedo Islands. Auk 15, 311–318. Anthony, A.W., 1900. Nesting habits of the Pacific coast species of the genus Puffinus. Auk 17, 247–252. Apps, P.J., 1983. Aspects of the ecology of feral cats on Dassen Island, South Africa. South African Journal of Zoology 18, 393–399. Berruti, A., 1986. The predatory impact feral cats Felis catus and their control on Dassen island. South African Journal of Antarctic Research 16, 123–127. BirdLife International, 2000. Threatened birds of the world. Lynx Edicions, BirdLife International, Barcelona, Cambridge. Bradley, J.S., Gunn, B.M., Skira, I.J., Meathrel, C.E., Wooller, R.D., 1999. Age-dependent prospecting and recruitment to a breeding colony of short-tailed shearwaters Puffinus tenuirostris. Ibis 141, 277–285. Brattstrom, B.H., 1990. Biogeography of the Islas Revillagigedo, Mexico. Journal of Biogeography 17, 177–183. Brattstrom, B.H., Howell, T.R., 1956. The birds of the Revilla Gigedo islands. Condor 58, 107–120. Brooke, M., 1990. The Manx shearwater. T & A D Poyser, London. Castellanos, A., Rodrı́guez-Estrella, R., 1992. La situación del zenzontle de Socorro (Mimodes graysoni). Ciencia y Desarrollo 18, 64–75. Caswell, H., 2001. Matrix population models, construction analysis, and interpretation. Sinauer, Sunderland. Collar, N.J., Gonzaga, L.P., Krabbe, N., Madroño Nieto, A.L., Naranjo, G., Parker III, T.A., Wege, D.C., 1992. Threatened birds of the Americas. The ICBP/IUCN Red Data Book (3rd edition), Part 2. International Council for Bird Preservation, Cambridge. Cruz, J.B., Cruz, F., 1987. Conservation of the Dark-rumped Petrel, Pterodroma phaeopygia, in the Galápagos islands, Ecuador. Biological Conservation 42, 303–312. Everett, W.T., 1988. Notes from Clarion Island. Condor 90, 512–513. Everett, W.T., Anderson, D.W., 1991. Status and conservation of the breeding seabirds on offshore Pacific islands of Baja California and the Gulf of California. In: Croxall, J.P. (Ed.), Seabird Status and Conservation: a supplement. International Council for Bird Preservation (Techn. Publ. 11), Cambridge, pp. 115–139. Hanna, G.D., 1926. Expedition to the Revillagigedo Islands, Mexico, in 1925, I. Proceedings of the California Academy of Sciences, 4th series 15, 1–113. Howell, S.N.G., Engel, S.J., 1993. Seabirds observation off Western Mexico. Western Birds 24, 167–181. Howell, S.N.G., Webb, S., 1989. Additional notes from Isla Clarion, Mexico. Condor 91, 1007–1008. Jehl, J.R., 1982. The biology and taxonomy of Townsend’s shearwater. Le Gerfaut 72, 121–135. Jehl, J.R., Parkes, K.C., 1982. The status of the avifauna of the Revillagigedo Islands, Mexico. Wilson Bulletin 94, 1–104. Kaeding, H.B., 1905. Birds from the West Coast of Lower California and adjacent islands. Condor 7, 105–111,134-138. Keith, B.S., Wilcox, C., Tershy, B.R., Croll, D.A., Donlan, J., 2002. The effect of feral cats on the population viability of black-vented shearwaters (Puffinus opisthomelas) on Natividad Island, Mexico. Animal Conservation 5, 217–223. Levin, G.A., Moran, R., 1989. The vascular Flora of Isla Socorro, Mexico. Memoirs San Diego Society of Natural History 16, 1–71. Llinas-Gutiérrez, J., 1994. Las aves marinas. In: Ortega-Rubio, A., Castellanos-Vera, A. (Eds.), La Isla Socorro, Reserva de la Biosfera Archipiélago de Revillagigedo, México. Centro de Investigaciones Biológicas del Noroeste S.C. Publicación No. 8. La Paz, Baja California Sur, México, pp. 277–297. Martı́nez-Gómez, J.E., Curry, R.L., 1995. First description of nest and eggs of the Socorro Mockingbird. Wilson Bulletin 107, 551–555. Martı́nez-Gómez, J.E., Curry, R.L., 1996. The conservation status of the Socorro Mockingbird Mimodes graysoni. Bird Conservation International 6, 271–283. Martı́nez-Gómez, J.E., Flores-Palacios, A., Curry, R.L., 2001. Habitat requirements of the Socorro Mockingbird Mimodes graysoni. Ibis 143, 456–467. McDonald, D.B., Caswell, H., 1993. Matrix methods for avian demography. In: Power, D.M. (Ed.), Current Ornithology 10. Plenum Press, pp. 139–185. McLellan, M.E., 1926. Expedition to the Revillagigedo islands, Mexico, in 1925, VI. Proceedings of the California Academy of Sciences, 4th series 15, 279–322. Perrins, C.M., 1991. Constraints on the demographic parameters of bird populations. In: Perrins, C.M., Lebreton, J.-D., Hirons, G.J.M. (Eds.), Bird Population Studies. Oxford University Press (Oxford Ornithology Series), pp. 190–206. Perrins, C.M., Harris, M.P., Britton, C.K., 1973. Survival of Manx shearwaters Puffinus puffinus. Ibis 115, 535–548. Ricklefs, R.E., 1991. Structures and transformations of life histories. Functional Ecology 5, 174–183. Rodrı́guez-Estrella, R., Arnaud, G., Álvarez Cárdenas, S., Rodrı́guez, A., 1991. Predation by feral cats on birds at Isla Socorro, Mexico. Western Birds 22, 141–143. Russell, R.W., 1999. Comparative demography and life history tactics of seabirds: implications for conservation and marine monitoring. American Fisheries Society Symposium 23, 51–76. Seabrook, W., 1990. The impact of feral cats, Felis catus, on the native fauna of Aldabra Atoll, Seychelles, Indian Ocean. Revue d’Ecologie la Terre et la Vie 45, 135–146. Spear, L.B., Ainley, D.G., Nur, N., Howell, S.N.G., 1995. Population size and factors affecting at-sea distriutions of four endangered procellariids in the tropical Pacific. Condor 97, 613–638. Telfer, T. C., 1983. The Hawaiian dark-rumped petrel and Newell’s Manx shearwater recovery plan. US Fish and Wildlife Service. Tomkins, R.J., 1985. Breeding success and mortality of Dark-rumped Petrels in the Galápagos, and control of their predators. In: Moors, P.J. (Ed.), Conservation of Island Birds. International Council for Bird Preservation (Techn. Publ. 3), Cambridge, pp. 159–175. Townsend, C.H., 1890. Scientific results of explorations by the US Fish Commission Steamer Albatross. Proceedings of the United States National Museum 13, 131–142. J.E. Martı´nez-Gómez, J.K. Jacobsen / Biological Conservation 116 (2004) 35–47 Veitch, C.R., 1985. Methods of eradicating feral cats from offshore islands in New Zealand. In: Moors, P.J. (Ed.), Conservation of island birds. International Council for Bird Preservation (Techn. Publ. 3), Cambridge, pp. 125–141. Veitch, C.R., 1989. The eradication of cats and sheep from Socorro Island. Northern Region Techical Report Series No. 11. New Zealand. 47 Walter, H.S., 1990. Small viable population: the red-tailed hawk of Socorro Island. Conservation Biology 4, 441–443. Wehtje, W., Walter, H.S., Rodrı́guez-Estrella, R., Llinas, J., Castellanos-Vera, A., 1993. An annotated checklist of the birds of Isla Socorro, Mexico. Western Birds 24, 1–16.