Biological Conservation 84 (1998) 35--45 Published by Elsevier Science Ltd Printed in Great Britain PII: S0006-3207(97)00077-3 0006-3207/98 $19.00+0.00 !IELSEVIER
LIFE-HISTORY AND VIABILITY ANALYSIS OF THE ENDANGERED HAWAIIAN STILT
J. Michael Reed,a* Chris S. Elphickb & Lewis W. OringC °Biological Resources Research Center, University of Nevada, Reno, 1000 Valley Rd, Reno, NV 89512 USA bEcology, Evolution, and Conservation Biology Program, University of Nevada, Reno, 1000 Valley Rd, Reno, NV 89512 USA C Ecology, Evolution, and Conservation Biology Program, and Department ofEnvironmental and Resource Sciences, University ofNevada, Reno, 1000 Valley Rd, Reno, NV 89512 USA
(Received 15 July 1996; revised version received 24 May 1997; accepted 29 May 1997)
[Hawaiian] stilt, ... .A very fine endemic bird which INTRODUCTION should not be allowed to become extinct or even rare. (Munro, (1946; p.46» Two primary problems in conservation biology are identifying species at risk of extinction and determining what can be done to reduce that risk. Population viabi Abstract lity analysis (PVA) is a tool that can be used to address The Hawaiian stilt Himantopus mexicanus knudseni is both problems. Using life-history data and their rela an endangered, endemic subspecies of black-necked stilt. tionships with environmental factors, PVA is used to We present life-history data required to perform popula estimate persistence probabilities of populations under tion viability analysis (PVA), and the results of a series different conditions (Shaffer, 1981; Salwasser et al., 1984; of PVAs under two scenarios, treating (a) the subspecies Gilpin and Soule, 1986; Marcot et al., 1986; Reed et al., as a single population, and (b) six subpopulations as a 1988; Woodruff, 1989; see Boyce, 1992 for a review). metapopulation. We performed sensitivity analyses on Data inadequacies and simplifying assumptions can model parameters and used results to address various limit confidence in the specific predictions of viability management options. Both basic models predicted that models (Caughley, 1994; Harcourt, 1995; Taylor, 1995). stilts would increase to fill available habitat with no By varying parameter conditions, however, PYA can be chance of a significant decline. Catastrophe, maximum used to explore the consequences of different manage age, and density-dependent reproduction had little effect ment schemes on model popUlation dynamics (Walsh, on population projections. Rapid declines in the probabil 1995), thereby providing insight useful to managers. In ity of stilt populations persisting occurred when clutch this paper, we use PYA to estimate extinction risk and to failure rate or first-year mortality rate increased above evaluate management options for an endangered shore 70%, or when adult mortality rate increased above 30%. bird, the Hawaiian stilt Himantopus mexicanus knudseni. Model predictions of mean population size at 200 years Hawaiian stilts are a subspecies of the black-necked tracked changes in carrying capacity. If current conditions stilt endemic to the Hawaiian islands. They are signifi change such that rates of clutch failure or stilt mortality cantly larger than their North American counterpart increase, population declines and eventual extinction (Coleman, 1981), and differ somewhat in plumage becomes more likely. Managers, therefore, should main characteristics (Wilson and Evans, 1893). Hawaiian tain predator control, limit water level fluctuations, and stilts are found on all five major islands (Hawai'i, maintain current habitat area. Downlisting is not war Kaua'i, Maui, Moloka'i, O'ahu) although their presence ranted because wetland management and predator control on the island of Hawai'i might be due to recent recolo are necessary for Hawaiian stilts to persist. Published by nization after several decades of absence (paton et al., Elsevier Science Ltd 1985; Banko, 1988). Stilts also are found in abundance on Ni'ihau, which shares birds seasonally with Kaua'i, Keywords: Hawaiian stilt, Himantopus, shorebird, and since 1989 on Lana'i at a newly available water extinction, PVA, demographic model. source (Engilis and Pratt, 1993). Hawaiian stilts forage in shallow water and nest on adjacent embankments (Coleman, 1981; Engilis and Reid, 1994). Historic and current population sizes have *To whom correspondence should be addressed at: Depart depended partly on certain agricultural practices that ment of Biology, Tufts University, Medford, MA 02155, USA, provide breeding and foraging grounds (e.g. taro, sugar Fax: 702-784-4583. cane runoff) (Griffin et al., 1989). Dependence on 35 36 J. M. Reed, C. S. Elphick, L. W.Oring agricultural habitats and practices, coupled with habitat conservation efforts to be focused appropriately. For conversion for housing and business, has resulted in a example, Crouse et at. (1987) found juvenile survival fragmented and reduced wetland landscape, particularly limited adult numbers in loggerhead sea turtles Caretta in coastal wetlands where stilts are confined (Handy and caretta, indicating conservation efforts should focus on Handy, 1972; Shallenberger, 1977; Coleman, 1981; juvenile survival rather than on egg production or hatch Griffin et aI., 1989). success. Because the most sensitive variables require the There are almost no published data on Hawaiian stilt most accurate data, sensitivity analyses also can be used ecology or population biology and only qualitative esti to focus researchers' energies on improving estimates of mates of population size before the 1940s. Henshaw, the most important variables (Reed et aI, 1993). We (1902) reported that stilts were common on the island of used results of this analysis to address particular O'ahu in the late 18oos, but by 1900 were very scarce. He management options. attributed the severe decline to overhunting. Although Our final goal (4) was to assess population growth Hawaiian stilt flesh was viewed as being of little value potential. State-wide population size in 1947 was esti for food, stilts were hunted even before white settlement mated to be 1000 (Schwartz and Schwartz, 1949). Only (Henshaw, 1902; Handy and Handy, 1972). Hunting six years earlier Munro, (1944) estimated it to be only continued to be legal until 1941 (Schwartz and Schwartz, 200. It has been supposed that Munro's estimate was 1949), and was probably a major factor in keeping low because this rate of growth was viewed as unlikely population sizes low during the late 1930s (Munro, 1938; (e.g., Schwartz and Schwartz, 1949; Fisher, 1951). We Shallenberger, 1977). Munro, (1944) estimated that used our population model to determine whether or not there were approximately 200 individuals in the early growth of this magnitude could have occurred. 1940s. Following cessation of hunting, numbers rose rapidly, and by 1947 there were 1000 individuals (Schwartz and Schwartz, 1949). Since then, Hawaiian METHODS stilt numbers have increased to their current value of approximately 1200 birds (Reed and Oring, 1993). Viability criteria and population structures Little is known of the population structures of We defined a popUlation as safe from extinction if there Hawaiian waterbirds, but evidence suggests that local was less than a 5% probability of its declining signifi stilt populations are connected through dispersal (Tel cantly in 200 years. Philosophically, we would have fer, 1971, 1972; Pyle, 1978; Telfer and Burr, 1978; preferred having no biologically significant decline as Engilis and Pratt, 1993; Reed et at., 1994). The Hawai our criterion, but we believe this cannot be determined a ian stilt population might exist as a metapopulation priori (Reed and Blaustein, 1997). In order to assess (Reed et at., 1994), which adds complexity to popula significance of an observed decline, we used a one tion processes (e.g. Murphy et at., 1990). Movement tailed, one-sample t-test. Our null hypothesis was that among populations affects demographics, population the population size at the end of 200 years (T = 200) dynamics, and genetics, and is the driving parameter in would be equal to, or greater than, the popUlation size metapopulation models (e.g. Hastings and Wolin, 1989; at the start (T = 0); our alternative hypothesis was that Hansson, 1991; Wu et at., 1993). Little is known about the population size at T = 200 would be significantly Hawaiian stilt dispersal patterns, except that they do less than that at T = O. Target times in population move among wetlands and islands (Munro, 1944; Engi projections are arbitrary (cf. Shaffer, 1981), and 200 lis and Pratt, 1993; Reed et al., 1994). years was chosen as a reasonable management time We have several goals in this manuscript: (1) we pre frame. Each iteration of a model was treated as a single sent life-history data required for performing popUla replicate and 140 iterations were used for each model. tion viability analyses, summarizing data from This sample size gives a statistical test with a power of unpublished studies and supplementing these data with I - {J = 0·80, when a = 0·05 and the desired effect size is our own research; (2) we used VORTEX (Lacy et at., assumed to be small (0·2) (calculated using correction 1995), a stochastic simulation model, to perform popu from a two-tailed test; Cohen, 1988 providing a strong, lation viability analyses for the endangered Hawaiian conservative test capable of detecting significant stilt. These analyses were done under two population declines. We ran two basic models using VORTEX structure scenarios: (a) treating the entire subspecies as Version 7 (Lacy et at., 1995): all birds in the state of a single isolated population, and (b) assuming a meta Hawai'i acting as a single population, and a six-popu population structure consisting of six islands as inter lation metapopulation model. Model parameters below connected subpopulations; (3) we performed sensitivity follow those required for running VORTEX. analyses on the life- history parameters in the viability analyses. Sensitivity analysis can be used to determine Model parameters which parameters most influence model output (e.g. Thomas et ai., 1990). This information is, perhaps, the Reproduction primary value of PVA to conservation biology. It iden Hawaiian stilts have a monogamous mating system, tifies crucial life-history stages or processes and allows typically begin breeding at age two, and have a Viability analysis of Hawaiian stilt 37 maximum brood size of four (Coleman, 1981; J.M.R., mates of first year survival of 0·53 and 0·60 for birds unpublished data). We used data from a wide variety of hatched in 1993 and 1994, respectively, and an estimate sources (Telfer, 1972; 1974; 1983; 1984; 1985; Ueoka et of 0·81 for second-year survival for birds hatched in al., 1976; Ohashi and Telfer, 1977; Dougherty et al., 1993. 1978; Ueoka and Telfer, 1980; Telfer et al., 1981, 1982; Because these estimates are based on very limited US Fish and Wildlife Service, unpublished data) to data, we also surveyed the literature for survival infor determine the proportions of females that produced mation on other large shorebird species to assess how broods of different sizes (n = 484 broods, across 15 good these estimates are likely to be. Minimum adult years, including five sites on three islands) (Fig. I). survival estimates for eight other species of large shore Summary data from other unpublished sources, added birds (Table 1) range between 0·61 and 0·92 to these data, give a mean brood size of 2·18 (n=982 (mean = O· 76), suggesting that our estimate of 0·81 for broods) and SO = 1·6 (n = 529), but a very non-normal Hawaiian stilts is not unreasonable. distribution (Fig. 1). Measures of variance in annual survival rates (i.e. We assumed that there was no density-dependence in return rates) exist only for a few species of shorebirds reproduction, that the sex-ratio was even, that all adult and range from 0·03 to 0·21 (Hilden, 1978; Safriel et al., males (~ 2 years old; see below) were in the breeding 1984; Barter, 1989; Holland and Yalden, 1991; Root et pool, and that variation in reproductive performance al., 1992; Paton, 1994; Peach et al., 1994; N. Warnock was not correlated with variation in survival. No data pers. comm.). For our model, we used the mean value exist for these parameters. Current predation problems (0·12). and flooding events occur with enough regularity that Finally, we assumed that male and female survival they are incorporated into observed variation in brood rates were equal, and that birds lived to a maximum age sizes. Effects of introducing exotics not currently in of 15 years. At least two birds banded between 1977- Hawai'i, for example, brown tree snakes Boiga irregu 1980 were alive in 1994 (JMR personal observations). It laris, are unknown but could have catastrophic effects is unknown whether they were banded as hatch-year or on reproduction and could be incorporated in future older birds, which creates a range of potential ages from models. 14 to greater than 18 years of age.
Mortality Population parameters Few data on the survival rates of Hawaiian stilts exist. We used the Hawaii Division of Wildlife's 1995 winter We therefore made survival estimates based on our own waterbird count population estimate for the initial limited banding of Hawaiian stilts and on published population size (total n = 1206 stilts). Carrying capaci research on other species of large shorebird. Reed and ties (K) were calculated as the maximum winter counts Oring (unpublished data) banded 30 pre-fledging for each island, with the K for the single-population Hawaiian stilts in 1993 and 83 in 1994. This sample of model being the sum of these maxima (total K= 1929, banded birds included individuals from all islands with Fig. 2); note that maxima could come from different large stilt populations. Searches for banded birds were years (Reed and Oring, 1993). VORTEX ws users to conducted weekly on O'ahu and monthly on the other model harvesting and supplementing the popUlation. major islands; these censuses covered virtually all avail We omitted both parameters from the model because able habitat. These resight data provide minimum esti- there is no source for supplementation nor legal harvest. We assumed a stable age-distribution. 40 Dispersal Birds move between Kaua'i and Ni'ihau seasonally 30 (Engilis and Pratt, 1993), so we treated these two islands
(.)> as a single population in our metapopulation model. C Hatch-year birds move between islands (Reed and ::sCD 20 Oring, unpublished data). Of 116 chicks banded in C" 1993/1994, seven individuals were seen to move between CD ... islands within 12 months of banding (Fig. 2). Birds are LL 10 known to have moved between most pairs of adjacent islands, and between O'ahu and Maui. We estimated dispersal rates by dividing the number of birds known o to move from one island to another by the number of o 1 2 3 4 marked birds on the source island. In our initial meta population model we used known inter-island annual Number of chicks movement rates, which ranged from 0 to 8·3%. We also Fig. 1. Distribution of number of chicks hatched from 484 assumed that adults of any age can disperse between Hawaiian stilt clutches. islands and that both sexes disperse at equal rates. 38 J. M. Reed, C. S. Elphick, L. W. Oring
Table 1. Estimates of the mean minimum adult survival rate (return rate) for large shorebird species Species Minimum survival rate Reference Eurasian oystercatcher Haematopus ostralegus 0·88 Boyd (1962), Goss-Custard et al. (1982), Safriel et al. (1984) American oystercatcher Haematopus palliatus 0·85 Nol (1985) American avocet Recurvirostra americana 0·92 Robinson and Oring (unpublished data) Pied avocet Recurvirostra avosetta 0·63 Boyd (1962) Eurasian curlew Numenius arquata 0·79 Boyd (1962), Evans (1991) Whimbrel Numenius phaeopus 0·69 Boyd (1962) Black-tailed godwit Limosa limosa 0·70 Boyd (1962) Bar-tailed godwit Limosa lapponica 0·61 Boyd (1962) Mean 0·76
2
~ 1
6
N95 Nmax 1170 1 Ni'ihau 14 239 2 Kaua'j 229 318 ? , 3 O'ahu 706 821 ------~ 4 Moloka'i 16 70 5 Lana'j 24 24 6 Maui 191 410 7 Hawai'i 28 47 100 Ian
Fig. 2. Hawaiian stilt movement patterns among Hawaiian islands, winter population size in 1995 (N95), and maximum population size recorded during winter censuses (Nmax). Solid arrows represent known movements; dashed lines represent suspected coloniza tion; ?s represent unknown source(s). Fractions are number of moves per number of banded individuals.
Based on the observation that one bird moved between example, does reducing reproductive success (analogous islands repeatedly over the course of 12 months, we to reducing predator control, in the case of Hawaiian assumed that the cost of dispersal was low and that stilts) have a significant effect on species persistence? If birds do not die while dispersing. not, then management of predators would be a poor use of limited resources for management. Sensitivity analysis In our sensitivity analysis we altered a number of Sensitivity analysis can be used to assess effects of inac model parameters (Table 2). We examined the effects of curacies of parameter estimates on model predictions. adding catastrophes and density-dependent reproduc To do this, a parameter is varied by a reasonable tion to the model. The impact of density-dependent amount (i.e. within a range of possible or likely values), reproduction was tested for using an equation provided while keeping other parameter estimates fixed. The in VORTEX: model is then rerun, and the effect on population dynamics determined. This procedure is repeated for each parameter of interest. A second reason to do sen peN) = {P(O) - [(P(O) - P(K» ( ~) Bn N: A sitivity analysis is to determine how much change in a parameter value is required for a significant change in population dynamics. This gives information on the where P( N) is the percent of females that breed when most important and effective management strategies by the population size is N; P(O) is the percent of females identifying thresholds below which a parameter should that breed when the population size is close to 0; P(K) not drop, and identifying where management efforts will is the percent of females that breed when the population have the greatest impact on population increase. For size is at carrying capacity; B describes the shape of the Viability analysis of Hawaiian stilt 39
T.bIe 2. Panlaeter dtaages used in seasitmty .....yses. For both the IDKle- ud meta..... tion models we giye the parameters we varied, the ~rent 'fIllues used, ud whether the parameter was indaded because of potendal data inaccaracles or to .ddress m.... gement questions Parameter Description of analysis Reason for conducting sensitivity analysis (a) Single-population model Catastrophe (hurricane) Probability of occurrence: Data uncertainty I 0·01 yr- ; impact = complete reproductive failure
Maximum age 20 years Data uncertainty Density-dependence (a) decrease in reproductive Data uncertainty success as the K is approached. Density-dependence parameter values: B = I, 2, 8, 16; A = o. (b) Allee effect. Allee parameter A = 0·5, 1, 2,4; B = 2. % of females producing no Varied 10% to 90% Management young (clutch failure)
Mortality (age 0 to I) Varied 10% to 90% Data uncertainty; management
Mortality (age> 1) Varied 5% to 60% Data uncertainty
K (carrying capacity) Halved and doubled Data uncertainty; management
Standard deviation in K Varied from 10% of K to 50% of Data uncertainty; management K (b) Metapopulation model connectivity (a) 0% and 1% annual Data uncertainty movement among all islands (b) I % movement among adjacent islands only, plus between O'ahu and Maui
curve relating percent breeding to population size; and A 1901 (SO = 88·6) individuals in 200 years. This popula defines the nature of the Allee effect (Lacy et al., 1995). tion size is not significantly different from the carrying We also varied the maximum breeding age, nest fail capacity (I = 0·32, p > 0·25). The probability of a decline ure rate, mortality rates, and the mean and standard in 200 years was 0·0%. When the model was rerun as a deviation of K. Finally, we varied dispersal rates in our metapopulation using observed data on movement metapopulation model. Parameter changes used to test among islands, the mean population size at 200 years sensitivity are given in Table 2. was slightly smaller (1833, SO = 92·0), but still indistin guishable from hypothesized carrying capacity (K) Population growth potential (t = 1·04, P > 0·1). Despite the occurrence of local extinc Finally, we used our single-population model to deter tions and recolonizations of the Lana'i sUbpopulation mine whether or not Munro's, 1944, 1941 population (six of each), the probability of subspecies decline over estimate of 200 individuals was unreasonable given 200 years remained at 0·0%. Population growth rates Schwartz and Schwartz's, 1949, 1947 estimate of 1000 for the single-population and metapopulation models, birds. We ran two models to assess this: (1) we used model prior to reaching carrying capacity, were r=0·18 parameters from our basic model with an initial popu (SO=0·19) and r=0·19 (SD=0·10), respectively. lation size of 200 and ran the model for six years; (2) we Sensitivity analyses indicated that the initial model assumed good breeding conditions, i.e. every breeding results were robust to most single- parameter modifica pair produced four chicks, and mortality from age 0 to I tions. Incorporating catastrophe and density-dependent was decreased to 33% (value in basic model was 43%). reproduction, and increasing the maximum age of birds into the single-population model did not change results significantly. In all cases the final popUlation size RESULTS increased to a level not significantly different from K (Table 3). The basic single-population model predicted that Changes in several model parameters exhibited Hawaiian stilts would increase in numbers to a mean of threshold changes in population persistence. As the 40 J. M. Reed, C. S. Elphick, L. W. Oring
Table 3. Compuisoas of Hawaiian stilt ... popaIation sizes at year 200 (N_) with earryiDg capacity (.I), aad with iIIiti8l population size (No) when N_ < Nf}o Population sizes iBdade all itentioas, ineIuding tIIo8e that do not last the entire 200 years
Parameter N200±SD N200
Catastrophe 1863± 156 0-42 >0·25 NA
Maximum age 1875±131 0·41 >0·25 NA
Density dependencea 1845± 156 0·54 >0·25 NA
Clutch failure rate 60% 1576±414 1·39 0·05
Juvenile mortality rate 70% 1482±441 1·01 >0·10 NA 80% 0·6±3·1 622·06 <0·0005 388·80 <0·0005
Adult mortality rate 30% 1637 ± 346 0·84 >0·10 NA 40% 0·8±9·0 213·30 <0·0005 133·30 <0·0005 K
50%K(965) 939±78 0·34 >0·25 NA 200%K(3858) 3766±223 0·41 >0·25 NA
SDofK 30%K(579) 1440±639 0·76 >0·10 NA 40%K(772) 293±595 2·75 <0·005 1·54 0·05
Meta-population model 1833±92 1·04 >0·10 NA
Connectivity 0% 1866±79 0·80 >0·10 NA I % (all islands) 1859± 109 0·64 >0·25 NA I % (adjacent) 1878±88 0·58 >0·25 NA aOf the eight models for density dependence, this is the result with the largest t value. bNA = test not applicable because mean popUlation size increased. percent of clutches that failed was increased from 70% For simulations where the mean popUlation size at to 80%, the probability of the subspecies persisting 200 years was lower than the initial population size, we decreased from 0·89 to 0·007 (Fig. 3(a». For those calculated one-tailed, single-sample t-tests to determine populations that persisted for 200 years in the model, whether or not the decline was significant. Highly sig mean size at year 200 starts to decline at 40% clutch nificant declines were found when clutch failure rate was failure. Even though there is a high popUlation persis raised to 80%, when juvenile mortality reached 80%, tence at 70% clutch failure, mean final population size is and when adult mortality was 40% (Table 3; Fig. 3). much smaller (Fig. 3(b». As juvenile mortality (prob Although population size distributions were skewed by ability of surviving to age 1) was increased from 70% to population extinctions, t-tests are robust to non-nor 80%, the probability of the subspecies persisting mality (Miller, 1986) decreased from 1·00 to 0·02 (Fig. 3(c». Again, mean Model predictions of mean population size at 200 population size at 200 years begins to decline at lower years tracked variations in carrying capacity. Specifi mortality values (Fig. 3(d». Finally, as annual adult cally, halving K to 965, gave a value of 939 (SD = 77 ·6) mortality was increased from 30% to 40%, the prob birds at 1200, while increasing K to 3858 resulted in a ability of the subspecies persisting decreased from 1·00 mean of 3766 (SD=222·7) birds (Table 3). Increasing to 0·01 (Fig. 3(e». The size of persisting populations at the standard deviation of K decreased population per 200 years also declined rapidly after adult mortality sistence, decreased mean size of persisting populations, increased above 30% (Fig. 3(0). and increased variance in population size for popula- Viability analysis of Hawaiian stilt 41
Persistence Population tions lasting 200 years (Fig. 4). With 40% standard probability size deviation in K, the mean final population size was a) 2000 1.0 b) almost significantly smaller than the starting population 0.8 1500 size (Table 3); with 50% standard deviation the popu 0.6 1000 lation failed to persist in all iterations of the model 0.4 1 (Fig. 4). This is not surprising, given that a 50% stan 500 0.2 dard deviation would predict K going to 0 frequently. 0.0 0 Degree of connectivity among islands in our metapo 0 20 40 60 80 100 0 20 40 60 80 100 pulation model had little effect on mean metapopuiation 0/0 of clutches that fail size (Table 3), and did not affect the probability of c) 1.0 d) 2000 \ metapopulation persistence (100% in all models). Sub 0.8 1500 populations went extinct only in the observed-movement 0.6 and no-dispersal models. In both cases, subpopulation 1000 0.4 extinctions were infrequent and occurred only on the 1 three islands with small carrying capacities (K < 70 for 0.2 500 each). Except for the no-dispersal model, all subpopu 0.0 1 0 0 20 40 60 80 100 0 20 40 60 80 100 lation extinctions were followed by recolonization. 0/0 juvenile mortality Finally, we examined the potential for rapid popula e) 1.0 f) 2000 tion expansion from relatively low numbers. Using our single-population model and basic parameter values, we 0.8 1500 found that a population of 200 birds increased with a 0.6 1000 growth rate of r=0·18 (SO=0·19), to a mean popula 0.4 tion size of 638 (SO = 269·8) in six years. Assuming 500 0.2 good breeding conditions (Le. no clutch failure and 0.0 reduced chick mortality) a growth rate of r=0·41 20 40 60 80 100 0 20 40 60 80 100 (SO = 0·11) was achieved, enabling a population of 200 0/0 adult mortality birds to increase to a mean population size of 1857 Fig. 3. Persistence probability and mean population size at (SO = 166·2) over six years. 200 years for varying values of (a)-{b) percent of clutches that fail, (c)-{d) percent juvenile mortality, and (e)-{f) percent adult mortality. Population sizes include only iterations that DISCUSSION persist the entire 200 years. The results of both our single-population and metapo
1.0 ...... -- _____ pulation models indicate that, if our parameter values a) and assumptions are reasonable, Hawaiian stilt popula 0.8 tions will not decline over 200 years and should increase to fill available habitat. Available habitat appears to be 0.6 key to Hawaiian stilts because it limits carrying capa 0.4 city. Although many larger wetlands are secure in Hawaii, most are unavailable to stilts for various rea 0.2 sons, such as overgrowth by woody vegetation (Engilis 0.0.j---,---,--,.-----.--., and Reid, 1994). Through wetland enhancement and o 10 20 30 40 50 restoration and subsequent management, more habitat would be available to stilts (e.g. Pyle, 1978; Engilis and 2000 b) Pratt, 1993). The high statistical power of our tests of population decline and the results of our sensitivity 1500 analyses allow reasonable confidence in this result. Sensitivity analyses indicated that our results are 1000 robust to possible single-parameter inaccuracies in SOO parameter estimates. We found that increases in percent of clutches that fail, percent juvenile mortality, percent 0+---,---,--,.-----.--1 adult mortality, and standard deviation in carrying o 10 20 30 40 50 capacity (K) can all lead to rapid declines in the prob sd of K
Henshaw, H. W. (1902) Birds of the Hawaiian Islands, Being a Safriel, U. N., Harris, M. P., Brooke, M. de L. and Britton, C. Complete List of the Birds of the Hawaiian Possessions with K. (1984) Survival of breeding oystercatchers Haematopus Notes on their Habits. Thos. G. Thrum, Honolulu. ostralegus. Journal of Animal Ecology 53,867-877. Hilden, O. (1991) Population dynamics in Temminck's stint Salwasser, H., Mealey, S. P. and Johnson, K. (1984) Wildlife Calidris temminckii. Oikos 30, 17-28. population viability: a question of risk. Transactions of the Hill, D. and Carter, N. (1991) An empirical simulation model North American Wildlife and National Resources Conference of an Avocet Recurvirostra avosetta population. Ornis 49,421-439. Scandinavica 22, 65-72. Schwartz, C. W. and Schwartz, E. R. (1949) A Reconnais Holland, P. K. and Yalden, D. W. (1991) Population dynam sance of the Game Birds in Hawaii. Div. Fish and Game, ics of common sandpipers Actitis hypoleucus breeding along Board of Commissioners of Agriculture and Forestry, an upland river system. Bird Study 38, 151-159. Honolulu. Lacy, R. c., Hughes, K. A. and Miller, P. S. (1995) VORTEX: Shaffer, M. L. (19S1) Inimum population sizes for species A Stochastic Simulation of the Extinction Process. Version 7 conservation. BioScience 31, 131-134. User's Manual. IUCN/SSC Conservation Breeding Specia Shallenberger, R. J. (1977) An ornithological survey of Hawai list Group, Apple Valley, MN. ian wetlands. Ahuimanu Productions. Report prepared on Marcot, B. G., Holthausen, R. and Salwasser, H. (1986) contract to the US Army Corps of Engineers. 2 V ols. Viable population planning. In The Management of Viable Taylor, B. L. (1995) The reliability of using population viabi Populations: Theory, Applications, and Case Studies, eds B. lity analysis for risk classification of species. Conservation A. Wilcox, P. F. Brussard, and B. G. Marcot, pp. 1-13. Biology 9, 551-55S. Center for Conservation Biology, Stanford University, CA. Telfer, T. C. (1971) Field investigation of native Hawaiian Miller, Jr, R. G. (1986) Beyond ANOVA, Basics of Applied waterbirds on the island of Kauai (S.L). Progress Report, Statistics. John Wiley and Sons, New York. Job No. VIII-C (1), Project No. W-15-1. Division of For Munro, G. C. (1938). Summary of the Hawaiian bird survey estry Wildlife, Honolulu. of 1935-1937. Unpublished ms, University of Hawaii library. Telfer, T. C. (1972) Field investigation of native Hawaiian Munro, G. C. (1944) Birds of Hawaii. Tongg Publ. Co., Hon waterbirds on the island of Kauai (S.L). Progress Report, olulu. Job No. VIII-C (2), Project No. W-15-2. Division of For Munro, G. C. (1990) The Hawaiian bird survey of 1935-37. estry Wildlife, Honolulu. 'Elepaio 7,22-24. Telfer, T. C. (1974) Field investigation of native Hawaiian Murphy, D. D., Freas, K. E. and Weiss, S. B. (1985) An waterbirds on the island of Kauai (S.L). Progress Report, environment-metapopulation approach to population via Job No. VIII-C (4), Project No. W-15-4. Division of For bility analysis for a threatened invertebrate. Conservation estry and Wildlife, Honolulu. Biology 4, 41-51. Telfer, T. C. (19S3) Evaluation of endangered waterbird habi Nol, E. (1985) Sex roles in the American oystercatcher. Beha tat improvements. Progress Report, Job No. R-I1I-E(a), viour 95, 232-260. Project No.s W-IS-R-8. Division of Forestry and Wildlife, Ohashi, T. J. and Telfer, T. C. (1977) Limited study of nesting Honolulu. by stilt on the islands of Maui, Oahu and Kauai. Progress Telfer, T. C. (1984) Evaluation of endangered waterbird habi Report, Job No. R-III-C, Project No. W-18-R-2. Division tat improvements. Progress Report, Job No. R-III-E(a), of Forestry and Wildlife, Honolulu. Project No.s W-1S-R-9. Division of Forestry and Wildlife, Paton, P. W. C. (1985) Survival estimates for snowy plovers Honolulu. breeding at Great Salt Lake, Utah. Condor 96, llO6-1109. Telfer, T. C. (1985) Evaluation of endangered waterbird habi Paton, P. W. C., Scott, J. M. and Burr, T. A. (1994) American tat improvements. Final Report, Job No. R-llI-E, Project coot and black-necked stilt in the island of Hawaii. Western No.s W-18-R-6, 7, 9, lO. Division of Forestry and Wildlife, Birds 16, 175-181. Honolulu. Peach, W. J., Thompson, P. S. and Coulson, J. C. (1978) Telfer, T. C. and Burr, T. A. (197S) Description of waterbird Annual and long-term variation in the survival rates of habitats as related to food availability and feeding behavior British lapwings Vanellus vanellus. Journal of Animal of endangered waterbird species on the islands of Kauai and Ecology 63, 60-70. Oahu. Progress Report, Job No. R-IlI-D, Project No. W Pyle, R. L. (1997) Hawaii bird observations March through IS-R-3. Division of Forestry Wildlife, Honolulu. July, 1978. 'Elepaio 39,63. Telfer, T. C., Ueoka, M. L., Morin, M. and Saito, R. S. (1981) Reed, J. M. and Blaustein, A. R. (1988) Biologically signifi Evaluation of endangered waterbird habitat improvements. cant popUlation declines and statistical power. Conservation Progress Report, Job No. R-III-E, Project No. W-1S-R- 6. Biology 11, 281-282. Division of Forestry and Wildlife, Honolulu. Reed, J. M., Doerr, P. D. and Walters, J. R. (1993) Minimum Telfer, T. c., Ueoka, M. L. and Saito, R. S. (1982) Evaluation viable population size of the red-cockaded woodpecker. of endangered waterbird habitat improvements. Progress Journal of Wildlife Management 52, 385-391. Report, Job No. R-III-E, Project No. W-18-R-7. Division Reed, J. M. and Oring, L. W. (1994) Long-term population of Forestry and Wildlife, Honolulu. trends of the endangered ae'o (Hawaiian stilt, Himantopus Thomas, J. W., Forsman, E. D., Lint, J. B., Meslow, E. C., mexicanus knudsem). Trans West Sect Wildl Soc 29, 54-60. Noon, B. R. and Verner, J. (1990) A Conservation Strategy Reed, J. M., Oring, L. W. and Silbernagle, M. (1993) Meta for the Northern Spotted Owl. US Gov. Printing Office, population dynamics and conservation of the endangered Portland, OR. Hawaiian stilt (Himantopus mexicanus knudsem). Transac Ueoka, M. L. and Telfer, T. C. (1980) Limited study of nest tions of the Western Regional Wildlife Society 30, 7-14. ing by stilt on the islands of Maui, Oahu, and Kauai. Pro Reed, J. M., Walters, J. R., Emigh, T. E. and Seaman, D. E. gress Report, Job No. R-llI-C, Project No. W-1S-R-5. (1992) Effective population size in red-cockaded woodpeck Division of Forestry and Wildlife, Honolulu. ers: population and model differences. Conservation Biology Ueoka, M. L., Dalrymple, J. S. and Telfer, T. C. (1976) 7,302-308. Limited study of nesting by stilt on the islands of Maui, Root, B. G., Ryan, M. R. and Mayer, P. M. (1984) Oahu and Kauai. Progress Report, Job No. R-IlI-C, Piping plover survival in the Great Plains. Journal of Field Project No. W-18- R-1. Division of Forestry and Wildlife, Ornithology 63, 10---15. Honolulu. Viability analysis of Hawaiian stilt 45
Walsh, P. D. (1995) PVA in theory and practice. Conservation Western and M. PearI, pp. 7~8. Oxford University Press, Biology',704-705. New York. Wilson, S. B. and Evans A. H. (1893) Aves Hawaiienses: Birds Wu, J., Vankat, J. L. and Barlas, Y. (1993) Effects of patch of the Sandwich Islands. Arno Press (1974), New York, NY. connectivity and arrangement on animal metapopulation Woodruff, D. S. (1989) The problems of conserving genes and dynamics: a simulation study. Ecological Modelling 65, species. In Conservation for the Twenty-First Century, eds D. 221-254.