kjeb-30-4-2-t1.jpg243.2KB

INTRODUCTION

Species extinctions or declines are deterministic processes of lowering population size by increasing mortality or decreasing fecundity. The reduction of population size involves many factors with a variety of complexities. In particular, small and isolated populations are more at risk of extinction by the additional threats of chance events, such as demographic, environmental, genetic stochasticities, and catastrophic events. Population viability analysis (PVA) uses computer simulations to assess the fates of populations under various scenarios of the deterministic forces and chance events. PVA has been widely used in conservation biology for large mammals, birds, fishes to invertebrates and plants (Nantel et al. 1996; Walters et al. 2002; Legault 2005; Schtickzelle et al. 2005; Haines et al. 2006). Initially, it was designed to estimate a minimum viable population size for a given period of time. However, in recent years the PVA is more extensively used to assess relative impacts of various management options, which is help to determine the most effective management strategy (Reed et al. 2002; Ball et al. 2003; McCarthy et al. 2003). Thus, PVA is a useful tool to conserve threatened species or to recover extirpated species through reintroductions. 

The Oriental White Stork (Ciconia boyciana) is an endangered species with a global population estimated at about 3000 individuals (Wetlands International 2006). They are therefore listed as ‘Endangered’ in the IUCN Red List of Threatened Species and there is great concern about their protection and conservation (Birdlife International 2001). The storks are representative wetland species, foraging in rice fields (these are artificial but well-developed wetlands), riversides, and various types of wetlands. As residents, they are now extinct in Korea from anthropogenic habitat destruction, over-hunting, pollution, food shortage, and other factors (Sonobe and Izawa 1987; Chan 1991; Collar et al. 2001). A re-introduction program for the Oriental White Storks is in progress in Korea. Many efforts for increasing the number of population size in captivity have been made from 1995, which was successful in restoring the populations with 75 individuals in 2009. Now the first re-introduction attempt is scheduled in 2014 at a historic breeding area, Yesan-gun, Chungchungnam-do, Korea. Yesan-gun as the releasing site was chosen based on the habitat suitability models considering land-use changes from the past 30 years and several habitat factors (Kim 2009). 

In general, re-introduction programs with incomplete preparation may cost time, energy and money and be unsuccessful. Developing PVA models under various ecological sceneries will help in estimating the risk of population extinction, identifying risk factors of extinction, and choosing relevant management options. In particular, in case of difficulties to obtain enough data (e.g. demographic field data of locally extinct species) through direct observation, an ideal approach is to adopt adaptive management strategy for optimal decision making to reduce uncertainty, where a baseline model developed based on present knowledge for other populations will be updated as new knowledge is added (Lacy and Clark 1993; William et al. 2002). 

In this paper, we presented a baseline model that incorporates the best available data for a re-introduction of the storks to the Yesan-gun area in 2013. We used this model to determine the sizes of releasing population and risk extinctions under different scenarios. If subsequent releases are necessary after re-introduction, we predicted the size of birds and timing of subsequent releases. As a result, our PVA models may provide insight into population dynamics, possible management actions, and future research hypotheses for the species. 

MATERIALS AND METHODS

1. Population information and PVA modeling

We developed a baseline model to re-introduce a captive breeding population to the Yesan area. We used the program VORTEX (Lacy 1993), which is the most widely used program for modeling low fecundity and long lifespan vertebrate population behavior. The program is individual-based Monte Carlo simulation to the effects of deterministic and genetic events, and each simulation iteration produces different result. Thus, we analyzed the population viability by projecting 1,000 simulations over 100 years for each set of parameters. Data for population modeling for detailed life history and breeding biology have been gathered from a captive breeding population over 13 years at Korea National University of Education, from a re-introduced wild population in Japan over five years, and from a literature review (Table 1). 

Table 1. Information on input values for obtaining a baseline simulation model of a stork population at Yesan area using Vortex.

2. Model parameters

Reproduction: the storks have a long-term monogamous mating system. Data from a captive breeding population show four years of females and three years of males to have first offspring. However, a male started to breed at age two in the re-introduced population of Japan, which was confirmed by gene research. We assumed that maximum age of reproduction was 15, and maximum number of progeny per year was six. From 20 nests (2.5±1.4 chicks/nest) of 11 breeding pairs, sex ratio at birth was 64% in females. From field populations of China and Japan, clutch sizes from 21 nests (3.0±1.6 eggs/nest) were 1 egg, 7.1%; 2 eggs, 10.7%; 3 eggs, 32.1%; 4 eggs, 28.6%; 5 and 6 eggs, 10.7% each. An average of 36.8% females out of total available adult females bred each year in captivity from 2004. 

Mortality: First-year mortality of re-introduced population of Japan was 65.0% (±32) from 14 nests, which quite differed from that of wild population of China as 78.7% (±32) from 10 nests. Thus we used the two data sets to be 70.7% (±31) as the mean first-year mortality rate. Second-year mortality rate was 12.7% (±2.2) from a Japan population, and we assumed that annual mortality rate of males and females older than third years was 10% (±3). 

Catastrophes: a main type of catastrophes in Korea for breeding storks is typhoon, which will severely influence the nest success and young chicks, rather than the survival of the adults because it occurs mainly in July, August, and September. The typhoons with more than 40.0 m sec^-1  maximum wind velocity that will be able to blow off peoples by the wind speed passed by the Korean Peninsula 2 times per 100 year. Thus, we assumed to kill 10% of the storks and to reduce reproduction by 30% during the year. 

Mate monopolization: as storks are monogamous, we specified only percentage of males available to breed per year. In captivity, males of 16.7% (3/18) to 31.8% (7/22) were participated in breeding in 2008, 2009 respectively. We used a plausible mean value of 24.3%. 

Initial population size: we plan to release 10 individuals (5 males and 5 females), where 4 individuals (2 males and 2 females) will be juveniles (second-year males and third-year females) and 6 individuals (3 males and 3 females) will be adults (third-year males and fourth-year females). 

Carrying capacity: Kim (2009) developed a habitat suitability model based on land-use changes and several habitat factors affecting potential nesting sites for re-introduction of the Oriental White Storks in Yesan-gun, where she suggested that a suitable habitat size for the storks within Yesan-gun is 34,985 ha. Approximately 3 or 5 pairs may be able to nest within a 3 km radius (2826 ha) if it is given a relevant habitat, food availability, and nest sites (Yang et al. 2007). Thus, we estimated the carrying capacity of Yesan-gun area to be 150 individuals (62 pairs+25 non-breeding immature storks). 

Other input parameters: we assumed no removing individuals during a simulation for management purposes. However, we planed additional releases in the future to increase gene diversity and to establish population stability in the field, and tested the supplementation with several possible values as a part of sensitivity analysis. As we do not have any available data on the impacts of inbreeding depression, we used default value (3.14 lethal equivalents; Ralls et al. 1988) provided by a VORTEX program. The value indicates 32% reduction in the survival of progeny of full-sib mating. 

3. Sensitivity analysis

To evaluate the sensitivity of PVA to change in the value of the parameters of the model, we performed sensitivity analysis. The sensitivity analysis helps to determine which model parameters play an important role in the population viability by showing how the model responds to small changes in the parameter values of interest. Once the parameters are known, we can develop the most effective management plans. Thus, we examined the sensitivity of stochastic population growth rate (S_s), deterministic population growth rate (S_d), population size (S_p), and extinction rate (S_e) to variation of the following seven variables: first-year mortality, adult female and male mortality, adult female breeding, carrying capacity, and adult female and male supplementation. Sensitivity of each variable was obtained as: 

Sensitivity=(Δx/x)/(ΔP/P)

Where Δx/x is the observed change in each state variable resulting from a change of ΔP/P in the parameter P (Jørgensen 1986). We selected the values of parameter P changing by -20% to +20% to assess sensitivity while holding other parameters constant. The large value of sensitivity indicates that PVA model highly affects to change in a particular parameter.

4. Corrected model

Based on the results of sensitivity analysis, we simulated a corrected model by changing parameters applicable to the population to suggest a best scenario to release the storks in the wild from 2014. 

RESULTS

1. The viability of the Oriental White Stork population

The re-introduced stork population was projected to rapidly increase to 107 birds in 15 years, stayed at over 100 birds for the next 14 years, and then slowly reduced to 49 birds in 100 years (Fig. 1). Mean growth rate (r) was 0.383 (±0.203) during the five years of supplementation and -0.0069 (± 0.1113) during the years without supplementation. The model predicted that the mean final population was female bias with the proportion of 63.5% (31 of 49 birds), and that there was a 12.5% (±0.01 SE) probability of extinction in 100 years with mean time to first extinction of 82.0 (±38.9) year. 

Fig. 1. Predicted population trajectory (±SE) of baseline model for an Oriental White Stork population. The population simulated repeatedly 1,000 times for 100 years. Carrying capacity was set to 150 individuals.

2. Sensitivity analysis

Sensitivity analysis showed that the baseline model population was most sensitive to two demographic variables: firstyear mortality and percent of adult female breeding, where stochastic population growth rate (S_s) and population size (S_p) were most sensitive to the percent of adult female breeding while deterministic population growth rates (S_d) and extinction rate (S_e) were most sensitive to the first-year mortality (Fig. 2, Table 2). In particular, the increasing level of firstyear mortality from a baseline model was more sensitive than the decreasing level of first-year mortality in three state variables: S_s, S_p, and S_e. Second level of sensitivity appeared in first-year mortality, percent of adult female breeding, adult female supplementation, and adult male mortality for S_s, S_d, S_p, and Se respectively. Changes in the adult female mortality, adult male and female supplementation on S_s , adult male mortality and carrying capacity on S_d, and adult female mortality and adult male supplementation on S_p  had relatively little effect on population risk. 

Table 2. Sensitivity analysis of seven parameters for stochastic population growth rates (S_s), deterministic population growth rates (S_d), population size (S_p), and extinction rate (S_e) with constant values of parameters changing by -20% to +20%. The sensitivity levels of five parameters for each state variable are shown graphically in Fig. 2.

Fig. 2. Comparative results of the Sensitivity Tests for the Oriental White Stork population predicted from the release in the field. The data graph shows (A) stochastic population growth rates, (B) deterministic population growth rates, (C) population size, and (D) extinction rate for models varied in levels of parameters by -20% to +20%. FYM: First-year mortality; FM: Adult female mortality; MM: Adult male mortality; PB: Percent of adult female breeding; CC: Carrying capacity; FS: Adult female supplementation; MS: Adult male supplementation.

3. Corrected model

We changed 1) first-year mortality from 70.7% to 65% (a re-introduced population of Japan), and 2) percent of adult female breeding from 36.8% (average value of 2004 to 2009) to 47.3% (2009 only). The corrected model population simulated repeatedly 1,000 times for 100 years was projected to increase to 131 birds in 22 years, stayed at over 100 birds for the next 55 years, then slowly reduced to 91 birds in 100 years (Fig. 3). The corrected model induced higher mean growth rate with 0.387 (±0.204) during the five years of supplementation and with 0.0121 (±0.0989) during the years without supplementation than the baseline model. Thus, the model predicted much lower probability of extinction (1.00% ±0.0031 SE) in 100 years with mean time to first extinction of 81.4 (±14.4) year. 

Fig. 3. Predicted population trajectories (±SE) of baseline model and corrected model for an Oriental White Stork population. The population simulated repeatedly 1,000 times for 100 years. Carrying capacity was set to 150 individuals.

DISCUSSION

Re-introduction is an integrated process that controls all potential risk factors to establish a species into an area (IUCN 1998). Understanding extinct risk and developing relevant management plans are prerequisite to guide reintroductions of extirpated species. The baseline model output suggests that the Oriental White Stork population could be driven to extinction ultimately under existing conditions. The population size was high enough to increase from 10 re-introduced birds to approximately 107 birds in 15 years. However, after the years the population was steady and slightly declining. This prediction assumed ten birds (5 females and 5 males) were supplemented every year for the next five years, and thus, mean growth rates in the model quite differed between the years with and without supplementation. 

Sensitivity analyses showed that two demographic variables (first-year mortality and percent of adult female breeding) had the greatest impacts on population persistence (S_s, S_d, S_p, and S_e). Corrected models that reflected the significance of the two demographic variables (-5.7% of first-year mortality and +10.5% of adult female breeding) predicted thev long-term population stability to be quite improved, where the extinction probability decreased from 12.5% to 1.0%. Unlike the effects of the two demographic variables, the variation in the carrying capacity and adult supplementation had much less influence on population persistence, which means relatively more changes in those variables than -20% to +20% are necessary to have the similar effects on population persistence. Thus, we should pay more attention to reduce first-year mortality and increase percent of adult female breeding for the long-term viability of this population. 

Our models have weakness in two ways: first, we considered habitats to be stable even if the potential environmental catastrophic assumed to be and habitat loss or modification is most common factor for limiting distribution of any species through time and space (Hoekstra et al. 2005). One reason is because no direct information is available. Another reason is that we are preparing for suitable habitat for food and nesting sites selected by the research results on the habitat suitability model for the re-introduction of the storks (Kim 2009). Habitat preparation includes establishment of fishways, organic farming, and biotope restoration of rice fields, which will be phased in before releasing in the field. Second, we do not know genetic effects on small population size, such as inbreeding depression, homozygosity. Even if some captive animals are inbred (Myroniuk and de Alvis 1990), we continuously imported wild storks from Russia (24%, 18/ 75) or exchanged individuals or eggs to other institutes. Till now, we restricted formation of breeding pairs with a genealogical tree in captivity. 

PVA simply reflects the natural world, and the accuracy of PVA depends on data quality. Even if high-quality data should be used to develop models, it is quite difficult to collect them adequately. Thus, the models should be repeated and reexamined with additional field data to improve the accuracy of PVA, which is essential to adaptive management as well as understanding of population dynamics in specific areas (Holling 1978; Salafsky et al. 2001). Based on the results of PVA, we need guidance for decision making to develop conservation and management plans, for which information about relative impacts of various deterministic and stochastic factors on the re-introduced population viability is necessary. 

At present, our models for stork re-introduction indicate this population will be stable by improving first-year mortality and adult female fecundity. In particular, as the population was predicted to be unable to reach carrying capacity (150 birds), management options could be further explored with alternative release strategies by not only improving the two demographic variables, but also increasing initial population size (e.g. 20 birds) or the number of adult supplementation with longer periods. 

ACKNOWLEDGEMENTS

This subject was supported by the Korea Ministry of Environment as “The Eco-technopia 21 project” and carried out with the support of “Cooperative Research Program for Agriculture Science & Technology Development (project No. PJ007522)”, Rural Development Administration, Republic of Korea.