Ecological knowledge reduces religious release of invasive species

Abstract

Globally, ceremonial wildlife release events, originated from the traditions of Buddhism and other Asian religions take place in large numbers every year, and have caused conservation concerns. These releases pose a paradox in that Buddhism is generally considered to have great respect for the environment and a desire to cause no harm to any living being, yet Buddhist wildlife releases have resulted in biological invasions. We explored this paradox by evaluating the release of two highly invasive species (American bullfrogs Lithobates catesbeianus and red‐eared slider turtles Trachemys scripta elegans) by 123 Buddhist temples surveyed across four provinces in China and validating survey results with intensive field surveys of release sites. We found that ecological knowledge of invasive species reduced the probability of release; conversely, market availability increased this probability. We suggest that this invasion paradox stems largely from a lack of invasive species knowledge and thus targeted public education about invasive species could be an effective strategy for preventing religious release of invasive species on a global scale.

Introduction

The potential for religion to play a positive role in promoting conservation biology is increasingly a topic of global interest (Bhagwat and Palmer 2009, Dudley et al. 2009, Hall et al. 2009, Bhagwat et al. 2011, Jenkins and Chapple 2011); however, the potentially negative influence of religious wildlife releases on biological invasions poses a conservation concern (Severinghaus and Chi 1999, Agoramoorthy and Hsu 2005, Liu et al. 2012). The release of wildlife for religious purposes has historically been a traditional practice in many religions of Asian origin (e.g., Buddhism, Taoism) and is especially prevalent in Buddhist doctrine (Agoramoorthy and Hsu 2005). Religious releases are most widespread in Asian countries with considerable Buddhist influence, but also occur in Canada, Australia, and the United States (Agoramoorthy and Hsu 2005, Shiu and Stokes 2008), where they are likely much more common than is known and have caused environmental problems (Agoramoorthy and Hsu 2005, Chan 2006, Liu et al. 2012). Although the overall chance of survival of animals released in novel environments may be low (Chan 2006), the chance that they will transmit diseases to wild populations or native species is high (Gilbert et al. 2012). Moreover, release of non‐native species can lead to biological invasions (Sherwood 2001, Agoramoorthy and Hsu 2005, Chan 2006, Agoramoorthy and Hsu 2007, Corlett 2010) and eradication of established invasive species is often impossible (Pimentel et al. 2005). According to the propagule pressure hypothesis of species invasions, high frequency of introduction events or high number of individuals or species introduced per event not only significantly increase the likelihood that introduced species will become established but can also increase impacts on native species (Cassey et al. 2004, Lockwood et al. 2009, Ricciardi et al. 2011). In keeping with this hypothesis, we recently found that high frequency of Buddhist religious release events increased the likelihood of establishment of feral populations of introduced bullfrogs (Liu et al. 2012), confirming that propagule pressure from religious releases can lead to biological invasions. Prevention of introductions is widely accepted as one of the best, most cost‐effective invasive species management strategies and evaluation of potential invasion pathways is an integral part of this approach (Wilson et al. 2009). Therefore, an understanding of factors influencing the release of invasive species by Buddhists could offer insights for timely prevention and conservation strategies.

However, little is known of the factors related to release of invasive species by Buddhists. It is important to recognize that Buddhist doctrine teaches deep respect for the environment, motivated by a desire to cause no harm to any living being (Nasr 1996). Buddhists are quite responsive to suffering and tend to eschew activities that lead to increased suffering. They believe that wildlife releases benefit the living creatures they release and therefore participation in releases could improve the karma of the releaser (Gyatso 2001, Geshe 2002); many may also believe that such good deeds can extend the life of their loved ones or remove life's obstacles. However, negative outcomes of releasing invasive species for Buddhist ceremonial purposes contrast sharply with the intentions of the releasers. Therefore, we hypothesize that these paradoxical negative outcomes could be largely unintentional and likely stem from a lack of ecological knowledge of negative impacts of the invasive species.

In this study, we explored factors in relation to Buddhist releases of two highly invasive model species, American bullfrogs (Lithobates catesbeianus = Rana catesbeiana; hereafter referred to as bullfrogs) and red‐eared slider turtles (Trachemys scripta elegans; hereafter referred to as sliders), in four provinces across China where Buddhist wildlife release is common practice. We selected these two model species for several reasons. First, these species are widely available in live markets and are therefore prime candidates for religious release across China, whereas other, less widely‐available species might only be candidates for local or regional release (Cheung and Dudgeon 2006, Ramsay et al. 2007, Liu and Li 2009). Second, these two species are easily recognizable due to their morphological characteristics—the large body size of the bullfrog and the red “ear” of the slider. Finally, these two species are extremely invasive and information provided by this study could help to prevent introductions of these species and thus contribute to critical efforts to conserve native species (Kats and Ferrer 2003, Ramsay et al. 2007).

Bullfrogs are native to eastern North America and have been introduced in many countries (Kraus 2009). In Mainland China, bullfrogs have been widely cultivated for food since the 1980s and were introduced via escapes from farming operations (Liu and Li 2009). The negative impacts of invasive bullfrogs on native frogs via competition, predation, reproduction disturbance, and disease transmission are well known and were summarized by Kraus (2009). Researchers continually find new impacts of invasive bullfrogs on native frogs; for example, a recent study found that bullfrogs alter the acoustic niche of white‐banded tree frogs (Hypsiboas albomarginatus) and could therefore reduce the reproductive success of this species (Both and Grant 2012). Moreover, recent work has supplied quantitative evidence that bullfrogs contributed to the global distribution of the deadly chytrid fungus that has been linked to amphibian declines (Liu et al. 2013).

Sliders are native to eastern North and Central America but have been extensively introduced to China through the exotic pet trade (Kraus 2009). The negative impacts of sliders on native species are less well known but include potential for genetic impacts on endemic turtles through hybridization (Kraus 2009). Furthermore, multiple enclosure experiments have shown that sliders are higly competitive compared with native turtles (Kraus 2009, Polo‐Cavia et al. 2010); however, negative impacts of competition from sliders have not been demonstrated in natural communities.

In this study we interviewed temple release event organizers at 123 Buddhist temples across China (Appendix: Fig. A1) and surveyed markets and release sites in order to evaluate the relative contribution of release event frequency, level of ecological knowledge of release organizers, temple size, market availability of the model species, human footprint and local resident disposable income to release of the invasive species (Table 1).

Description of factors included as explanatory variables in generalized linear mixed models with release of our model invasive species (i.e., bullfrogs or sliders) as the response variable; for each factor, sampling unit is temple (n = 123).

Methods

Study area

We visited temples (Appendix: Fig. A1) and markets in four geographically and socioeconomically diverse provinces across China (Zhejiang, Hubei, Sichuan, Yunnan) during late April to July of 2008 and 2009. We focused our survey effort in Yunnan province, which boasts a multi‐ethnic population whose diverse religious practices span at least three Buddhist traditions, all of which share a strong custom of wildlife release (Zhao 2001). Yunnan is also a conservation priority due to its location at the convergence of the Indo‐Burma and South‐Central China biodiversity hotspots (Myers et al. 2000). In Sichuan province, our survey effort was reduced when an earthquake damaged roads in 2008.

Temple surveys

Buddhist traditional wildlife releases may be performed by individuals desiring to remove obstacles confronting their families or prolong their loved ones' lives or by large, organized groups of people seeking to improve their karma during temple events or festivals (Fig. 1). We focused on organized release events because the ad hoc nature of individual releases makes them virtually impossible to quantify (Severinghaus and Chi 1999); as a result, it is also impossible to compare the impacts of organized versus individual releases of wildlife.

Typical Buddhist release ceremony held in Shandong Province, China. Upper left: A temple organizer leads followers in prayer while walking to the release site—note that red and orange bins in the van contain animals for release. Upper right: Bullfrogs await release. Lower left: Followers release animals en masse at the release site. Center: A follower reverently touches a bullfrog. Lower right: Recently released bullfrogs swim at the release site. Photographs provided by Zunwei Ke.

We visited Buddhist temples (n = 123; Appendix: Fig. A1) to collect data on temple‐sponsored release ceremonies by interviewing the designated release event organizer, who is solely responsible for fund raising, purchasing animals for release, and record keeping (Liu et al. 2012). We interviewed each release organizer (n = 123; one per temple) using an 8‐question questionnaire designed to address known biases of interview surveys in ecological and conservation research (White et al. 2005) (Appendix), and used bullfrog and slider photos to ensure accurate identification. Although it has been suggested that the general public are usually unable to recognize invasive species without prior education (Somaweera et al. 2010), this was not a concern in our study because both bullfrogs and sliders were familiar, widely available food or pet species (Cheung and Dudgeon 2006, Ramsay et al. 2007, Liu and Li 2009).

In order to obtain objective and accurate responses, we did not reveal any ecological information about the two study species to temple organizers during the interview process. We first asked whether the temple has ever released the species in the photo (Appendix: Q1). Then, to elicit information about the release organizer's ecological knowledge on the two study species, we asked open‐ended questions about differences that might distinguish these species from other species sold in the market (Appendix: Q2–Q3) and about sources from which knowledgeable organizers acquired information about invasive species (Appendix: Q4). Finally, we collected information about temple release sites, release frequency, where released species were purchased, and other freshwater species that have previously been released (Appendix: Q5–Q8). Because temple size might be related to the number of devotees releasing wildlife and thus might influence the amount of funds raised for purchasing animals for release, we obtained information about temple size (i.e., area) from the temple brochure or used GPS (Magellan eXplorist210, Santa Clara, CA, USA) to map temple boundaries for area estimation during the survey.

Release site surveys

In order to validate the accuracy of Buddhist temple organizers' reports of invasive species releases, we also conducted intensive field surveys at release sites each of which was used by only one temple for wildlife release to compare temple reports of releases with observations of the two species at reported release sites, because their presence might indicate release activity by temples. We excluded release sites that were used by multiple temples for release events because it was impossible to identify the source of the animals released by different temples. We used teams of two observers to search for the model invasive species for three consecutive days using standard visual and auditory (bullfrogs only) survey techniques and a stratified random sampling design (Liu et al. 2012). We performed all slider surveys during peak basking hours (9 am–3 pm) (Ramsay et al. 2007) and searched for bullfrogs at night (7 pm–10 pm)—when they are most active and likely to be calling—with the aid of a 12‐volt DC lamp (Jaeger 1994). We excluded from our analysis all sites with history of bullfrog farming within a 1 km radius, in order to reduce confounding effects due to aquaculture farming operations, which are a potential source of introductions (Liu and Li 2009, Liu et al. 2012). Although bullfrog dispersal over distances greater than 1km may be possible (Austin et al. 2004), a previous study based on robust statistical analysis found that distance to the nearest bullfrog farm was not an important predictor of bullfrog presence in our study area (Liu et al. 2012). Consequently, we believe that the 1 km distance was sufficient to control for influence of bullfrog farming activities. We did not find any evidence of current or former slider farming activities in our study areas.

Market surveys

We visited all live markets and pet stores (n = 135) near each temple; markets ranged from small, privately‐owned operations to large, national chains. We categorized model invasive species as available in the temple region (i.e., city or county) if they were sold in at least one market. A small proportion (3.7%) of temples reported purchasing animals for release in the vicinity of distant release sites; we excluded those temples from our analysis to ensure accurate evaluation of species availability.

Socioeconomic factors

We considered the potential for economic influence on release of invasive species by including resident disposable income in the temple region in our analysis; we gleaned these data from the 2009 Chinese National Statistic Year Book (National Bureau of Statistics of China 2009). We also included local human footprint index—a composite variable integrating urbanization, population density, transportation networks, and other human activities—in our analysis to reflect the degree of local anthropogenic influence (Sanderson et al. 2002) (data available at http://www.ciesin.columbia.edu/wild_areas/).

Statistical analysis

We conducted a Chi‐square test to compare occurrence of bullfrogs and sliders, separately, between release and non‐release sites (i.e., according to temple organizers' reports) in order to validate reports of release by temple organizers. We used t‐tests (for continuous variables), Mann‐Whitney U tests (for ordinal variables) and χ² tests of independence (for categorical variables) to compare characteristics between temples that did or did not release the model invasive species. We log‐transformed resident disposable income and temple size variables in order to improve normality before analysis. We used generalized linear mixed models (GLMM) with a binomial error structure and a logit link function to identify which factors were most influential in predicting whether Buddhist temple organizers released our model invasive species. Models used release of at least one invasive species as the binary response variable (i.e., release = 1, no release = 0) and included ecological knowledge, release frequency, temple size, market availability, and regional socioeconomic factors (i.e., human footprint index and resident disposable income) as explanatory variables. Based on survey answers, we classified ecological knowledge as a dichotomous variable for statistical analysis—respondents were categorized as either (1) having ecological knowledge of their impacts (e.g., “these are non‐native exotic species that might have negative effects on native habitat”) or (0) lacking ecological knowledge of these species' impacts (e.g., “they are just the native common frog/turtle similar with other species sold in the market” or “these are not from China, but have no differences with other species in the market”). Because there may be regional similarity in release activities, we included region as a random factor in analyses to account for the possibility of spatial pseudoreplication arising from the presence of multiple temples per region (Bolker et al. 2009).

We created GLMM using all possible combinations of the six variables (i.e., total = 63 models, or 2⁶ − 1) and used a model averaging approach to estimate the effect of each variable. This approach is robust to analyze the relative importance of each variable to the response variable (Burnham and Anderson 2002) and uses an entire set of models to make a more reliable inference than selecting a single best model (Whittingham et al. 2006). We calculated relative importance of each variable by summing Akaike weights across all GLMM based on the Akaike's Information Criterion (AIC) (Burnham and Anderson 2002). Because we were interested in the relative importance of each pruned variable, we excluded models with interaction terms from the calculation of relative variable importance. Furthermore, due to collinearity problems among our predictors (Table 2), we applied hierarchical partition analysis to evaluate the unique and shared variance of each variable (Mac Nally 2002). We determined the statistical significance of the unique deviance of each variable using a randomization test (n = 100) based on an upper confidence limit of 0.95 (a pseudo Z‐scores > 1.65). We conducted all analyses in R (Version 2.15.1, R Development Core Team 2012); we used lmer function in the lme4 package in R to perform GLMM analysis and used the dredge and model.avg functions in the MuMIn package to perform the model averaging analysis. We performed the hierarchical partitioning analyses using the hier.part and rand.hp functions in the hier.part package.

Pearson correlation coefficient (r) among predictor variables. Bold values indicate a significant correlation with a significance (two‐tailed) level of α = 0.01 (**) or α = 0.05 (*).

Results

Species released by surveyed temples

Approximately 71.5% (88/123) of the Buddhist temple organizers surveyed reported releasing at least one of the model invasive species; the remaining temples reported no release of either species. Of the 88 temples releasing these invasive species, 55.7% (49/88) reported releasing bullfrogs and 80.7% (71/88) reported releasing sliders; 36.4% (32/88) reported releasing both species. In addition to bullfrogs and sliders, temple organizers reported selecting various other species for release, including native turtles and fishes such as Asian swamp eels (Monopterus albus), Oriental weatherfish (Misgurnus anguillicaudatus), and several Cyprinid species. Temple organizers reported selecting a balanced proportion of all species available in the market in order to abide by the Buddhist doctrine of equality of all beings; they began to release bullfrogs and sliders as soon as they were available in local live markets. We therefore obtained data on timing of bullfrog and slider availability from local Aquatic Culture Departments and validated these data using published references on the introduction history of the bullfrog and sliders in China (Li and Li 1997, Liu et al. 2010). We determined that religious releases of bullfrogs and sliders in China likely began in the 1980s when a nationwide expanded farming effort began, resulting in widespread availability of both species in local markets.

Release site validation by field surveys

Temple organizers reported selecting release sites based primarily on water body permanence—still water reservoirs and ponds and flowing streams or rivers were preferentially selected over temporary waters that were deemed unfavorable for the released animals' survivals. Due to the importance of this selection criterion, release sites did not change over time. We identified 69 release sites at each of which only one temple released wildlife, including 50 sites with release of at least one invasive species reported by temples, and 19 sites with no invasive species release reported.

Field surveys revealed that bullfrogs were present at the majority of sites where release of this species was reported (13/22 = 59.1%), as well as at 4.3% (2/47) of sites with no release of bullfrogs reported. Similarly, sliders were present at the majority of sites where release of this species was reported (31/41 = 75.6%), and at 32.1% (9/28) of sites with no release of sliders reported. Occurrence of each of these species was significantly higher at sites where temples reported having released that species (Chi‐square test: bullfrogs, χ² = 26.5, df = 1, P < 0.001; sliders, χ² = 12.9, df = 1, P < 0.001), suggesting that the reported release of invasive species by temples was highly indicative of actual release activity.

Factors related to religious release of the model invasive species

A significantly greater proportion of temple organizers reporting no release of these invasive species had some ecological knowledge of these species and their potential impacts than did organizers at temples reporting release of invasive species (χ² = 37.5, df = 1, P < 0.001). Approximately 82% of release organizers from temples that reported no release of the invasive species had some ecological knowledge as compared to only 18% of organizers from temples that reported releasing these invasive species. Resident disposable income in the region of temples that reported releasing invasive species was significantly lower than around temples reporting no release (t test, t = 2.38, P = 0.019). Moreover, market availability of the invasive species was also significantly greater near temples that reported releasing these invasive species (χ² = 21.3, df = 1, P < 0.001) than availability in the vicinity of those reporting no release. However, there were no significant differences in release frequency (Mann‐Whitney U test, Z = −1.94, P = 0.053), local human footprint index (t test, t = 1.10, P = 0.304), and temple size (log‐transformed, t = −0.29, P = 0.754) between temples reporting the release of invasive species and those reporting no release.

Model averaging analysis revealed that the release organizer's knowledge of the invasive species had the greatest relative importance to the release of the invasive species, followed by market availability of the invasive species (Table 3). There was a negative relationship between the probability of invasive species release and the release organizers' level of ecological knowledge of the species. In contrast, market availability and probability of invasive species release were positively related (Table 3). The hierarchical partition analysis confirmed the results of model averaging; release organizer knowledge and market availability were the two most important variables and each had a significant effect on the probability of invasive species release by organizers (Table 3). Together, the two variables explained more than 80% of the total deviance (Table 3).

A summary of model averaging and hierarchical partitioning analyses based on the GLMM with release of invasive species (i.e., yes/no, as reported by each temple) as the response variable and environmental or socioeconomic factors as explanatory variables.

Temple release organizers acquired knowledge of invasive species through several sources. The majority of organizers reported that they acquired knowledge of invasive species from newspaper (43.9%) and internet (25.2%) sources. Additional sources included television or radio (11.4%) and word of mouth (4.9%). Some organizers (14.6%) reported that they acquired ecological knowledge of invasive species from more than one pathway.

Discussion

Our study is the first effort to evaluate factors influencing traditional Buddhist release of non‐native species. Our findings support the hypothesis that paradoxical release of invasive species for Buddhist ceremonial purposes is likely unintentional, and largely stems from a lack of knowledge of the negative impacts of invasive species. Most release organizers surveyed in our study had no knowledge of the invasive species. Naïve organizers were more likely to release invasive species, whereas a much larger percentage of release organizers with ecological knowledge did not release the invasive species. Moreover, both model averaging and hierarchical partition analyses showed that knowledge of the invasive species, which was negatively correlated with release of the invasive species, was the most important factor and plays a key role in influencing Buddhist releases of invasive species.

Higher occurrence of both bullfrogs and slider turtles at release sites suggests that the introduction of invasive species as a result of this traditional Buddhist practice is extensive and widespread in China, corroborating previous findings and concerns raised by previous studies (Agoramoorthy and Hsu 2005, Liu and Li 2009, Liu et al. 2012). The model invasive species were also detected at some release sites where temples reported no invasive species release. These occurrences could be partly the result of dispersal from other release sites and distant farms or isolated releases by individuals, which cannot be quantified accurately (Severinghaus and Chi 1999). Religious release—whether organized or by individuals—is only one of several invasion pathways that have been implicated in introductions of these two model species. In addition to religious releases, introductions via the pet trade and farming for food also contribute to the distribution of these species (Kraus 2009, Liu and Li 2009). Although one previous study found releases from bullfrog farming facilities to be relatively rare (i.e., <2%; Liu and Li 2009), the relative importance of each of these pathways has not been quantified. We suggest that the unique and relative contribution of different introduction pathways to species invasions may warrant future investigation and could enhance efforts to manage and prevent species invasions; however, quantification of ad hoc releases (i.e., religious releases, pet releases) will likely remain impossible without stringent market regulation.

Market availability was another important predictor of invasive species release. Buddhists believe in the equality of all live beings and therefore they usually purchase individuals of all live species available in local markets rather than designate specific species for release (Chen 2005). Bullfrogs and sliders are widely available in live markets across China (Cheung and Dudgeon 2006, Ramsay et al. 2007, Liu and Li 2009) and are therefore liable to be included in most release events, unless religious groups are aware of the ecological impacts of these species. However, it is important to recognize that local presence of these species in the wild (e.g., populations established due to releases) might be related to local market availability and, in turn, likelihood of release during temple events. Therefore, sale of wild‐caught animals in markets could have a circular, confounding effect on studies related to establishment success of released animals; we suggest that future studies should investigate the source of animals for sale in markets (i.e., wild‐caught vs. captive‐bred).

The financial status of Buddhist followers has also been shown to influence animal release events (Severinghaus and Chi 1999). For example, people in middle‐income classes are most likely to participate in prayer animal release in Taiwan (Severinghaus and Chi 1999). Although regional resident disposable income was lower around temples reporting release of the invasive species, this difference in income explained little variation in the release of the invasive species in China, possibly because the temple organizers, rather than local release participants, decided which species to release. However, the influence of resident income on religious release of wildlife may differ among cultures; therefore, we recommend that future studies of religious release in other regions should still consider income as a potential factor. Similarly, temple size and human footprint around a temple were unlikely to influence release of invasive species. However, larger temples might release more animals, and thus influence establishment success of released animals through increased propagule pressure (Liu et al. 2012). Therefore, the relationship between temple size and establishment success of released species may warrant future investigation. Release frequency is an important predictor of establishment of feral populations of invasive species released for Buddhist ceremonial purposes (Liu et al. 2012), but was not related to whether or not a temple released invasive species.

These findings support the importance of the doctrine of equality of all beings in determining use of species for release (Chen 2005); without ecological knowledge of invasive species, all species are released equally, and size of the temple, local income, and degree of urbanization (i.e., human footprint) do not override this doctrine. However, Buddhists' desire to do no harm is very strong and therefore our finding that lack of knowledge of negative impacts of invasive species was the most important predictor of release of invasive species by Buddhist temples implies that dissemination of knowledge can be an effective strategy for preventing releases and may be crucial for future management efforts. The importance of education in preventing release of non‐native species has also been addressed by the Convention on Biological Diversity (CBD); our findings expand the implications of the CBD policy and identify potential conservation implications of reducing impacts of traditional wildlife release practices. Furthermore, our results suggest that joint efforts by government agencies and conservation organizations to foster international educational campaigns targeted to Buddhists might effectively address this introduction pathway.

Campaigns should initially target temple release organizers and Buddhist teachers who prescribe or support release of species, in order to have the greatest influence on reducing releases of invasive species. Our survey results suggest that local newspaper and internet media are important means by which temple release organizers acquired knowledge of invasive species. Therefore, we speculate that dissemination of invasive species information primarily through these media in Buddhist‐dominated communities could prove effective, although television and radio may be especially useful for reaching remote regions where temple organizers have less access to the internet. It is also important to make invasive species information widely available to all who might release non‐native species, including Buddhist followers and pet hobbyists; the use of broad media strategies can help to meet this goal. Finally, traditional release practices must be modernized to incorporate the idea that all species are not equal because some—outside of their native environment—are harmful and therefore known invasive species should not be released under any circumstances.

In light of the large number of religious release events held every year around the world (Liu et al. 2012), which could negatively impact biodiversity, we suggest that there is an urgent need for regulations pertaining to release of non‐native wildlife. Many countries, including China, currently lack such regulations. Regulations should focus on prohibiting release of non‐native wildlife for any purpose, including organized Buddhist release events, and enforcement of regulations is essential to their effectiveness. Because market availability is an important predictor of invasive species release for religious purposes, regulation of market trade in these species is also an important aspect of a comprehensive management plan. In addition, markets are a prime location for spreading information about the potential negative impacts of releasing invasive species. Finally, although we suggest that the findings from this study conducted in Mainland China are likely to be broadly applicable in other regions with a large Buddhist population, additional research will be needed to evaluate regional differences in the influence of Buddhist traditional release events on wildlife invasions.

Acknowledgments

We thank Y. Luo for helping with the fieldwork. Funding was provided by the National Natural Science Foundation of China (31200416, 31172111). We also thank J. Wilson, B. Gratwicke and three anonymous reviewers for constructive comments that have greatly improved this manuscript. This study was conducted under the approval of the Animal Care and Ethics Committee, Institute of Zoology, Chinese Academy of Sciences.

Supplemental Material

Appendix

Questionnaire used to interview release event organizers for factors relate to release our model invasive species at each temple.

• Q1. 

  Have you ever released the species in the photo?

  Yes □ No □

• Q2. 

  Are there any differences of these species with other species sold in the market?

  Yes □ No □

• Q3. 

  What are the differences of these species with other species sold in the market?

• Q4. 

  How did you acquire information about the species in the photo?

  Newspaper □ Internet □ Television (radio) □ Word of mouth □

• Q5. 

  What other aquatic species have you released?

• Q6. 

  Where will you purchase the animals used to be released?

  Local markets □ Markets in the vicinity of distant release sites □

• Q7. 

  How many times do you organize release events each year?

• Q8. 

  What sites do you choose for organized release events?

Fig. A1. Distribution of surveyed temples (n = 123; shown as dark triangles) across four provinces in China; some points are superimposed.