Cangsong CHEN， Jia YANG， Yunke WU， Zhongyong FAN， Weiwei LU， Shuihua CHEN，*and Lipeng YU

1College of Agriculture and Biotechnology， Zhejiang University， Hangzhou 310012， Zhejiang， China

2Zhejiang Museum of Natural History， Hangzhou 310014， Zhejiang， China

3Department of Ecology and Evolutionary Biology， Cornell University， Ithaca 14853， New York， USA

4The Administration Bureau of Longwangshan Natural Reserve， Anji 313301， Zhejiang， China

?

The Breeding Ecology of a Critically Endangered Salamander， Hynobius amjiensis （Caudata： Hynobiidae）， Endemic to Eastern China

Cangsong CHEN1，2， Jia YANG2， Yunke WU3， Zhongyong FAN2， Weiwei LU2， Shuihua CHEN2，*and Lipeng YU4

1College of Agriculture and Biotechnology， Zhejiang University， Hangzhou 310012， Zhejiang， China

2Zhejiang Museum of Natural History， Hangzhou 310014， Zhejiang， China

3Department of Ecology and Evolutionary Biology， Cornell University， Ithaca 14853， New York， USA

4The Administration Bureau of Longwangshan Natural Reserve， Anji 313301， Zhejiang， China

Hynobius amjiensis is a critically endangered salamander species （IUCN Red List） endemic to eastern China. It currently has three known populations： one in Longwangshan， Zhejiang Province （type locality）， and two in Qingliangfeng between Anhui and Zhejiang Provinces. We examined the relatively unstudied breeding ecology of this species in the fi eld and at laboratory from March 2007 to May 2014. Adult males and females were year-round terrestrial， except for the February-April breeding season. During this period， we captured only a total of 16 breeding adults （11 males and 5 females）. As few as 100 breeding females were estimated based on the number of egg sacs observed since 2007. This number was signifi cantly reduced from the estimated number between 1992 and 1998. Males（mean total length = 16.21 cm， mean body mass = 18.8 g） were slightly smaller than females （16.51 cm， 19.2 g）. Size of breeding pools ranged from 0.2 m2to 1.2 m2（0.1-1.2 m depths）. Each female deposits a pair of egg sacs by attaching the adhesive tips of the sacs to aquatic plants or dead twigs. Fifteen pairs of egg sacs had an average length of 28.6 cm and a diameter of 3.3 cm. On average， each egg sac contained 75 eggs with a diameter of 0.3 cm. Our fi eld survey revealed that H. amjiensis used oviposition sites in small， cool， and weakly acidic pools at high elevations （1 300-1 600 m） where peat moss was abundant. Reduction in wetland size and disappearance of suitable breeding pools suggest that this salamander species is under threat of extinction， particularly at Longwangshan， where 5 of the 9 breeding pools have either dried up or disappeared. Combined size of the remaining 4 pools is less than 2 m2. We urge immediate implementation of more effective conservation measures and suggest that preservation priority should be given to habitat that contains suitable breeding pools.

Amji's salamander， oviposition site， reproductive traits， population decline， habitat destruction， preservation priority

1. Introduction

Breeding ecology of amphibians generally involves oviposition site， breeding period， sex ratio， sexual dimorphism， egg deposition， and so on （Zou， 1993）. Such information is fundamental to the conservation and restoration of rare species or species that are experiencingsignifi cant population decline. The choice of oviposition site is also important in the evolution of a species' life history （Resetarits， 1996）. It will affect development，growth， and the survival rate of the offspring， which determine the success of reproduction （Refsnider and Janzen， 2010； Resetarits， 1996； Skelly， 2001）. Many amphibians can intentionally choose oviposition site instead of laying eggs at random locations to augment the survivorship of their offspring （Crump， 1991； Hagman et al.， 2006）.

Hynobius amjiensis is a salamander species in the family Hynobiidae described from Longwangshan，Zhejiang Province by a Chinese herpetologist， Huiqing Gu （Gu， 1992）. Currently， this species is only known from Qianmutian of Longwangshan （type locality） and two sites on Qingliangfeng between Anhui and Zhejiang Provinces （Fu et al.， 2003； Li et al.， 2013）. IUCN Red List considers H. amjiensis as critically endangered（Gu and Lau， 2004； Wang and Xie， 2004）. Destruction of natural habitat， climate change， and invasion of predators all could contribute to the population decline in H. amjiensis， which may has as few as 92 to 186 reproductive females between 1992 to 1998 （Gu et al.， 1999）. To bring back this species from the edge of extinction， the China State Forestry Administration listed H. amjiensis as an extremely threatened species in 2012's national biodiversity survey. The Department of Forestry of Zhejiang Province and the administrative office of the Longwangshan Nature Reserve also implement conservation measures trying to restore the population. However， critical knowledge about the breeding ecology of H. amjiensis is still incomplete. Gu et al. （1999） documented a few breeding characteristics of this salamander in a discussion of population size and dynamics. Ye （2012） studied ecological factors of the microhabitat utilized by H. amjiensis， which may affect its choice of oviposition sites as well. Because H. amjiensis provides minimal， if any， parental care for their eggs， location of the oviposition site will affect the development， growth， and overall survival rate of the offspring. The lack of correct information impedes the implementation of conservation policy， which includes， for example， what kind of habitat should be given the highest priority for preservation. Therefore，we studied the breeding ecology of H. amjiensis to advance our understanding on its reproduction and breeding characteristics. This study is signifi cant toward the preservation of oviposition sites， increasing larvae survival rate under laboratory conditions， and restoring the wild population.

2. Materials and Methods

2.1 Study sites The study sites were located on Qianmutian of Longwangshan， Zhejiang Province （type locality： 30°23′ N， 119°23′ E） and Qingliangfeng between Zhejiang （site 1： 30°05′ N， 118°51′ E） and Anhui Province（site 2： 30°06′ N， 118°50′ E）. These are major mountains of the Tianmushan Mountain Range， which still has relatively untouched old growth forest with dominant species as oak trees （Quercus fabri， Q. stewardii），rhododendrons （Rhododendron fortunei）， and pine trees（Pinus taiwanensis）. Wetland is found on high elevations surrounded by peat moss and blady grass （Imperata cylindrica）.

2.2 Field investigation and data collection We performed extensive survey for H. amjiensis and its egg sacs in wetland on Qianmutian of Longwangshan from 2007 to 2014 （early February to late March between 2007 and 2011 and early February to early May between 2012 and 2014）， covering the area of about 20 000 m2. We also surveyed the two sites on Qingliangfeng during early February to late April from 2011 to 2014， covering the area of about 10 000 m2. Observance of salamanders or egg sacs suggests population persistence in the surveyed area. We determined 3 physical parameters in study pools： （1） water pH with a portable pH meter （Hanna Instruments， Padova， Italy）； （2） water temperature at a 5 cm depth with a thermometer； and （3） water depth as estimated by inserting a measuring rod. The paired egg sacs were whitish transparent with their adhesive tips attaching to aquatic plants or dead twigs. Thus it was easy to locate the egg sacs in a breeding pool. Length of the egg sac was measured along its cephalo-caudal axis. We counted the number of eggs in each sac and measured their diameters with digital calipers to the nearest 0.01 cm. Because the egg sac does not dissolve until late May and each female only lays one pair of egg sacs each year，the total number of paired egg sacs laid that year would equal the number of breeding females at that site （Gu et al.， 1999）. We sexed all individuals captured during this survey， measured their total length with digital calipers to the nearest 0.01 cm， and weighed them for body mass with a digital balance to the nearest 0.1 g. Because in hynobiid salamanders such as the genera Hynobius and Salamandrella， an aquatic-phase male generally has an elongated tail， we cannot accurately compare total lengths，including tail lengths， between males and females during the breeding season （Hasumi， 2010； Hasumi and Borkin，2012）. We therefore need some caution against comparing the total lengths between the sexes in H. amjiensis. We released all salamanders after examination. We performed preliminary capture-mark-recapture using biochips attached to two males in 2008 but never recaptured them in following years （C-S. Chen， unpublished）. So it is less likely that we examined the same salamander in different years.

2.3 Laboratory observation We conducted two hatching experiments under laboratory conditions in March 2007（11 eggs sacs） and March 2011 （4 egg sacs）. Each time，egg sacs were equally divided and separated into 50containers measuring 27 × 21 × 8 cm3（length × width × height） that had gravels and aquatic plants. Water was 4 cm deep and at room temperature （8.5-22oC）. When larvae hatched successfully， we measured their total length from the 1stweek to the 10thweek with digital calipers to the nearest 0.01 cm. After the larvae completed metamorphosis， we returned them to the original survey sites.

3. Results

3.1 Breeding habitat and time Hynobius amjiensis inhabits wetland at elevations between 1 300-1 600 m. It breeds in small pools （usually ＜ 1 m2； this is the top open area of the pool） with a depth of 0.1-1.2 m. Pool substrate consisted of a thick layer of silt and dead leaves. Aquatic plants were abundant. Water was clear， cold， and weakly acidic （Table 1）. Pools were not covered by tree canopy. Dense peat moss surrounded the pools （Figure 1）. This species seemed to breed only in pools covered with dense peat moss. Small pools in adjacent areas （similar elevations） without peat moss did not have egg sacs. The breeding season started in mid-February and ended in early April. In our surveys， the earliest time we observed newly deposited egg sacs was on February 10 and the latest was on April 8. Most eggs were laid from late February to early March.

3.2 Breeding characteristics When a gravid female was ready to lay eggs， she pushed the egg sacs out of the cloaca only for a small portion， which had the adhesive tip attaching to aquatic plants or dead twigs. The female then wiggled its body forward. This movement pulled out the rest of the egg sacs. During this process， the male grasped the female with his forelimbs， then shifted his forelimbs to the egg sacs， and fi nally pushed the cloacal region of the female away with his hind limbs. This helped pull the egg sacs out of the female. Meanwhile，the male ejaculated seminal fl uid on the sacs. The female left the breeding pool shortly after the eggs were laid，whereas the male stayed for additional 6-7 days.

When being just laid， the pair of egg sacs stuck together， each about 6 cm long and 1.3 cm in diameter（Figure 2）. Soon they started to absorb water and considerably increased in size. The two egg sacs gradually separated from each other. After one week， each fully hydrated egg sac was around 31 cm long and 2.5 cm in diameter （Figure 3）. When larvae were hatched， the egg sac measures 28.2 ± 5.1 cm in length with a diameter of 3.3 ± 0.2 cm （n = 15）. The number of eggs contained in each sac was 75 ± 5 （n = 15）. Individual egg measured 0.3 ± 0.02 cm in diameter （n = 75）.

3.3 Adults measurements During the eight-year survey period， we captured a total of 16 adults. Longwangshan：3 in March 2008， 1 in February 2011， and 2 in February 2013； Qingliangfeng， Zhejiang： 3 in March 2011 and 5 in March 2013； and Qingliangfeng， Anhui： 2 in March 2013. The sample consisted of five females and 11 males. Females measured 16.51 ± 1.12 cm in total length and weighed 19.2 ± 3.1 g. Males （Figure 4） measured 16.21 ± 1.07 cm in total length and weighed 18.8 ± 2.7 g. Therefore， females were generally slightly larger than males. We did not perform statistical tests on the signifi cance of the difference due to the very small sample size.

Figure 1 Two breeding pools of Hynobius amjiensis at Longwangshan.

Figure 2 Egg sacs （just after deposition） of Hynobius amjiensis in the breeding pool.

Figure 3 Several pairs of 12-day old egg sacs of Hynobius amjiensis.

3.4 Breeding population size From 2007 to 2014， the number of paired egg sacs was relatively stable at Longwangshan （Table 2） with an average of 26.75 pairs. Since 2011， we found many more egg sacs at the Zhejiang site of Qingliangfeng. However， the number plummeted from over 90 pairs in previous years to only 65 in 2014. In that year， many egg sacs were either disturbed or pulled out of water by people. There were only a few egg sacs found at the Anhui site of Qingliangfeng. Therefore，given that the number of paired egg sacs should equal to the number of females that reproduce that year （Gu et al.，1999）， the total number of breeding females at the three sites combined is likely only around 100.

3.5 Embryo and larval development We observed embryo and larval development both in the field and under laboratory conditions （15 egg sacs）. After fertilization， gastrulation started in 5-6 days （gastrula stage） and neurulation （neurula stage） occurred in 8 or 9 days. The embryo can be recognized as a larva in 14-15 days. In 17 days， hatchlings started swimming out of the egg sac. Two days later the egg sac was empty. In the fi eld， over 90% of egg sacs hatched successfully； for the 15 egg sacs reared under laboratory conditions， all except one， which totally failed to develop， had a hatching rate over 95%.

Larvae were hatched with three pairs of feathery external gills and one pair of balancers. Their total length measured 1.3 ± 0.12 cm （n = 75）. We did not observe when the balancers were detached from a larva. A bud of forelimbs appeared on day 19， and four digits became visible at the top of the forelimbs 30 days after fertilization. A bud of hind limbs was visible 33 days after fertilization， and digits of the hind limbs started differentiating 47 days after fertilization. Larval growth rate was separated into three stages： （1） relatively slow growth in the first five weeks； （2） an abrupt increase in size between the 5th and 7th weeks； and （3） slow growth again into the metamorphosis stage （Figure 5）. After 7 weeks the external gills started being absorbed； costal groove became conspicuous. In 10 weeks all larvae completed metamorphosis.

4. Discussion

Many species of the genus Hynobius found on in the mainland of China reproduce in the winter. For example，H. guabangshanensis breeds from mid-November to next January （Guo et al.， 2008）； and H. yiwuensis breeds from mid-December to next February （Fei et al.， 2006）. However， the breeding season of H. amjiensis is from mid-February to early April. Although Gu et al. （1999）documented a breeding season between November and March， we found that breeding pools at all three surveyed areas （Longwangshan and two sites at Qingliangfeng）were still frozen in January because of cold weather at high elevations. Therefore， H. amjiensis tends to adapt a late breeding season similar to the Japanese hynobiids，which occur at much higher latitudes （Sparreboom， 2014）.

Figure 4 Male Hynobius amjiensis found at the edge of the breeding pool.

Table 1 Breeding pools used by Hynobius amjiensis at the three survey sites.

Table 2 Number of paired egg sacs observed in each locality during fi eld survey.

Mating behavior of H. amjiensis is similar to that of other species of Hynobius （Guo et al.， 2008； Hasumi，1994； Park et al.， 1996； Sparreboom， 2014） except that male-male competition was unseen probably due to its extremely small population size. However， the use of oviposition site is unique. Comparison among all three study sites indicates that H. amjiensis tends to lay eggs in small pools with clear， cool， and weakly acidic water. The pools need to be surrounded by dense peat moss，under which adults also usually hide. Dense vegetation often provides adequate protection to amphibian larvae（Resetarits and Wilbur， 1989）. Wetland on adjacent mountains without abundant peat moss is not used by H. amjiensis for breeding. This suggests that the use of oviposition sites is not random. Those oviposition sites are all at high elevations， with winter air temperature as low as -20.6oC. Adults of H. amjiensis can be found active inside the breeding pool （even after the surface water is frozen） or near the edge of the pool from December to March. We found that male H. amjiensis may stay in the breeding pool for a few days after mating. Females leave the breeding pool soon after eggs are laid. It is possible that males will wait for additional gravid females to come to the breeding pool. Alternatively， the male may guard the egg sacs from other conspecific adults or predators. However， Hasumi （2015） declared that parental care has not yet been demonstrated in any hynobiid species.

Figure 5 Average larvae growth chart of Hynobius amjiensis.

Our survey indicated that， at Qianmutian of Longwangshan， the number of breeding females was relatively stable （21-35） throughout those 8 years（2007-2014）. However， this number is considerably smaller than what was documented between 1992 and 1998， when as many as 186 females had laid eggs in the breeding pools in 1992 （Gu et al.， 1999）. The wetland size shrunk from 20 000 m2（1998） to merely 10 000 m2（present）. Breeding pools concentrated in the center of the wetland are disappearing. Gu et al. （1999） documented 9 breeding pools at Longwangshan， 5 of which have either dried out or disappeared. The total size of the remaining 4 pools combined is less than 2 m2. Tremendous reduction of oviposition sites as well as adult habitat is likely the largest threat to H. amjiensis. Even after the discovery of additional populations on the adjacent Qingliangfeng， the total number of breeding females is still very low （around 100）. In addition， invasive aquatic macro-invertebrate or fish can compete or prey upon salamander larvae. For example， Asian carps （Carassius auratus） had beenintroduced into one of the breeding pools by local people，who occasionally discarded unwanted fi sh into the water system of the wetland. Fortunately， we stopped this action through rounds of public education on conservation of H. amjiensis in local villages. Furthermore， illegal collection possibly from amateurs and hobbyists poses another threat to this extremely vulnerable salamander species. Gu et al.（1999） suggested that one illegal collection in February 1994 greatly reduced the number of reproductive adults. In February 2014， almost all egg sacs on Qingliangfeng at the Zhejiang side were either disturbed or pulled out of water by people， which may result in complete reproductive failure for that year.

In conclusion， our finding of the characteristics of oviposition sites in H. amjiensis suggests that the highest preservation priority should be given to habitat that contains suitable breeding pools. We need to implement immediate measures to stop further reduction of breeding pools in those known occurrence localities. Given the extreme small number as well as size of those pools， a single environmental event （e.g.， drought） can extirpate the entire species. Invasive species also impose great threat to the survival of larvae through predation and competition. Furthermore， laws against illegal collection must be strictly enforced. Only with combined efforts from the government， research institutes， and public， can this rare of the rarest salamander be brought back from the edge of extinction.

Acknowledgements We thank Shuyan ZHANG and Rui GUO （Zhejiang Qingliangfeng National Nature Reserve）for their field assistance. We are grateful to supporting from the administrative office of Longwangshan Nature Reserve and the administrative office of Zhejiang Qingliangfeng National Nature Reserve. Dr. Jianping JIANG and two anonymous reviewers provided comments that much improved the manuscript. This work was funded by the Zhejiang Provincial Natural Science Foundation （LQ12C06001）.

References

Crump M. L. 1991. Choice of oviposition site and egg load assessment by a tree frog. Herpetologica， 47： 308-315

Fei L.， Hu S.， Ye C.， Huang Y. 2006. Fauna Sinica， Amphibia， Vol. 1. Beijing： Science Press

Fu C.， Rao R.， Wu J.， Chen J.， Lei G. 2003. Effects of density and food availability on growth and cannibalism in basin-raising larval salamanders （Hynobius amjiensis）. Zool Res， 24： 186-190（In Chinese with English abstract）

Gu H. 1992. A new species in the genus Hynobius， H. amjiensis. In：China Zoological Society （Ed.）， Zoological Science Research. Beijing： Chinese Forestry Press， 39-43

Gu H.， Mao X.， Wang J.， Du Z.， Lou X. 1999. Research on population number and dynamics of Hynobius amjiensis. Sichuan J Zool， 18： 104-106 （In Chinese with English abstract）

Gu H.， Lau M. W. N. 2004. Hynobius amjiensis. The IUCN Red List of Threatened Species. Version 2015.2. ＜www.iucnredlist. org＞

Guo K.， Mi X.， Deng X. 2008. Breeding ecology of Hynobius guabangshanensis. Chinese J Ecol， 27： 77-82 （In Chinese with English abstract）

Hagman M.， Shine R. 2006. Spawning site selection by feral cane toads （Bufo marinus） at an invasion front in tropical Australia. Austral Ecol， 31： 551-558

Hasumi M. 1994. Reproductive behavior of the salamander Hynobius nigrescens： monopoly of egg sacs during scramble competition. J Herpetol， 28： 264-267

Hasumi M. 2010. Age， body size， and sexual dimorphism in size and shape in Salamandrella keyserlingii （Caudata： Hynobiidae）. Evol Biol， 37： 38-48

Hasumi M. 2015. Social interactions during the aquatic breeding phase of the family Hynobiidae （Amphibia： Caudata）. Acta Ethol， 18： 243-253

Hasumi M.， Borkin L. J. 2012. Age and body size of Salamandrella keyserlingii （Caudata： Hynobiidae）： a difference in altitudes， latitudes， and temperatures. Org Divers Evol， 12：167-181

Li Y.， Wu X.， Fang G.， Gu C.， Xue H.， Wu X. 2013. Hynobius amjiensis found in Anhui province， China. Chinese J Zool， 48：526-528 （In Chinese with English abstract）

Park S-R.， Park D-S， Yang S. Y. 1996. Courtship， fighting behaviors and sexual dimorphism of the salamander， Hynobius leechii. Korean J Zool， 39： 437-446

Refsnider J. M.， Janzen F. J. 2010. Putting eggs in one basket：Ecological and evolutionary hypotheses for variation in oviposition-site choice. Annu Rev Ecol Evol Syst， 41： 39-57

Resetarits W. J. Jr. 1996. Oviposition site choice and life history evolution. Am Zool， 36： 205-215

Resetarits W. J. Jr.， Wilbur H. M. 1989. Choice of oviposition site by Hyla chrysoscelis： role of predators and competitors. Ecology， 70： 220-228

Skelly D. K. 2001. Distributions of pond-breeding anurans： an overview of mechanisms. Israel J Zool， 47： 313-332

Sparreboom M. 2014. Salamanders of the Old World： The Salamanders of Europe， Asia and Northern Africa. Zeist，Netherlands： KNNV Publishing

Wang S.， Xie Y. 2004. China Species Red List， Vol 1. Beijing：Higher Education Press

Ye L. 2012. Research of the survival status and habitat characteristics of Anji Hynobiid during the breeding season. Master Thesis， Zhejiang Normal University， China

Zou S. 1993. Study contents and methods for the breeding ecology in anurans. Biol Bull， 28： 9-10

*

Shuihua CHEN， curator， from Zhejiang Museum of Natural History， Hangzhou， China， with his research focusing on animal ecology and conservation biology.

E-mail： chensh@zmnh.com

9 July 2015 Accepted： 19 December 2015