Genetic Diversity and Population History of the Red Panda (Ailurus fulgens) as Inferred from Mitochondrial DNA Sequence Variations (2024)

,

Article history

Abstract

Introduction

The red panda (Ailurus fulgens) (also known as the lesser panda) is one of the earth's living fossils. Its ancestor can be traced back to tens of millions of years ago with a wide distribution across Eurasia (Mayr 1986 ). Fossils of the red panda have been unearthed from China in the east to Britain in the west (Hu 1990a ). However, due to recent environmental destruction, the red panda is becoming an endangered species and has drawn a lot of attention in the conservation efforts, being rated as one of the flagship species (Hu 1990a ; Wei and Hu 1992 ; IUCN red list of threatened animals, 1996: http://www.wcmc.org.UK/species/animals/animal_redlist.html). The red panda lives in the bamboo forests of the Himalayan and Heng-Duan Mountains. Its current habitat extends through Nepal, Bhutan, Myanmar, and Southwestern China (Tibet, Yunnan, and Sichuan provinces), overlapping with the distribution of the giant panda (Gao 1987 ). Molecular phylogenetic studies showed that as an ancient species in the order Carnivora, the red panda is relatively close to the American raccoon (family Procyonidae) and may be either a monotypic family or a subfamily within the procynonid (Mayr 1986 ; Zhang and Ryder 1993 ; Slattery and O'Brien 1995 ).

Genetic variation in a sample is informative in studying population DNA history. Patterns of mismatch distribution and phylogenetic analyses among genes have been utilized to delineate population processes (Slatkin and Hudson 1991 ; Rogers and Harpending 1992 ; Nee et al. 1994 ; Moritz 1995 ; Glenn, Stephan, and Braun 1999 ). In addition, several methods were also developed to estimate population parameters and to test biological hypotheses (Watterson 1975 ; Tajima 1983, 1989 ; Fu and Li 1993 ; Fu 1994, 1996, 1997 ). Compared with its relative the giant panda, the red panda has not received sufficient attention in population genetic studies, partly due to the difficulty in obtaining large samples for such studies, a difficulty which is also common for many other endangered species. Here, we report the first study of mitochondrial DNA sequence variations in a large sample of red pandas.

Materials and Methods

DNA Samples

A total of 74 samples were collected, including blood samples (16), hair samples (16), and dried leather samples (42). Due to degradation, DNA extractions were successful for only 21 of the 42 dried leather samples (table 1 ). Therefore, the total number of DNA samples was reduced to 53. Both of the two subspecies were included, with five of them being Ailurus fulgens fulgens and the others being Ailurus fulgens styani (table 1 ). The blood and hair samples were obtained from the Chongqing Zoo and Chengdu Zoos of China, and their wild origins were known. Blood samples were anticoagulated with heparin and stored at −70°C before DNA extraction. The hair samples were collected by plucking and stored at −70°C. The dried leather samples were obtained from collections of the Kunming Institute of Zoology, Chinese Academy of Sciences, and stored at −70°C after sampling. The 53 red pandas were originally from 8 different geographic locations in the Sichuan and Yunnan provinces of China (fig. 1 ). Although efforts were made to avoid sampling related individuals, the relationships among animals in the sample were generally unknown.

DNA Extraction, Polymerase Chain Reaction, and Sequencing

DNA extractions from blood samples follow the standard phenol-chloroform method. The fresh hair and dried leather samples were first treated with proteinase K at 56°C for 2 h and then incubated with 10% Chelex 100 (Bio-Rad) at 98°C for 30 min. After centrifugation at a high speed (10,000 rpm) for 10 min, the supernatants were collected and directly used as DNA templates for PCR (Walsh 1990 ). The PCR was conducted by predenaturing at 94°C for 2 min, cycling at 94°C for 1 min, 56°C for 1 min, and 72°C for 1 min for 35–40 cycles, and a final extension at 72°C for 5 min. The primer sequences are CAC CAT CAA CAC CCA AAG CTG (forward) and TTC ATG GGC CCG GAG CGA G (reverse), which amplify a 276-bp fragment located upstream of the mtDNA D-loop region. The PCR products were purified through low-melting-point agarose gel electrophoresis. Sequencing was conducted on an ABI377 automatic sequencer with both forward and reverse primers.

Phylogenetic Analysis and Statistical Tests of Neutrality

For phylogenetic analysis, parsimony (PAUP, version 3.1.1; Swofford 1993 ) and median-joining network analyses (Bandelt, Forster, and Röhl 1999 ) were used. The hom*ologous sequence of the raccoon (Procyon lotor), the closest living relative of the red panda, was included as an outgroup. The pairwise mismatch distribution was generated using Arlequin, version 2.000 (Schneider, Roessli, and Excoffier 2000). The essential population parameter θ was estimated using Watterson's (1975) estimate, Tajima's (1983) estimate, and Fu's (1994) UPBLUE estimate. Watterson's estimate is based on the number of segregating sites among the sequences. Tajima's estimate is based on the calculation of the mean number of pairwise differences of the sequences, while Fu's UPBLUE estimate is done by incorporating the genealogical information of the sequences. A statistical test of neutrality was carried out using Fu's (1997)F_S test. Strictly speaking, all three of these estimators of θ are based on the infinite-sites model (Watterson 1975 ; Tajima 1983 ; Fu 1997 ). Since the sequences generated in this study are from the D-loop region that has mutation hot spots, the infinite-sites model is violated to some extent. To minimize the effect of violation of the model on the estimation of θ, as well as statistical tests of neutrality, we inferred all the required information for parameter estimation and neutrality testing from the parsimony analysis. This was done by first reconstructing a parsimony tree from the sequences and then inferring the required information from the tree. For example, to infer the total number of mutations in the sample, we counted the total number of steps in the parsimony tree. For each pair of sequences, the distance needed for UPBLUE could easily be computed from the parsimony tree as well.

Fu's F_S test of neutrality was used to infer the population history of the red panda. The F_S value tends to be negative when there is an excess of recent mutations, and therefore a large negative value of F_S will be taken as evidence against the neutrality of mutations, an indication of deviation caused by population growth and/or selection.

Results and Discussion

D-Loop Sequence Variations in the Red Panda

A total of 236 bp of the sequence of the D-loop upstream region was generated from the 53 samples, with 22 of them from the Yunnan population and 31 from the Sichuan population. The aligned sequences are shown in figure 2 , including the hom*ologous segment of the raccoon. There are 17 variant sites; 16 of them are transitions and 1 is a transversion (fig. 2 ). A total of 25 haplotypes were obtained from the 53 individual sequences, with 13 from the Sichuan population and 12 from the Yunnan population, respectively (table 2 ). Considering the nonrecombinant nature and high mutation rate of mtDNA, multiple recurrent mutations were responsible for the excessive number of haplotypes observed in the red panda. Among the 25 haplotypes, 18 of them were singletons (9 in Yunnan and 9 in Sichuan), indicating a high level of recent sequence diversity. Gene diversity was estimated to be 0.93 ± 0.02 based on Nei's (1987) method.

Mismatch Distribution and Phylogenetic Analysis

The pairwise sequence difference among the 53 red panda sequences was calculated using Arlequin, version 2.000 (Schneider, Roessli, and Excoffier 2000), and the mismatch distribution is shown in figure 3 . The pairwise differences range from 0 to 12 substitutions. Interestingly, the mismatch distribution is a better fit to a bell-like curve of a population undergoing exponential growth than a typical L-shaped one at equilibrium (Slatkin and Hudson 1991 ; Rogers and Harpending 1992 ). The pairwise sequence differences among the 25 haplotypes and the raccoon sequence are shown in table 3 .

Furthermore, phylogenetic analysis was performed with PAUP, version 3.1.1 (Swofford 1993 ). Based on the parsimony rule, we obtained a total of 13 equal most-parsimonious trees (tree length = 74, tree length among ingroups = 37). The strict consensus tree is shown in figure 4a . As revealed, the consensus tree demonstrated a very shallow phylogenetic structure among haplotypes. The starlike phylogeny in figure 4a again indicates the signature of population expansion in the red panda (Slatkin and Hudson 1991 ; Moritz 1995 ). We also constructed a network using the median-joining method (Bandelt, Forster, and Röhl 1999 ). Similarly, the haplotypes from the Sichuan and Yunnan populations were mixed together, and no phylogenetic inference could be made from the network in view of either geographic distribution or subspecies of the red panda (fig. 4b ).

Tests for Population Expansion

We conducted neutrality tests in two ways. First, all the 53 sequences were considered as one population, in which a total of 13 most-parsimonious trees existed. Second, based on the geographic information, the 53 red pandas were separated into two subpopulations, the Sichuan population (31 individuals) and the Yunnan population (22 individuals). Phylogenetic analyses using parsimony generated 25 and 160 equal most-parsimonious trees for the Sichuan and Yunnan populations, respectively. As explained earlier, special care was made to reduce bias in our analysis by inferring all of the required information from the parsimony analyses. Since hom*oplasy in the data did not seem to be severe (fig. 4b ), the parsimony trees should recover most mutations in the sample, and the influence of hom*oplasy on our analyses should be minimal. In addition, Fu (1994) showed that there is little difference in θ estimates from different most-parsimonious trees. The results of the θ estimations and the neutrality tests are summarized in table 4 .

Fu's F_S test of neutrality, based on 5,000 simulated samplings, was significant at the 5% level (F_S = −7.573) for the total population, a strong indication of population expansion, which was already implicated by the mismatch and phylogenetic analyses. However, when the Sichuan and Yunnan populations were analyzed separately, no significant F_S values were obtained. The F_S value of the Yunnan population was still negative (F_S = −2.283) while that for the Sichuan population was positive. Hence, the Sichuan population seems to be relatively stable, and the Yunnan population shows a tendency for population growth (Fu 1997 ). We also applied several other statistical tests, including Tajima's (1989) and Fu and Li's (1993) tests (results not shown). None of them were able to reject the null hypothesis. This was likely due to a lack of power in these tests for population expansion (Fu 1997 ).

It is interesting to note that different estimators of θ put different weights on mutations occurring in different time periods. The UPBLUE puts heavy emphasis on recent mutations, thus revealing relatively recent population process, while Tajima's estimator put more weights on ancient mutations, therefore reflecting ancient population events (Fu 1997 ). Hence, a comparison of the two estimates could give some clues as to how population size has changed over time. Since θ = 2Nμ for the mitochondrial genome, the ratio of population size change is positively correlated with the θ values given a constant mutation rate. Table 4 shows that for the total population, the UPBLUE estimate is about two times as large as that of the Tajima estimate, indicating that the population size has been at least doubled recently. A similar situation was also seen in the Yunnan population (UPBLUE θ/Tajima's θ = 1.889), but not in the Sichuan population (UPBLUE θ/Tajima's θ = 1.105).

According to the fossil record, the red panda diverged from its common ancestor with bears about 40 MYA (Mayr 1986 ). With this divergence, by comparing the sequence difference between the red panda and the raccoon, the observed mutation rate for the red panda was calculated to be on the order of 10⁻⁹ for the D-loop region, which is apparently an underestimate compared with the average rate in mammals (Li 1997 ). This underestimation is probably due to multiple recurrent mutations in the D-loop region, as the divergence between the red panda and the raccoon is extremely deep.

It should be noted that population expansion may not be the only explanation for a significant F_S test (Fu 1997 ). Other evolutionary forces, e.g., genetic hitchhiking and background selection, can also lead to similar patterns of variation. However, we did not observe any obvious population subdivision in the phylogenetic analysis, and we have not seen any data showing selection pressure on the mitochondrial DNA genome of the red panda, especially considering the noncoding nature of the D-loop region. Furthermore, selection would likely produce similar polymorphism patterns in the Sichuan and Yunnan populations, which is not the case in our observations. Therefore, the data presented in this study suggest that population expansion is the most likely cause of the significant F_S test for the red panda.

It should also be noted that no shared haplotypes were observed between the Sichuan and Yunnan populations. This is probably due to either the sample size in this study or an implication of limited genetic divergence between these two populations, even though it was not observed in the phylogenetic analysis. The Yangtze River, the second largest river in China, lining between the Sichuan and Yunnan provinces could serve as a natural barrier in recent history (fig. 1 ). However, how complete the separation could be is unclear. According to the F_S tests shown above, the effective population size of the Sichuan population is larger and more stable than that of the Yunnan population. Therefore, historically, Sichuan might be the homeland of the red panda, and population growth might have led to a southward expansion to Yunnan.

It is well known that genetic diversity exists in natural populations and is considered the raw material of evolution. When a population grows rapidly, genetic variations will be accumulated and maintained and in the long run will be beneficial to the success of this species. It has been reported that rare and endangered animal species usually show extremely low levels of genetic variation, which were interpreted as one of the critical reasons leading to extinction (O'Brien et al. 1985 ; Su et al. 1994 ; Wayne 1994 ). In this study, we showed that the red panda harbors a considerable amount of genetic variation resulting from both a relatively large effective population size and a recent population expansion, although its population size has been decreasing in the past several decades due to human activity. For the conservation of this endangered species, our results are encouraging. With a high level of genetic variation, the red panda would be more viable than its relative the giant panda, a well-known species with extremely low genetic variation (Su et al. 1994 ). This comparison coincides with the field observation and the ex situ breeding of both endangered animals, for which the newborn death rate is much higher for the giant panda than that for the red panda in the field, and the breeding of the red panda is much more successful than that of the giant panda (Hu 1990a, 1990b ). Therefore, as long as efforts are made to protect the natural habitats, the recovery of red panda populations should be expected.

Supplementary Material

GenBank accession numbers are AF294229—AF294253 (see fig. 2 for the sequence alignment).

Wolfgang Stephan, Reviewing Editor

1

Keywords: red panda mitochondrial DNA D-loop sequence diversity neutrality test population expansion

2

Address for correspondence and reprints: Bing Su, Human Genetics Center, University of Texas–Houston, 6901 Bertner Avenue, Houston, Texas 77030. bsu@sph.uth.tmc.edu .

We are grateful to Dr. David S. Woodruff for providing lab resources for part of the sequencing work. Dr. Ya-ping Zhang provided the primer and the raccoon sequences. We also thank Hongguang Hu, Menghu Wu, Guangxin He, Lisong Fei, and Fuwen Wei for providing samples. This project was supported by the Yunnan Natural Science Foundation, the National Natural Science Foundation of China, and the Chinese Academy of Sciences.

literature cited

Issue Section:

Regular Article

Email alerts

Email alerts

Citing articles via

More from Oxford Academic