A tail of two pandas- whole genome k-mer signature analysis of the red panda (Ailurus fulgens) and the Giant panda (Ailuropoda melanoleuca)

Abstract

Background: The red panda (Ailurus fulgens) is a riddle of morphology, making it hard to tell whether it is an ursid, a procyonid, a mustelid, or a member of its own family. Previous genetic studies have given quite contradictory results as to its phylogenetic placement.

Results: A recently developed whole genome-based algorithm, the Whole Genome K-mer Signature algorithm was used to analyze the genomes of 28 species of Carnivora, including A. fulgens and several felid, ursid, mustelid, one mephitid species. This algorithm has the advantage of holistically using all the information in the genomes of these species. Being a genomics-based algorithm, it also reduces stochastic error to a minimum. Besides the whole genome, the mitochondrial DNA from 52 mustelids, mephitids, ursids, procyonids and A. fulgens were aligned to draw further phylogenetic inferences. The results from the whole genome study suggested that A. fulgens is a member of the mustelid clade (p = 9·10^{- 97}). A. fulgens also separates from the mephitid Spilogala gracilis. The giant panda, Ailuropoda melanoleuca also clusters away from A. fulgens, together with other ursids (p = 1.2·10^{- 62}). This could be due to the geographic isolation of A. fulgens from other mustelid species. However, results from the mitochondrial study as well as neighbor-joining methods based on the sequence identity matrix suggests that A. fulgens forms a monophyletic group. A Maximum Likelihood tree suggests that A. fulgens and Ursidae form a monophyletic group, although the bootstrap value is weak.

Conclusions: The main conclusion that we can draw from this study is that on a whole genome level A. fulgens possibly belongs to the mustelid clade, and not an ursid or a mephitid. This despite the fact that previously some researchers classified A. fulgens and A. melanoleuca as relatives. Since the genotype determines the phenotype, molecular-based classification takes precedence over morphological classifications. This affirms the results of some previous studies, which studied smaller portions of the genome. However, mitochondrial analyses based on neighbor-joining and maximum likelihood methods suggest otherwise.

Keywords: Giant panda; Mephitid; Mustelid; Pearson correlation; Procyonid; Red panda; Ursid; Whole genome k-mer signature.

Fig. 1

Heatmap depicting group relationships for 28 species based on results from the WGKS algorithm. Brighter colors represent species pairs which are in the same group, with a PCC value closer to 1. Darker colors represent species pairs which are in different group, with a PCC less than 1

Fig. 2

UPGMA-based hierarchical tree for the 28 species based on PCC values. Ursids, felids, and mustelids form separate clades, with S. gracilis in its own group

Fig. 3

Heatmap depicting group relationships for 52 carnivore species based on alignment of the mitochondrial genome using the online MUSCLE software. Brighter colors represent species pairs which are in the same group, with a sequence identity closer to 1. Darker colors represent species pairs which are in different group, with a sequence identity closer to 0

Fig. 4

UPGMA-based hierarchical tree for the 52 species analyzed in the mtDNA study, based on sequence identity metrics. Mustelids and ursids form two large clades, and mephitids, procyonids forming two small groups. A. fulgens and A. fulgens styani appear either to form their own clade, or loosely associate with mustelids

Fig. 5

Hierarchical tree constructed using the Neighbor-Joining method using the MEGA-X software. The bootstrap consensus tree was inferred from 1000 replicates. Branches corresponding to partitions reproduced in less than 50% bootstrap replicates are collapsed. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test are shown next to the branches. This analysis involved 52 nucleotide sequences, with a total of 17,440 positions in the final dataset

Fig. 6

Hierarchical tree inferred using the Maximum Likelihood method and Tamura-Nei model using the MEGA-X software. The tree with the highest log likelihood (− 251,658.67) is shown. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using the Tamura-Nei model, and then selecting the topology with superior log likelihood value. The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The proportion of sites where at least one unambiguous base is present in at least one sequence for each descendent clade is shown next to each internal node in the tree. This analysis involved 52 nucleotide sequences, with a total of 17,440 positions in the final dataset