Biodiversity Tasmanian tiger. What happened to the Tasmanian tiger? What did the Tasmanian tiger really eat? The Last Tasmanian Tiger As the Tasmanian tiger population started diminishing, people realized what was happening. Choose one of the following categories to see related pages: Endangered Animals. Share this Page. You can follow Lindsay VanSomeren Twitter. Common Name: Tasmanian tiger. Classification Kingdom:. Species Range. Science Newsletter:. Full List of our Videos.
Teaching Biology? How to Make Science Films. Read our Wildlife Guide. On the Trail of the Egret. Tips for Shooting Smoke Grenade Photos. Pacific Sleeper Shark: Giant of the Deep. The Burmese Python - A docile ish giant. Australia's Most Dangerous Creatures. White-nosed Syndrome in Bats. Gluten and You.
Arctic Tundra Biome. The Science of Breadmaking. How does water move in plants? What is the molecular clock? Plants Can Make Memories. How does your stomach work?
Whale Poo and You. Bringing Back the Woolly Mammoth. Learn to Love Lichens. Where did the Zika Virus come from? Snow Leopard: Ghost of the Mountains. How long can people live? Make a Handheld Gimbal out of your Phantom 3 Drone. How the Body Adjusts to Altitude. Collecting Bugs in Sweden. Although the large head was dog- or wolf-like, the tail was stiff and the legs were relatively short. Body hair was dense, short and soft, to 15mm in length. It had short ears about 80 mm long that were erect, rounded and covered with short fur.
Jaws were large and powerful and there were 46 teeth. Adult male Thylacine were larger on average than females. The female Thylacine had a back-opening pouch. The litter size was up to four and the young were dependent on the mother until at least half-grown.
Interestingly, males also had a back-opening, partial pouch. The Thylacine was mainly nocturnal or semi-nocturnal but was also out during the day. The animal moved at a slow pace, generally stiff in its movements. The Thylacine hunted singly or in pairs and mainly at night. Thylacines preferred kangaroos and other marsupials , small rodents and birds. They were reported to have preyed on sheep and poultry after European colonisation, although the extent of this was almost certainly exaggerated.
For example, this was perpetuated, intentionally or otherwise, by a series of famous photos taken by Harry Burrell. Get our monthly emails for amazing animals, research insights and museum events. At one time the Thylacine was widespread over continental Australia, extending north to New Guinea and south to Tasmania.
In recent times it was confined to Tasmania where its presence has not been established conclusively for more than seventy years. In Tasmania the species was best known from the north and east coast and midland plains region rather than from the mountains of the south-west. Although the precise reasons for extinction of the Thylacine from mainland Australia are not known it appears to have declined as a result of competition with the Dingo and perhaps hunting pressure from humans.
The Thylacine became extinct on the Australian mainland not less than years ago. CREs such as promoters and enhancers regulate spatial and temporal gene expression and have a modular structure. Consequently, many CREs may have greater evolutionary flexibility than their target genes and these elements have been suggested as preferred targets of natural selection Cis -regulatory evolution may be particularly important for driving morphological adaptations 43 , which involve changes in body patterning.
We suggest that future genome-wide studies on the molecular basis of convergent phenotypic evolution should explore the contribution of cis -regulatory changes to this phenomenon. The Carnivora is divided into two suborders, the Feliformia and Caniformia, which together comprise over living species Among these are the well-known large-bodied carnivores such as lions, wolves and hyenas, but also omnivores some bears , insectivores aardwolf and bat-eared fox and a diversity of small- to medium-sized mustelids, raccoons, civets, mongooses and others with varying ecologies.
Members of the other major clade of carnivorous eutherians, the extinct Creodonta, ranged from cat- to hyena-sized and were the most successful and abundant predators of the early Tertiary period, an ecological niche later dominated by carnivorans Among metatherians, carnivory here defined as the killing and comminuting of vertebrate prey, sensu 46 has evolved independently in the extinct stem order Sparassodonta for example, Thylacosmilus atrox from South America, the extinct Australian diprotodontian family Thylacoleonidae, and in members of the Australian order Dasyuromorphia.
To match morphological and ecological variation in carnivorans, they included the insectivorous numbat Myrmecobius fasciatus and other generalists outside of Dasyuromorphia, plus five extinct marsupials: the thylacine and Miocene Nimbacinus dicksoni Thylacinidae ; the highly specialized predator and largest carnivorous Australian mammal Thylacoleo carnifex comparable to a large felid and its smaller more primitive relative Wakaleo vanderleuri , both thylacoleonids; and the most morphologically conservative dasyurid known, Barinya wangala 5.
Here we further expand that 3D cranial dataset 6 by an additional 35 species, including members of Canis and Vulpes the eutherian genera to which the thylacine is most commonly compared 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , dasyuromorphs and New World didelphids, 8 families of Diprotodontia the clade containing carnivorous thylacoleonids but also kangaroos, wallabies, koalas, possums, gliders, and wombats , and other insectivorous australidelphids from the orders Microbiotheria, Notoryctemorphia and Peramelemorphia.
The majority of newly added taxa are relatively small 0. Herbivorous diprotodontids were included to increase metatherian morphospace to a level comparable to that of the sampled eutherians, since marsupial crania have been shown to be significantly less disparate in shape than for eutherians They were also sampled to improve ancestral state reconstructions of cranial shape for Diprotodontia, which were used in convergence tests between the thylacine and other carnivorous marsupials such as T.
Additional landmarks were generated from 28 newly sampled museum specimens and 13 cranial surface models made available through DigiMorph. All scanned specimens were adult males unless otherwise noted Supplementary Table 4. Landmarks were placed on the newly generated models in Landmark Editor Institute of Data Analysis and Visualization, UC Davis, USA , with landmark locations matching those in the previously published analysis 6 , so that the datasets could be combined.
In total, thirty homologous landmarks were digitized on the dorsal and lateral surfaces of the cranium Supplementary Fig. Because the goal of the original study was to examine cranial convergence related to diet and feeding ecology, landmarks were focused on facial, dental and zygomatic regions, including muscle attachment sites such as the sagittal crest 6.
However, we note that landmarks on the palatal surface are lacking, which may miss other aspects of cranial morphology related to diet. The final combined data set consisted of individuals from species, including two representatives of the thylacine Supplementary Table 4. Geometric information was extracted from the landmark coordinates by a generalized Procrustes fit The resulting Procrustes coordinates representing the symmetric component of shape variation after translating, scaling and rotating all individuals to a common average were extracted and used as shape variables in all analyses.
The asymmetric component, which is typically the focus of studies on developmental integration and modularity 56 , was disregarded. Individual size information, preserved in the centroid size, was calculated as the square root of the sum of squared distances of landmarks to the centre of their configuration.
Species with multiple specimens in the dataset were represented by their mean Procrustes shape and centroid size CS values. To examine evolutionary trends in cranial morphology, we generated a simplified phylogenetic tree of mammals using a direct supertree approach Published phylogenies of eutherian and metatherian species were pruned to match our taxon sampling and combined at the root using the bind.
Extinct taxa were placed according to published morphological analyses and, in some cases, phylogenetic analyses including ancient DNA Supplementary Fig.
Although phylogenetic positions of many of the fossils are based on cranial material, and are therefore not strictly independent of the cranial shape data, the majority of coded cranial characters are highly detailed and involve qualitative descriptions of specific cranial elements not likely to be captured in our landmark data.
Dominant patterns of cranial shape variation were identified by principal component analysis PCA. To determine evolutionary patterns of cranial shape change, species PC scores were mapped onto the phylogeny and weighted-square change parsimony was used to reconstruct the morphological state of ancestral nodes The mapped tree was then projected back into morphospace to visualize patterns of phenotypic evolution.
Inspection of the PCA plots, together with examination of relative changes in landmark positions along each axis aided in interpretations of evolutionary changes in cranial shape. For statistical analyses of phenotypic variation, we constructed analysis of variance models for individual effects based on the Procrustes distance among species, known as Procrustes ANOVA Each species was assigned to a general diet type carnivore, herbivore, insectivore, omnivore; Supplementary Table 4 , and the relative effects of size represented by centroid size and diet on cranial shape were compared.
Given the extensive range of body sizes in our data set, centroid size was log-transformed prior to analysis Due to the non-independence of related species 64 , tests were also run in a phylogenetic context using our mammalian supertree Supplementary Fig.
A distance-based phylogenetic generalized least squares model D -PGLS , equivalent to phylogenetically independent contrasts 65 , was constructed separately for each of the above effects. Shape data were permutated across the tips of the phylogeny, and trait values predicted under a Brownian motion BM model of evolution were compared to those observed to assess statistical significance. This model assumes that phenotypic changes are independent from time step to time step, and that phenotypic variation increases proportionally with time A complete list of statistical results is given in Supplementary Table 6.
Under this approach, convergence is implied when two or more taxa have evolved to be more similar to one another than their ancestors were to each other Unlike other convergence methods like SURFACE 68 , which identify convergent shifts in selective regimes among lineages, C 1 -C 3 measure the increase in phenotypic similarity between taxa compared to that between the most divergent species in their lineages, without assuming an adaptive process.
Therefore, the more dissimilar the ancestor and the more similar the descendants, the greater the strength of the convergence. Values range from 0 to 1, with 0 indicating that lineages are as dissimilar as they have ever been, and 1 meaning complete evolutionary convergence, that is, descendants are indistinguishable. From C 2 , the proportion of convergence relative to the total evolution L tot. For all measures, values closer to 1 indicate greater morphological similarity.
C 1 —C 3 were estimated for all pairwise comparisons between the thylacine and species of Canis and Vulpes , the two eutherian lineages which the thylacine most superficially resembles 46 , 47 , 50 , We also estimated convergence values between the thylacine and its closest carnivorous marsupial relatives, the extinct Nimbacinus dicksoni and Barinya wangala , as well as with the closely related insectivorous taxon, Myrmecobius fasciatus Fig.
Statistical significance of each measure was determined by simulation of the parameters derived from the observed data on our phylogeny using a BM model of evolution As an independent assessment of phenetic relationships among taxa, we additionally conducted an unweighted pair group average UPGMA hierarchical cluster analysis on the matrix of Procrustes distances between species using PAST software v3. All other analyses were performed using MorphoJ v1.
The sampled individual was a female, approximately 1 month old at the time of death. She was part of a litter of four young, three of which, C male , C female and C female are preserved intact at Museums Victoria.
The fourth pouch young, C male was sectioned for microscopy in We extracted and purified genomic DNA using a method similar to that described previously Samples were then centrifuged and the aqueous phase removed to a new tube; this procedure was repeated until the interphase was clear. The isolated DNA was fragmented as expected 73 , and contained kilobase length fragments as determined by electrophoresis using the BioAnalyzer high sensitivity DNA assay Agilent; Supplementary Fig.
In short, thylacine genomic DNA was sheared with a Covaris ultrasonicator. The libraries were then amplified with 6 cycles of PCR modified from 8 cycles. Data were depleted of microbial and mitochondrial sequences by mapping to two databases one containing all Ensembl microbial and fungal genomes and another containing the thylacine and human mitochondrial genomes using bwa mem v.
Reads that mapped to the contaminant database were removed from the dataset and not subjected to further analysis. To further validate that retained reads were composed largely of thylacine DNA, reads were mapped against the repeat-masked genomes of three extant marsupials: the Tasmanian devil Sarcophilus harrisii 10 , the tammar wallaby Macropus eugenii 11 and the gray short-tailed opossum Monodelphis domestica 12 as well as the complete, unmasked genome of the Tasmanian devil assembly versions given in Supplementary Table 8.
The percentage of reads mapping uniquely to each reference genome was graphed Supplementary Fig. The vast majority of reads mapped to the unmasked assembly of the Tasmanian devil, the closest relative to the thylacine available, indicating very low levels of microbial contamination. The percentage of mapped reads was observed to decline with increasing phylogenetic distance between the reference genome and the thylacine, indicating that the majority of reads were thylacine in origin Supplementary Fig.
Briefly, residual adaptor sequences were trimmed and reads shorter than 25 nucleotides were removed, and overlapping paired reads were collapsed using AdapterRemoval v. Mapping of collapsed and paired reads was performed using the mem algorithm with default parameters in bwa v.
Reads with mapping quality below 25 were removed. Finally, mapDamage 2. The only damage pattern characteristic of ancient DNA sequence data was a slight increase of depurination G only immediately before the reads start Supplementary Fig. A reference-based assembly was generated by producing a consensus sequence of thylacine read data using samtools v0. For comparative data, reference-based assemblies were also produced for five wild canids wolf, coyote, golden jackal, red fox and arctic fox.
The same pipeline was used to produce the wild canid reference-based assemblies as was used for the thylacine, however data was mapped to the repeat-masked dog genome Ensembl CanFam3.
These elements were previously selected to computationally screen the genome sequences of dasyuromorph species for phylogenetic markers, and returned 25 diagnostic markers The corresponding loci of these markers were used in the present study as queries for blastn BLAST version 2.
All but 3 of them had orthologous sequences in the thylacine. To search for additional diagnostic markers to further clarify the phylogenetic position of the thylacine within Dasyuromorphia, we also conducted bioinformatics and experimental screens to test all possible tree topologies:. To test hypothesis no. We randomly selected 75 of the most conserved to enable reliable PCR loci from each of the three datasets. These loci were manually aligned, and sequences from the short-tailed opossum taxid , the Tammar wallaby taxid , and the koala taxid were aligned along with these where possible Supplementary Table These three hypothesis-testing screens yielded 11 phylogenetically informative markers present in the numbat and dasyurids but absent in the thylacine Supplementary Fig.
To remove X-linked scaffolds, we started with a list of 2, putatively X-linked scaffolds in the tammar wallaby reference MEUG3. By mapping Tasmanian devil scaffolds to the tammar wallaby reference we could therefore identify X-linked scaffolds for the Tasmanian devil and thylacine.
Pask, personal communication resulted in larger numbers of scaffolds remaining unmatched 25, of 35, We split individual scaffolds at the occurrence of one or more Ns. Resulting contigs were named for their scaffold of origin with an additional numerical suffix for example, the 12th contig in scaffold GL The resulting whole-genome alignment was filtered to retain contigs that mapped to one of the MEUG3.
This process identified putatively X-linked devil scaffolds. Given the small size of marsupial sex chromosomes 94 , we expect that this list contains a high number of scaffolds falsely marked as X-linked. The exclusion of these scaffolds is thus considered to represent a conservative autosomal set. After the removal of short and X-linked scaffolds we were left with scaffolds, which had mean read depths for the Tasmanian devil and the thylacine of Scaled population sizes, and the times to which they apply, were inferred using MSMC v1.
MSMC produces an output that is scaled by the per-generation mutation rate, requiring estimates of the generation time and the per-generation mutation rate in order to rescale plots into real time. We assume a generation time of three years for both the Tasmanian devil and the thylacine.
We first attempted to estimate the per-year mutation rate by dividing the number of observed substitutions between Tasmanian devil and thylacine by twice the time to their most recent common ancestor. Because of the large divergence time and low contiguity of the Tasmanian devil reference, only conserved regions of the genomes map to each other and as a consequence this strategy produced an unreasonably low estimate 1. In mammals, the number of recombinations per-generation is strongly predicted by the number of chromosome arms The Tasmanian devil has six metacentric autosomes, leading to the expectation that we should observe 12 recombinations per-generation.
For consistency with the scaled recombination rate estimate, we also infer the demography from the 50th iteration. By multiplying the ratio of the scaled mutation rate to scaled recombination rate by the per-generation recombination rate implied by the karyotype we were able to estimate a per-generation mutation rate for scaling the MSMC output.
The final mutation rate estimates are:. Therefore, we expect that the true thylacine and Tasmanian devil mutation rates may be slower than those we have estimated above. Mutation rate scaling is linear, and the effect of scaling with a slower mutation rate is to shift the demographic curve farther back in time and higher up on the population size axis. Consequently, the reconstructions for each species in Fig.
A typical MSMC bootstrap is performed by first concatenating MSMC input files for all chromosomes, then sampling chunks with replacement until the sum of the sampled chunks is an equivalent genome size. A script for this can be found in the msmc-tools git repository. This bootstrap strategy accounts for: 1 variability in demographic inference along the chromosomes; and 2 variability due to a small rate of false recombinations, induced at chromosome boundaries and at chunk boundaries.
For a chromosome level reference assembly, the impact of 2 should be minor. Thus, the impact of 2 for our assembly is likely to be major, and not representative of falsely detecting recombinations from our data. We instead choose to sample scaffolds, with replacement, until the total original data size is reached. Because 1 should also include variability in erroneously detected recombinations, we believe this resampling strategy is more appropriate for data mapped to a scaffold level assembly.
We did bootstrap replicates for each of the Tasmanian devil and the thylacine. Thylacine and wild canid protein-coding DNA sequences CDS were extracted from the reference-based assembly using gffread and the Ensembl 84 annotations of the Tasmanian devil and dog genomes respectively. All CDS sequences for high-confidence one-to-one orthologues to Tasmanian devil genes from 21 vertebrates inclusive of the devil were downloaded from Ensembl 84 BioMart Supplementary Table 8 and filtered for the best available CDS sequence per gene using custom scripts If no complete CDS was available, the longest sequence beginning with a start codon was accepted.
Otherwise the longest available partial sequence was selected. Filtered sequences were grouped by orthology using ParaAT1. In total, 11, groups of orthologous genes were retained for comparative genomic analyses Supplementary Table 9. We used published phylogenies of all included species to construct a fixed tree topology , used in all comparative genomic analyses Supplementary Fig.
To test the frequency of homoplasious substitutions across mammals, we collected all genes from our previous analyses which contained representative sequences for all therian mammals and the platypus as an outgroup.
Gene trees for the genes retained after filtering were used to construct a consensus tree with Mesquite v3. We then counted the observed number of homoplasious substitutions for all pairwise comparisons between all nodes that do not share a most recent ancestral node in the fixed tree topology.
We then compared it to the number of expected homoplasious substitutions under the JTT-F site model 29 , using a python script generously provided by Z.
Zou and J. Zhang University of Michigan. To identify genes containing convergent and parallel substitutions between the thylacine and the canids, ancestral protein sequences were reconstructed using CodeML 31 Supplementary Table 25 using the fixed tree topology described above Supplementary Fig.
The thylacine was compared to the ancestral sequence for all canids, as this represents the node in which adaptive substitutions are likely to have resulted in features shared by all canids. We used previously described definitions for parallel and convergent changes Briefly, for each amino acid position, the thylacine and the reconstructed ancestral sequence for the extant canids were compared to each other and to their most recent ancestors respectively.
Positions that were identical between the thylacine and canid sequence, but different from both of their respective ancestors, were deemed to be homoplasious. Homoplasious substitutions in which the ancestors shared an identical residue were considered to be parallel, while residues that differed between the ancestors were considered to be convergent.
The modified model A Supplementary Table 26 was compared to the null Supplementary Table 27 assuming the fixed tree topology described above Supplementary Fig. Four likelihood ratio tests were performed on nucleotide alignments of orthologous genes using different foreground branches; the thylacine, the branch leading to the subfamily Caninae, the Tasmanian devil and the branch leading to the family Bovidae the latter two acting as controls.
The total number of genes analysed for each foreground branch varies slightly thylacine: 10,; Caninae: 10,; Tasmanian devil: 10,; Bovidae: 9, due to the lack of representative sequences available for a given foreground branch in some alignments, and because PAML discards all alignment columns containing gaps.
Genes under positive selection in multiple lineages were found by comparing positively selected genes from each foreground branch Supplementary Tables 12 — 13 , 15 — 18 and 28 — The resulting lists of HGNC symbols for the thylacine and canids were then separately analysed for enrichment of KEGG pathways, the results of which are contained in Supplementary Fig. Further information on experimental design is available in the Life Sciences Reporting Summary.
Custom PERL scripts were used for data parsing and to identify convergent and parallel amino acid substitutions. Paddle, R. Press, Cambridge, Lambeck, K.
Sea level change through the last glacial cycle. Science , — Bininda-Emonds, O. The delayed rise of present-day mammals. Nature , — Losos, J. Convergence, adaptation, and constraint. Evolution 65 , — Article Google Scholar. Wroe, S. Convergence and remarkably consistent constraint in the evolution of carnivore skull shape. Evolution 61 , — Goswami, A. Biting through constraints: cranial morphology, disparity and convergence across living and fossil carnivorous mammals. B , — Menzies, B.
Limited genetic diversity preceded extinction of the Tasmanian tiger. Miller, W. The mitochondrial genome sequence of the Tasmanian tiger Thylacinus cynocephalus. Genome Res. Krajewski, C. Phylogenetic relationships of the thylacine Mammalia: Thylacinidae among dasyuroid marsupials: evidence from cytochrome b DNA sequences.
B , 19—27 Murchison, E. Genome sequencing and analysis of the Tasmanian devil and its transmissible cancer.
Cell , — Renfree, M. Genome sequence of an Australian kangaroo, Macropus eugenii , provides insight into the evolution of mammalian reproduction and development. Genome Biol. Mikkelsen, T. Genome of the marsupial Monodelphis domestica reveals innovation in non-coding sequences. Jonsson, H. Bioinformatics 29 , — Bensley, B.
On the evolution of the Australian Marsupialia: with remarks on the relationships of marsupials in general. S2 , 83— Zemann, A. Ancestry of the Australian termitivorous numbat. Kuritzin, A. Incomplete lineage sorting and hybridization statistics for large-scale retroposon insertion data. PLoS Comput.
0コメント