Evolutionary biology Genetic data from the extinct giant rat royalsocietypublishing.org/journal/rsbl from Tenerife (Canary Islands) points to a recent divergence from mainland relatives Research Pere Renom1, Toni de-Dios1,2, Sergi Civit3, Laia Llovera1, Alejandro Sánchez- Gracia3, Esther Lizano1,4, Juan Carlos Rando5, Tomàs Marquès-Bonet1,4,6,7, Cite this article: Renom P et al. 2021 Genetic Gael J. Kergoat8, Isaac Casanovas-Vilar4 and Carles Lalueza-Fox1 data from the extinct giant rat from Tenerife 1 (Canary Islands) points to a recent divergence Institute of Evolutionary Biology (CSIC-Universitat Pompeu Fabra), Barcelona 08003, Spain 2 University of Tartu, Institute of Genomics, Estonian Biocentre, Tartu 51010, Estonia Downloaded from https://royalsocietypublishing.org/ on 23 January 2022 from mainland relatives. Biol. Lett. 17: 3 Departament of Genetics, Microbiology and Statistics-Institut de Recerca de la Biodiversitat (IRBio), Universitat 20210533. de Barcelona, Barcelona 08028, Spain https://doi.org/10.1098/rsbl.2021.0533 4 Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Barcelona 08193, Spain 5 Departamento de Biología Animal, Edafología y Geología, Universidad de La Laguna, La Laguna 38206, Tenerife, Spain 6 Received: 14 October 2021 Catalan Institution of Research and Advanced Studies (ICREA), Barcelona 08010, Spain 7 CNAG-CRG, Centre for Genomic Regulation, Barcelona Institute of Science and Technology (BIST), Barcelona Accepted: 26 November 2021 08036, Spain 8 CBGP, INRAE, IRD, CIRAD, Institut Agro, Univ Montpellier, Montpellier, France Td-D, 0000-0001-9260-8846; SC, 0000-0002-2511-4579; LL, 0000-0002-5836-593X; AS-G, 0000-0003-4543-4577; EL, 0000-0003-3304-9807; JCR, 0000-0001-7397-869X; Subject Areas: TM-B, 0000-0002-5597-3075; GJK, 0000-0002-8284-6215; IC-V, 0000-0001-7092-9622; evolution, taxonomy and systematics CL-F, 0000-0002-1730-5914 Evolution of vertebrate endemics in oceanic islands follows a predictable Keywords: pattern, known as the island rule, according to which gigantism arises in ancient DNA, body mass, gigantism, insular originally small-sized species and dwarfism in large ones. Species of extinct evolution, molecular phylogeny, rodents insular giant rodents are known from all over the world. In the Canary Islands, two examples of giant rats, †Canariomys bravoi and †Canariomys tamarani, endemic to Tenerife and Gran Canaria, respectively, disappeared soon after human settlement. The highly derived morphological features Authors for correspondence: of these insular endemic rodents hamper the reconstruction of their evol- Isaac Casanovas-Vilar utionary histories. We have retrieved partial nuclear and mitochondrial e-mail: isaac.casanovas@icp.cat data from †C. bravoi and used this information to explore its evolutionary Carles Lalueza-Fox affinities. The resulting dated phylogeny confidently places †C. bravoi e-mail: carles.lalueza.fox@gmail.com within the African grass rat clade (Arvicanthis niloticus). The estimated diver- gence time, 650 000 years ago (95% higher posterior densities: 373 000– 944 000), points toward an island colonization during the Günz–Mindel interglacial stage. †Canariomys bravoi ancestors would have reached the island via passive rafting and then underwent a yearly increase of mean body mass calculated between 0.0015 g and 0.0023 g; this corresponds to fast evolutionary rates (in darwins (d), ranging from 7.09 d to 2.78 d) that are well above those observed for non-insular mammals. 1. Introduction The Canary Islands are located northwest off the coast of Africa, with their nearest island (Fuerteventura) being only separated from the continent by about 100 km (figure 1b). Although this volcanic archipelago was never con- Electronic supplementary material is available nected to the mainland by any land bridge or island chain, colonization of online at https://doi.org/10.6084/m9.figshare. c.5744039. © 2021 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited. (a) Homo sapiens (b) 2 Draa River royalsocietypublishing.org/journal/rsbl Arvicanthis niloticus †Canariomys bravoi species complex 1492–1650 g Arvicanthis niloticus 89–130 g (c) Biol. Lett. 17: 20210533 †Malpaisomys Lanzarote †Canariomys bravoi insularis La Palma Cueva del 50 km Downloaded from https://royalsocietypublishing.org/ on 23 January 2022 Viento Barranco de La Gomera las Moraditas Fuerteventura 2 cm El Hierro Tenerife †Canariomys Gran Canaria tamarani Figure 1. (a) Outline drawings and body mass for A. niloticus and †C. bravoi, along with a human reference. (b) Current distribution of A. niloticus in Africa (data from IUCN, 2008). The inset shows the Canary Islands and the distribution of extinct endemic rodents. Sampling sites are indicated for Tenerife. (c) Size differences between a typical A. niloticus representative (left) and †C. bravoi (right) as illustrated by their crania and femora (specimens are curated in IMEDEA and DZUL collections with numbers 12758 (A. niloticus) and 3199 (†C. bravoi), respectively). The latter has not been subjected to DNA analysis. terrestrial organisms from the mainland was favoured by dominant oceanic currents. This archipelago offers a unique 2. Material and methods opportunity to study the colonization and diversification of (a) The samples multiple groups of organisms, such as birds, reptiles or We performed DNA extraction from 12 mandibles: two from small mammals [1]. Among the latter, there are three Barranco Moraditas and 10 more from Cueva del Viento known striking examples of gigantism: the lava mouse of (figure 1b). Specimens used for extractions were deposited in Fuerteventura and Lanzarote (†Malpaisomys insularis) and the Vertebrate collection (DZUL) of Departamento de Biología the extinct giant rats of Tenerife and Gran Canaria (†Canari- Animal, Edafología y Geología de la Universidad de La omys bravoi and †Canariomys tamarani, respectively). Laguna (Tenerife) with the following inventory numbers: CB-1 (DZUL 3200); CB-2 (DZUL 3201); CB-3 (DZUL 3202); CB-4 The Tenerife giant rat was described by Crusafont-Pairó & (DZUL 3203); CB-5 (DZUL 3204); CB-6 (DZUL 3205); CB-7 Petter [2] after the discovery of numerous specimens in Qua- (DZUL 3206); CB-8 (DZUL 3207); CB-9 (DZUL 3208); CB-10 ternary sites. Subsequent studies explored its diet, ecology, (DZUL 3209); CB-11 (DZUL 3210) and CB-12 (DZUL 3211). body mass and extinction causes, and also tentatively Cueva del Viento site is a 17 km-long system of volcanic lava assessed its phylogenetic affinities based on dental traits tubes formed 0.17–0.13 Ma [7] and situated in the north side of (e.g. [3,4]). †Canariomys bravoi shows a set of traits character- Tenerife at 700 m above sea level. The animals went into the istic of insular rodents, including gigantism, a robust skeleton cave through a small pit fall that acted like a trap. Bones were and high-crowned teeth (figure 1c). It became extinct after the found in connection, showing the absence of transport after fourth century BCE, likely in relation to the arrival of Canar- deposition. Previous calibrated radiocarbon ages of †C. bravoi ian indigenous people [4]. †Canariomys tamarani also became samples from this site are between 17 300 and 2150 cal BP [4,8]. extinct soon after the arrival of the first settlers while †M. The samples from Barranco de las Moraditas were recovered from a small cave infilling in basaltic materials of Quaternary insularis survived until the beginning of the fourteenth cen- age at the east of Tenerife [7]. The median age reported for tury, when Europeans reached the archipelago [5]. Ancient another †C. bravoi sample from this site is 2310 cal BP [4]. mitochondrial DNA (mtDNA) from †M. insularis, showed close affinities to the extant genus Mus and pointed to a 6.9 Ma divergence date (genetic data were obtained by means of the traditional polymerase chain reaction method). (b) DNA extraction, mitochondrial DNA capture and Hot and humid thermal conditions hamper the retrieval of ancient genetic data [6]. Without this information, it is dif- library preparation ficult to unravel the affinities of highly modified extinct All DNA extraction and initial library preparation steps were performed in a dedicated clean laboratory, physically isolated species such as †C. bravoi to their mainland smaller relatives. from the laboratory used for post-PCR analyses; no previous Here, we managed to retrieve partial nuclear and mtDNA work on extinct or extant rodents was ever conducted in our lab- data from two †C. bravoi specimens; we subsequently used oratory. Strict protocols were followed to minimize the amount this information to provide divergence age estimates and of human DNA in the ancient DNA laboratory, including wear- phylogenetic relationships for this lineage and determine ing a full-body suit, sleeves, shoe covers, clean shoes, facemask, the rate of increase in body size of this insular rodent. hair net and double gloving. First, teeth samples were UV irradiated (245 nm) for 10 min supplementary material, figures S3 and S4). We further validated 3 and the outermost surface of the teeth was scraped off with a it with PMD tools [12], a statistical tool designed to isolate drill engraving cutter, followed by another UV irradiation in endogenous from contaminant DNA sequences; the PMD score royalsocietypublishing.org/journal/rsbl order to exclude the surface DNA contamination. Second, distribution obtained is shifted toward positive values (electronic approximately 30 mg of tooth cementum was obtained by supplementary material, figures S5 and S6), which is character- drilling at low speed (5000 r.p.m.) with a new engraving cutter. istic of ancient samples. Several precautions were taken to DNA extraction from teeth powder was performed following account for the low coverage and the existence of single DNA the method of Dabney et al. [9]. The teeth powder samples, reads: the ends of the reads were trimmed to eliminate potential including an extraction blank, were added to 1 ml of extraction post-mortem damage and C–T and G–A substitutions were only buffer (final concentrations: 0.45 M EDTA, 0.25 mg ml−1 protein- considered when they were shared with other rodent species. ase K, pH 8.0), resuspended by vortexing and incubated at 37°C We further inferred a time-calibrated Bayesian phylogenetic overnight on rotation. The remaining tooth powder was then pel- tree, relying on three secondary calibration points based on the leted by centrifugation in a bench-top centrifuge at maximum results of Aghová et al. [13]. Bayesian inference (BI) was used to speed (16 100g). The supernatant was added to 10 ml of binding estimate the phylogenetic relationships and node ages using the buffer (final concentrations: 5 M guanidine hydrochloride, 40% BEAST v. 2.6.5 package [14]. The multiple sequence alignment Biol. Lett. 17: 20210533 (vol/vol) isopropanol, 0.05% Tween-20 and 90 mM sodium acet- was built using MAFFT software [15]. We first aligned arvicanthin ate ( pH 5.2)) and purified on a High Pure Extender column sequences and then we added to this alignment the sequence of (Roche). DNA samples were eluted with 45 µl of EDTA TE †C. bravoi using the –add option in MAFFT. Best fit model buffer ( pH 8.0). However, 10 samples failed to yield quantifiable of nucleotide substitution for this alignment was elected with Downloaded from https://royalsocietypublishing.org/ on 23 January 2022 DNA after extraction and only two from Cueva del Viento (CB-4 jModelTest [16] based on the Bayesian information criterion. and CB-10) were further selected for library building. To infer the time-calibrated phylogeny, we used the Bayesian A total of 35 µl of each DNA extract was used for library prep- uncorrelated lognormal relaxed clock (ULRC) model imple- aration in three sequential reactions: end-repair, adapter ligation, mented in BEAST v. 2 [17]. We used a coalescent model tree and nick fill-in; following the BEST protocol [10]. DNA extract prior with a constant population size [18]. We set three palaeon- from CB-4 was used for DNA-library preparation prior to Illumina tological calibration points at different nodes of the tree: a Mus/ sequencing; the resulting library was amplified by PCR with two Rattus divergence between 11.6 and 13.8 Ma [13,19], an uniquely barcoded primers and used for shotgun sequencing. Arvicanthis/Lemniscomys divergence between 6.1 and 8.5 Ma Both libraries were purified with a 1× AMPure clean (Beckman [20,21] and a basal node of the Arvicanthini of 8.5–9.2 Ma [13,20]. Coulter) and eluted in 25 µl of low EDTA TE buffer (pH 8.0). The Tree Model was set to a birth–death speciation process Library size and concentration were determined with the Agilent [22] to account more accurately for extinct and missing lineages. DNA 7500 Kit on the 2100 BioAnalyzer instrument. The DNA We used BEAST 2 to run 180 million generations of the model to libraries were sequenced using HiSeq400 of Illumina platform sample trees from the posterior distribution (each 5000 gener- (Illumina, USA) in Macrogen, Inc. biotechnology company. ations). After examining effective sample sizes (ESS) and the After library preparation, sample CB-10 was enriched for traces for posterior, prior and likelihood with the tool Tracer mtDNA sequences with the use of commercially biotinylated [23], we discarded the first 20% of trees from the analysis probes for mouse mtDNA (MYbaits). Prior to hybridization, (burn-in proportion). Finally, we generated the tree with the DNA library (approx. 500 ng) was brought to 7 µl using a median age estimates and 95% higher posterior densities (95% Speedvak concentrator. Two consecutive hybridizations were HPD bars) using tree annotator tool (distributed with the conducted with the myBaits Capture Kit (Arbor Biosciences) BEAST v. 2 package). Convergence of runs was assessed by according to the manufacturer’s manual v. 4.01. The hybridiz- examining the ESS of parameters, using the recommended ation reaction was carried out at 65°C for 24 h in a final threshold of 200 [17]. volume of 30 µl. Captured targets were recovered with Dyna- beads MyOne Streptavidin C1 magnetic beads (Invitrogen), followed by bead–bait binding and washing according to the manufacturer’s recommendations. After the first round of enrich- (d) Evolutionary rates We calculated evolutionary rates for the body mass (in grams) ment, post captured amplification was performed using PCR increase from A. niloticus to †C. bravoi. Mean body mass for primers IS5 and IS6. All of the captured material was concen- A. niloticus (sexes combined) is taken from Monadjem et al. [24], trated to 7 µl and used for the second round of hybridization. while estimated mean body mass for †C. bravoi is taken from Mon- The second hybridization was performed under the same con- cunill-Solé et al. [25] and is based on multiple regressions ditions and the final captured pool was amplified with P5 and considering dental, cranial and postcranial measurements. P7 indexed primers compatible for Illumina sequencing [11]. Evolutionary rates are calculated using the simple classic Sample CB-10 was radiocarbon dated to 2800 ± 30 years BP equation by Haldane [26]: (Beta-598676).   ln(x2) r ¼ ln(x1)  , ð2:1Þ Dt (c) Phylogenetic analysis where, r is the rate of change (in darwins, d); x1 and x2 are the All resulting DNA reads from samples CB-4 and CB-10 were initial and final value of the analysed variable, respectively; mapped (edit distance equal to 0.0001) to Mus musculus and Δt is the amount of time elapsed. The calculations are carried (MN964117.1), Rattus rattus (NC_012374.1) and Arvicanthis niloti- for the minimum and maximum divergence dates between cus (CM022273.1) mtDNA genomes. Mapped reads were A. niloticus and †C. bravoi. The age of the oldest †C. bravoi fossils subsequently blasted and only reads specific to rodent mitogen- (17 300 cal BP [4,8]) is taken as the endpoint of the size increase omes were retained. Additionally, DNA reads from CB-4 were trend. Calculated evolutionary rates are compared to those mapped against the A. niloticus nuclear genome (NCBI: derived for other mammals (e.g. [27,28]) as well as to those for txid61156) with standard aDNA edit distance (0.01). †M. insularis, which is included because it represents another The authenticity of the generated sequences was confirmed case of murid gigantism in the same archipelago. Body mass with the observation of the typical post-mortem ancient DNA for †M. insularis is estimated from published cranial and post- damage at the end of the DNA reads (electronic supplementary cranial measurements [29], applying described equations [25]. material, figure S2) and length fragmentation pattern (electronic Estimated mean body mass for †M. insularis is compared to 16 14 12 10 8 6 4 2 0 4 Ma Miocene Pliocene Pleist. H epoch royalsocietypublishing.org/journal/rsbl middle late early late Early M L Rattus rattus Mus musculus 14.16 Otomys sp. (Mount Kenya) Otomys typus Otomys irroratus 12.79 8.51 Golunda ellioti 10.45 Oenomys hypoxanthus Hybomys trivirgatus 8.51 Hybomys planifrons Dephomys defua Stochomys longicaudatus 9.65 Hybomys univittatus Biol. Lett. 17: 20210533 Hybomys sp. Micaelamys namaquensis Thallomys paedulcus Grammomys poensis Downloaded from https://royalsocietypublishing.org/ on 23 January 2022 Thamnomys kempi Aethomys chrysophilus Aethomys nyikae Aethomys kaiseri 8.11 Aethomys hindei Grammomys selousi 9.07 Grammomys dolichurus Grammomys macmillani Grammomys surdaster Dasymys incomtus Lamottemys okuensis Rhabdomys dilectus 8.57 Desmomys harringtoni 6.95 Mylomys dybowskii Pelomys fallax Lemniscomys barbarus Lemniscomys striatus Lemniscomys rosalia 5.46 Lemniscomys bellieri Arvicanthis rufinus Arvicanthis ansorgei 4.51 Arvicanthis nairobae Arvicanthis abyssinicus 2.66 Arvicanthis somalicus Arvicanthis grp niloticus C2–C4 (Sahel) 0.65 †Canariomys bravoi 2.42 Arvicanthis grp niloticus (Masai Mara) Figure 2. Molecular phylogenetic tree of the Muridae with the mitochondrial DNA data. Median ages are indicated at the nodes while error bars (grey shading) at nodes correspond to the 95% highest posterior density (HPD) intervals of age estimates. Purple circles at nodes indicate posterior probabilities greater than 95%. NCBI codes for each mitogenome can be found at electronic supplementary material, table S1. Table 1. Mapping statistics of †C. bravoi mitochondrial and nuclear DNA reads. CB-4 corresponds to shotgun sequencing and CB-10 to mtDNA capture and sequencing. mtDNA reads were mapped following the procedure described in the Methods section; nuclear reads from CB-4 were mapped against the Arvicanthis niloticus nuclear genome (NCBI:txid61156). sequenced read mapped unique q20 mapped reference specimen (mtDNA) pairs reads reads BLAST reads bases recovered CB-4 119 102 486 281 104 41 1826 10.78% CB-10 3 746 833 10 745 55 45 3387 7.21% specimen (nuclear sequenced read mapped unique reads unique q20 mapped average DNA) pairs reads reads bases coverage CB-4 119 102 486 111 084 40 434 35 593 1 616 176 0.0006X that of its closest relatives, Mus (Coelomys) pahari and Mus (Coelomys) crociduroides from Southeastern Asia [30] with calcu- 3. Results and discussion lations formulated for minimum and maximum divergence A total of 1 616 176 nucleotides mapped to Arvicanthis dates between †M. insularis and Mus (Coelomys) spp. The oldest nuclear genome, representing only a 0.0006× depth of cover- †Malpaisomys fossils have been dated to 32 000 cal BP [30], age but proving that genomic retrieval of †C. bravoi is a which is taken as the age for the end of the size increase trend. possible—albeit a challenging—task. A total of 2627 mapped mtDNA nucleotides were aligned melitensis (13.5 heavier than its putative most closely related 5 to a large dataset of rodent mitogenomes (table S1) and sub- species, the garden dormouse Eliomys quercinus [11]) and gen- royalsocietypublishing.org/journal/rsbl sequently used for the phylogenetic analysis (table 1). The erally well above values inferred for Pleistocene giant resulting dated phylogeny confidently places †C. bravoi murines from several Mediterranean islands (usually 2–3 within the Arvicanthis genus, in the A. niloticus species com- times heavier than their mainland ancestors [34,35]). It is plex [30] (figure 2); it is closely related to a specimen from also far greater than that calculated for †Malpaisomys, which Masai Mara (Kenya) and is more distinctly related to a speci- is almost four times heavier (90 g) than its mainland relatives men belonging to the C2–C4 lineage that is distributed across (24 g) [30]. It is difficult to assess when the process of size the Sahel (both Arvicanthis specimens correspond to cryptic increase was achieved because the age of the oldest †C. species in the A. niloticus species complex [31]). This unex- bravoi fossils is not well constrained and their earliest date pected placement parallels that of M. insularis, which was for the start of gigantism is necessarily after the splitting found to cluster between members of the genus Mus [32], from a mainland Arvicanthis lineage. According to our data, despite its uncommon dental and skeletal traits. †Canariomys the resulting temporal range would suppose a yearly increase and †Malpaisomys belong to different murine tribes (Murini of mean body mass of just between 0.0015 g and 0.0023 g Biol. Lett. 17: 20210533 and Arvicanthini), thus indicating that their origin must be (considering maximum and minimum splitting dates, traced to different ancestors and likely different colonization respectively). This corresponds to evolutionary rates (in dar- events (the former being much more recent). wins (d)), ranging from 7.09 d to 2.78 d that are well above Downloaded from https://royalsocietypublishing.org/ on 23 January 2022 The expansion of Arvicanthis species through North Africa those observed for non-insular mammals (usually less than was heavily influenced by Pleistocene climatic fluctuations 1 d). By contrast, calculated evolutionary rates for †Malpais- [33,34]. When environmental conditions changed and the omys are in the range of those of mainland mammals (0.22– Sahara region dried up, different Arvicanthis populations 0.16 d). We must remark that our †Canariomys evolutionary were cornered in areas of grassland and savannah habitats rate estimates represents a minimal estimate and assumes a far apart each other. The current patchy distribution of mem- constant rate of change. However, large body size may have bers of the A. niloticus species complex includes the Nile not been achieved at a constant rate but soon after coloniza- River up to the great African lakes, the whole strip of the tion and stabilized from then on. This would imply even a Sahel and some isolated populations surviving in Pleistocene faster initial growing-size rate for †Canariomys. refuges such as the Hoggar mountains (southern Algeria) However, only the retrieval of additional †Canariomys (figure 1b). A previous molecular study indicated that the A. nuclear genome-wide data would further clarify its evol- niloticus species complex likely originated in eastern Africa as utionary history and would also allow the identification of early as 2 Ma and differentiated in genetically distinct lineages the genomic regions under selection that might be respon- from east to west as a result of Pleistocene climatic cycles [33]. sible for the conspicuous physical differences observed The divergence time between †C. bravoi and its closest between this extinct lineage and its living relatives. Arvicanthis relative is estimated at 650 000 years ago (95% Data accessibility. Genetic data generated are deposited in the Dryad HPD intervals: 373 000–944 000 years ago) (figure 2). This database ( project https://doi.org/10.5061/dryad.9s4mw6mhn). interval basically includes the Günz and Mindel glaciations, Authors’ contributions. P.R. and C.L.-F. conceived the project; J.C.R. col- as well as the Günz–Mindel interglacial. This interglacial lected and studied the skeletal samples; L.L. and E.L. performed appears as the most probable period for the colonization of experimental work; T.d.-D., P.R., S.C. and A.S.-G. performed compu- Tenerife by †Canariomys ancestors. Interglacial periods tational work; G.J.K., T.M.-B. and I.C.-V. helped in the interpretation altered the monsoon regime and increased rainfall across of the results; T.M.-B. and C.L.-F. coordinated experimental and com- putational teams; P.R., C.L.-F., G.J.K. and I.C.-V. wrote the Africa. Satellite images of the Draa ancient river bed, which manuscript with inputs from all co-authors. All authors agree to be drains the anti-Atlas and flows right in front of the Canary held accountable for the content therein and approve the final version Islands, suggests it must have been a river with a flow of of the manuscript. more than 400 m3 s−1 that period and probably dragged logs and masses of vegetation to the sea on which some †Canariomys ancestors might have drifted away. Competing interests. The authors declare no conflict of interest. This relatively recent split date points to a rapid evol- Funding. C.L.-F. is supported by a PGC2018-0955931-B-100 grant utionary process associated with gigantism. Body mass of (MCIU/AEI/FEDER, UE) of Spain; T.M.-B. is supported by funding A. niloticus ranges between 89 and 130 g (mean 114 g) [24], from the European Research Council (ERC) (grant agreement no. whereas the estimated weight of †C. bravoi was 1492–1650 g 864203), BFU2017-86471-P (MINECO/FEDER, UE), ‘Unidad de Exce- lencia María de Maeztu’, funded by the AEI (CEX2018-000792-M), (mean 1571 g) [25], which is almost 14 times heavier (we Howard Hughes International Early Career and Generalitat de Cata- are assuming that the body size of the A. niloticus ancestor lunya (CERCA Programme and 2017 SGR 880); I.C.-V. is supported that originated the Canariomys lineage had the same weight by grant nos. I+D+i PID2020-117289GBI00 funded by MCIN/AEI/ than the extant A. niloticus). This size increase is comparable 10.13039/501100011033/, the MINECO (RYC-2013-12470) and the to that of the extinct Sicilian giant dormouse †Leithia Generalitat de Catalunya (CERCA Programme and 2017 SGR 116). References 1. Juan C, Emerson BC, Oromí P, Hewitt GM. 104–109. (doi:10.1016/s0169-5347(99) nov., sp. nov. (Rongeurs, Muridés). Mammalia 28, 2000 Colonization and diversification: 01776-0) 608–611. (doi:10.1515/mamm.1964.28.4.607) towards a phylogeographic synthesis for 2. Crusafont-Pairo M, Petter F. 1964 Un muriné géant 3. Michaux J, Hautier L, Hutterer R, Lebrun R, Guy F, the Canary Islands. Trends Ecol. Evol. 5, fossile des îles Canaries: Canariomys bravoi gen. García-Talavera F. 2012 Body shape and life style of the extinct rodent Canariomys bravoi (Mammalia, Murinae) set of fossil calibrations to improve the temporal 26. Haldane JBS. 1949 Suggestions as to quantitative 6 from Tenerife, Canary Islands (Spain). C. R. Palevol. 11, phylogenetic framework of murid rodents measurement of rates of evolution. Evolution royalsocietypublishing.org/journal/rsbl 485–494. (doi:10.1016/j.crpv.2012.06.004) (Rodentia: Muridae). Mol. Phylogenetics Evol. 128, 3, 51–56. (doi:10.1111/j.1558-5646.1949. 4. Rando JC, Alcover JA, Galván B, Navarro JF. 2014 98–111. (doi:10.1016/j.ympev.2018.07.017) tb00004.x) Reappraisal of the extinction of Canariomys bravoi, 14. Bouckaert R et al. 2019 BEAST 2.5: an advanced 27. Gingerich PD. 1983 Rates of evolution: effects of the giant rat from Tenerife (Canary Islands). software platform for Bayesian evolutionary analysis. time and temporal scaling. Science 222, 159. Quat. Sci. Rev. 94, 22–27. (doi:10.1016/J. PLoS Comput. Biol. 15, e1006650. (doi:10.1371/ (doi:10.1126/science.222.4620.159-a) QUASCIREV.2014.04.013) journal.pcbi.1006650) 28. MacFadden BJ. 1986 Fossil horses from ‘Eohippus’ 5. Rando JC, Alcover JA, Navarro JF, García-Talavera F, 15. Katoh K, Standley DM. 2013 MAFFT multiple (Hyracotherium) to Equus: scaling, Cope’s Law, and Hutterer R, Michaux J. 2008 Chronology and causes sequence alignment software version 7: the evolution of body size. Paleobiology 12, of the extinction of the lava mouse, Malpaisomys improvements in performance and usability. Mol. 355–369. (doi:10.1017/S0094837300003109) insularis (Rodentia: Muridae) from the Canary Biol. Evol. 30, 772–780. (doi:10.1093/molbev/ 29. Hutterer R, López-Martínez N, Michaux J. 1998 A Islands. Quat. Res. 70, 141–148. (doi:10.1016/j. Mst010) new rodent from quaternary deposits of the Canary yqres.2008.04.012) 16. Posada D. 2008 jModelTest: phylogenetic model Islands and its relationships with Neogene and Biol. Lett. 17: 20210533 6. García-Garcerà M, Gigli E, Sanchez-Quinto F, Ramirez O, averaging. Mol. Biol. Evol. 25, 1253–1256. (doi:10. recent murids of Europe and Africa. Palaeovertebrata Calafell F, Civit S, Lalueza-Fox C. 2011 Fragmentation of 1093/molbev/msn083) 18, 241–262. contaminant and endogenous DNA in ancient samples 17. Drummond AJ, Ho SYW, Phillips MJ, Rambaut A. 30. Denys C, Taylor PJ, Burgin CJ, Aplin KP, Fabre P-H, Downloaded from https://royalsocietypublishing.org/ on 23 January 2022 determined by shotgun sequencing; prospects for 2006 Relaxed phylogenetics and dating with Haslauer R, Woinarski JCZ, Breed WG, Menzies JI. human palaeogenomics. PLoS ONE 6, e24161. (doi:10. confidence. PLoS Biol. 4, e88. (doi:10.1371/journal. 2017 . Species accounts of Muridae. In Handbook of 1371/journal.pone.0024161) pbio.0040088) the mammals of the world. 7. Rodents II, handbook 7. Ancochea E, Fuster JM, Ibarrola E, Cendrero A, Coello J, 18. Heled J, Drummond AJ. 2009 Bayesian inference of of mammals of the world (eds DE Wilson, TE Lacher Hernán F, Cantagrel JM, Jamond C. 1990 Volcanic species trees from multilocus data. Mol. Biol. Evol. Jr, RA Mittermeier), pp. 598–884. Barcelona, Spain: evolution of the island of Tenerife (Canary Islands) in 27, 570–580. (doi:10.1093/molbev/msp274) Lynx Edicions. the light of new K-Ar data. J. Volcanol. Geotherm. Res. 19. Jacobs LL, Flynn LJ. 2005 Of mice again: the Siwalik 31. Bryja J, Colangelo P, Lavrenchenko LA, Meheretu Y, 44, 231–249. (doi:10.1016/0377-0273(90)90019-C) rodent record, murine distribution, and molecular Šumbera R, Bryjová A, Verheyen E, Leirs H, Castiglia 8. Crowley BE, Yanes Y, Mosher SG, Rando JC. 2019 clocks. In Interpreting the past: essays on human, R. 2019 Diversity and evolution of African rats Revisiting the foraging ecology and extinction primate, and mammal evolution in honor of David (Muridae: Arvicanthis)—from radiation in East history of two endemic vertebrates from Tenerife, Pilbeam (eds DE Lieberman, RJ Smith, J Kelley), pp. Africa to repeated colonization of northwestern and Canary Islands. Quaternary 2, 10. (doi:10.3390/ 63–80. Leiden, The Netherlands: Brill Academic southeastern savannas. J. Zool. Syst. Evol. Res. 57, quat2010010) Publishers, Inc. 970–988. (doi:10.1111/jzs.12290) 9. Dabney J et al. 2013 Complete mitochondrial 20. Geraads D. 2001 Rongeurs du Miocène supérieur de 32. Pagès M, Chevret P, Gros-Balthazard M, Hughes S, genome sequence of a Middle Pleistocene cave bear Ch’orora, Ethiopie: Murinae, Dendromurinae et Alcover JA, Hutterer R, Rando JC, Michaux J, Hänni reconstructed from ultrashort DNA fragments. Proc. conclusions. Palaeovertebrata 30, 89–109. C. 2012 Paleogenetic analyses reveal unsuspected Natl Acad. Sci. USA 110, 15 758–15 763. (doi:10. 21. Manthi FK. 2007 A preliminary review of the rodent phylogenetic affinities between mice and the 1073/pnas.1314445110) fauna from Lemudong’o, southwestern Kenya, and extinct Malpaisomys insularis, and endemic rodent 10. Carøe C, Gopalakrishnan S, Vinner L, Mak SST, its implication to the late Miocene of the Canaries. PLoS ONE 7, e311123. (doi:10.1371/ Sinding MHS, Samaniego JA, Wales N, Sicheritz- paleoenvironments. Kirtlandia 56, 92–105. journal.pone.0031123) Pontén T, Gilbert MTP. 2018 Single-tube library 22. Gernhard T. 2008 The conditioned reconstructed 33. Dobigny G, Tatard C, Gauthier P, Ba K, Duplantier preparation for degraded DNA. Methods Ecol. Evol. process. J. Theor. Biol. 253, 769–778. (doi:10.1016/ JM, Granjon L, Kergoat GJ. 2013 Mitochondrial and 9, 410–419. (doi:10.1111/2041-210X.12871) j.jtbi.2008.04.005) nuclear genes-based phylogeography of Arvicanthis 11. Kircher M, Sawyer S, Meyer M. 2012 Double 23. Rambaut A, Drummond AJ, Xie D, Baele G, Suchard niloticus (Murinae) and Sub-Saharan open habitats indexing overcomes inaccuracies in multiplex MA. 2018 Posterior summarization in Bayesian Pleistocene history. PLoS ONE 8, e77815. (doi:10. sequencing on the Illumina platform. Nucleic Acids phylogenetics using Tracer 1.7. Syst. Biol. 67, 1371/journal.pone.0077815) Res. 40, e3. (doi:10.1093/nar/gkr771) 901–904. (doi:10.1093/sysbio/syy032) 34. Lomolino MV, van der Geer AA, Lyras GA, 12. Skoglund P, Northoff BH, Shunkov MV, Derenvianko AP, 24. Monadjem A, Taylor PJ, Denys C, Cotterill FPD. 2015 Palombo MR, Sax DF, Rozzi R. 2013 Of mice Pääbo S, Krause J, Jakobsson M. 2014 Separating Rodents of Sub-Saharan Africa. A biogeographic and and mammoths: generality and antiquity of the endogenous ancient DNA from modern day taxonomic synthesis. Berlin, Germany: De Gruyter. island rule. J. Biogeogr. 40, 1427–1439. (doi:10. contamination in a Siberian Neandertal. Proc. Natl Acad. 25. Moncunill-Solé B, Jordana X, Marín-Moratalla N, 1111/jbi.12096) Sci. USA 111, 2229–2234. (doi:10.1073/pnas. Moyà-Solà S, Köhler M. 2014 How large are the 35. Gingerich PD. 2001 Rates of evolution on the time 1318934111) extinct giant insular rodents? New body mass scale of the evolutionary process. In Microevolution 13. Aghová T, Kimura Y, Bryja J, Dobigny G, Granjon L, estimations from teeth and bones. Integr. Zool. 9, rate, pattern, process (eds AP Hendry, MT Kinnison), Kergoat GJ. 2018 Fossils know it best: using a new 197–212. (doi:10.1111/1749-4877.12063) pp. 127–144, Dordrecht, The Netherlands: Springer.