Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Contents lists available at ScienceDirect Palaeogeography, Palaeoclimatology, Palaeoecology journal homepage: www.elsevier.com/locate/palaeo Tooth and long bone scaling in Sardinian ochotonids (Early Pleistocene-Holocene): Evidence for megalodontia and its palaeoecological implications Blanca Moncunill-Solé a, b, 1, *, Caterinella Tuveri c, Marisa Arca c, Chiara Angelone a, d, e a Dipartimento di Scienze, Università degli Studi Roma Tre, 00146 Roma, Italy b Departamento de Física e Ciencias da Terra, Universidade da Coruña, 15008 A Coruña, Spain c Soprintendenza Archeologia, Belle Arti e Paesaggio Per le Prov. di Sassari e Nuoro, 08100 Nuoro, Italy d Institut Català de Paleontologia Miquel Crusafont (ICP), Campus de la Universitat Autònoma de Barcelona, 08193 Cerdanyola del Vallès, Spain e Key Laboratory of Vertebrate Evolution and Human Origins, Institute of Vertebrate Paleontology and Palaeoanthropology, Chinese Academy of Sciences, 100044 Beijing, China A R T I C L E I N F O A B S T R A C T Editor: Howard Falcon-Lang Body size is a useful character to unravel the biology of fossil taxa and, indirectly, the palaeoenvironment in which they lived. However, the reliability of size proxies is debated, particularly among insular endemics in Keywords: which large teeth relative to body size have been observed. To shed light on this issue, here we compute sta­ Body size tistical models to assess: (i) teeth and long bone scaling and (ii) dentition size evolutionary dynamics in the Dental abrasion anagenetic lineage Prolagus figaro-P. sardus from the late Early Pleistocene-Holocene of Sardinia, Italy. Post­ Eco-evolutionary adaptations cranial bones, particularly tibial epiphyses, are shown to be the most suitable parameters for prediction of body Life-history theory Postcanine occlusal surface mass in Prolagus and other lagomorph taxa. By contrast, p3 should not be used for size reconstructions, whereas Prolagus m1 width provided the highest correlation with long bones. On the other hand, the postcanine occlusal surface of Sardinian Prolagus (m1 and toothrow area) appeared significantly larger than those of extant lagomorphs, pointing to a higher resistance to tooth wear and a more durable permanent dentition. It seems likely that megalodontia might have evolved in Sardinian Prolagus to cope with the abrasive diet found on islands (food habit hypothesis). Nonetheless, we cannot exclude that other biological features, such as long longevity, could be co-driving forces in this fossil lineage (life history hypothesis). The results achieved throw light on size and scale evolution in lagomorphs, as well as refine the eco-evolutionary responses of small mammals to insular regimes. 1. Introduction Moncunill-Solé, 2021). Historically, the BM of extinct lagomorphs were estimated using rodent regression curves (Deng, 2009), although de­ Body mass (BM), a proxy of size, has a central position in the scriptions of these methodological procedures were rarely provided. In immense network of biological variables of organisms. It is correlated recent years, the body size evolution of fossil lagomorphs has evoked with an extensive set of morphological, behavioral, physiological, great scientific interest, and accurate BM regression models have been ecological and life history traits (Peters, 1983; Calder III, 1984; Schmidt- developed and used in this mammalian order (Quintana Cardona, 2005; Nielsen, 1984) and, ultimately, it has a decisive role in the evolution of Quintana et al., 2011; van der Geer et al., 2013; Quintana Cardona and organisms (Purvis and Orme, 2005). In this regard, the weight estima­ Moncunill-Solé, 2014; Moncunill-Solé et al., 2015, 2016a, 2016b; tions of extinct species provide essential clues to understand their Winkler et al., 2016; Moncunill-Solé, 2021). palaeobiology, as well as their roles in palaeocommunities and palae­ To deepen the knowledge of the Island Rule in small mammals oenvironments (Palombo, 2009; Lyras et al., 2010; van der Geer et al., (generalities, mechanisms, and drivers), the BM variation of the ana­ 2013; Lomolino et al., 2013; Moncunill-Solé et al., 2014, 2016a, 2016b; genetic lineage of Prolagus from Sardinia, Italy (Prolagus figaro López * Corresponding author at: Departamento de Física e Ciencias da Terra, Universidade da Coruña, 15008 A Coruña, Spain. E-mail addresses: blanca.moncunill@udc.es (B. Moncunill-Solé), caterinella.tuveri@beniculturali.it (C. Tuveri), marisa.arca@beniculturali.it (M. Arca), chiara. angelone@uniroma3.it (C. Angelone). 1 Postal address: Departamento de Física e Ciencias da Terra, Universidade da Coruña, Campus da Zapateira, Rúa da Fraga 10, 15008 A Coruña, Spain. https://doi.org/10.1016/j.palaeo.2021.110645 Received 14 July 2021; Received in revised form 2 September 2021; Accepted 2 September 2021 Available online 9 September 2021 0031-0182/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Martínez in López Martínez and Thaler, 1975 - P. sardus [Wagner, Fiori SD (Carbonia), Grotta Teulada (Teulada), Dragonara cave (= 1829]) was addressed. Angelone et al. (2008) described an evolutionary Grotta della Medusa; Alghero), and S’Omu e S’Orku (Arbus) (Fig. 1A) asymptotic size increase, considering teeth and bones from several sites (Gliozzi et al., 1984; Abbazzi et al., 2004; Floris et al., 2009; Boldrini located in Monte Tuttavista (Orosei, Italy). Eventually, Moncunill-Solé et al., 2010; Palombo and Rozzi, 2014; Palombo et al., 2017). For details et al. (2016a) considered the size variation of the lower third premolar about the stratigraphy and the faunal assemblages relative to the above (p3) and long bones (femur, humerus and tibia) in correspondence with mentioned fossil sites, see Boldrini (2009) and references therein. The the turnover in which the Middle Pleistocene-Holocene species P. sardus Monte Tuttavista fissures infillings X3 and X4 are correlated to the Capo replaced its ancestor, the latest Pliocene/earliest Pleistocene-late Early Figari/Orosei 1 faunal subcomplex of the Nesogoral Faunal Complex and Pleistocene species P. figaro. The authors provided evidence of a pro­ to the Orosei 2 faunal subcomplex of the Microtus (Tyrrhenicola) Faunal gressive BM increase during the anagenetic lineage, but also identified Complex (latest Pliocene to late Early Pleistocene). The Monte Tutta­ wide discrepancies between teeth and long bones as BM proxies. Similar vista fissures infillings XIr and VIb6, Cava Alabastro, Grotta dei Fiori SD, inconsistencies have also been observed in other taxa, and a strong Grotta Teulada and Dragonara Cave belong to the Dragonara faunal debate is ongoing with regard to the reliability of size proxies in extinct subcomplex of the Microtus (Tyrrhenicola) Faunal Complex (Middle species (Gould, 1975; Jungers, 1990; Millien and Bovy, 2010; Mon­ Pleistocene to Holocene). Absolute dates are available for Dragonara cunill-Solé et al., 2015). Especially in the case of insular endemics, re­ Cave (22.39–21.91 ka) and S’Omu e S’Orku (8.7–8.5 ka) (Fig. 1B) searchers have noted large teeth relative to body size in extinct taxa, but (Palombo et al., 2017; Zoboli et al., 2018). this issue has not been studied in detail (Maglio, 1973; Moyà-Solà and The studied material is curated in the Soprintendenza dei Beni Köhler, 1997; Angelone, 2005; Moncunill-Solé et al., 2015, 2016a). Archeologici per le Province di Sassari e Nuoro (Nuoro, Sardinia, Italy) A decade after the publication of the first regression models for and the Laboratory of Vertebrate Palaeontology of the Department of estimating the size of extinct lagomorphs and in view of the above, a Science, Università degli Studi Roma Tre (Roma, Italy). A detailed list of statistical review is required. In this regard, the present study has two extant lagomorph species (ochotonids and leporids), with associated main objectives: (i) to assess the scaling relationship among skeletal data (sample size, collection, and reference), used for comparative elements (teeth and long bones) and to determine the validity and purposes is provided in Supplementary Data (Appendix A). reliability of these features as size proxies in the order Lagomorpha; (ii) to evaluate the evolutionary trends of dentition size to BM in an extinct insular lagomorph. The long-lasting lineage P. figaro - P. sardus from 2.2. Measurements Sardinia (Early Pleistocene-Holocene) is considered the most suitable taxon for carrying out the present research. This lineage stands out for Teeth were measured using a Leica MZ6 modular stereomicroscope recording significant morphological and dimensional variations in its combined with a digital camera Leica DFC290. The following mea­ dental and osteological components, and the remains (teeth and bones) surements were taken: p3 width (Wp3), p3 length (Lp3), m1 width of both species are extremely abundant, well preserved, and quite easy (Wm1), m1 length (Lm1) and length of lower toothrow (TRL) (Fig. 2A- to identify from the rest of the fossil small mammal assemblage. The C) (López-Martínez, 1989; Quintana Cardona, 2005; Quintana and results of the present study will contribute to improving the biological Agustí, 2007; Hordijk, 2010; Moncunill-Solé et al., 2015). Values of the knowledge of fossil lagomorphs and to understanding the size diversity areas of the occlusal surface of p3 (p3AA = Lp3 x Wp3), m1 (m1AA = and the scale variations of this mammalian order. In addition, they will Lm1 x Wm1), and lower toothrow (TRLAA = Wm1 x TRL) were sub­ shed light on the eco-evolutionary dynamics of small mammals in sequently calculated (Moncunill-Solé et al., 2015). On postcranial bones, insular regimes. the following measurements were taken with a manual caliper: humerus length (HL), proximal humeral antero-posterior diameter (HAPDp), 2. Material and methods distal humeral transversal diameter (HTDd), distal humeral antero- posterior-diameter (HAPDd), femur length (FL), proximal femoral 2.1. Material transversal diameter (FTDp), distal femoral transversal diameter (FTDd), tibia length (TL), proximal tibial antero-posterior diameter A total of 2997 isolated fossil elements belonging to P. figaro and (TAPDp), proximal tibial transversal diameter (TTDp), and distal tibial P. sardus were assessed, including teeth (p3 and m1), hemimandibles, transversal diameter (TTDd) (Fig. 2D-F) (Quintana Cardona, 2005; and postcranial bones (femora, humeri and tibiae) of adult individuals Moncunill-Solé et al., 2015). (Table 1, S1). It was impossible to identify teeth and bones from the The above listed measurements are commonly used for predicting same specimen, because in the field the remains were not in anatomical the size of extinct lagomorphs and they lack taxonomic value (Quintana position. The study did not consider sexual size dimorphism due to lack Cardona, 2005; Moncunill-Solé et al., 2015). The specific differentiation of such kind of differences in extant and extinct lagomorph taxa (Lu between the two Sardinian ochotonids (P. figaro and P. sardus) is based et al., 2014; Pelletier, 2019). on the morphology and complexity of the p3 enamel (López Martínez The remains were retrieved from 9 fossiliferous sites of Sardinia and Thaler, 1975). Other postcranial elements, such as proximal tarsals, (Italy): the fissures infillings X3, X4, XIr, VIb6 of the Monte Tuttavista are considered good BM proxies in mammals (Tsubamoto, 2014). karstic complex (Orosei), Cava Alabastro (Fluminimaggiore), Grotta dei However, they were found in low amounts in the studied fossiliferous sites, and the few recovered were in poor conditions. Table 1 Number of assessed remains belonging to P. figaro and P. sardus. Last column and row indicate the total material per element and site respectively. Sites Element X3 X4 Cava Alabastro XIr Grotta dei Fiori SD Grotta Teulada VI banco 6 Dragonara cave S’Omu e S’Orku Total p3 7 161 33 52 9 7 51 54 82 456 Teeth m1 1 51 33 135 16 5 0 141 111 493 Lower toothrow 0 15 15 108 8 6 0 146 119 417 Humerus 14 0 53 57 33 11 45 347 167 727 Postcranial Femur 14 13 7 68 10 17 19 178 73 399 Tibia 5 11 14 39 6 6 9 375 40 505 Total 41 251 155 459 82 52 124 1241 592 2997 2 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Fig. 1. A, Sardinian location map (Italy) showing the studied fossiliferous sites (white triangles) and the main cities (black dots). Data derive from Gliozzi et al. (1984), Abbazzi et al. (2004), Floris et al. (2009), Melis et al. (2013), Palombo and Rozzi (2014), and Palombo et al. (2017). Study sites are protected but exact location information will be shared with qualified researchers upon request. B, Chronological distribution of Sardinian lagomorph taxa, based on Angelone et al. (2020). Chronological position of fossil sites derives from Palombo (2009), Palombo and Rozzi (2014) and Zoboli et al. (2018). Those underlined have been absolute dated, whereas the others have a relative position. 2.3. Scaling analysis use of means is a common procedure carried out in historical and cutting-edge allometric investigations for not biasing the sample (Peters, Following the elastic similarity model (ESM), McMahon (1973, 1983; Calder III, 1984; Schmidt-Nielsen, 1984; Christiansen, 1999b; 1975) proposed that limb bone lengths (L) should scale to BM1/4, Copes and Schwartz, 2010; Millien and Bovy, 2010). Data were log whereas limb bone diameters (D) or least circumferences should do to transformed for analysis. The relationships among linear dimensions of BM3/8. In artiodactyls (especially in ungulates), this model was fitted teeth (L and W) and postcranial bones (L and D) were assessed by means suitable, but the empirical data of other taxa suggested geometric sim­ of correlation tests: Pearson correlation coefficient (r, parametric sta­ ilarity (L α D α BM1/3) (Alexander et al., 1979; Biewener, 1983; Garcia tistics) and Spearman correlation coefficient (ρ, non-parametric statis­ and da Silva, 2006 and references therein). When the size of specimens tics), which range from − 1 to 1. Extreme scores (− 1 or + 1) indicate a was considered a critical factor, results showed that long bones of small perfect linear relationship (negative or positive respectively), whereas mammals (BM < 20 kg) were geometrically similar and those of large score of 0 means no relationship. Area variables (p3AA, m1AA and mammals followed ESM (Economos, 1983; Christiansen, 1999a). TRLAA) were not considered in this assessment, because original ones (L Twenty years later, Garcia and da Silva (2004, 2006) proposed a new and W) are tested. scaling model which considered that bending and compressive stresses Bivariate linear regression tests (Ordinary Least Squares, OLS, Model were the forces that governed long-bone allometry. Accordingly, L and D I) are used in allometric studies to examine the dependency degree of should scale to BM0,37 in small-sized mammals. On the other hand, there variables and their scaling relationship (Christiansen, 1999b). This test is an historical debate about teeth area scaling. Following geometric was only carried out on those variables that showed significant corre­ similarity model, areas should scale to BM2/3. Gould (1975), alterna­ lation relationships. This methodology allowed to find the line (equa­ tively, suggested that postcanine occlusal area (PCOA ≈ TRLAA), which tion) that describes the variables’ relationship with the smallest error. was directly related to food processing and mammal’s metabolic needs, The result is an eq. Y = a + bX, where Y is the dependent variable, a is should scale as basal metabolic rate did (PCOA α BM3/4). However, the intercept with the y-axis, b is the slope, and X is the independent empirical evidences showed different relationships among the taxa. The variable. Scaling relationship was evaluated by means of slope values. b results obtained for rodents pointed to geometric similarity, but similar = 1 indicates an equal growth rate of variables (isometry), whereas b ∕ =1 analyses were never tested in rabbits, hares, or pikas (Copes and suggests an allometric relationship (positive or hyperallometric when b Schwartz, 2010). Moncunill-Solé et al. (2015) have carried out the most > 1, and negative or hypoallometric when b < 1) (Huxley and Teissier, complete scaling study based on extant lagomorphs, assessing 12 species 1936). We evaluated b considering its 95% bootstrapped confidence of ochotonids and 36 of leporids. Their results showed that teeth linear intervals (CI) (N = 1999). Isometry was established when CI comprised dimensions of ochotonids scale to 0.37 (Lm1 α BM0,37, Wm1 α BM0,38 the value 1, whereas allometry (negative or positive) was established and TRL α BM0,36) and areas to 0,71 (m1AA α BM0,70, TRLAA α BM0,72). when this value was not included. On the other hand, they found that in long bones L α BM0,43 and D α In addition, multiple regressions (Y = a + b1X1 + b2X2 + … + bkXk; BM0,39 in lagomorphs. where k is the total number of independent variables) were carried out For the assessment of scaling, the first step was to carry out a sta­ using Partial Least Square Regression (PLS or PLSR) models. This data- tistical descriptive analysis of the morphometric variables. The dimension reduction test was considered suitable in the present case normality (Shapiro-Wilk test) and the homogeneity of variances (Lev­ because the number of variables was higher than number of cases ene’s test) were also assessed. Arithmetic averages of morphometric (Höskuldsson, 1988). This technique reduces the number of predicting measurements were calculated for the 9 studied fossiliferous sites. The variables to a set of components (orthogonal factors, predictors), which 3 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 A D E F HA Lp3 PD p FTDp TTDp Wp3 HL B Lm1 HAPDd HTDd FL TL Wm1 C TRL FTDd TTDd Fig. 2. Measurements taken on Prolagus remains. A, p3: Lp3, length of p3; Wp3, width of p3. B, m1: Lm1, length of m1; Wm1, width of m1. C, hemimandible: TRL, toothrow length. D, humerus: HL, humerus length; HAPDp, proximal humeral antero-posterior diameter; HAPDd, distal humeral antero-posterior diameter; HTDd, distal humeral transversal diameter. E, femur: FL, femur length; FTDp, proximal femoral transversal diameter; FTDd, distal femoral transversal diameter. F, tibia: TL, tibia length; TAPDp, proximal tibial antero-posterior diameter; TTDp, proximal tibial transversal diameter; TTDd, distal tibial transversal diameter. Illustrated material belongs to P. sardus from Dragonara cave. describe the maximum correlation between explanatory and response see results Section 3.2. for details): variables, and, in addition, removes the multicollinearity (Mateos- observed m1AA Aparicio Morales and Caballero Domínguez, 2009; Šerbetar, 2012). This TSIm1AA = predicted m1AA from the BM test was fitted using the R package “plsdepot” (Sánchez, 2016). The significance level of all mentioned tests was fixed at 0.05. This Values of TSI > 1 indicate an increase of teeth size compared with the statistical methodology is extensively described in Quinn and Keough average (control) extant species (= megalodontia); whereas values of (2002) and McDonald (2014). All the analyses were performed using the TSI < 1 mean microdontia. Predicted teeth areas of extinct populations software Jamovi version 0.9.2.3 (The Jamovi Project, 2019), Past 3.25 were obtained using the allometric models described by Moncunill-Solé (Hammer et al., 2001) and R 3.6.1. (R Core Team, 2019). et al. (2015), which were established independently by extant ochoto­ nids (log BM = 0.566 + 1.397 log TRLAA and log BM = 1.454 + 1.418 log m1AA) and leporids (log BM = 0.462 + 1.659 log TRLAA and log BM 2.4. Teeth size index (TSI) = 1.890 + 1.470 log m1AA). The BM of each specimen was calculated using tibial diameters (TAPDp, TTDp and TTDd), which are the pa­ Teeth size indices were calculated to evaluate evolutionary trends in rameters that show the best correlation with BM in Prolagus (see present the dentition of Sardinian Prolagus. They were based on the postcanine results, Moncunill-Solé et al., 2016a). megalodontia (= megadontia) quotient established by McHenry (1988) ANCOVA tests were used to evaluate statistically the m1AA and in the study of extinct primates and early hominins. This index was TRLAA differences among groups adjusted for the size (covariate: BM, calculated considering: (i) the area of lower premolars and molars TAPDp, TTDp and TTDd). Data were log transformed for analysis. The (TRLAA): established categories or groups (fixed factor) were: (i) Sardinian Pro­ observed TRLAA lagus, (ii) extant ochotonids, and (iii) extant leporids (Supplementary TSITRLAA = predicted TRLAA from the BM Data, Appendix A). Whilst it would be interesting to consider mainland Prolagus species in the study, the general absence of their long bones and (ii) the area of the m1 (m1AA, teeth that showed the highest cor­ (poor and in bad conditions) prevents it. The significance level was fixed relation coefficient and significant dependency with postcranial bones, 4 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 at 0.05 and the tests were carried out using the software Jamovi version Table 3 0.9.2.3 (The Jamovi Project, 2019). Humerus measurements (in mm) of P. figaro and P. sardus per fossiliferous site: mean, standard deviation (SD) and number of specimens (N). 3. Results HL HAPDp HTDd HAPDd Mean 35.80 8.36 6.51 3.66 3.1. Descriptive analysis X3 (SD) (2.530) (0.826) (0.631) (0.404) Prolagus N 5 7 12 12 The material of Prolagus analyzed in this paper was not fairly figaro Mean – – – – X4 (SD) distributed among sites. Dragonara, the richest assemblage, provided N 0 0 0 0 more than 1200 fossils, whereas in other localities the number of re­ Mean 34.30 5.94 3.38 mains was significantly lower (e.g. X3 or Grotta Teulada) (Tables 2–5). Cava – (SD) (1.150) (0.417) (0.228) Alabastro The relative abundance of each bone/teeth was not uniform either. In N 3 0 43 53 general, the most frequently recovered bone was the humerus, in Mean 38.20 8.82 6.59 3.87 XIr (SD) (1.480) (0.577) (0.360) (0.215) particular those in which distal epiphysis was preserved (Table 3). N 47 42 53 54 Regarding teeth, in some cases p3 was the most abundant (X4 or VIb6), Mean 39.10 8.30 6.11 3.53 Grotta dei but in other sites m1 predominated (XIr or S’Omu e S’Orku) (Table 2). (SD) (2.820) (0.707) (0.457) (0.365) Fiori SD X3 and Grotta dei Fiori SD localities were characterized by the N 3 2 31 32 Mean 6.21 3.51 smallest teeth dimensions (p3 and m1), whereas the largest ones were Prolagus Grotta – – (SD) (0.561) (0.398) retrieved from Dragonara. Considering TRL, the specimens from Grotta sardus Teulada N 0 0 7 7 dei Fiori SD, XIr and Grotta Teulada were the smallest and S’Omu e Mean 38.70 8.98 6.63 3.92 S’Orku ones the largest (Table 2). The smallest humeral values were VI banco 6 (SD) (1.730) (0.696) (0.341) (0.246) observed in Cava Alabastro specimens and the largest ones in VIb6, N 44 41 44 44 Mean 39.80 8.64 6.81 4.03 Dragonara and S’Omu e S’Orku (Table 3). By contrast, Cava Alabastro Dragonara (SD) (1.940) (0.570) (0.462) (0.315) and Dragonara individuals showed the largest femora, whereas those cave N 48 57 318 334 from VIb6, X3 and X4 the smallest (Table 4). In the case of tibia, it was Mean 45.00 7.93 6.44 3.69 S’Omu e observed a stable pattern in all the assessed variables: X3 specimens had S’Orku (SD) (6.360) (0.833) (0.429) (0.311) N 2 3 152 158 the smallest values and those from Dragonara the largest (Table 5). It is worth to notice that some variables did not follow a normal Variables with a non-normal distribution are underlined, while those without distribution (e.g. humerus variables), and/or their variances were not homogeneity of variances among groups are in bold. See ages of fossil sites in homogenous among groups (e.g. p3 variables) (Tables 2–5, Supple­ Table 2. mentary Data, Appendix A). In these cases, the subsequent statistical tests were non-parametric. Also noteworthy is that some histograms 3.2. Correlation, bivariate regression and PLS analyses seem not to be unimodal distributions (Supplementary Data, Appendix A), but they probably result from the low number of specimens exam­ The study showed significant correlations (p < 0.05) among some of ined in these populations (e.g. X3 or Grotta dei Fiori SD). the assessed variables (Fig. 3), as well as dependency (Table 6; Table 2 Teeth measurements (in mm) of P. figaro and P. sardus per fossiliferous site: mean, standard deviation (SD) and number of specimens (N). Lp3 Wp3 p3AA Lm1 Wm1 m1AA TRL TRLAA Mean 2.11 1.86 4.01 2.73 (− ) 2.43 (− ) 6.62 (− ) – – X31 (SD) (0.383) (0.272) (1.280) Prolagus N 7 7 7 1 1 1 0 0 figaro Mean 2.59 2.53 6.53 2.52 2.75 6.95 12.50 34.80 X41 (SD) (0.320) (0.322) (1.520) (0.228) (0.284) (1.250) (0.678) (5.320) N 151 141 130 51 44 44 15 12 Mean 2.77 2.85 7.90 2.45 2.80 6.94 12.30 32.80 Cava Alabastro2 (SD) (0.188) (0.173) (0.886) (0.212) (0.293) (1.180) (0.736) (4.620) N 25 31 24 28 32 27 15 8 Mean 2.66 2.90 7.72 2.73 3.08 8.46 12.90 40.10 XIr3 (SD) (0.177) (0.284) (1.120) (0.184) (0.232) (1.100) (0.664) (4.480) N 52 49 49 133 116 114 108 86 Mean 2.32 2.43 5.79 2.25 2.56 5.87 11.10 29.40 Grotta dei Fiori (SD) (0.427) (0.386) (2.080) (0.215) (0.302) (1.280) (0.901) (6.690) SD4 N 9 9 9 16 13 13 8 5 Mean 2.66 3.03 8.05 2.68 2.86 7.75 12.30 35.60 Prolagus Grotta Teulada5 (SD) (0.116) (0.183) (0.763) (0.091) (0.259) (0.854) (0.533) (3.790) sardus N 7 6 6 5 4 4 6 4 Mean 2.06 2.24 4.62 – – – – – VI banco 66 (SD) (0.134) (0.177) (0.591) N 51 48 48 0 0 0 0 0 Mean 2.73 3.05 8.36 2.93 3.24 9.52 13.00 42.40 7 Dragonara cave (SD) (0.152) (0.227) (0.972) (0.167) (0.219) (1.020) (0.599) (4.050) N 54 53 53 140 133 132 146 124 Mean 2.65 2.76 7.48 2.83 3.03 8.58 13.20 40.40 S’Omu e S’Orku8 (SD) (0.151) (0.219) (0.865) (0.146) (0.247) (1.050) (0.632) (4.560) N 70 71 59 107 64 60 119 56 Ages: 1 latest Pliocene - Early Pleistocene; 2 Middle -?Late Pleistocene; 3?Late Pleistocene; 4 Middle Pleistocene (>350 ka); 5 Late Pleistocene; 6 Middle or Late Pleistocene; 7 Late Pleistocene (22,390–21,910 ka cal BP); 8 Holocene. Variables with a non-normal distribution are underlined, while those without homogeneity of variances among groups are in bold. 5 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Table 4 Lm1) rarely correlated significantly (Lp3 was only correlated with Wp3 Femur measurements (in mm) of P. figaro and P. sardus per fossiliferous site: and Wm1; and Lm1 with TRL). The established dependency among teeth mean, standard deviation (SD) and number of specimens (N). variables was mostly isometric (p < 0.05, b = 1), with the exception of FL FTDp FTDd Wm1/Wp3 and TRL/Lm1 (Table 6). By contrast, a general lack of cor­ Mean 50.20 9.37 7.81 relation was observed among HL and FL and other postcranial variables. X3 (SD) (3.020) (1.090) (1.330) This pattern was also followed by some epiphyses of stylopods (HAPDp Prolagus N 2 10 5 and FTDd) (Fig. 3). In the case of distal humerus diameters, they showed figaro Mean – 9.55 7.43 a high dependent correlation (r = 0.952, p < 0.05, b = 1). Both variables X4 (SD) (0.746) (0.484) were also correlated with tibial variables (r = 0.762–0.943), although it N 0 10 3 Mean 9.72 9.70 was only isometric dependent (p < 0.05) in the case of HAPDd with TL Cava 54.10 (− ) and TTDd. Regarding FTDp, it was positively correlated with tibial (SD) (1.380) (0.819) Alabastro N 1 5 3 measurements in an isometric way (r = 0.766–0.895, b = 1) (Fig. 3). Mean 48.90 10.80 8.59 Strong and significant correlations (r = 0.892–1.000) were also identi­ XIr (SD) (2.470) (0.567) (0.568) N 52 55 58 fied assessing tibial variables (Fig. 3, Table 6). In general, they corre­ Mean 10.10 8.16 lated isometrically, excepting TAPDp and TL (hyperallometric) and Grotta dei TTDd and TTDp (hypoallometric). In other words, as tibia increased in – (SD) (1.180) (0.850) Fiori SD N 0 5 5 length, it became proportionally more robust and tibial head increased Mean 48.80 10.50 8.84 Prolagus Grotta at a faster rate than distal epiphysis. Finally, a general lack of significant (SD) (2.550) (0.821) (0.590) sardus Teulada N 2 11 8 and dependent relationships was observed among teeth variables and Mean 48.40 10.50 8.48 those from postcranium, with few exceptions (Fig. 3, Table 6). Most of VI banco 6 (SD) (2.700) (0.726) (0.464) them corresponded to Wm1 (r = 0.786–0.976), which was isometrically N 19 19 19 related with femoral and tibial variables (b = 1, excepting TTDp b < 1). Mean 50.90 11.20 9.28 Dragonara This means that when TTDp and Wm1 increase in size, the latter grow at (SD) (3.180) (0.944) (0.724) cave a lower rate (Table 6). N 39 126 113 Mean 49.40 10.40 8.70 The first two components yielded by PLSR models were good proxies S’Omu e S’Orku (SD) (2.750) (0.866) (0.500) of morphometric variables (total variance explained = 61,53-99,83%). N 5 40 39 Notably when Wm1, FTDp, TL, TAPDp and TTDd were considered the Variables with a non-normal distribution are underlined, while those without dependent variable, the variance explained was over 95%. By contrast, homogeneity of variances among groups are in bold. See ages of fossil sites in the lower values were obtained when FL and FTDp were assessed (not Table 2. exceeding 70%). In general, Grotta Teulada and XIr were the fossil sites less correlated with first and second component of PLSR models, con­ trasting with Dragonara and Cava Alabastro results. The lack of associ­ Table 5 ation among populations stated us that they were well differentiated. All Tibia measurements (in mm) of P. figaro and P. sardus per fossiliferous site: morphometric variables were well represented in these two first axes mean, standard deviation (SD) and number of specimens (N). (high explained variances). In this regard, the circle of correlations of TL TAPDp TTDp TTDd the models allowed to identify strong relationships among them. In Mean 48.50 6.68 6.19 6.25 general, Wm1, FTDp and tibial measurements were disposed together in X3 (SD) (2.080) (0.748) (0.427) (0.488) the same morphospace. This also happened considering: (i) HAPDd and Prolagus N 2 5 5 2 HTDd, and (ii) Lp3 and Wp3. On the contrary, the other variables had figaro Mean 8.73 8.32 6.99 X4 (SD) – (0.777) (0.842) (0.693) more erratic locations. This was particular true in the case of FL, whose N 0 6 6 5 correlations were never associated to other traits. Cava Mean – 7.42 7.80 6.61 Accordingly to the present results, m1 was the dental element with a (SD) (0.487) (0.424) (0.408) higher and significant correlation and dependency with postcranial el­ Alabastro N 0 5 5 9 Mean 53.40 8.76 9.28 7.57 ements. In the case of long bones, the distal part of the humerus and XIr (SD) (2.610) (0.582) (0.576) (0.458) tibial epiphyses showed the best intercorrelations. N 39 36 34 38 Grotta dei Mean – – – 6.65 3.3. TSIs and ANCOVA tests (SD) (0.582) Fiori SD N 0 0 0 6 Mean 49.90 7.80 8.20 7.23 Overall, Sardinian Prolagus populations showed TSIm1AA and TSITR­ Prolagus Grotta (SD) (− ) (− ) (− ) (0.403) LAA values >1 considering both ochotonids and leporids as control sardus Teulada N 1 1 1 6 groups (Table 7). Only in two samples (X4 and Grotta dei Fiori SD), Mean 54.50 8.99 9.71 7.94 TSIm1AA was <1 when ochotonids were the reference group (0.97 and VI banco 6 (SD) (2.500) (0.478) (0.271) (0.576) N 9 8 8 9 0.92 respectively). This means that Sardinian Prolagus populations had Mean 56.40 9.35 9.89 8.02 much larger post-canine teeth (the double or threefold larger) than ex­ Dragonara (SD) (3.230) (0.775) (0.749) (0.602) pected for an average modern leporid of their sizes (ANCOVA test, p < cave N 71 124 85 303 0.05) (Fig. 4; Supplementary Data, Appendix A). When considering Mean 52.10 8.51 9.15 7.36 S’Omu e ochotonids, these increases were slighter (3%–42% larger) but attestable (SD) (3.250) (0.709) (0.769) (0.588) S’Orku N 4 29 22 16 (ANCOVA test, p < 0.05) (Fig. 4; Supplementary Data, Appendix A). Accordingly, the results pointed that Sardinian Prolagus was a mega­ Variables with a non-normal distribution are underlined, while those without lodont lineage when compared to extant ochotonids and leporids. homogeneity of variances among groups are in bold. See ages of fossil sites in Table 2. Supplementary Data, Appendix A). Significant correlations were observed among teeth variables, espe­ cially when Wm1 was assessed. On the contrary, tooth lengths (Lp3 and 6 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 A B Spearman’s ρ Pearson’s r HAPDp HAPDd HAPDp HAPDd TAPDp TAPDp HTDd HTDd FTDp FTDd TTDp TTDd FTDp FTDd TTDp TTDd Wm1 Wm1 Wp3 Wp3 Lm1 Lm1 TRL TRL Lp3 Lp3 HL HL FL TL FL TL Lp3 1 Lp3 1 Wp3 * Wp3 * Lm1 Lm1 Wm1 * * Wm1 * * TRL * * * TRL * * HL 0.333 HL 0.333 HAPDp HAPDp HTDd * HTDd * HAPDd * HAPDd * * FL FL FTDp * -0.333 FTDp * * -0.333 FTDd FTDd * * TL * * * TL * * * TAPDp * * * TAPDp * * * * * TTDp * * * * * TTDp * * * * * * TTDd * * * * * * -1 TTDd * * * * * * * * -1 Fig. 3. Pearson’s r (A) and Spearman’s ρ (B) correlation matrix among variables of teeth, hemimandible and postcranial bones taken on Sardinian Prolagus. In the upper triangle, colors and size of circles are indicative of the correlation values, from 1 (blue and green, respectively) to − 1 (red and purple). In the lower triangle, asterisks (*) show significant correlations (p < 0.05). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 6 Parameters of the linear bivariate regression models (Y = a + bX) between teeth and postcranial measurements (which correlated significatively). Y X a b p-value N r2 RMSE BIC VIF CI b Scaling Wp3 Lp3 − 0.1 1.299 <0.001 9 0.813 0.0289 − 31.7 1 0.741–1.857 b¼1 Wm1 Lp3 0.103 0.859 0.009 8 0.704 0.0212 − 32.7 1 0.303–1.414 b¼1 Wm1 Wp3 0.24 0.501 0.007 8 0.731 0.0202 − 33.5 1 0.197–0.805 b<1 TRL Lm1 0.855 0.575 0.005 7 0.823 0.0098 − 39.1 1 0.268–0.882 b<1 TRL Wm1 0.798 0.643 0.009 7 0.777 0.011 − 37.5 1 0.247–1.040 b¼1 Lm1 HTDd − 0.836 1.566 0.024 7 0.605 0.0209 − 28.5 1 0.306–2.826 b¼1 Lm1 HAPDd 0.112 0.343 0.827 5 0.019 0.0395 − 13.3 1 − 9.17 b=1 Wp3 FTDp − 1.35 1.75 0.067 9 0.402 0.0516 − 21.2 1 − 3.808 b=1 Wm1 FTDp − 0.863 1.304 0.008 8 0.718 0.0207 − 33.1 1 0.487–2.122 b¼1 Lp3 FTDd − 0.272 0.718 0.149 9 0.273 0.0395 − 26 1 − 2.095 b=1 Wp3 FTDd − 0.785 1.289 0.058 9 0.424 0.0507 − 21.6 1 − 2.688 b=1 Wm1 TL − 2.46 1.7 0.036 5 0.816 0.0185 − 20.9 1 0.218–3.186 b¼1 Wm1 TAPDp − 0.18 0.701 0.011 7 0.705 0.0187 − 30 1 0.241–1.161 b¼1 Wm1 TTDp − 0.088 0.594 <0.001 7 0.953 0.0082 − 41.5 1 0.441–0.746 b<1 Wm1 TTDd − 0.498 1.119 <0.001 8 0.9 0.0124 − 41.6 1 0.746–1.493 b¼1 TRL TTDd 0.574 0.608 0.061 7 0.538 0.0158 − 32.4 1 − 1.3 b=1 HAPDd HTDd − 0.458 1.272 <0.001 8 0.931 0.0065 − 51.9 1 0.927–1.618 b¼1 TAPDp HTDd − 0.355 1.541 0.166 7 0.345 0.0389 − 19.7 1 − 4.88 b=1 TTDp HTDd − 0.314 1.538 0.341 7 0.181 0.06 − 13.7 1 − 7.52 b=1 TTDd HTDd − 0.186 1.293 0.077 8 0.431 0.0278 − 28.4 1 − 2.97 b=1 TL HAPDd 1.153 0.981 0.016 6 0.802 0.0099 − 33 1 0.304–1.660 b¼1 TAPDp HAPDd 0.104 1.416 0.056 7 0.552 0.0322 − 22.4 1 − 2.936 b=1 TTDp HAPDd 0.025 1.587 0.151 7 0.365 0.0528 − 15.5 1 − 4.821 b=1 TTDd HAPDd 0.204 1.15 0.025 8 0.593 0.0236 − 31 1 0.198–2.101 b¼1 TL FTDp 0.912 0.792 0.035 6 0.712 0.0119 − 30.8 1 0.093–1.490 b¼1 TAPDp FTDp − 0.39 1.292 0.044 8 0.518 0.0319 − 26.2 1 0.047–2.538 b¼1 TTDp FTDp − 1.1 2.01 0.011 8 0.636 0.0347 − 24.8 1 0.659–3.363 b¼1 TTDd FTDp − 0.419 1.262 0.002 9 0.763 0.017 − 41.2 1 0.633–1.891 b¼1 TAPDp TL − 2.65 2.07 0.004 6 0.901 0.0153 − 27.7 1 1.120–3.030 b>1 TTDp TL − 3.87 2.79 0.017 6 0.795 0.0315 − 19.1 1 0.824–4.762 b¼1 TTDd TL − 1.69 1.49 0.009 6 0.847 0.014 − 28.8 1 0.608–2.363 b¼1 TTDp TAPDp − 0.243 1.279 <0.001 8 0.898 0.0198 − 33.8 1 0.849–1.709 b¼1 TTDd TAPDp 0.219 0.698 <0.001 8 0.836 0.0142 − 39.1 1 0.390–1.007 b¼1 TTDd TTDp 0.358 0.539 <0.001 8 0.908 0.0107 − 43.7 1 0.367–0.711 b<1 Columns: Y (log-transformed dependent variable), X (log-transformed independent variable), a (intercept), b (slope), p-value (of the regression model), N (number of assessed populations), r2 (adjusted coefficient of determination), RMSE (root mean square error), BIC (Bayesian information criterion), VIF (Variance Inflation Factor), CI b (95% bootstrapped confidence intervals of b, N = 1999), Scaling (b = 1 indicates isometry, b < 1 indicates hypoallometry, and b > 1 indicates hyperallometry). Bold type indicates significant p-values. 4. Discussion palaeontologists can collect from small mammals. In lagomorphs, loose teeth are frequent, whereas complete toothrows (hemimandibles or 4.1. Best BM proxy for extinct lagomorphs: A compromise between maxillae) are unusual and postcranial bones are rare (López-Martínez, availability and suitability 1989). In this regard, the taxonomy of most extinct lagomorph species, as well as their body size inferences, have been based on teeth remains. Teeth are the most common, sometimes quasi-exclusive remains that The m1 is historically the preferred tooth for estimating BM in large and 7 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Table 7 vertebrates. Hindlimb bones are slightly modified by lifestyle in extant TSIm1AA and TSITRLAA of the different populations of Sardinian Prolagus species. ochotonids (Reese et al., 2013), and the sizes of their epiphyses (TAPDp, Site TSIm1AA TSITRLAA TTDp, TTDd and FTDp) provided the highest intercorrelation values (Fig. 3). In addition, their growth plates (tibia and femur) count among Ochotonidae Leporidae Ochotonidae Leporidae the last to close in lagomorphs (Geiger et al., 2013), suggesting that they X3 1.42 2.97 – – represent better the adult body size. This is well in line with previous X4 0.97 2.06 1.09 1.63 Cava Alabastro 1.12 2.37 1.19 1.75 studies, which identified these bones as the most suitable BM proxies in XIr 1.06 2.26 1.13 1.72 lagomorphs (Quintana et al., 2011; Moncunill-Solé et al., 2015, 2016a). Grotta dei Fiori SD 0.92 1.94 1.03 1.52 Instead, the size variation of some other traits cannot be mainly attrib­ Grotta Teulada 1.06 2.25 1.09 1.64 utable to a scaling increase/decrease of the body plan (e.g. FL or VI banco 6 – – – – HAPDp). Several researchers have noted that, in extant ochotonids, the Dragonara cave 1.07 2.28 1.07 1.65 S’Omu e S’Orku 1.13 2.4 1.19 1.8 dimensional variation in these latter traits would be more related to their specialized lifestyle, rather than to their BM (Reese et al., 2013; Values of TSI > 1 indicate megalodontia; whereas values of TSI < 1 mean Young et al., 2014). Consequently, they will not be suitable for BM re­ microdontia. constructions of past species. Considering tibia epiphyses, the body masses of P. figaro and P. sardus were 350 g (ca. 250–460 g) and 500 g small extinct mammals (Gingerich et al., 1982; Legendre and Roth, (ca. 400–650 g) respectively (Moncunill-Solé et al., 2016a), although 1988; Legendre, 1989; Damuth and MacFadden, 1990; Millien and significant fluctuations over time were identified. Bovy, 2010; Freudenthal and Martín-Suárez, 2013; Moncunill-Solé The scaling relationships among linear dimensional variables (L, W et al., 2014, 2015, 2016c). However, in lagomorphs, loose m1s are very and D) of teeth and long bones belonging to Sardinian Prolagus lineage difficult to identify for non-specialists. Instead, p3 dimensions (the most were mainly isometrics (L α D; Table 6). Therefore, our empirical results recognizable tooth at taxonomic level) have traditionally been used to support both: (i) the geometric similarity approach (L α D α BM1/3) gain a rough idea of the size of extinct lagomorph taxa. Undeniably, the (Economos, 1983), and (ii) the Garcia-Silva model (L α D α BM0,37) tiny-toothed mainland taxon Prolagus praevasconiensis Ringeade, 1979 of (Garcia and da Silva, 2004, 2006), but disaffirm the ESM (McMahon, the early Miocene of Switzerland, at date the oldest known species of the 1973, 1975). Moncunill-Solé et al. (2015) established that longitudinal genus, surely had a smaller BM than the species with the largest p3, the measurements (of long bones and teeth) scale to BM0,36–0,39 in ochoto­ insular taxon P. imperialis Mazza, 1987 of the Messinian of Italy (Gar­ nids, with the exception of lengths of long bones (L α BM0,43). In this gano Peninsula). However, lagomorph researchers suspected that p3/ regard, L increases a faster rate than expected by theoretical models. BM scaling relationship was not sufficiently accurate. A preliminary Because the dimensions of bones are strongly depending on limb usage approach for shedding light on this issue was carried out by Moncunill- (Garcia and da Silva, 2004, 2006), the particular locomotion of lago­ Solé et al. (2016a), who evidenced different BM scaling patterns in p3 morphs (most of them are very specialized racers and jumpers) probably and long bones in some populations of P. figaro and P. sardus. Para­ is key to defining the scaling relationships in this mammalian order. At doxically, P. figaro was commonly considered quite larger than its direct present and considering the abovementioned, data of extant and extinct descendent P. sardus (on the basis of Lp3), but BM estimations based on lagomorphs likely fit better the Garcia-Silva model. postcranial bones demonstrated that it was just the opposite (Moncunill- Solé et al., 2016a: Fig. 2). 4.2. Megalodontia: Is it a response to ecological stressors of islands? In accordance with this, the large data set used in the present study revealed a lack of significant correlations between the size of p3 and Lagomorphs show a dentition perfectly adapted to a fibrous diet postcranial bones (p > 0.05), demonstrating statistically a decoupling (Varga, 2014). The sharpened incisors are used as plant vertical cutters, between the size variation of such skeletal remains. A similar disasso­ whereas posterior cheek teeth rows (premolars and molars) grind the ciation has already been noted in other large and small mammals food before it is swallowed. The jaw movements are essential for a (Damuth and MacFadden, 1990; Egi, 2001; Millien and Bovy, 2010; correct food processing by occlusal surfaces of post-canine teeth, and, Moncunill-Solé et al., 2014, 2015). However, unexpectedly, the study of ultimately, the food intake should ensure a correct nutrition of the or­ Sardinian Prolagus also established iso- and hypoallometric statistical ganism (Quesenberry and Carpenter, 2011; Varga, 2014). In this regard, relationships between m1 (W) and postcranial bones (mainly tibia and the size (occlusal area or crown height) and shape (pattern of enamel femur) (p < 0.05; Supplementary Data, Appendix A). In lagomorphs, due bands and complexity) of cheek teeth are subjected to selection, and to their larger evolutionary variability, p3 and other premolars (upper or their relative values give us clues to the biology and ecology of the lower) are used with taxonomic purposes (López-Martínez, 1989). By species (Lucas, 2004; Famoso et al., 2013). contrast, m1 (located in the middle of the toothrow) shows a more Sardinian Prolagus species showed a significant larger occlusal area conservative morphology and structure (López-Martínez, 1989) and it is of post-canine teeth than extant relatives when size effect was removed not subjected to biting forces, which are concentrated at the front of (Table 7, Fig. 4). In comparison with leporids, m1AA of these extinct toothrow (Mazza and Zafonte, 1987). This more “stable” position of m1 taxa was proportionally much larger (Fig. 4A). Regarding the dental could entail a narrower and more persistent scaling trend between its formulae of both groups, leporid lower toothrow includes a bi-lobed m3 size and BM. As mentioned above, in most lagomorphs, p4 and m1 show (Hillson, 2005), whereas Prolagus lacks independent m3 (though having a similar overall appearance, what increases the difficulty to distinguish instead a tri-lobed m2). Consequently, the presence of a larger m3 in them. However, in the case of Prolagus, m1 is characterized by a more leporids could entail that each tooth occupies a smaller area (keeping curved shaft that may help discriminate it. anyway into account that m3 is indeed quite small). On the other hand, Postcranial dimensions (excepting HL, HAPDp, FL and FTDd) were the m1AA of Sardinian Prolagus was also larger than in extant ochoto­ also significantly related (p < 0.05, r > 0.762), notably in the case of nids (Table 7, Fig. 4A), but, in this case, both groups have the same lower femur (FTDp) and tibia (TL, TAPDp, TTDp, TTDd) (variation explained toothrow pattern (Dawson, 1969; Hillson, 2005). Similarly, when >95%, sharing morphospace). This strong scaling relationship is asso­ TRLAA was considered, the results also identified a larger post-canine ciated to the weight-bearing function that they carry out (Moncunill- area in Sardinian Prolagus (Table 7, Fig. 4B). This latter value is also Solé et al., 2015). By contrast, teeth experienced more adaptive evolu­ independent of the number of teeth. Hence, these larger occlusal areas tion, and the variation on their size would be related to other biological are not solely the result of the size increase that Sardinian ochotonids factors, such as diet (Damuth, 1990). In this regard, skeletal long bones, experienced in insular regimes (Island Rule). Other ecological factors when available, are preferable as BM proxies in lagomorphs and other should play a pivotal role in its selection. 8 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Fig. 4. Bivariate scatterplots between log-transformed body size proxies (BM and tibial epiphyses dimensions) and post-canine areas (A, C, E, G: m1AA; B, D, F, H: TRLAA), considering extant lagomorph species (leporids and ochotonids) and Sardinian Prolagus. ANCOVA test identified significant differences (p < 0.05) between Sardinian Prolagus and extant lagomorphs (ochotonids and leporids), with the exception of panel H where these are only with ochotonid group. Legend: green triangles: extant ochotonids; orange squares: extant leporids; and violet circles: Sardinian Prolagus. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) The mastication process evolved to increase the relative surface of the lifespan of the individual, leading even to mastication rate decli­ food particles, allowing a better action of digestive enzymes and a higher nation consequence of severe wear. In this regard, an increase of rate of energy acquisition (Lucas, 2004). Thus, evolutionarily, it was occlusal surface of cheek teeth is described as an evolutionary response established a direct association between rate of energy requirement of to a long-term abrasiveness (Lucas et al., 1985; Lucas, 2004). mammals and need of mastication. Post-canine teeth (= the masticatory Lagomorphs are highly selective in food choice, and they usually unit) are responsible for food crushing and shearing, and an increase of feed on the most tender and succulent regions of vegetables or plants their occlusal area enhances the chance of food to be fractured at a (buds and young leaves, grasses, or flowers; Quesenberry and Carpenter, smaller scale (Lucas, 2004). In mammals, it is stated an evolutionary 2011). Studying the p3 of insular endemic species of Prolagus, Angelone relationship among cheek tooth size, body size and dietary preferences (2005) defined two indices: (i) hypsodonty degree (ibid., Tab. 3, p. 20), (considering the extent, stickiness and abrasiveness of food surface) at which evaluates the absolute value of the tooth shaft height not biased the specific level (Fortelius, 1985; Lucas, 2004). Whereas the stickiness by the overall teeth size; and (ii) density index (ibid., Tab. 1, p. 18), of food changes during mouthful processing, abrasiveness acts during which quantifies the enamel complexity. Two different patterns were 9 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 noticed in Sardinian species. Prolagus figaro showed an hypsodonty de­ leporids and ochotonids) when the size effect was removed. Besides, gree similar to congeneric species but a high density index (more com­ Sardininan Prolagus dentition is hypsodont (P. sardus) and with a more plex occlusal surface enamel pattern). On the contrary, P. sardus showed complex enamel pattern (P. figaro). Such dental features have been the highest, out of scale value of hypsodonty degree among the assessed noted in herbivores to face diet abrasiveness, which allow that tooth last species (both insular endemics as well as from mainland), but a quite for a longer time (food habits hypothesis). In accordance with this and low density index. Both modifications have been noted in small and considering the ecology of the studied species, megalodontia may be an large herbivores to cope with diet abrasiveness (Jernvall and Fortelius, evolutionary response to the particular stressors of islands (more abra­ 2002; Fortelius et al., 2002; Famoso et al., 2013, 2016). In accordance sive diet). At the moment other biological and ecological factors, such as with these results, the diet of past insular endemics is described as more a longer longevity, cannot be excluded (life history hypothesis). abrasive, as consequence of the limited-resource ecosystems that they inhabited (especially in arid periods) (Alcover et al., 1999; Hautier et al., Declaration of Competing Interest 2009; Winkler et al., 2013). Frequently, they had to resort to fallback resources and they rooted, overgrazed or near-surface browsed, intro­ The authors declare that they have no known competing financial ducing soil particles (grit) in the ingestion (van der Geer, 2014). In this interests or personal relationships that could have appeared to influence regard, the sister species from the Gargano palaeo-archipelago (Prolagus the work reported in this paper. arpicenicus Mazza, 1987 and P. imperialis) have been described as grassland dwellers, well adapted to an abrasive diet (De Giuli et al., Acknowledgments 1986). Although at present an in-depth study on the diet of Sardinian Prolagus lacks (isotopic or wear analysis), all these evidences suggest a The authors thank Howard Falcon-Lang (Editor), Alexandra van der significant feeding shift towards a more abrasive ingestion in Sardinian Geer and one anonymous referee for their helpful comments and advice, Prolagus. This diet change could have been the evolutionary trigger of which helped us improve this research. This work was supported by megalodontia. Hypsodonty and megalodontia enhance the resistance of Consellería de Cultura, Educación e Universidade, Xunta de Galicia, tooth wear (Lucas, 2004), whereas an increase of enamel complexity Spain (BM-S, ED481B 2018/046, Axudas á etapa postdoutoral da Xunta suggests a more efficient food processing but also a lower wear (Pérez- de Galicia 2018 – Modalidade A), Visiting Professor grant of the Presi­ Barberia and Gordon, 1998; Kaiser et al., 2010; Famoso et al., 2013). dent’s International Fellowship Initiative of the Chinese Academy of These eco-evolutionary adaptations maximize the longevity of the per­ Science (CA), Spanish Agencia Estatal de Investigación and the Euro­ manent dentition and it seems likely that they respond to the same pean Regional Development Fund of the European Union (CA, CGL2016- ecological stressors of the insular energy-limited environments. 76431-P), CERCA Program of the Generalitat de Catalunya (CA), and On the other hand, Angelone (2005) criticized the interpretation that Grant to Department of Science, Roma Tre University (CA, MIUR-Italy diet shift is the single driver of such dental modifications in insular Dipartimenti di Eccellenza, ART. 1, C. 314-337L. 232/2016). Funding small-sized endemics (food habits hypothesis), due to the low likelihood for open access charge: Universidade da Coruña/CISUG. that the same type of environment (arid, dry, open ecosystems, with harshly limited resources, and phytolith-rich plants) occurs in all Appendix A. Supplementary data islands. Other biological and ecological factors might have had a main selective role, such as an extended life (life-history hypothesis) (Jordana Supplementary data to this article can be found online at https://doi. et al., 2012; Angelone et al., 2017). Some authors have researched the org/10.1016/j.palaeo.2021.110645. life history of some insular small mammals, and have assessed the relationship of the slow life history with hypseolodonty and hypsodonty References dental features (Moncunill-Solé et al., 2016b; Angelone et al., 2017). However, the palaeobiological data of the Sardinian Prolagus are Abbazzi, L., Angelone, C., Arca, M., Barisone, G., Bedetti, C., Delfino, M., Kotsakis, T., currently rather limited to draw further conclusions. Also relevant will Marcolini, F., Palombo, M.R., Pavia, M., Piras, P., Rook, L., Torre, D., Tuveri, C., Valli, A.M.F., Wilkens, B., 2004. Plio-Pleistocene fossil vertebrates of Monte be assessing cheek teeth size and BM of the coeval mainland congenerics Tuttavista (Orosei, Eastern Sardinia, Italy), an overview. Riv. Ital. Paleontol. Stratigr. to rule out possible phylogenetic effects, and evaluating the teeth size 110, 681–706. relative to body mass in the Garganic Prolagus to strength the results and Alcover, J.A., Pérez-Obiol, R., Yll, E.I., Bover, P., 1999. The diet of Myotragus balearicus conclusions obtained in the present research. Bate 1909 (Artiodactyla: Caprinae), an extinct bovid from the Balearic Islands: evidence from coprolites. Biol. J. Linn. Soc. 66, 57–74. The evolutionary dynamics of teeth size observed here contribute to Alexander, R.M., Jayes, A.S., Maloiy, G.M.O., Wathuta, E.M., 1979. Allometry of the limb highlight again the inaccuracy of teeth as size proxy in lagomorphs, bones from shrews (Sorex) to elephant (Loxodonta). J. Zool. 189, 305–314. because their variations response to shifts in BM but also to ecological Angelone, C., 2005. Evolutionary trends in dental morphology of the genus Prolagus (Ochotonidae, Lagomorpha) in the Mediterranean Islands. In: Alcover, J.A., Bover, P. stressors (Famoso et al., 2013). (Eds.), Proceedings of the International Symposium “Insular Vertebrate Evolution: The Palaeontological Approach”, Monografies de la Societat d’Història Natural de les 5. Conclusions Balears, 12. Societat d’Història Natural de les Balears, Palma de Mallorca, pp. 17–26. Angelone, C., Tuveri, C., Arca, M., López Martínez, N., Kotsakis, T., 2008. Evolution of Prolagus sardus (Ochotonidae, Lagomorpha) in the Quaternary of Sardinia Island Several eco-evolutionary adaptations have been described in fossil (Italy). Quat. Int. 182, 109–115. insular ochotonids regarding adult body size, teeth features, lifestyle as Angelone, C., Čermák, S., Rook, L., 2017. New insights on Paludotona, an insular endemic lagomorph (Mammalia) from the Tusco-Sardinian palaeobioprovince (Italy, well as life history. The present paper contributes to this topic testing Turolian, late Miocene). Riv. Ital. Paleontol. Stratigr. 123, 455–473. tooth and long bone scaling, as well as dentition size evolutionary dy­ Angelone, C., Moncunill-Solé, B., Kotsakis, T., 2020. Fossil Lagomorpha (Mammalia) of namics in the anagenetic lineage of Sardinian Prolagus (Early Pleistocene Italy: systematics and biochronology. Riv. Ital. Paleontol. Stratigr. 126, 157–187. Biewener, A.A., 1983. Allometry of quadrupedal locomotion: the scaling of duty factor, to Holocene). Size variation of p3 is not correlated to other skeletal re­ bone curvature and limb orientation to body size. J. Exp. Biol. 105, 147–171. mains, calling into question the palaeobiological and palaeoecological Boldrini, R., 2009. Ecomorfologia e Geochimica Isotopica, Quale Supporto per Analisi reconstructions based on it. In fact, m1 was the only teeth item which Paleoambientali? Il caso dei Micromammiferi Quaternari Della Sardegna.. PhD Dissertation Università degli Studi di Roma La Sapienza, Roma, p. 322. showed isometric dependent correlations with long bones, attributable Boldrini, R., Palombo, M.R., Iacumin, P., Melis, R.T., 2010. The Middle Pleistocene to its more conservative position. Long bones, particularly tibia epiph­ fossiliferous sequence of Grotta Dei Fiori (Sardinia, Italy): multidisciplinary analysis. yses, were stated as the most reliable BM proxies in Prolagus and prob­ Boll. Soc. Paleontol. Ital. 49, 123–134. ably in extinct lagomorphs. Calder III, W.A., 1984. Size, Function and Life History. Dover Publications, Mineola, p. 431. Sardinian Prolagus was identified as a megalodont lineage, whose Christiansen, P., 1999a. Scaling of mammalian long bones: small and large mammals cheek teeth are significatively larger than those of extant relatives (both compared. J. Zool. 247, 333–348. 10 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Christiansen, P., 1999b. Scaling of the limb long bones to body mass in terrestrial López Martínez, N., Thaler, L., 1975. Biogéographie, évolution et compléments à la mammals. J. Morphol. 239, 167–190. systématique du groupe d’Ochotonides Piezodus-Prolagus (Mammalia, Lagomorpha). Copes, L.E., Schwartz, G.T., 2010. The scale of it all: postcanine tooth size, the taxon- Bull. Soc. Géol. Fr. (Série 7) 17, 850–866. level effect, and the universality of Gould’s scaling law. Paleobiology 36, 188–203. López-Martínez, N., 1989. Revisión Sistemática y Biostratigráfica de los Lagomorpha Damuth, J., 1990. Problems in estimating body masses of archaic ungulates using dental (Mammalia) del Terciario y Cuaternario de España.. PhD Dissertation Universidad measurements. In: Damuth, J., MacFadden, B.J. (Eds.), Body Size in Mammalian Complutense de Madrid, Madrid, p. 342. Paleobiology: Estimations and Biological Implications. Cambridge University Press, Lu, D., Zhou, C.Q., Liao, W.B., 2014. Sexual size dimorphism lacking in small mammals. Cambridge, pp. 229–253. North-Western J. Zool. 10, 53–59. Damuth, J., MacFadden, B.J., 1990. Body Size in Mammalian Paleobiology: Estimations Lucas, P.W., 2004. Dental Functional Morphology: How Teeth Work. Cambridge and Biological Implications. Cambridge University Press, Cambridge, p. 412. University Press, Cambridge, p. 355. Dawson, M.T., 1969. Osteology of Prolagus sardus, a Quaternary ochotonid (Mammalia, Lucas, P.W., Corlett, R.T., Luke, D.A., 1985. Plio-Pleistocene hominids: an approach Lagomorpha). Palaeovertebrata 2, 157–190. combining masticatory and ecological analysis. J. Hum. Evol. 14, 187–202. De Giuli, C., Masini, F., Torre, D., Boddi, V., 1986. Endemism and bio-chronological Lyras, G.A., van der Geer, A.A.E., Rook, L., 2010. Body size of insular carnivores: reconstructions: the Gargano case history. Boll. Soc. Paleontol. Ital. 25, 267–276. evidence from the fossil record. J. Biogeogr. 37, 1007–1021. Deng, T., 2009. Late Cenozoic environmental changes in the Linxia Basin (Gansu, China) Maglio, V.J., 1973. Origin and evolution of the Elephantidae. Trans. Am. Philos. Soc. as indicated by cenograms of fossil mammals. Vert. PalAs. 10, 282–298. (New Series) 63, 1–149. Economos, A.C., 1983. Elastic and/or geometric similarity in mammalian design? Mateos-Aparicio Morales, G., Caballero Domínguez, A.J., 2009. La regresión por J. Theor. Biol. 103, 167–173. mínimos cuadrados parciales: orígenes y evolución. In: Basulto Santos, J., García del Egi, N., 2001. Body mass estimates in extinct mammals from limb bones dimensions: the Hoyo, J.J. (Eds.), Historia de la Probabilidad y la Estadística. Universidad de Huelva, case of north American hyaenodontids. Palaeontology 4, 497–528. Huelva, pp. 441–448. Famoso, N.A., Feranec, R.S., Davis, E.B., 2013. Occlusal enamel complexity and its Mazza, P., 1987. Prolagus apricenicus and Prolagus imperialis: two new ochotonids implications for lophodonty, hypsodonty, body mass, and diet in extinct and extant (Lagomorpha, Mammalia) of the Gargano (southern Italy). Boll. Soc. Paleontol. Ital. ungulates. Palaeogeogr. Palaeoclimatol. Palaeoecol. 387, 211–216. 26, 233–243. Famoso, N.A., Davis, E.B., Feranec, R.S., Hopkins, S.S.B., Price, S.A., 2016. Are Mazza, P., Zafonte, F., 1987. Phyletic and ecologic considerations on the Gargano hypsodonty and occlusal enamel complexity evolutionarily correlated in ungulates? (Southern Italy) Prolagus (Ochotonidae, Lagomorpha, Mammalia). Boll. Soc. J. Mamm. Evol. 23, 43–47. Paleontol. Ital. 26, 221–231. Floris, R., Melis, R.T., Mussi, M., Palombo, M.R., Iacumin, P., Usai, A., Mascia, F., 2009. McDonald, J.H., 2014. Handbook of Biological Statistics, 3rd ed. Sparkly House La presenza umana nella Sardegna centro occidentale durante l’Olocene antico: il Publishing, Baltimore Maryland, p. 299. sito di S’Omu e S’Orku (Arbus, VS). In: Atti della XLIV Riunione Scientifica McHenry, H.M., 1988. New estimates of body weight in early hominids and their dell’Istituto Italiano di Preistoria e Protostoria, 23–28 novembre 2009, vol. III. significance to encephalization and megadontia in “robust” australopithecines. In: Istituto Italiano di Preistoria e Protostoria, Cagliari, pp. 999–1004. Grine, F.E. (Ed.), Evolutionary History of the Robust Australopithecines. Routledge, Fortelius, M., 1985. Ungulate cheek teeth: development, functional and evolutionary New York, pp. 133–148. interrelationships. Acta Zool. Fenn. 180, 1–76. McMahon, T.A., 1973. Size and shape in biology. Science 179, 1201–1204. Fortelius, M., Eronen, J., Jernvall, K., Liu, L.P., Pushkina, D., Rinne, J., Tesakov, A., McMahon, T.A., 1975. Allometry and biomechanics: limb bones in adult ungulates. Am. Vislobokova, I., Zhang, Z., Zhou, L., 2002. Fossil mammals resolve regional patterns Nat. 109, 547–563. of Eurasian climate change over 20 million years. Ecol. Res. 4, 1005–1016. Melis, R.T., Ghaleb, B., Boldrini, R., Palombo, M.R., 2013. The Grotta dei Fiori (Sardinia, Freudenthal, M., Martín-Suárez, E., 2013. Estimating body mass of fossil rodents. Scr. Italy) stratigraphical successions: a key for inferring palaeoenvironment evolution Geol. 145, 1–130. and updating the biochronology of the Pleistocene mammalian fauna from Sardinia. Garcia, G.J.M., da Silva, J.K.L., 2004. On the scaling of mammalian long bones. J. Exp. Quat. Int. 288, 81–96. Biol. 207, 1577–1584. Millien, V., Bovy, H., 2010. When teeth and bones disagree: body mass estimation of a Garcia, G.J.M., da Silva, J.K.L., 2006. Interspecific allometry of bone dimensions: a giant extinct rodent. J. Mammal. 91, 11–18. review of the theoretical models. Phys. Life Rev. 3, 188–209. Moncunill-Solé, B., 2021. Eco-evolutionary adaptations of ochotonids (Lagomorpha: Geiger, M., Forasiepi, A.M., Koyabu, D., Sánchez-Villagra, M.R., 2013. Heterochrony and Mammalia) to islands: new insights into late Miocene pikas from the Gargano post-natal growth in mammals – an examination of growth plates in limbs. J. Evol. palaeo-archipelago. Biol. J. Linn. Soc. 132, 400–413. Biol. 27, 98–115. Moncunill-Solé, B., Jordana, X., Marín-Moratalla, N., Moyà-Solà, S., Köhler, M., 2014. Gingerich, P.D., Smith, B.H., Rosenberg, K., 1982. Allometric scaling in the dentition of How large are the extinct giant insular rodents? New body mass estimations from primates and prediction of body weight from tooth size in fossils. Am. J. Phys. teeth and bones. Integr. Zool. 9, 197–212. Anthropol. 58, 81–100. Moncunill-Solé, B., Quintana, J., Jordana, X., Engelbrektsson, P., Köhler, M., 2015. The Gliozzi, E., Malatesta, A., Palombo, M.R., 1984. Upper Pleistocene small mammal weight of fossil leporids and ochotonids: body mass estimation models for the order associations in the is Oreris area (Iglesiente, SW Sardinia). Geol. Romana 23, Lagomorpha. J. Zool. 295, 269–278. 121–129. Moncunill-Solé, B., Tuveri, C., Arca, M., Angelone, C., 2016a. Comparing the body mass Gould, S.J., 1975. On the scaling of tooth size in mammals. Am. Zool. 15, 351–362. variations in endemic insular species of the genus Prolagus (Ochotonidae, Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statistics software Lagomorpha) in the Pleistocene of Sardinia (Italy). Riv. Ital. Paleontol. Stratigr. 122, package for education and data analysis. Palaeontol. Electron. 4, XIX–XX. 23–36. Hautier, L., Bover, P., Alcover, J.A., Michaux, J., 2009. Mandible morphometrics, dental Moncunill-Solé, B., Orlandi-Oliveras, G., Jordana, X., Rook, L., Köhler, M., 2016b. First microwear pattern, and paleobiology of the extinct Balearic dormouse Hypnomys approach of the life history of Prolagus apricenicus (Ochotonidae, Lagomorpha) from morpheus. Acta Palaeontol. Pol. 54, 181–194. Terre Rosse sites (Gargano, Italy) using body mass estimation and paleohistological Hillson, S., 2005. Teeth. Cambridge Manuals in Archaeology, 2nd ed. Cambridge analysis. C. R. Palevol. 15, 235–245. University Press, Cambridge, p. 373. Moncunill-Solé, B., Jordana, X., Köhler, M., 2016c. How common is gigantism in insular Hordijk, K., 2010. In: Perseverance of pikas in the Miocene: Interplay of Climate and fossil shrews? Examining the “Island Rule” in soricids (Mammalia: Soricomorpha) Competition in the Evolution of Spanish Ochotonidae (Lagomorpha, Mammalia). from Mediterranean Islands using new body mass estimation models. Zool. J. Utrecht University, Utrecht, p. 232. PhD Dissertation. Linnean Soc. 178, 163–182. Höskuldsson, A., 1988. PLS regression methods. J. Chemom. 2, 211–228. Moyà-Solà, S., Köhler, M., 1997. The phylogenetic relationships of Oreopithecus bambolii Huxley, J.S., Teissier, G., 1936. Terminology of Relative Growth. Nature 137, 780–781. Gervais, 1872. C. R. Acad. Sci. Paris 324, 141–148. Jernvall, J., Fortelius, M., 2002. Common mammals drive the evolutionary increase of Palombo, M.R., 2009. Body size structure of Pleistocene mammalian communities: what hypsodonty in the Neogene. Nature 417, 538–540. support is there for the “island rule”? Integr. Zool. 4, 341–356. Jordana, X., Marín-Moratalla, N., DeMiguel, D., Kaiser, T.M., Köhler, M., 2012. Evidence Palombo, M.R., Rozzi, R., 2014. How correct is any chronological ordering of the of correlated evolution of hypsodonty and exceptional longevity in endemic insular Quaternary Sardinian mammalian assemblages? Quat. Int. 328-329, 136–155. mammals. Proc. R. Soc. B 279, 3339–3346. Palombo, M.R., Antonioli, F., Lo Presti, V., Mannino, M.A., Melis, R.T., Orru, P., Jungers, W.L., 1990. Problems and methods in reconstructing body size in fossil Stocchi, P., Talamo, S., Quarta, G., Calcagnile, L., Deiana, G., Altamura, S., 2017. The primates. In: Damuth, J., MacFadden, B.J. (Eds.), Body Size in Mammalian late Pleistocene to Holocene palaeogeographic evolution of the Porto Conte cave: Paleobiology: Estimation and Biological Implications. Cambridge University Press, clues for a better understanding of human colonization of Sardinia and faunal Cambridge, pp. 103–118. dynamics during the last 30 ka. Quat. Int. 439 Part A 117–140. Kaiser, T.M., Fickel, J., Streich, W.J., Hummel, J., Clauss, M., 2010. Enamel ridge Pelletier, M., 2019. Morphological diversity of wild rabbit populations: implications for alignment in upper molars of ruminants in relation to their natural diet. J. Zool. 281, archaeology and palaeontology. Biol. J. Linn. Soc. 128, 211–224. 12–25. Pérez-Barberia, F.J., Gordon, I.J., 1998. Factors affecting food comminution during Legendre, S., 1989. Les communautés de mammifères du Paléogène (Eocène supérieur et chewing in ruminants: a review. Biol. J. Linn. Soc. 63, 233–256. Oligocène) d’Europe occidentale: structures, milieux et évolution. Münch. Geowiss. Peters, R.H., 1983. The Ecological Implications of Body Size. Cambridge University Press, Abh. (A) 16, 1–110. Cambridge, p. 344. Legendre, S., Roth, C., 1988. Correlation of carnassial tooth size and body weight in Purvis, A., Orme, C.D.L., 2005. Evolutionary trends in body size. In: Carel, J.C., Kelly, P. recent carnivores (Mammalia). Hist. Biol. 1, 85–98. A., Christen, Y. (Eds.), Deciphering Growth. Springer-Verlag, Berlin, pp. 1–18. Lomolino, M.V., van der Geer, A.A., Lyras, G.A., Palombo, M.R., Sax, D.F., Rozzi, R., Quesenberry, K.E., Carpenter, J.W., 2011. Ferrets, Rabbits, and Rodents. Clinical 2013. Of mice and mammoths: generality and antiquity of the island rule. Medicine and Surgery, 3rd ed. Elsevier Inc, Missouri, p. 608. J. Biogeogr. 40, 1427–1439. Quinn, G.P., Keough, M.J., 2002. Experimental Design and Data Analysis for Biologists. Cambridge University Press, New York, p. 556. 11 B. Moncunill-Solé et al. Palaeogeography, Palaeoclimatology, Palaeoecology 582 (2021) 110645 Quintana Cardona, J., 2005. Estudio Morfológico y Funcional de Nuralagus rex The Jamovi Project, 2019. Jamovi (Version 0.9) [Computer Software]. https://www. (Mammalia, Lagomorpha, Leporidae). Universitat Autònoma de Barcelona, jamovi.org. Barcelona, PhD Dissertation, p. 515. Tsubamoto, T., 2014. Estimating body mass from the astragalus in mammals. Acta Quintana Cardona, J., Moncunill-Solé, B., 2014. Hypolagus balearicus Quintana, Bover, Paleontol. Pol. 59, 259–265. Alcover, Agustí & Bailon, 2010 (Mammalia: Leporidae): new data from the Neogene van der Geer, A.A., Lyras, G.A., Lomolino, M.V., Palombo, M.R., Sax, D.F., 2013. Body of Eivissa (Balearic Islands, western Mediterranean). Geodiversitas 36, 283–310. size evolution of palaeo-insular mammals: temporal variations and interspecific Quintana, J., Agustí, J., 2007. Los mamíferos insulares del Mioceno medio y superior de interactions. J. Biogeogr. 40, 1440–1450. Menorca (islas Baleares, Mediterráneo occidental). Geobios 40, 677–687. van der Geer, A.A.E., 2014. Parallel patterns and trends in functional structures in extinct Quintana, J., Köhler, M., Moyà-Solà, S., 2011. Nuralagus rex, gen. et sp. nov., and island mammals. Integr. Zool. 9, 167–182. endemic insular giant rabbit from the Neogene of Minorca (Balearic Islands, Spain). Varga, M., 2014. Textbook of Rabbit Medicine, 2nd ed. Butterworth-Heisenmann J. Vertebr. Paleontol. 31, 231–240. Elsevier Ltd, Edinburgh, p. 512. R Core Team, 2019. R: A Language and Environment for Statistical Computing. R Wagner, J.A., 1829. Über die Knochenbrekzie in Sardinien und die darin gefundenen Foundation for Statistical Computing, Vienna. https://www.R-project.org/. Thiere, so wie über einige andere hieher gehörige Erscheinungen. Archiv für die Reese, A.T., Lanier, H.C., Sargis, E.J., 2013. Skeletal indicators of ecological Gesammte Naturlehre 15, 10–31. specialization in pika (Mammalia, Ochotonidae). J. Morphol. 274, 585–602. Winkler, A.J., Winkler, D.A., Harrison, T., 2016. Forelimb anatomy of Serengetilagus Ringeade, M., 1979. Découverte de nouveaux Lagomorphes dans le Miocène inférieur praecapensis (Mammalia: Lagomorpha): a Pliocene leporid from Laetoli, Tanzania. d’Aquitaine et implications biostratigraphiques. Bull. Inst. Géol. Bassin d’Aquitaine Hist. Biol. 28, 252–263. 26, 111–157. Winkler, D.E., van den Hoek Ostende, L.W., Schulz, E., Calandra, I., Gailer, J.P., Sánchez, G., 2016. Package ‘plsdepot’: Partial Least Squares (PLS) Data Analysis Landwehr, C., Kaiser, T.M., 2013. Dietary divergence in space and time – Lessons Methods. Version 0.1.17. https://cran.r-project.org/web/packages/plsdepot/plsdep from the dwarf-goat Myotragus balearicus (Pleisto-Holocene, Mallorca, Spain). ot.pdf. Mamm. Biol. 78, 430–437. Schmidt-Nielsen, K., 1984. Scaling: Why Is Animal Size So Important? Cambridge Young, J.W., Danczak, R., Russo, G.A., Fellmann, C.D., 2014. Limb bone morphology, University Press, Cambridge, p. 256. bone strength, and cursoriality in lagomorphs. J. Anat. 225, 403–418. Šerbetar, I., 2012. Partial least squares regression analysis: example of motor fitness data. Zoboli, D., Zedda, M., Pillola, G.L., Palombo, M.R., 2018. Does a relationship exist Croat. J. Educ. 14, 917–932. between palaeopathologies and insularity? A case study of some bones of Prolagus sardus (Wagner, 1829) from Sardinia (Italy). Alp. Mediterr. Quat. 31, 75–86. 12