Environmental control on the biogeographical distribution of Desmanella (Soricomorpha, Mammalia) in the Miocene of the Iberian Peninsula

Authors


Abstract

Abstract:  This paper reports the first record of Desmanella (Mammalia, Soricomorpha) from the Granada Basin in southern Iberian Peninsula, which represents its south-westernmost occurrence in all Eurasia. It is a controversial taxon whose systematic assignment has been discussed for a long time. This genus belongs to the family Talpidae, a group of insectivores that include extant moles, shrew moles and desmans. Desmanella was very abundant in the late Miocene and early Pliocene of Eurasia, including several basins of northern Iberian Peninsula, but it has not been found until now in southern Iberia. Previous studies have shown that the faunal record and the climatic conditions during the Neogene in the Granada Basin were different from other areas of the Iberian Peninsula. The particular distribution of Desmanella supports the hypothesis that specific climatic features (wetter than neighbouring areas) prevailed in the Granada Basin during the late Turolian (late Miocene).

T he Mio–Pliocene record of insectivores from the Iberian Peninsula was poorly known until the end of the 1970s, when several authors started to study this group of mammals (Van den Hoek Ostende and Furió 2005). The number of works dealing with insectivores increased considerably from 2000 onwards, mainly because of its palaeoecological applications. However, studies on fossil Talpidae from Iberia are certainly scarce (Martín Suárez et al. 2001; Van den Hoek Ostende 2003; Agustíet al. 2005; Casanovas-Vilar et al. 2006; Furió 2007; Minwer-Barakat et al. 2008; Furióet al. 2011b).

The Granada Basin is located in the central sector of the Betic Mountain Range (southern Iberia) (Text-fig. 1); it is composed of marine and continental Neogene sediments, which began to accumulate during the Tortonian. The continental sediments of the basin have yielded a very complete Miocene and Pliocene record of fossil small mammals. Rodents have been studied in detail, and a precise biozonation is established (García-Alix et al. 2007ac, 2008ad). In contrast, the insectivores from the basin, because of their scarce biostratigraphic interest, are poorly known: only Parasorex (Erinaceidae; Mein and Martín Suárez 1993) and Archaeodesmana (Desmaninae; Martín Suárez et al. 2001) have been studied in detail.

Figure TEXT‐FIG. 1..

 Geographical and geological context of the Granada basin and situation of the studied sectors (modified after Braga et al. 1990); 1: Cantera de Pulianas sector; 2: La Dehesa sector.

The distribution of the insectivore faunas in Europe is controlled by environmental conditions (humidity and/or temperature) and linked to latitudinal gradients (Van Dam 2004; Furióet al. 2011a). In this paper, the distribution of Desmanella on the Iberian Peninsula is discussed, based on the updated record from several levels in the Granada Basin. The record of Desmanella in the Iberian Peninsula is mainly from northern basins of Iberia (especially Teruel and Catalonia), but it is also present in the Júcar Basin (Albacete; Mein et al. 1978) and Crevillente Basin (Alicante; Martín Suárez and Freudenthal 1998). The studied material represents the south-westernmost record of all Eurasia. There is no reference until now of its presence in the neighbouring Guadix Basin, despite its completeness record of Mio–Pliocene small mammals (Minwer-Barakat et al. 2008, 2009a, b, 2010). Moreover, there is no record of Desmanella in the Granada Basin, except for the late Turolian. The fossil record of the genus indicates that it was adapted to the wetter conditions of Central Europe rather than to those drier that prevailed in Iberia during most of the Miocene (Furióet al. 2011a). Thus, it is likely that this taxon had very specific ecological requirements, and its presence reflects that humid conditions prevailed at that time.

Geological Setting

The Granada Basin is located in the central sector of the Betic Mountain Range, overlying the contact between the Internal and External Zones of the range, mainly constituted by metamorphic and sedimentary rocks, respectively (Text-fig. 1). Its infilling consists of marine and continental sediments: after the deposit of marine Tortonian sediments, continental sedimentation began in the latest Tortonian (middle Turolian), associated with fluvio-lacustrine systems (García-Alix et al. 2008a).

Remains of Desmanella have been found at eight fossil levels in the Granada Basin. They are situated in two different sectors: the Cantera de Pulianas sector (levels: PUR-23, PUR-24A, PUR-25, and PUR-25A), located in the eastern area of the basin, and the Dehesa sector (levels: DHS-1, DHS-4A, DHS-4B and DHS-15B), in the south-western area of the basin. The sediments of these sectors are related with deltaic complexes (García-Alix et al. 2008a).

The studied sediments were deposited in the late Turolian, being the fossil levels of the Dehesa sector (Paraethomys meini Zone) slightly younger than those of the Cantera de Pulianas sector (Apocricetus alberti Zone; García-Alix et al. 2008a).

Material and Methods

In this paper, we will describe in detail the richest sample of Desmanella from the Granada Basin: that from the level PUR-24A; however, because of the absence of p4 in this level, we have added the description of the p4 (three specimens) from DHS-15B. Finally, we will describe some different features of specimens from other levels. Material and measurements have been compiled in Table 1.

Table 1.   Measurements (in mm) of the teeth of Desmanella aff. dubia from the studied levels.
LevelsLengthWidth
NtNMin.MeanMax.NtNMin.MeanMax.
  1. Nt is the total number of specimens (measurable and not measurable).

p4
 DHS-15B331.081.121.16330.700.730.76
m1
 DHS-15B11 1.65 11 1.22 
 DHS-4B11 1.66 11 1.22 
 DHS-4A11 1.78 11 1.15 
 PUR-25A1    11 1.22 
 PUR-25221.531.551.57221.251.261.27
 PUR-24A221.481.521.55221.221.241.26
 PUR-23651.541.571.63661.111.201.32
m2
 DHS-15B11 1.80 11 1.17 
 PUR-2511 1.74 11 1.18 
 PUR-24A331.601.721.82331.141.251.35
 PUR-23221.711.741.77221.271.371.47
m3
 DHS-15B11 1.36 11 0.85 
 PUR-25A221.351.371.39220.890.981.06
 PUR-2511 1.43 11 0.98 
 PUR-24A221.401.421.44220.970.991.01
 PUR-23431.271.301.31440.800.880.92
P4
 DHS-15B11 1.54 11 1.24 
 PUR-25321.311.381.4431 1.19 
 PUR-24A541.211.301.39551.051.101.16
M1
 DHS-4A2    21 1.74 
 PUR-25A1    1    
 PUR-25422.052.142.22421.931.961.98
 PUR-24A322.002.082.16321.661.681.70
 PUR-23221.922.062.20221.831.831.83
M2
 DHS-11    11 1.99 
 DHS-15B21 1.66 221.881.911.93
 PUR-25A321.621.671.72321.911.951.99
 PUR-25221.661.681.70221.941.971.99
 PUR-24A331.671.681.68331.881.962.01
 PUR-232    2    
M3
 DHS-15B11 0.84 11 1.33 
 PUR-25A11 0.97 11 1.53 
 PUR-2511 0.91 11 1.36 
 PUR-24A220.940.960.99221.441.501.55
 PUR-23220.960.970.98221.461.491.52

The nomenclature used in the descriptions of the teeth of Desmanella is that of Rümke (1974), except for the posterior lingual cusp of the upper molars. There are different opinions about the correct name for this cusp in the Talpidae. Some authors (Rümke 1974; Van den Hoek Ostende 1989; among others) use the term hypocone, as in other insectivore families. However, we follow the opinion of other specialists (Hutchison 1974; Engesser 1980; Thenius 1989; among others), who noted that the correct term is not hypocone but metaconule, because the hypocone originates from the posterior cingulum, whereas the posterior lingual cusp in the M1 and M2 of Desmanella is connected to the protocone. To employ a single terminology, and according with the latter authors, we use the term metaconule also for the M3, although in this dental element, the referred cusp is indeed connected to the posterior cingulum. The cusp situated between the protocone and the metaconule is called accessory cusp.

Lengths and widths have been measured as defined by Furió (2007), only considering total lengths and widths of the teeth (Text-fig. 2). The method proposed by Rümke (1974) considers too many variables of doubtful utility (L2 and L3 in the upper molars, and L1, L2, W1 and, W2 in the lower ones) and does not define the exact orientation of the teeth. The absence of an unequivocal orientation of the teeth can lead to a significant inaccuracy of the measurements. The method defined by Furió (2007) is based in that of Rümke (1974), but it provides one reference line to orientate the teeth and measure length and width perpendicularly. It must be mentioned that the reference line in M1 is drawn connecting the highest points of the metacone and the paracone in occlusal view. This could result in shorter lengths of the first upper molars than previously measured in other studies. Other than that, the measurements obtained using this method and that of Rümke (1974) are not expected to differ substantially in absolute values but rather in variability, and we can compare quite confidently our data with those from previous papers. To avoid problems, we encourage other authors to use this method from now in future works dealing with teeth of Desmanella.

Figure TEXT‐FIG. 2..

 Measurement method from Furió (2007). The reference lines are marked by continuous lines. Length (L) and width (W) are marked by dotted lines.

Measurements were taken with a Wild M7S binocular microscope, equipped with a Sony Magnescale LM12 digital measuring device. Measurements are given in units of 1 mm. Photographs were made with the FEI ESEM QUANTA 400 of the ‘Centro Andaluz de Medio Ambiente’ (Granada) in low vacuum mode. The specimens are kept in the ‘Departamento de Estratigrafía y Paleontología’ of the University of Granada, Spain.

Institutional abbreviations.  DEPUGR, ‘Departamento de Estratigrafía y Paleontología’ of the University of Granada, Spain.

Systematic Palaeontology

Family TALPIDE Fischer von Waldheim, 1814
Subfamily UROPSILINAE Dobson, 1883

Genus DESMANELLA Engesser, 1972

Type species. Desmanella stehliniEngesser, 1972.

Desmanella aff. dubiaRümke, 1976
Text-figure 3

Figure TEXT‐FIG. 3..

Desmanella aff. dubia: A, left p4 DEPUGR DHS-15B 48. B, right p4 DEPUGR DHS-15B 50. C, left m1-m2 DEPUGR PUR-23 188. D, right m1 DEPUGR PUR-24A 119. E, right m2 DEPUGR PUR-24A 122. F, right m3 DEPUGR PUR-24A 123. G, left m3 DEPUGR PUR-25A 106. H, right P4 DEPUGR PUR-24A126. I, left M1 DEPUGR PUR-24A 131. J, left M1 DEPUGR PUR-25 82. K, right M2 DEPUGR PUR-24A 134. L, left M2 DEPUGR PUR-25 86. M, left M3 DEPUGR PUR-23 197. N, right M3 DEPUGR PUR-24 137. Scale bar represents 1 mm.

Description

The p4 has a quasi-oval outline. The protocone is high and sharp; its posterior flank is concave. There is a short cristid ending in an incipient hypoconid. The anterior cingulid is weak, and the posterior one is well developed. Anterior and posterior cingulids are connected by a very thin labial cingulid. There are two roots (one anterior and one posterior).

The m1 has the trigonid narrower than the talonid. The protoconid is higher, more inclined and more lingually placed than the hypoconid. The paraconid is very reduced. The oblique cristid ends near the metaconid. The anterior cingulid is well developed, but the posterior one is weak; both are connected by a thin labial cingulid. There is a weak lingual cingulid under the trigonid basin. The parastylid is very much reduced or absent. There is a well-developed, hook-shaped entostylid. There are two roots (one anterior and one posterior).

Trigonid and talonid are similar in width in the m2. The oblique cristid does not reach the top of the metaconid. The protoconid is the highest cusp. The paraconid is very much reduced (crest-shaped). There is a continuous cingulid, well developed on the anterior border and narrower on the labial and posterior parts. There is a weak lingual cingulid under the trigonid basin. The parastylid is reduced to a weak thickening of the anterior cingulid. There is a well-developed, hook-shaped entostylid. There are two roots (one anterior and other posterior).

In the m3, the talonid is notably narrower and lower than the trigonid. There is a continuous cingulid from the anterior to the posterior border along the labial side; its anterior part is the best developed. The oblique cristid reaches the base of the metaconid. Some specimens have a narrow lingual cingulid under the trigonid basin, which weakly extends posteriorly. The parastylid is reduced to a weak expansion of the anterior cingulum, and there is no entostylid. There are two roots.

The P4 has a subtriangular outline, with two concavities in the anterior and posterior parts. The paracone is high; the protocone is rounded and well individualized. There is no anterocrista. The posterocrista extends along the posterior side of the paracone and ends in a very small metacone, connected to the posterior cingulum in some specimens. The cingulum borders the entire base of the tooth, except for the labial side, where it is much reduced or even interrupted. This cingulum is well developed in the anterior part of the tooth, forming a parastyle. There are three roots (one anterior, one posterior and one labial).

The M1 is trapezoidal, with a concave posterior outline. The metacone is the highest cusp; its posterior arm extends to the postero-labial corner of the teeth, ending in the metastyle. The metacone is displaced posteriorly with respect to the metaconule. The protocone is separated from the protoconule and metaconule; the accessory cusp between protocone and metaconule is undistinguishable because of the bad preservation of the labial zone of the teeth. The protocone is situated in the same axis than the paracone. The mesostyle is notched (almost split). The parastyle is well developed and continues labially in a cingulum that borders the paracone, and lingually in a cingulum that reaches the base of the protoconule. Paracone, mesostyle and metacone delimit two labial depressions, which are partially closed by a weak labial cingulum. Roots are not preserved.

The M2 has a subrectangular outline with a weak concavity on its posterior border. The highest cusps are the labial ones; the metacone is slightly higher than the paracone. The lingual cusps are well individualized, but less than in M1. The metaconule is approximately in the same axis as the metacone, and the protocone is posterior with respect to the paracone. The protoconule is rounded and attached to the protocone. The metaconule is connected to the protocone by a crest. The anterolabial and posterolabial corners of the tooth are broken. The mesostyle is undivided. There are two labial depressions, delimited by paracone, mesostyle and metacone, which have a weak cingulum in their labial border. The parastyle is hook-shaped and continues in an anterior cingulum, which ends at the base of the protoconule. Roots are not preserved.

The M3 has a subtriangular outline. The paracone is the highest cusp; its anterior arm extends labially, ending in a small hook-shaped parastyle, which continues in an anterior cingulum. The metacone has no posterior arm expansion. The protocone is large. The protoconule is reduced to a thickening of the anterior arm of the protocone. The metaconule is low and attached to the lingual base of the metacone; it is separated from the protocone by a deep valley. The mesostyle is undivided. Mesostyle and paracone delimit a labial depression. Roots are not preserved.

The metastyle in unworn M1 from the other levels, such as PUR-25 83, is sometimes hook-shaped. In some M1, the accessory cusp is reduced to a small thickening in the posterior arm of the protocone, and the mesostyle may be split, or notched. The M3 from other samples has three roots (anterior, posterior and lingual); the mesostyle is divided into the specimens PUR-23 197 and DHS-15B 57.

Comparisons

The studied specimens have some features that might resemble those of the genus AsthenoscapterHutchison, 1974, such as the typical concave posterior outline in M1 and M2. However, the strong lingual relief because of the inflation of the lingual cusps in M1 and M2, the location of the parastyle in M1 (aligned with the paracone) and the reduction of the talonid in lower molars (especially in m3) differ from Asthenoscapter but agree with genus Desmanella.

The oldest record of Desmanella (Desmanella sp.) is from the late Oligocene and found in southern Germany (Van den Hoek Ostende 1989). During the Miocene, the genus was widespread in Europe and Asia, and it became restricted to central and western Europe in the Pliocene (Rzebik-Kowalska 2002, 2009; Furió 2007). For comparisons, we focus on the Miocene and Pliocene European species of Desmanella.

The oldest Miocene species of de genus are D. gudrunaeVan den Hoek Ostende and Fejfar, 2006 (early Orleanian, MN 3; central Europe) and D. engesseriZiegler, 1985 (early–middle Orleanian, MN 3–4; central Europe), which differ from our material mainly in their smaller size. The outline of the M1 differs too: the posterior concavity is very much marked in D. gudrunae, whereas the posterior outline of D. engesseri is almost straight; the concavity of our specimens is intermediate. The mesostyle in the upper molars of D. gudrunae is undivided; on the contrary, in the M1 and M3 of D. aff. dubia, it may be undivided, notched or split. The posterior contour of the metaconule in M1 and M2 is much more rounded in D. gudrunae.

The teeth of D. stehliniEngesser, 1972 (middle-late Astaracian, MN 7–8; central and south-western Europe) are smaller than the studied specimens. As in D. engesseri, the M2 and, specially, the M1 have a less concave posterior border than those of D. aff. dubia. The accessory cusp is more developed in the M2 of D. stehlini.

The specimens of D. rietscheliStorch and Dahlmann, 2000 (early Turolian, MN 11; central Europe) are, in general, slightly smaller than those of the studied populations. The main difference is the better development of the accessory cusp in the upper molars of D. rietscheli. The cingulid of the lower molars of D. rietscheli is more developed than in D. aff. dubia from the Granada Basin, and the posterior cingulum in the M1 and M2 is somewhat larger.

The studied specimens are smaller than those of D. woelfersheimensisDahlmann, 2001 (late Ruscinian, MN 15; central Europe).

The main difference with D. gardiolensisCrochet, 1986 (early Villanyian, MN 16; western Europe) is the large size of its M1. The cingulum of the P4 differs too: it is very much developed in D. gardiolensis, so the posterior outline of the premolar is straight; this posterior outline is markedly concave in the P4 from the Granada Basin. The accessory cusp is more voluminous in the upper molars of D. gardiolensis than in those studied in this paper.

The closest similarity is found with the contemporaneous species D. crusafontiRümke, 1974 (early Vallesian–late Turolian, MN 9–MN 13; south-western Europe) and D. dubiaRümke, 1976 (early Ruscinian, MN 14; eastern and western Europe). The studied material is, in general, larger than D. dubia and smaller than D. crusafonti. The most important difference with D. dubia is the size; our specimens are larger than those of D. dubia from Pikermi (Rümke 1976), Maramena (Doukas et al. 1995), Can Vilella, and Romanyà d’Empordà (Furió 2007), among others. The size is within the range of D. crusafonti; the studied specimens are, in general, close to the lowest values of D. crusafonti, except for the P4, which is larger in D. aff. dubia from Granada. The size, outline and distribution of the cusps are similar in our P4 and those of D. dubia; in D. crusafonti, the anterior and posterior concavities of the outline are less marked. The development of the cingula (in upper and lower teeth) is similar to that of D. dubia; the cingula of the studied teeth are slightly less developed than those from the type locality (Pikermi), but similar to those from other localities, like Maramena (Doukas et al. 1995) or Can Vilella (Furió 2007). On the contrary, the cingula are poorly developed in D. crusafonti. The oblique cristid is lingually displaced (towards the metacone) in the studied specimens and in D. dubia. There is a lingual cingulid under the trigonid basin in the lower molars of D. dubia from Maramena (Doukas et al. 1995), and in those of D. aff. dubia from Granada; this cingulum is absent in D. crusafonti. Summarizing, the studied material resembles that of D. dubia from different European sites, but the existence of some differences does not allow a conclusive-specific assignment. Therefore, we determine this material as Desmanella aff. dubia.

Remarks on the Genus Desmanella

The systematic position of the genus Desmanella within the family Talpidae is under discussion (Engesser 1972; Hutchison 1974; Rümke 1974; Storch 1978; Engesser 1980; Ziegler 1985; Storch and Dahlmann 2000; Van den Hoek Ostende 2001a; Ziegler 2003; Van den Hoek Ostende and Fejfar 2006). Since its original description by Engesser (1972), the genus has been referred to Desmaninae (Engesser 1972; Hutchison 1974), Uropsilinae (Rümke 1974; Engesser 1980; Ziegler 1985; Van den Hoek Ostende 2001a; Van den Hoek Ostende and Fejfar 2006), Talpinae (Storch 1978), Urotrichini (Ziegler 2003), and it has been even regarded as Talpidae incertae sedis (Storch and Dahlmann 2000).

The controversy in the relationship of Desmanella within the family is probably due to an over-interpretation of some dental characters. However, there is solid evidence that the humerus of Desmanella is of uropsiline-type (Van den Hoek Ostende and Fejfar 2006). Moreover, molecular studies indicate that Eurasian ambulatory moles (i.e. subfamily Uropsilinae) represent a basal divergent evolutionary line from the rest of talpids (Shinohara et al. 2003). Because the talpid fossil record extends back to the Eocene–Oligocene, it is quite unlikely that the fossil record of the subfamily is not represented by anything else than Uropsilus. If the genera with a primitive type of humerus like Mystypterus, Theratiskos, Asthenoscapter, Mygatalpa and Desmanella are uropsilines as well (as suggested by Van den Hoek Ostende 2001a), most of this gap between Oligocene and Recent times is filled up. Added to the morphological similarities, this makes much more sense than an apparent blank of about 40 million years in the evolutionary history of the group.

Palaeoecological Discussion

The palaeoecological requirements of Desmanella are hitherto unknown. As a first approach, Van den Hoek Ostende (2001b) pointed out that the presence of Desmanella was not affected by the variation in humidity conditions, in view of its uninterrupted occurrence despite the climatic fluctuations inferred by Van Dam (1997) in the Teruel Basin. Nevertheless, some authors have pointed out that in the northern area of the Iberian Peninsula, the general climatic conditions were more humid than in the southern ones (Furióet al. 2011b). Thus, in the Teruel Basin, there were some permanent wet environments despite the fluctuations in humidity, enough to allow the persistence of Desmanella. On the contrary, southern Spanish basins had, in general, more arid conditions. Thus, changes in the humidity values had an important impact in the environments, limiting the presence of taxa particularly sensitive to some ecological conditions. Such was probably the case of Desmanella. In fact, the absence of Desmanella in the southern basins had already been highlighted by Agustíet al. (2006), with an apparent relationship with the impoverishment of the Spanish insectivore communities southwards. In view of the current data, it does not seem to be true anymore, because Desmanella was certainly present in the Granada Basin, one of the southernmost sedimentary basins in Iberia with Turolian faunas. There are, nonetheless, some remarks to be made on such an unexpected occurrence.

During the Miocene, the latitudinal distribution of Desmanella and some other talpids was certainly influenced by the mean annual rainfalls and the general moisture of the environments (Furióet al. 2011a). The co-occurrence of Talpa, Archaeodesmana and Desmanella during the Turolian (MN 11–MN 13) in the Teruel Basin (see Van Dam et al. 2001) was indeed preceded in the latest Aragonian (MN 7–8, late Astaracian) and the Vallesian by the common association of Talpa with Desmanella in the Catalan basins (NE Iberia). Not by chance, these Catalan basins were characterized by greater levels of humidity than the inner Iberian ones (see Agustí 1990; Casanovas-Vilar and Agustí 2007), where Desmanella was clearly missing during this time interval.

We find a similar case in the southern Spanish basins of Granada and Guadix during the Turolian; Desmanella was absent in the neighbouring Guadix Basin. García-Alix et al. (2008e) pointed out that the Guadix Basin has drier conditions than the Granada Basin, as nowadays, mainly due to the ‘rain-shadow’ effect of the Sierra Nevada and the Sierra de los Filabres mountains to the south of the Guadix Basin. In this way, the abundant precipitations in the Granada Basin must have favoured the development of lakes and associated densely vegetated environments. But these humid conditions were not present all along the Turolian.

In fact, the presence of Desmanella in the Granada Basin is restricted to a few fossil levels (see García-Alix et al. 2008a), thus covering just a short interval of the late Turolian (Messinian). In the rest of known sites, covering the middle Turolian (latest Tortonian), latest Turolian (latest Messinian) and Ruscinian (Pliocene), Desmanella is missing. There are several facts indicating that the levels where Desmanella comes from were deposited coinciding with an increase of the wet conditions in the basin.

Evidences of a higher level of humidity for this short time interval are found in the sedimentary record of the basin. The remains of Desmanella from Canteras de Pulianas sector (sites PUR-23, PUR-24A, PUR-25 and PUR-25A) are found in lignitic lutites deposited in the floodplain of the deltaic complex of the north-eastern margin of the basin. The material from the sector of La Dehesa (sites DHS-1, DHS-4, DHS-4B and DHS-15B) was extracted from fine-grained sediments of interdistributary bays and floodplain of the deltaic complex of the south-western margin of the basin (García-Alix et al. 2008a). Both sectors correspond to a moment in which a large lake occupied almost the whole basin (García-Alix et al. 2008a), in contrast to the preceding braided fluvial systems typical of drier environmental conditions. In few words, Desmanella arrived to the basin coinciding with the development of lacustrine systems (García-Alix et al. 2008a, e). The levels of the basin that have not yielded remains of Desmanella correspond to other sedimentary settings and rather reveal that more arid conditions were developed in the middle Turolian, latest Turolian and Ruscinian. The Granada Basin was then transformed into an unfavourable environment for Desmanella.

Extra evidences of a higher degree of moisture in the environment are provided by the associated fauna. The fossil remains of Desmanella co-occur with other taxa with clear preferences for wet environments. This is the case of Castoridae and Desmaninae (García-Alix et al. 2007b, 2008e), which have been identified in the Granada Basin. Beavers and aquatic moles, both strongly dependent on the existence of aquatic environments, were absent in the Guadix Basin during the late Turolian. In contrast, taxa with arid preferences found in the Guadix Basin, such as gerbillids (Myocricetodon and Debruijnimys; Minwer-Barakat et al. 2009a), have not been recorded in the Granada Basin.

Attending to such co-occurrences, one would expect Desmanella to be also adapted to aquatic environments. However, as previously noted, we do know nowadays that the humerus of this talpid does not show any specialization to dive or swim as the ones of Desmaninae or Castoridae. Desmanella, as the extinct uropsilines Asthenoscapter and Mygatalpa and the extant genus Uropsilus, shows a slender, but not flattened humerus more typical of a terrestrial quadruped than of a specialized burrower or diver (Furióet al. 2011a). Demanella was probably a litter burrower, strongly limited in distribution by the existence of soft soils with abundant organic matter in which it could scratch searching for small invertebrates.

In a similar way, the distribution of Desmanella could also be conditioned by the small size of its species. Gureev (1979) noticed that the extant genus Uropsilus is too small to retain enough fat to survive a long winter, so its species are probably active during all the year. Desmanella could have been similarly restricted to mild climates like most shrews.

Thus, the restricted distribution of Desmanella all along the Miocene can be confidently explained considering that its species avoided extreme climates, with clear preferences for mild temperatures and rather high levels of humidity. The presence of Desmanella in the Granada Basin during the late Turolian agrees with this view. The arrival of the genus to the south-westernmost zone of Eurasia was possible because of the existence of particular climatic conditions, and other factors controlled by these, like vegetation, which contributed to develop favourable environments for Desmanella.

Conclusions

Desmanella is an enigmatic talpid, whose affiliation to the subfamilies Desmaninae, Uropsilinae or Talpinae has been considered doubtful by some authors during decades; however, nowadays, molecular studies and skeletal evidences suggest its allocation in the subfamily Uropsilinae.

We have followed a new measurement method for the teeth of Desmanella (that of Furió 2007), instead of that of Rümke (1974). Although this method is based on that of Rümke, it considers fewer variables (only length and width), and it requires the exact orientation of the teeth. These measurements and those of Rümke (1974) do not differ substantially. Owing to its simplicity and accuracy, we recommend this method for future works dealing with Desmanella.

The new finding of Desmanella from the Granada Basin extends its geographic range to the south of the Iberian Peninsula during the late Miocene, and it represents the south-westernmost record of all Eurasia. The climatic, sedimentary and palaeogeographic changes in the Granada Basin (García-Alix et al. 2008a, e) had an effect on biotopes and habitats, and in view of the data from the Granada Basin, we propose a positive relationship between wet conditions and the presence of Desmanella, which may be strongly enhanced in areas where the climatic changes are more extreme, such as southern Iberian Peninsula. That ‘ecological control’ could explain its absence in other neighbouring southern basins, such as the Guadix Basin.

Acknowledgements.  This study was supported by the program ‘Consolider Ingenio 2010’ (CSD 2006-00041), the research groups RNM0190 and RMN309 of the ‘Junta de Andalucía’, and the Project ‘Grandes simios fósiles (Hominoidea) del Mioceno del área mediterránea: origen, paleobiología y evolución’, HOPE, Ministerio de Ciencia e Innovación, CGL2008-00325/BTE. A. G. A. was also supported by a Juan de la Cierva contract from the Spanish ‘Ministerio de Ciencia e Innovación’. M. F. received support from the SYNTHESYS Project http://www.synthesys.info/, which is financed by European Community Research Infrastructure Action under the FP6 ‘Structuring the European Research Area Programme’, by means of the SYNTHESYS Application NL-TAF-2790 ‘Diversity and distribution of Insectivora (Mammalia) in the Iberian Peninsula from the Middle to the Upper Miocene’. We thank I.M. Sánchez-Almazo for taking the photographs (CEAMA, Granada, Spain). Comments and suggestions by the reviewers P. Mein, B. Rzebik-Kowalska, and B. Engesser and by the Editor are kindly acknowledged.

Editor. Marcello Ruta

Ancillary