Historical background The fossil site of Can Llobateres (Late Miocene), in the municipality of Sabadell (Catalonia, NE Spain), has figured prominently in the study of Miocene mammals from Europe—being considered the reference locality of MN9 (Mein, 1990)—with emphasis on fossil hominoids. The site was discovered in 1926 (Crusafont Pairó, 1969; Alba et al., 2011a, 2011b), and the first accounts of its fauna were published in the 1940s (Villalta Comella and Crusafont Pairó, 1943; Crusafont Pairó and Villalta Comella, 1948). Hominoid dental remains, assigned to the dryopithecine1Hispanopithecus laietanus, were first discovered there in 1958 (Crusafont Pairó, 1958; Crusafont Pairó and Hürzeler, 1969), and additional hominoid remains were subsequently recovered during the 1960s (Crusafont Pairó and Hürzeler, 1961, 1969; Crusafont Pairó, 1965; Golpe Posse, 1982, 1993; Moyà-Solà et al., 1990). Crusafont Pairó and Hürzeler, 1961, Crusafont Pairó and Hürzeler, 1969 reported two additional species from the site—never figured or described, and hence considered nomina nuda (Szalay and Delson, 1979; Alba, 2012; Alba et al., 2012a)—while a fourth hominoid was subsequently reported based on an isolated male upper canine (Crusafont Pairó and Golpe Posse, 1973), being attributed to Sivapithecus indicus.

In 1981, additional dental hominoid material was recovered from the lower levels of the site (Begun et al., 1990; Golpe Posse, 1993), known as Can Llobateres 1 (CLL1). From 1992 onward, a partial cranium (Moyà-Solà and Köhler, 1993, 1995; Begun, 1994) and skeleton (Moyà-Solà and Köhler, 1996; Almécija et al., 2007; Pina et al., 2012; Tallman et al., 2013; Susanna et al., 2014) were discovered on the upper levels of the site, termed Can Llobateres 2 (CLL2), which were last excavated in 1997. In 2010, excavations were resumed again at CLL1 by a team led from the Institut Català de Paleontologia Miquel Crusafont (ICP), and additional hominoid teeth were found (Alba et al., 2012a), along with plant remains that enabled a better characterization of the paleoenvironment (Marmi et al., 2012). The locality of CLL1 was last excavated in 2015, owing to the paucity of the fossiliferous levels that delivered most of the hominoid dental remains in previous years. Begun et al. (1990) and most subsequent authors attributed the whole hominoid sample from Can Llobateres to Dryopithecus laietanus (Harrison, 1991; Moyà-Solà and Köhler, 1993, 1995, 1996; Begun, 1994; Ribot et al., 1996) or, more recently, H. laietanus (Cameron, 1997, 1998, 1999, 2004; Almécija et al., 2007; Moyà-Solà et al., 2009; Begun, 2009; Alba, 2012; Alba et al., 2012a).

Geological background and taphonomical remarks Can Llobateres is located in the Vallès sector of the Vallès-Penedès Basin (Catalonia, Spain; Fig. 1), an elongated half-graben about 100 km in length and 12–14 km in width, parallel to the Catalan coastline near Barcelona, and bounded by the Catalan Coastal Ranges (for an updated review, see Casanovas-Vilar et al., 2016a). The Miocene sedimentary infill of the Vallès-Penedès Basin has been divided into four main litostratigraphic units, Can Llobateres belonging to the Upper Continental Units, which range from the Middle to the Late Miocene (Casanovas-Vilar et al., 2016a). In particular, Can Llobateres is located within distal facies of the alluvial fan system of Castellar del Vallès (Agustí et al., 1996, 1997), consisting of a short (∼20 m-thick) sequence, mostly defined by mudstones (claystones and siltstones), as well as polymictic breccias and conglomerates (Marmi et al., 2012; Alba et al., 2012a). The locality of CLL1 is located in the lower section of the sequence, characterized by organic matter and abundant mudstones, indicating a poorly drained alluvial plain that would have favored the development of small shallow lakes and ponds (Agustí et al., 1996; Alba et al., 2011a, 2011b, 2012a).

No taphonomic study has been performed at CLL1, but Begun et al. (1990) made some sedimentological and taphonomic remarks based on the 1981 campaign, while Marmi et al. (2012) and Alba et al. (2012a) provided far more detailed sedimentological descriptions and additional taphonomic details based on the 2010–2011 campaigns. The former authors distinguished two ‘sedimentary units’ at CLL1, which they interpreted as corresponding to two different ‘fluvial cycles’ (Begun et al., 1990): the lower one, characterized by finer sediments indicative of a low-energy depositional environment, would have yielded taxa associated with humid and forested environments (such as primates and suids); the upper one, in turn, would be characterized by coarser sediments (including channel deposits) indicative of a higher-energy depositional environment and, according to these authors, would have yielded taxa indicative of more open conditions (such as hipparionin equids) that could have been transported from greater distances. Subsequent fieldwork at the site (Alba et al., 2012a) confirmed that primates are apparently restricted to the lower ‘unit’ recognized by Begun et al. (1990) but evinced a greater stratigraphic and taphonomic complexity (Marmi et al., 2012; Alba et al., 2012a), in which four stratigraphic bodies may be discerned.

The bottom of the CLL1 sequence (not available to Begun et al., 1990) is composed by layers of reddish to brown silts, sands, and conglomerates, indicative of less humid conditions than the ‘classical’ layers of CLL1, probably representing the end of a preceding depositional cycle (Alba et al., 2012a). The lower ‘unit’ of Begun et al. (1990), in turn, comprises variously colored mudstones (mostly fine-grained clays) with abundant micromammal and some macromammal remains (including the hominoid teeth recovered in 1981 and 2011), plus mollusk shells, silicified figs, and other fragmentary and poorly preserved plant remains. The lower layers (1a and 1b) are blackish and become lighter in color (light gray to greenish) toward the top (layer 2), although at different locations, layers 1b and 2—the ones that yielded primate remains—are completely light brown (Marmi et al., 2012; Alba et al., 2012a). The top layers (3a and 3b) consist of greenish to yellowish clays with some coarser sediments, in which gastropod shells are poorly preserved and vertebrate remains sparser than in the underlying layers. The upper ‘unit’ of Begun et al. (1990) begins with a paleochannel (layer 4a) of varying thickness composed of much coarser sediments (with decreasing granulometry from bottom to top), which at some point erodes some or all of the aforementioned layers and which locally yielded more abundant large mammal remains. Overlying these channel deposits, there is a thick (>2.5 m) sedimentary package (layer 4b/4p) of alternating green clays, greenish to orange silts, and conglomerates, which locally preserve abundant plant macroremains (Marmi et al., 2012). A fourth sedimentary body can be distinguished at the top of the CLL1 sequence, including a paleochannel (layer 5a) that erodes layer 4b and an overlying 6.5 m-thick package (layer 5b) of multiple episodes of ocher to red clays, silts, and conglomerates, indicative of paleosol formation and well-aerated conditions (Marmi et al., 2012).

Pending more detailed analyses, and contrary to Begun et al.'s (1990) previous assessment, the 2010–2015 fieldwork campaigns (D.M.A., pers. obs.) failed to confirm differences in faunal composition between the primate-bearing layers and the paleochannel deposits (other than the apparent lack of primates in the latter). For example, suids are not restricted to the primate-bearing layers (contra Begun et al., 1990) but are also present in the paleochannel deposits (layer 4a) and in the layer 4b that overlies them. Similarly, Hippotherium remains are more abundant in layer 4a but are also present in most of the remaining layers, both above (4b) and below (3b, 2, and 1b), thus including the primate-bearing ones. Finally, the fact that forest-adapted taxa are not restricted to the latter is best exemplified by the recovery of a Tapirus tooth from the bottom of layer 4a (D.M.A., pers. obs.). Coupled with the fact that detailed stratigraphic provenance is not recorded for most of the remains recovered before 2010, this makes impossible to conclusively ascertain whether, in fact, CLL1 as a whole mixes faunal elements from different environments due to differential transport among various fossiliferous layers.

On the other hand, different degrees of transport depending on the layer are confirmed by available data. In particular, the presence of isolated teeth from two single individuals (hominoid and cervid) scattered over a few square meters in one of the clay layers that yielded primate remains suggests minimal transport, probably under water (Alba et al., 2012a). By contrast, the paleochannel combines fragmentary and rounded remains of multiple large mammals with better-preserved and larger macromammal remains (including disarticulated dentognathic and postcranial remains of a single Hippotherium individual; D.M.A., pers. obs.)—overall indicative of transport in higher-energy conditions from longer distances than in the primate-bearing layers as previously concluded by Begun et al. (1990). Finally, the plant remains from layer 4p (Marmi et al., 2012) are interpreted as a parautochtonous assemblage accumulated by wind and short water transport in a shallow-water epositional environment. All this preliminary taphonomic evidence combined suggests that the fossil assemblages from CLL1 are representative of the fauna and flora present at and near the depositional environment, although it cannot be discounted that some taxa preferentially inhabited areas farther away from it, as some fossiliferous layers evince greater degrees of transport than others.

Chronological and paleoenvironmental background On biostratigraphic grounds, CLL1 is correlated to the Cricetulodon hartenbergeri–Progonomys hispanicus interval local subzone, whereas CLL2 is correlated to the Cricetulodon sabadellensis + Progonomys hispanicus concurrent range local subzone (Casanovas-Vilar et al., 2016b). In addition, the Can Llobateres sequence records three magnetozones, with reverse polarity in the lower and upper parts and normal polarity in the middle (Agustí et al., 1996, 1997). In particular, CLL1 is correlated to C4Ar.3r, whereas CLL2 is correlated to C4Ar.2r (Agustí et al., 1996, 1997), with interpolated ages of 9.76 and 9.62 Ma, respectively (Casanovas-Vilar et al., 2016b). Based on the absence of Progonomys from CLL1, it has traditionally been considered that the Can Llobateres sequence records the early to the late Vallesian transition, with CLL1 and CLL2 being correlated to MN9 and MN10, respectively (Agustí et al., 1996, 1997; Alba et al., 2012a); indeed, CLL1 is considered the reference locality for MN9 (Mein, 1990; de Bruijn et al., 1992; Casanovas-Vilar et al., 2016b). However, Progonomys remains have been found in older sites from other Iberian basins, placing the lower boundary of MN10 (as defined by the first common occurrence of this genus) at 9.98 Ma (Hilgen et al., 2012; Van Dam et al., 2014). Accordingly, under a strictly biostratigraphic approach to MN units, CLL1 must be correlated to MN10 instead of MN9 (Casanovas-Vilar et al., 2016b; Alba et al., 2018). It is unlikely that the paleobiodiversity of CLL1 is inflated by time averaging as most of the remains come from a short stratigraphic interval of ∼2 m (Alba et al., 2012a). Based on the average sedimentation rate for the Vallesian of the Vallès-Penedès Basin (20 cm/kyr; Garcés et al., 1996), this would merely represent a time interval of ∼10 kyr—in rough agreement with the difference between the interpolated ages of CLL1 and CLL2 (∼140 kyr; Casanovas-Vilar et al., 2016b), which are separated by less than 20 m of stratigraphic distance.

Paleoenvironmental inferences have previously been drawn for CLL1 based on both mammals and plants, using different methods—see summary in Table 1 and the Discussion for additional details. The plant remains (Álvarez Ramis, 1975; Sanz de Siria Catalán, 1993, 1994; Alba et al., 2011b; Marmi et al., 2012), which enable a reliable reconstruction of the paleoenvironment that H. laietanus inhabited, are generally in agreement with the conclusions drawn from the large and/or small mammals (Nagatoshi, 1987; Köhler, 1993; Andrews, 1996; Hernández Fernández et al., 2003; Costeur, 2005; Casanovas-Vilar and Agustí, 2007), which indicate the presence of a humid and closed forest paleoenvironment at the depositional area but further hint at the existence of relatively more open woodlands nearby (Marmi et al., 2012). Some paleoenvironmental conclusions have also been derived for CLL1 from the numerical analyses of its mammalian assemblage composition (Andrews, 1996; Hernández Fernández et al., 2003), and ecometrics in general have been applied to this site as part of studies that aimed to reconstruct continental patterns of Neogene paleoprecipitation (Fortelius et al., 2002; Eronen et al., 2009, 2010a; Kaya et al., 2018) or net primary production (Toivonen et al., 2022). However, no dedicated ecometric approaches to paleoenvironment reconstruction have been performed thus far. Given that H. laietanus is the latest hominoid recorded from mainland Western Europe, the reconstruction of its paleoenvironment is highly significant for understanding the local extinction of hominoids and other mammalian taxa in Europe during the Late Miocene (Casanovas-Vilar et al., 2011).

Ecometrics is a trait-based approach based on the study of functional morphological features of organisms that attempts to quantify links between the distribution of those traits across biotic communities and specific environmental factors as well as to analyze their dynamics through space and time in the fossil record (Fortelius et al., 2002; Eronen et al., 2010b; Vermillion et al., 2018). Ecometric traits are measurable macroscopic features that are known to represent their function in relation to local environmental conditions (Eronen et al., 2010b)—either due to adaptive reasons or to use/plasticity (ecophenotypic features). The most typical ecometric traits of vertebrates describe dental morphology, limb proportions, and body mass. Candidate traits for ecometric modeling must be at least preliminarily known to be associated through their functional relationship with the local environmental conditions, including the dominant temperature, precipitation, or dominant vegetation type (Eronen et al., 2010b; Vermillion et al., 2018). However, the exact associations are captured computationally when fitting statistical models to link the traits of communities with their environmental conditions. Due to its functional perspective, ecometrics has sometimes been referred to as a taxon-free approach to highlight that the ecology of fossil organisms is not inferred from that of their nearest living relatives and that ecometric approaches do not rely on presence or absence of any particular individual taxa. Theoretically, ecometric approaches can be completely taxon blind—e.g., analyzing random samples of teeth that are found at localities without considering any taxonomic information. However, in practice, for robustness and potentially larger samples, most ecometric studies, including the present work, analyze the distribution of traits over species at localities, sometimes on global-scale datasets (e.g., Liu et al., 2012; Žliobaitė et al., 2018). Ecometrics, thus, does not focus on individual organisms but deals with the functional composition of communities, thereby enabling comparisons between present and past communities. Of course, there is no way to statistically test for predictions about the past (retrodictions), which makes it necessary to use different approaches and see how they compare. Because of its relatively fast and nondestructive sampling, dental ecometrics has been particularly used to make large-scale inferences about paleoenvironmental and biotic changes across different geographic and temporal scales, and different tailored global and regional models have been developed for different types of analysis (Eronen et al., 2010b; Liu et al., 2012; Fortelius et al., 2016; Žliobaitė et al., 2016; Oksanen et al., 2019).

In this study, we aim at predicting multiple climatic characteristics of CLL1, for which we use the dental trait scoring scheme reported in Žliobaitė et al. (2016), termed functional crown types (FCT). This scheme was devised to potentially increase the scope of environmental predictions particularly in the Late Miocene and Plio-Pleistocene. The scheme is based on a modular system called crown types (Jernvall, 1995), which was designed to be applicable to mammals generally, primarily focusing on the shape of unworn teeth. Functional crown types, in contrast, focus on characteristics of how teeth wear while in use. The FCT scheme includes seven variables (Žliobaitė et al., 2016; Galbrun et al., 2018). Ecometric approaches based on hypsodonty (HYP; relative crown height) and loph count (Fortelius et al., 2002, 2016; Eronen et al., 2010a, 2010c; Liu et al., 2012; Oksanen et al., 2019) have been used to estimate paleoprecipitation and productivity in the Old World from the Miocene through the Pleistocene, as well as temperature in the Pleistocene. However, the set of seven variables of the FCT scheme offers potential for estimating a broader set of climatic conditions, including seasonality characteristics and reoccuring extremities of climatic conditions (Žliobaitė et al., 2016). Here, we apply regional and global ecometric approaches based on the FCT of herbivorous large mammals from CLL1 to refine the previous paleoenvironmental inferences on the habitat of H. laietanus.