INTRODUCTION
Assuming that large mammals were a relevant resource for humans during the early Pleistocene (e.g., Binford, Reference Binford1981, Reference Binford1985; Marean, Reference Marean1989; Gaudzinski and Roebroeks, Reference Gaudzinski and Roebroeks2000; Roebroeks, Reference Roebroeks2001; McNabb, Reference McNabb2007; Speth, Reference Speth2010), the ability of hominins to obtain meat was conditioned by prey abundance, their ecological characteristics, and the intensity of competition with other carnivorous mammals for this trophic resource. The percentage of animal resource in the diet of recent hunter-gatherer populations is variable (Cordain et al., Reference Cordain, Miller, Eaton, Mann, Holt and Speth2000), representing between 30 and 60% of their nutritional intake (Jenike, Reference Jenike2001; Leonard et al., Reference Leonard, Robertson and Snodgrass2007). Binford (Reference Binford2001) showed that in recent hunter-gatherer populations hunting represents between 11 and 89% (mean 38%) of their food resources; gathering, between 0.01 and 88% (mean 45%); and fishing, between 0 and the 70% (mean 18%). Animal food, large game in particular, is generally considered a key resource for the Paleolithic hunter-gatherer populations (Bunn and Ezzo, Reference Bunn and Ezzo1993; Mann, 2000; Bunn and Pickering, Reference Bunn and Pickering2010; Domínguez-Rodrigo et al., Reference Domínguez-Rodrigo, Bunn, Mabulla, Baquedano, Uribelarrea, Pérez-González and Gidna2014), although this does not imply that the different Pleistocene Homo species were strictly carnivorous. On the contrary, species in the genus Homo were omnivorous, and they likely included a significant amount of plant food in their diets, as highlighted by Hardy et al. (Reference Hardy, Radini, Buckley, Blasco, Copeland, Burjachs, Girbal, Yll, Carbonell and Bermúdez de Castro2017) and Prado-Nóvoa et al. (Reference Prado-Nóvoa, Mateos, Zorrilla-Revilla, Vidal-Cordasco and Rodríguez2017). The availability of resources and the competition with carnivores have been repeatedly proposed as key limiting factors for the early European human populations (e.g., Turner, Reference Turner1992; Martínez-Navarro and Palmqvist, Reference Martínez-Navarro and Palmqvist1996; Arribas and Palmqvist, Reference Arribas and Palmqvist1999; Palombo, Reference Palombo2007, Reference Palombo2010, Reference Palombo2013; Madurell-Malapeira et al., Reference Madurell-Malapeira, Minwer-Barakat, Alba, Garcés, Gómez, Aurell-Garrido, Ros-Montoya, Moyà-Solà and Berástegui2010b; Manzi et al., Reference Manzi, Magri and Palombo2011; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). Turner (Reference Turner1992) and Palombo (Reference Palombo2010) proposed that competition with scavengers and/or predators could have delayed human expansion throughout Europe during the early Pleistocene because of the assumed limited technological capabilities of those first European humans. Palombo (Reference Palombo2014) suggested that the changes in the structure of the mammal paleocommunities, together with an enlarged prey spectrum, played an important role in the success of human settlements at that time. Conversely, Meloro and Kovarovic (Reference Meloro and Kovarovic2013) suggested that human arrival modified the structure of the mammal paleocommunities. Some scholars propose that the early Pleistocene human populations were strongly dependent on the scavenging of ungulate carcasses (Turner, Reference Turner1992; Martínez-Navarro and Palmqvist, Reference Martínez-Navarro and Palmqvist1996; Arribas and Palmqvist, Reference Arribas and Palmqvist1999; Espigares et al., Reference Espigares, Martínez-Navarro, Palmqvist, Ros-Montoya, Toro, Agustí and Sala2013). Saber-toothed felids likely yielded a high amount of carrion in the latter part of the early Pleistocene paleoecosystems, given their dental characteristics and killing capabilities (Marean, Reference Marean1989; Turner, Reference Turner1992; Martínez-Navarro and Palmqvist, Reference Martínez-Navarro and Palmqvist1995; Palmqvist et al., Reference Palmqvist, Martínez-Navarro and Arribas1996, Reference Palmqvist, Martínez-Navarro, Toro, Espigares, Ros-Montoya, Torregrosa and Pérez-Claros2005, Reference Palmqvist, Torregrosa, Pérez-Claros, Martínez Navarro and Turner2007, Reference Palmqvist, Martínez-Navarro, Pérez-Claros, Torregrosa, Figueirido, Jiménez-Arenas, Espigares, Ros-Montoya and De Renzi2011; Arribas and Palmqvist, Reference Arribas and Palmqvist1999). The giant hyena (Pachycrocuta brevirostris) was likely a powerful competitor for Homo in the search for carrion, and pack-hunting wild dogs and large felids, among others, were strong hunting competitors (Antón et al., Reference Antón, Galobart and Turner2005; Madurell-Malapeira et al., Reference Madurell-Malapeira, Minwer-Barakat, Alba, Garcés, Gómez, Aurell-Garrido, Ros-Montoya, Moyà-Solà and Berástegui2010b; Palombo, Reference Palombo2010).
Climatic conditions changed in the course of the early to middle Pleistocene transition (Schneider and Root, Reference Schneider and Root1998; Shackleton, Reference Shackleton1995; Maslin and Ridgwell, Reference Maslin and Ridgwell2005) during the so-called mid-Pleistocene Revolution (MPR) (Maslin and Ridgwell, Reference Maslin and Ridgwell2005). The MPR promoted the renewal of the mammalian faunal complex, with the appearance of new carnivores in Europe and a progressive increase of the herbivore richness (e.g., Turner, Reference Turner1992; Azanza et al., Reference Azanza, Palombo and Alberdi2004; Rodríguez et al., Reference Rodríguez, Alberdi, Azanza and Prado2004; Cuenca-Bescós et al., Reference Cuenca-Bescós, Rofes and García-Pimienta2005; Meloro et al., Reference Meloro, Raia and Barbera2007; Palombo, Reference Palombo2007; Raia et al., Reference Raia, Meloro and Barbera2007; Meloro, Reference Meloro2011b). Thus, it may be speculated that Homo had better access to carcasses after the MPR than in the former period, and that human weapons and tool kits became more effective, facilitating its dispersal across Europe (Palombo, Reference Palombo2010; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). Nevertheless, the debate around the dispersal of human populations in Europe during the middle Pleistocene is still open. Several authors provide arguments in favor of an African or Near East source of hominins and technological modes, suggesting that Homo antecessor populations were replaced by a new, more technologically advanced African hominin species (the Acheulean [Mode 2] technology) (Lordkipanidze et al., Reference Lordkipanidze, Jashashvili, Vekua, Ponce de León, Zollikofer, Rightmire and Pontzer2007; Santonja and Pérez-González, Reference Santonja and Pérez-González2010; Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011; Pérez-Claros et al., Reference Pérez-Claros, Jiménez-Arenas and Palmqvist2015). Carbonell et al. (Reference Carbonell, Mosquera, Rodríguez, Sala and Made1999) suggest the replacement of human populations was slow, initially with two technological modes (Mode 1 and Mode 2) coexisting in Europe. Carbonell et al. (Reference Carbonell, Sala Ramos, Rodríguez, Mosquera, Ollé, Vergès, Martínez-Navarro and Bermúdez de Castro2010) argue that the Acheulean spread throughout Eurasia because of the demographic growth in both the Near East and Africa. However, the origin of the European Acheulean is much debated because of a lack of data on the (a) arrival of new traditions, (b) technological changes because of contact with new hominin groups, and (c) local origin of this technological mode (Mode 2 or Acheulean) (e.g., see Moncel et al., 2016c; Mosquera et al., Reference Mosquera, Ollé, Saladié, Cáceres, Huguet, Rosas and Villalaín2016). Whatever the causes and processes, the human expansion was evident around 0.6–0.5 Ma, when the number of archaeological sites increased significantly and the northern latitudes began to be frequently inhabited (e.g., Thieme, Reference Thieme1997; Mania and Vlcek, Reference Mania and Vlcek1999; Roberts and Parfitt, Reference Roberts and Parfitt2000), showing the different paleoanthropological and cultural features that define the European middle Pleistocene (Doronichev and Golovanova, Reference Doronichev and Golovanova2010; Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011; Ollé et al., Reference Ollé, Mosquera, Rodríguez, Lombera-Hermida, de, García-Antón, García-Medrano and Peña2013). However, although the Acheulean culture was undoubtedly established in Europe around 0.6–0.5 Ma, a number of older lithic assemblages, dating back to the end of the early Pleistocene, exhibit early Acheulean or later Mode 1 features (Scott and Gibert, Reference Scott and Gibert2009; Barsky and de Lumley, Reference Barsky and de Lumley2010; Santonja and Pérez-González, Reference Santonja and Pérez-González2010; Barsky et al., Reference Barsky, Garcia, Martínez, Sala, Zaidner, Carbonell and Toro-Moyano2013; Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013, Reference Mosquera, Ollé, Saladié, Cáceres, Huguet, Rosas and Villalaín2016; Walker et al., Reference Walker, López-Martínez, Carrión-García, Rodríguez-Estrella, San-Nicolás del-Toro, Schwenninger and López-Jiménez2013; Vallverdú et al., Reference Vallverdú, Saladié, Rosas, Huguet, Cáceres, Mosquera and Garcia-Tabernero2014; Moncel et al., Reference Moncel, Arzarello, Boëda, Bonilauri, Chevrier, Gaillard, Forestier, Yinghua, Sémah and Zeitoun2016a, Reference Moncel, Arzarello, Boëda, Bonilauri, Chevrier, Gaillard, Forestier, Yinghua, Sémah and Zeitoun2016b), although the age of some them is debatable (Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011).
Several researchers and studies defend a depopulation or discontinuity in the human occupation of Europe during the early to middle Pleistocene transition (Moncel, Reference Moncel2010; Santonja and Pérez-González, Reference Santonja and Pérez-González2010; Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011; Moncel et al., Reference Moncel, Despriée, Voinchet, Tissoux, Moreno, Bahain, Courcimault and Falguères2013; Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013, Reference Mosquera, Ollé, Saladié, Cáceres, Huguet, Rosas and Villalaín2016; Vallverdú et al., Reference Vallverdú, Saladié, Rosas, Huguet, Cáceres, Mosquera and Garcia-Tabernero2014). Environmental factors, such as the extinction of saber-toothed cats, have been frequently suggested as possible causes of the discontinuity in human occupation or the disappearance of Mode 1 technology in Europe (Martínez-Navarro and Palmqvist, Reference Martínez-Navarro and Palmqvist1995, Reference Martínez-Navarro and Palmqvist1996; Arribas and Palmqvist, Reference Arribas and Palmqvist1999; Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011; Palmqvist et al., Reference Palmqvist, Martínez-Navarro, Pérez-Claros, Torregrosa, Figueirido, Jiménez-Arenas, Espigares, Ros-Montoya and De Renzi2011). The existence of several human migration waves into Europe during the early Pleistocene has also been proposed (O’Regan, Reference O’Regan2008; Agustí et al., Reference Agustí, Blain, Cuenca-Bescós and Bailon2009; Carbonell et al., Reference Carbonell, Sala Ramos, Rodríguez, Mosquera, Ollé, Vergès, Martínez-Navarro and Bermúdez de Castro2010; Made and Mateos, Reference Made and Mateos2010; Muttoni et al., Reference Muttoni, Scardia and Kent2010; Muttoni et al., 2015; Palombo, Reference Palombo2010, Reference Palombo2013; O’Regan et al., Reference O’Regan, Turner, Bishop and Lamb2011; Bermúdez de Castro et al., Reference Bermúdez de Castro, Martinón-Torres, Blasco, Rosell and Carbonell2013, 2016; Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013, Reference Mosquera, Ollé, Saladié, Cáceres, Huguet, Rosas and Villalaín2016; Carotenuto et al., Reference Carotenuto, Tsikaridze, Rook, Lordkipanidze, Longo, Condemi and Raia2016).
According to the hypothesis that human presence in Europe was conditioned by competition with carnivores (Martínez Navarro, Reference Martínez Navarro1992; Arribas and Palmqvist, Reference Arribas and Palmqvist1999; Palombo, Reference Palombo2007, Reference Palombo2010; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012), the discontinuity in human occupation during the early to the middle Pleistocene transition could be explained by high competition intensity among the secondary consumers, including humans, in this period. Competition might make access to meat more difficult for humans, making their presence in the paleocommunities less stable. Predator/prey ratios have been traditionally used in paleontology to measure the intensity of competition in paleocommunities (Stiner, Reference Stiner1992; Palombo and Mussi, Reference Palombo and Mussi2006; Raia et al., Reference Raia, Meloro and Barbera2007; Palmqvist et al., Reference Palmqvist, Pérez-Claros, Janis and Gröcke2008; Croitor and Brugal, Reference Croitor and Brugal2010; Feranec et al., Reference Feranec, García, Díez and Arsuaga2010; Palombo, Reference Palombo2010; Meloro and Clauss, Reference Meloro and Clauss2012; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). This ratio is a measure of the efficiency of a guild of secondary consumers to consume prey biomass; lower values of the predator/prey ratio indicate that the guild of secondary consumers of a community is less efficient at using prey resources than the guild in another community with higher values, because in the former case the community requires more biomass of primary consumers to obtain the same number or biomass of secondary consumers (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González and Rodríguez2016a). In some studies, the predator/prey ratios are based on species numbers, whereas in others the ratios are based on biomass and density estimations. Several studies showed a similar pattern of variation in the predator/prey ratios in the European paleocommunities from the latter part of the early Pleistocene (late Villafranchian) to the latter part of the middle Pleistocene (late Galerian) (Meloro et al., Reference Meloro, Raia and Barbera2007; Raia et al., Reference Raia, Meloro and Barbera2007; Croitor and Brugal, Reference Croitor and Brugal2010; Palombo, Reference Palombo2010): an abrupt decrease in the predator/prey ratio (based on number of species) from the late Villafranchian (1.2 Ma) to the early Galerian (0.9 Ma), a slight increase in the middle Galerian (0.6 Ma), and, finally, a slight decrease in the late Galerian (0.45 Ma) to a value that was maintained approximately constant during the Aurelian (0.3 Ma). This pattern could be interpreted as a reduction of the efficiency of the predator guilds from the late Villafranchian to the Galerian with a slight increase in the middle Galerian; therefore, more resources were necessary to maintain the same number or biomass of secondary consumers in the Galerian than in the Epivillafranchian. The efficiency of a guild of secondary consumers is not directly related to competition intensity inside the community, because the efficiency depends on different aspects like the diet and prey preferences of the secondary consumer species, the composition of the secondary and primary consumers guilds, or the mortality rate of subadults of the primary consumer species. Thus, predator/prey ratios are not enough to measure the competition intensity inside the guild of secondary consumers. Hence, it is necessary to use others indexes to test the hypothesis that the discontinuity in the human occupation of Europe was because of a high competition for meat among the secondary consumers at that time (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González and Rodríguez2016a).
Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Rodríguez, Martín-González, Goikoetxea and Mateos2013) developed a quantitative model to estimate the availability of resources for secondary consumers and to study predator/prey relationships, which was inspired by previous studies (Bermúdez de Castro et al., Reference Bermúdez de Castro, Díez Fernández-Lomana, Mosquera Martínez, Nicolás Checa, Pérez Pérez, Rodríguez Méndez and Sánchez Marco1995; Fariña, Reference Fariña1996; Palmqvist et al., Reference Palmqvist, Gröcke, Arribas and Fariña2003; Vizcaíno et al., Reference Vizcaíno, Fariña, Zárate, Bargo and Schultz2004, Reference Vizcaíno, Bargo, Kay, Fariña, Giacomo, Perry, Prevosti, Toledo, Cassini and Fernicola2010). That model was applied, at a local scale, to study the large mammal paleocommunity from the TD6 level of the Gran Dolina site (Sierra de Atapuerca, Spain). In Atapuerca, human settlements are documented between 1.2 and 0.9 Ma and between 0.5 and 0.25 Ma, but there is no evidence of human presence for the period between 0.9 and 0.5 Ma (Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013). The same methodology was used to compare the competition intensity among secondary consumers at Atapuerca in two time periods, represented by two fossil assemblages from the Gran Dolina site. Those assemblages were TD6, dated to 900 ka and with human presence, and TD8, dated to ~600 ka and without human presence. This methodology was used to evaluate competition intensity as a possible explanation for the discontinuity of human presence at a local scale (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González, Blasco, Rosell and Rodríguez2014b). Competition intensity was higher at TD8 than at TD6, and Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Mateos, Martín-González, Blasco, Rosell and Rodríguez2014b) concluded that this could be a factor relevant to determine human occupation opportunities. An extension of that study was carried out by Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Rodríguez, Martín-González and Mateos2017) showing that TD6 exhibited exceptionally low competition intensity, suggesting that the TD6 paleocommunity likely included a large felid species that was not recorded in the fossil assemblage. However, the hypothesis that the lack of human presence in the TD8 assemblage might be related to high competition intensity inside the secondary consumer guild was not rejected. Moreover, this methodology was also applied to the study of the large mammal assemblages from two Orce sites (Spain), Barranco León D and Fuente Nueva 3, which preserve the oldest evidence of human presence in western Europe (Toro-Moyano et al., Reference Toro-Moyano, Martínez-Navarro, Agustí, Souday, Bermúdez de Castro, Martinón-Torres and Fajardo2013). The analyses showed higher values of competition intensity in Orce than those recorded in the analyzed levels from the Gran Dolina and Galería sites of Atapuerca (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Palmqvist, Rodríguez, Mateos, Martín-González, Espigares, Ros-Montoya and Martínez-Navarro2016b). In spite of this, intraguild competition did not impede human settlement of Orce at 1.4 Ma. Using this methodology in the site of Venta Micena (Orce), a recent study (Rodríguez-Gómez et al. 2017) suggests that the human settlement in Orce was probably not a matter of ecological opportunity. Competition intensity in those studies was estimated via several indexes based on the estimated and expected densities of the occurring species and accounting for the ecological characteristics of the potential prey and prey preferences of the carnivores. These indexes have been shown to represent competition intensity in recent and past communities better than simple predator/prey ratios (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González and Rodríguez2016a).
Our aim is to extend the analyses carried out by Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Mateos, Martín-González, Blasco, Rosell and Rodríguez2014b, Reference Rodríguez-Gómez, Rodríguez, Martín-González and Mateos2017) to a continental scale to test whether variations in competition intensity and availability of trophic resources acted as limiting factors for the human occupation of Europe. For this, we use the competition indexes described by Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Mateos, Martín-González and Rodríguez2016a). Our analysis focuses on three time intervals of the equivalent duration: the oldest interval is roughly defined by the beginning of the Jaramillo subchron and the Matuyama-Brunhes boundary (1.1–0.8 Ma) and represents the period when populations with Oldowan technologies were well established in southern Europe (Carbonell et al., Reference Carbonell, Bermúdez de Castro, Parés, Pérez-González, Cuenca-Bescós, Ollé and Mosquera2008; Toro-Moyano et al., Reference Toro-Moyano, Martínez-Navarro, Agustí, Souday, Bermúdez de Castro, Martinón-Torres and Fajardo2013); the second interval (0.8–0.5 Ma) coincides with an apparent decrease in human presence in Europe; and the third interval (0.5–0.2 Ma) coincides with an expansion and increased intensity of human occupation at the continental scale. We compare the competition intensity in the European ecological communities among these three periods using several indexes of intraguild competition to test whether the apparent depopulation of Europe from 0.8 to 0.5 Ma may be related to an increased difficulty in access to trophic resources.
MATERIALS
We selected local faunas as units of analysis because we consider them to be the best analogues of the biological communities. The term “paleocommunity” is often applied in paleoecology to the fossil fauna of a given age found in a large geographic area (e.g., the Italian peninsula, England, or central Europe) at a given period (Raia et al., Reference Raia, Piras and Kotsakis2005, Reference Raia, Meloro and Barbera2007, Reference Raia, Carotenuto, Meloro, Piras, Barbera and Kotsakis2009; Meloro et al., Reference Meloro, Raia and Barbera2007; Meloro, Reference Meloro2011a; Meloro and Clauss, Reference Meloro and Clauss2012). Although it is not our intention here to discuss whether that use of the term “paleocommunity” is appropriate, what is clear is that those paleocommunities are not adequate units of analysis to evaluate competition inside a guild. For competition to exist, the potential competitors should coincide in time and space, and a paleocommunity may include species living, for whatever reason, in different environments or in different areas inside the region. Thus, information on European local faunal assemblages (LFAs) (longitude 10°00.00’W to 45°00.00’E and latitude 30°00.00’N to 55°00.00’N”) dated from the latter part of the early Pleistocene to the latter part of the middle Pleistocene (1.1–0.2 Ma) was compiled from published sources. Initially, LFAs with mammal species weighing more than 10 kg were selected, and those with no reliable dating were rejected, yielding a total of 98 LFAs from 71 localities. Because complete or nearly complete faunas are required for this analysis, the LFAs were filtered according to the number of prey and predator species in the assemblage. Only those LFAs with both a number of prey and a number of carnivore species above the median for the 98 faunal assemblages (>8 primary consumer species, >4 secondary consumer species, and at least 12 species by LFA) were selected (Supplementary Tables 1 and 2). This criterion was met by 36 LFAs distributed in three groups, which correspond to the following three time intervals: interval 1, from 1.1 to 0.8 Ma; interval 2, from 0.8 to 0.5 Ma; and interval 3, from 0.5 to 0.2 Ma. Thus, we have 9 faunal assemblages in the first group, 7 in the second group, and 20 in the third group (Table 1, Fig. 1).
(color online) Geographic distribution of local faunas included in the analyses for the 1.1 to 0.2 Ma time intervals. There are 9 local faunas in interval 1 (1.1–0.8 Ma, top); 7 in interval 2 (0.8–0.5 Ma, middle); and 20 in interval 3 (0.5–0.2 Ma, bottom).
The early and middle Pleistocene faunal assemblages used in this study were distributed in three time intervals (see text) and assigned locality codes (LCs). The number of macromammal species of primary consumers (N1) and secondary consumers (N2) are indicated for each assemblage. HE indicates that the assemblage includes evidence of human presence.
We reviewed all faunal lists and applied uniform taxonomic criteria (see Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012 and references therein) to obtain a taxonomically consistent database. Our analysis was restricted to mammal species of more than 10 kg because they constitute the portion of the food web that allegedly included hominins (Binford, Reference Binford1981, Reference Binford1985; Marean, Reference Marean1989; Díez et al., Reference Díez, Fernández-Jalvo, Rosell and Cáceres1999; Gaudzinski and Roebroeks, Reference Gaudzinski and Roebroeks2000; Roebroeks, Reference Roebroeks2001; Speth, Reference Speth2010; Saladié et al., Reference Saladié, Huguet, Díez, Rodríguez-Hidalgo, Cáceres, Vallverdú, Rosell, Bermúdez de Castro and Carbonell2011; Lozano et al., Reference Lozano, Mateos and Rodríguez2016). Primary consumer species included in this study belong to the families Bovidae, Castoridae, Cercopithecidae, Cervidae, Elephantidae, Equidae, Hippopotamidae, Hystricidae, Rhinocerotidae, and Suidae. The secondary consumers belong to the families Canidae, Felidae, Hominidae, Hyaenidae, and Ursidae. The family Mustelidae and the genus Vulpes were excluded because their diet is mainly based on small mammals (Seebeck, Reference Seebeck1978; Carbone et al., Reference Carbone, Mace, Roberts and Macdonald1999).
To perform our analyses, it was necessary to estimate several physiological and life history parameters for every primary consumer species, such as adult body mass, body mass at birth, litter size, breeding interval, age at reproductive maturity, growth rate, and life span. Only adult body mass was required for secondary consumers. We obtained the values of these physiological variables for recent species from the PanTHERIA database (Jones et al., Reference Jones, Bielby, Cardillo, Fritz, O’Dell, Orme and Safi2009). Values for the adult body mass of fossil species were obtained from the literature (Weers, Reference Weers1994; Alberdi et al., Reference Alberdi, Prado and Ortiz-Jaureguizar1995; Koufos et al., Reference Koufos, Kostopoulos and Sylvestrou1997; Rodríguez, Reference Rodríguez1997; Made, Reference Made1998; Nowak, Reference Nowak1999; Collinge, Reference Collinge2001; Crégut-Bonnoure and Spassov, Reference Crégut-Bonnoure and Spassov2002; Athanassiou, Reference Athanassiou2003; Palmqvist et al., Reference Palmqvist, Gröcke, Arribas and Fariña2003; Brugal and Fosse, Reference Brugal and Fosse2004; Prado et al., Reference Prado, Alberdi, Azanza and Rodríguez2004; Antón et al., Reference Antón, Galobart and Turner2005; Breda and Marchetti, Reference Breda and Marchetti2005; Crégut-Bonnoure and Tsoukala, Reference Crégut-Bonnoure and Tsoukala2005; Kahlke and Gaudzinski, Reference Kahlke and Gaudzinski2005; Crégut-Bonnoure and Dimitrijevic, Reference Crégut-Bonnoure and Dimitrijevic2006; Croitor and Brugal, Reference Croitor and Brugal2007; Meloro et al., Reference Meloro, Raia and Barbera2007; Fostowicz-Frelik, Reference Fostowicz-Frelik2008; Carotenuto, Reference Carotenuto2009; Jones et al., Reference Jones, Bielby, Cardillo, Fritz, O’Dell, Orme and Safi2009; Lister and Stuart, Reference Lister and Stuart2010; Madurell-Malapeira et al., Reference Madurell-Malapeira, Alba, Moyà-Solà and Aurell-Garrido2010a; Jiménez-Arenas et al., Reference Jiménez-Arenas, Pérez-Claros, Aledo and Palmqvist2014; Made et al., Reference Made, Stefaniak and Marciszak2014; Arsuaga et al., Reference Arsuaga, Carretero, Lorenzo, Gómez-Olivencia, Pablos, Rodríguez and García-González2015). For species with living representatives and without body mass values, we took the values of the living populations. Sometimes, taxa were identified in the original source to the genus, subfamily, or family level only. In these cases, we computed the mean adult body weight of the species present in Europe in that time interval in the corresponding taxonomic group (genus, subfamily, or family). For instance, the adult body mass of a taxon identified in an LFA as “Equus indet.” was estimated as the mean body weight of the species in the genus Equus occurring in that period. In the case of Homotherium sp., the mean body mass of Homotherium latidens and Homotherium crenatidens was used (Brugal and Fosse, Reference Brugal and Fosse2004; Antón et al., Reference Antón, Galobart and Turner2005), although likely they may be considered a single species (Antón et al., Reference Antón, Salesa, Galobart and Tseng2014). A least squares regression equation was computed for each primary consumer family or subfamily to estimate each physiological variable from the mean body weight (Supplementary Table 3). The equation was computed only when data for at least four species in the family were available. Only those regression equations in which body mass explained more than 80% of the variance in the dependent variable were used. If body mass explained less than 40% of the variance, we took the median for the family or subfamily. If it explained between 40 and 80%, we estimated the value of the physiological variable as the median for the species in the family or subfamily with a body weight similar to that of the species studied.
METHODS
The model
We investigated the distribution of meat resources (i.e., primary consumer biomass) among secondary consumers in the Pleistocene using a mathematical model that estimates the amount of primary consumer biomass available for the secondary consumers in a community (total available biomass, or TAB) and the requirements of secondary consumers (total demanded biomass, or TDB) (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Martín-González, Goikoetxea, Mateos and Rodríguez2014a). These concepts were already dealt with and discussed by Prevosti and Vizcaíno (Reference Prevosti and Vizcaíno2006) for paleocommunities from South America. A summary description of the model components is provided subsequently; for a detailed formal description of this model, see Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Rodríguez, Martín-González, Goikoetxea and Mateos2013). The model was written and executed in Matlab R2009b.
TAB
Our model was developed on the assumption that all of the variations in population size and composition may be taken as oscillations around a mean value that is constant through time (i.e., population fluctuations are randomly distributed above and below the value of this mean), an assumption that is widely accepted in population dynamics studies (Owen-Smith, Reference Owen-Smith2010). We represented the average long-term condition of every population using a Leslie matrix (Leslie, Reference Leslie1945, Reference Leslie1948). Leslie matrices are used in population dynamics to represent a population structure at different times and to describe its oscillations. We conditioned Leslie matrices to obtain the average structure of a population that was stable (i.e., population size should be constant from year to year) and stationary (i.e., the age structure should be constant from year to year over time).
The input data of the model are species-specific physical and physiological variables, including adult body mass, body mass at birth, litter size, breeding interval, age at reproductive maturity, growth rate, and life span. The population profiles obtained from this model for every primary consumer population provide estimates of the average sustainable biomass output by age classes, which were eventually translated into body-size intervals. Biomass output by age interval was obtained from the annual mortality rates obtained from the Leslie matrix. Each potentially dead individual of a primary consumer species was assigned to one of six size categories according to its average body mass at the age of death: 10–45, 45–90, 90–180, 180–360, 360–1000, or >1000 kg (see Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). The biomass made available for secondary consumers by each single primary consumer population was obtained as the sum of the biomass of all dead individuals. Sustainable primary consumer biomass output was obtained as the sum of the biomass outputs in each size category from each primary consumer population.
Combining the mortality profiles obtained from the Leslie matrix with the mean body size per age class and the estimated population density of the species, the sustainable biomass output (total biomass output, or TBO) can be computed. A size-specific “wastage factor” (modified from Viljoen, Reference Viljoen1993) is included in the model to account for the fact that a carcass includes a variable amount of nonedible tissues (e.g., horns, bones, and hide), which are included in the TBO provided by the model. Thus, this percentage of nonedible biomass is subtracted from the TBO to obtain the final amount of biomass available to secondary consumers, or total biomass or TAB, which is also distributed by body mass classes. See Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Rodríguez, Martín-González, Goikoetxea and Mateos2013) for the computational details.
The model yields several population profiles for each species, corresponding to different mortality rates. We selected extreme values with maximum and minimum pressure on subadults (or maximum and minimum mortality rates) that produce minimum and maximum TAB levels, respectively (TAB-m and TAB-M, respectively).
The model solutions are not dependent on population size: thus, we needed to estimate the population density of each primary consumer species. We used the equation provided by Damuth (Reference Damuth1981) for European mixed temperate forest ecosystems for all primary consumer species included in this study:
$${\rm log}\left( D \right)\, {\equals}\, {\minus}0.79\, {\asterisk}\, {\rm log}\left( m \right)\, {\plus}\, 4.33,r^{2} \, {\equals}\, 0.94,$$
where D is the population density (individuals/km2), and m is the body mass (g). We used Damuth’s (Reference Damuth1981) equation instead of Silva and Downing’s (Reference Silva and Downing1995) equation because the former shows better goodness-of-fit values (Jones et al., Reference Jones, Bielby, Cardillo, Fritz, O’Dell, Orme and Safi2009) than the latter, according to a chi-square test.
TDB
Carnivore-demanded resources should be estimated as a first step in evaluating resource distribution among secondary consumers. The secondary consumer intake rate was estimated using the equation reported by Farlow (Reference Farlow1976):
$${\rm log}\,I\, {\equals}\, \left( {0.69686\,\pm\,0.01276} \right)\log \left( m \right)\, {\plus}\, 0.27747,r^{2} \, {\equals}\, 0.97,$$
where I is the intake rate (km/day), and m is the body mass (g). We used the maximum value for the slope (0.70962) to estimate the maximum demanded biomass for secondary consumer species. Some adjustments were made for some secondary consumers according to their inferred dietary preferences based on Rodríguez et al. (Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012) and Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Rodríguez, Mateos, Martín-González and Goikoetxea2012) (and references cited therein) (Table 2). We estimated that large-mammal flesh represented 20% of the energetic requirements of Canis arnensis, Canis etruscus, and Canis mosbachensis; 10% for Lynx pardinus and Lynx sp.; 80% for Lynx issidorensis; 98% for Chasmaporthetes lunensis, Crocuta crocuta, Crocuta sp., and Pachycrocuta brevirostris (because 2% of the total requirements could be obtained from bone marrow); 45% for Homo antecessor, Homo heidelbergensis, and Homo sp. because 45% is the average animal resource consumption of recent hunter-gatherer populations (Jenike, Reference Jenike2001; Leonard et al., Reference Leonard, Robertson and Snodgrass2007); 75% for Hyaena sp.; 10% for Ursus arctos, Ursus deningeri, Ursus dolinensis, Ursus etruscus, Ursus sp., and Ursus thibetanus; and 1% for Ursus spelaeus. For other secondary consumers, we assumed that flesh represented 100% of their energetic requirements (Fig. 2A).
Graphic representation of the computation steps to obtain the total biomass demanded (TDB) and the proportional predation pressure (PPP) used to distribute the total available biomass among secondary consumers (SCs). (A) Percentage of meat of large mammals (>10 kg) in the diet of the different species in the guild of SCs of a hypothetical assemblage. Color bars and the numbers inside them indicate the percentage of the diet consisting of meat of large mammals. Black bars indicate the part of the diet represented by other food resources (e.g., mammals weighing <10 kg, birds, reptiles, amphibians, or plant resources). (B) Biomass demanded by each SC population from the primary consumer (PC) species (kcal/km2/yr) in different body-size (BS) categories. The total requirements of each population are distributed among the six categories according to the prey preference profile or percentage of predation of the SC (see text and Table 2). These total requirements would be met in an optimal condition with maximum densities. (C) TDB by BS categories of PCs (kcal/km2/yr). Each bar represents the sum of the biomasses demanded by all the SCs from each BS category of PCs, according to panel B. (D) PPP (%) for each BS category of PCs according to the estimations obtained in panel C. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
The guild of secondary consumers present in the set of European assemblages included in this study (1.1–0.2 Ma) with their body mass in kilograms and their energetic requirements in kilocalories per square kilometer per year. The requirements of secondary consumers were corrected according to their diet by multiplying the total requirements of each species by a correction factor. The last six columns represent the preference, expressed in percentage, of each species of secondary consumer for the primary consumers in different body-size categories (see text).
The annual energetic requirements of a carnivore population per square kilometer are obtained by multiplying the individual annual intake by the population density. The equation provided by Damuth (Reference Damuth1993) for African flesh eaters was used to estimate the typical secondary consumer density because it included species more similar to those present in European Pleistocene fauna:
$${\rm log}\left( D \right)\, {\equals}\, {\minus}0.64\,{\times}\log \left( m \right)\, {\plus}\, 2.23,r^{2} \, {\equals}\, 0.36,$$
where D is the population density (individuals/km2), and m is the body mass (g). As for the primary consumer, we used Damuth’s (Reference Damuth1993) equation instead of Silva and Downing’s (Reference Silva and Downing1995) equation for the same reasons.
As in the case of TAB, TDB was distributed over the same six body-size categories based on the inferred prey-size preferences of each predator and based on the behavior of their living relatives (Table 2) (Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). The preference of a predator for a body-size category is represented by the percentage of predation (PD) that this size category was presumed to represent in its diet. If a predator was presumed to be unable to kill prey in a given size category and to not consume carrion, a PD of 0 was assigned to the predator in that size category (Fig. 2B and C).
Distribution of TAB between secondary consumers
The distribution of TAB among secondary consumers is based on the proportional predation pressure (PPP ij ) of each species in each body-size category (Fig. 2D). PPP ij represents the relative amount of biomass demanded by the jth secondary consumer species from the ith primary consumer body-size category and is calculated as the proportion of the total amount of biomass demanded from a prey body-size category by all carnivores that corresponds to the requirements of a single carnivore species. PPP ij incorporates intraguild competition in the model (Fig. 2D). A detailed formal description of resource distribution computation is available in Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Rodríguez, Martín-González, Goikoetxea and Mateos2013). For a numerical example, we refer the reader to Supplementary Table 4.
Analytic methods
Predator/prey ratios have been traditionally used in paleontology to measure the intensity of competition in the fossil faunas of the past (Palombo and Mussi, Reference Palombo and Mussi2006; Raia et al., Reference Raia, Meloro and Barbera2007; Palmqvist et al., Reference Palmqvist, Pérez-Claros, Janis and Gröcke2008; Croitor and Brugal, Reference Croitor and Brugal2010; Feranec et al., Reference Feranec, García, Díez and Arsuaga2010; Palombo, Reference Palombo2010; Meloro and Clauss, Reference Meloro and Clauss2012; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012; Stiner, Reference Stiner1992; but see Volmer and Hertler, Reference Volmer and Hertler2016). Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Mateos, Martín-González and Rodríguez2016a) discussed the utilization of predator/prey ratio versus estimated/expected density indexes to measure competition intensity among secondary consumers and concluded that the indexes based on estimated/expected densities were more accurate than predator/prey indexes. Therefore, we used estimated/expected density indexes (the global competition index [GCI] and the global competition index biomass [GCIB]) to compare the competition intensity in the different faunal assemblages selected in our analysis (see Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González, Blasco, Rosell and Rodríguez2014b, Reference Rodríguez-Gómez, Rodríguez, Martín-González and Mateos2017).
The GCI is obtained from the following:
$${\rm GCI}\, {\equals}\, 1{\minus}\Big( {{{\mathop \sum\nolimits_{j\, {\equals}\, 1}^n Ds_{j} } \mathord{\Big/ {\vphantom {{\mathop \sum\nolimits_{j\, {\equals}\, 1}^n Ds_{j} } {\mathop \sum\nolimits_{j\, {\equals}\, 1}^n Dx_{j} }}} \Big. \kern-\nulldelimiterspace} {\mathop \sum\nolimits_{j\, {\equals}\, 1}^n Dx_{j} }}} \Big),$$
where Dx j is the expected density for secondary consumer species j obtained from the allometric equation in Damuth (Reference Damuth1993). Ds j is the estimated density for the secondary consumer species j obtained from the model. The GCI index shows to what degree the secondary consumers satisfy their requirements according to their population densities in a given environment. GCIs with TAB-m and TAB-M are denoted as GCI-m and GCI-M, respectively.
The GCIB is computed as follows:
$${\rm GCIB}\, {\equals}\, 1\,{\minus}\left( {{{\mathop{\sum}\nolimits_{j\, {\equals}\, 1}^n {Ds_{j} } \, {\times}\, W_{j} } \mathord{\left/ {\vphantom {{\mathop{\sum}\nolimits_{j\, {\equals}\, 1}^n {Ds_{j} } \, {\times}\, W_{j} } {\mathop{\sum}\nolimits_{j\, {\equals}\, 1}^n {Dx_{j} } {\times}W_{j} }}} \right. \kern-\nulldelimiterspace} {\mathop{\sum}\nolimits_{j\, {\equals}\, 1}^n {Dx_{j} } {\times}W_{j} }}} \right),$$
where W j is the body mass of the jth species. The GCIB relates the estimated and expected biomasses of the secondary consumer species and provides information about the type of secondary consumers in different paleoecosystems according to body size. We obtained one GCIB for TAB-m and another for TAB-M (GCIB-m and GCIB-M, respectively). For computational examples of both indexes, please refer to Supplementary Tables 5 and 6.
These indexes provide information on the degree of fulfillment of the secondary consumers’ requirements and, thus, the degree of competition intensity in the ecosystem compared to an ideal condition in which all species would reach optimal densities and maximum population biomass. For both the GCI and GCIB, higher competition among secondary consumers is represented by values close to 1, whereas values close to 0 indicate reduced competition where all species would reach densities near to their maximum population density (see White et al., Reference White, Ernest, Kerkhoff and Enquist2007). Thus, these indexes provide information about the competition intensity in the ecosystem with regard to an ideal condition in which all species reach optimal densities, because the Dx j values are taken as references to evaluate the Ds j values.
To compare competition intensity in different time intervals, we used the nonparametric Mann-Whitney U-test to detect differences in the median value of these two indexes. Similarly, the Mann-Whitney U-test was also used to test for differences in TAB and the TDB between time intervals.
RESULTS
We obtain minimum and maximum TAB for each LFA in the three intervals of this study, the predator requirements in optimal conditions or expected requirements (TDB), and the distribution of TAB among secondary consumers (Table 3). According to our results, on average the TAB was higher in the LFAs from the interval 0.8 –0.5 Ma (interval 2) than it was before or after this period, considering TAB-m or TAB-M (Fig. 3A and B, Table 4). However, these differences are statistically significant only when comparing intervals 2 and 3 (0.5–0.2 Ma) (Table 5).
Boxplot representation of total available biomass (TAB) and total demanded biomass (TDB) for the three intervals of this study (i.e., 1.1–0.8 Ma [interval 1], 0.8–0.5 Ma [interval 2], and 0.5–0.2 Ma [interval 3]) in kcal/km2/yr. (A) Minimum TAB (TAB-m). (B) Maximum TAB (TAB-M). (C) TDB.
A number of different parameters are estimated by the model for from each faunal assemblage. TAB-m, minimum total available biomass; TAB-M, maximum total available biomass; TDB, total demanded biomass; GCI-m, global competition index with minimum TAB; GCI-M, global competition index with maximum TAB; GCIB-m, biomass global competition index for minimum TAB; GCIB-M, biomass global competition index for maximum TAB. Units for TAB-m, TAB-M, and TDB are kcal/km2/yr. GCI-m, GCI-M, GCIB-m, and GCIB-M are parameters without units because their values are parts per unit.
Median values of different parameters (see Table 3) in the three time intervals considered in this study (i.e., 1.1–0.8, 0.8–0.5, and 0.5–0.2 Ma), with all faunal assemblages (Total), for faunal assemblages with human presence (with Homo), and for faunal assemblages without human presence (without Homo). Units for TAB-m, TAB-M, and TDB are kcal/km2/yr. GCI-m, GCI-M, GCIB-m, and GCIB-M are parameters without units because their values are parts per unit.
Exact P values from Mann-Whitney U-test comparing a number of parameters (see Table 3) among the three time intervals of this study (1.1–0.8, 0.8–0.5, and 0.5–0.2 Ma). The time intervals are compared two by two with this statistical test (i.e., column “1 and 2” compares the values of the parameters between the first and the second time intervals. The P values in bold show significant differences: *, P<0.05; **, P<0.01; ***, P<0.005.
A gradual tendency toward a reduction in the requirements of secondary consumers (TDB) from the interval 1.1–0.8 Ma (1,379,501 kcal/km2/yr) to the intervals 0.8–0.5 Ma (862,091 kcal/km2/yr) and 0.5–0.2 Ma (650,748 kcal/km2/yr) is apparent (Table 4, Fig. 3C), although the differences are statistically significant only between intervals 1 and 3 (Table 5). This might be explained by a decrease in the number of secondary consumers and/or by a change in the guild of secondary consumers because of the appearance of species with lower meat requirements. The three intervals show a similar absolute minimum value of TDB of ~400,000 kcal/km2/yr (Fig. 3C). This is likely an artifact of our method because we selected sites with a minimum number of secondary consumer species. This tendency is not observed for the maximum values of TDB; the value of maximum TDB is lower in interval 2 than in the first and third intervals.
Concerning the competition indexes, GCI was significantly higher during 1.1–0.8 Ma than in the two younger intervals (Table 4 and 5), both with TAB-m (Fig. 4A) and with TAB-M (Fig. 4B), in accordance with the results from previous studies (Meloro et al., Reference Meloro, Raia and Barbera2007; Raia et al., Reference Raia, Meloro and Barbera2007; Croitor and Brugal, Reference Croitor and Brugal2010; Palombo, Reference Palombo2010). Differences in GCI between the second and third intervals are not significant (Table 5). Similarly, GCIB was not significantly different comparing the second and third time intervals, but it was significantly higher in the LFAs for 1.1–0.8 Ma (Fig. 4C and D, Table 5). Thus, competition intensity for meat resources between the secondary consumer species was higher in the latter part of the early Pleistocene than in the middle Pleistocene, but it was similar during 0.8–0.5 Ma and 0.5–0.2 Ma. The three intervals include some LFAs with an absolute minimum value of competition intensity (0.0). Intervals 1 and 3 show LFAs with absolute maximum values higher than the second interval. In summary, the same pattern in GCI and GCIB along the three time intervals is observed for TAB-m and for TAB-M, although, as it may be expected, with less competition when resources are more abundant (TAB-M).
Boxplot representation of global competition index (GCI) and global competition index biomass (GCIB) values for the three intervals of this study (i.e., 1.1–0.8 Ma [interval 1], 0.8–0.5 Ma [interval 2], and 0.5–0.2 Ma [interval 3]), in two scenarios: with minimum and maximum total available biomass (TAB) in all assemblages. GCI and GCIB take the value of 0 when there is no competition among secondary consumers because all species reach their requirements. When GCI and GCIB are 1, competition among the secondary consumers species is maximum. (A) GCI-m (GCI with minimum TAB). (B) GCI-M (GCI with maximum TAB). (C) GCIB-m (GCIB with minimum TAB). (D) GCIB-M (GCIB with maximum TAB).
Considering the 36 LFAs selected for our study, Homo was present in 67% of them during 1.1–0.8 Ma, in 43% during 0.8–0.5 Ma, and in 55% during 0.5–0.2 Ma. The presence of Homo in a site does not imply that the assemblage includes hominin fossils; human presence may also be inferred from other evidence such as the existence of lithic tools (see Rodríguez et al., Reference Rodríguez, Mateos, Martín-González and Rodríguez-Gómez2015). We explore the relationship between the presence of Homo in an LFA and the values of the competition indexes GCI and GCIB in the three time intervals in Figures 5 and 6. Both the GCI and GCIB indexes show a wide variation in the three intervals when all sites are considered, although the variation is smaller in the second interval. There are not significant differences in GCI or GCIB among the three intervals considering only the faunal assemblages with human presence. It is apparent, however, that the presence of Homo is not recorded in the LFA with the highest GCI value (Hundsheim, N/A=0.73; Fig. 5). Indexes for LFAs with human presence were similar in the three intervals. There are significant differences in GCI and GCIB between the first and second intervals considering LFAs without Homo with TAB-m and TAB-M, but there are not differences between the second and the third intervals.
Graphic distribution of global competition index (GCI) values in each local faunal assemblage for the three intervals of this study (i.e., 1.1–0.8 Ma [interval 1], 0.8–0.5 Ma [interval 2], and 0.5–0.2 Ma [interval 3]), in two scenarios: with minimum (GCI-m; top) and maximum (GCI-M; bottom) total available biomass in all assemblages. The first column (Total) contains all assemblages, the second column (with Homo) includes assemblages with evidence of the presence of Homo, and the third column (without Homo) includes assemblage lacking this evidence. GCI is equal to 0 when competition intensity is minimum and 1 when it is maximum. Locality codes used are as shown in Table 1.
Graphic distribution of the global competition index biomass (GCIB) values in each local faunal assemblage for the three intervals of this study (i.e., 1.1–0.8 Ma [interval 1], 0.8–0.5 Ma [interval 2], and 0.5–0.2 Ma [interval 3]), in two scenarios: with minimum (GCIB-m; top) and maximum (GCIB-M; bottom) total available biomass in all assemblages. The first column (Total) contains all assemblages, the second column (with Homo) includes assemblages with evidence of the presence of Homo, and the third column (without Homo) includes assemblages lacking this evidence. GCIB is equal to 0 when competition intensity is minimum and 1 when it is maximum. Locality codes used are as shown in Table 1.
DISCUSSION
Our results do not support the hypothesis that competition among secondary consumers was higher between 0.8 and 0.5 Ma (interval 2) than in the previous (1.1–0.8 Ma) and later (0.5–0.2 Ma) intervals. On the one hand, the TAB (TAB-m and TAB-M) was higher in interval 2 than in the other two intervals, and thus, there was more biomass available to secondary consumers. In contrast, we found no significant differences in the trophic requirements of secondary consumers (TDB), which appeared to decrease through time, being highest during 1.1–0.8 Ma. Herbivores are well known to have increased in size during this period, and a significant number of primary consumer species were free, or almost free, of predators (Meloro et al., Reference Meloro, Raia and Barbera2007; Raia et al., Reference Raia, Meloro and Barbera2007; Croitor and Brugal, Reference Croitor and Brugal2010; Meloro and Clauss, Reference Meloro and Clauss2012; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). On the other hand, the competition intensity was at its highest during the first interval and at its lowest in the second interval. The variability in the intensity of competition inside an interval could be a reflection of the diversity of paleocommunity structures, and it is moderately higher in the first than in the other two intervals. The second interval shows the lowest diversity of LFAs, suggesting that it was more homogeneous from an ecological point of view. The competition intensity in the guild of secondary consumers during the latter part of the early Pleistocene and the middle Pleistocene seems to be characteristic of each period. Therefore, each time interval could have its own properties from the point of view of competition and ecological interactions. The pattern of change in TAB is similar to the pattern observed in the predator/prey ratio in previous studies, which rose in the middle Galerian (Meloro et al., Reference Meloro, Raia and Barbera2007; Raia et al., Reference Raia, Meloro and Barbera2007; Croitor and Brugal, Reference Croitor and Brugal2010; Palombo, Reference Palombo2010) (see “Introduction”). Thus, the increased efficiency of the secondary consumer guild during the middle Galerian could be related to an increase in the available biomass (TAB) for this guild. Note that the pattern observed for the predator/prey ratios is not directly comparable with the changes in the GCI and GCIB indexes because of their different conceptualization and nature (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González and Rodríguez2016a).
Considering the three periods together, Homo was present only in LFAs in which the secondary consumers satisfied at least one-third of their requirements (GCI and GCIB are <0.67). Moreover, this threshold (GCI=0.67) is obtained in the middle unit from the Vallparadís site (Spain), but human presence in this LFA is controversial (Martínez et al., Reference Martínez, Garcia, Carbonell, Agustí, Bahain, Blain and Burjachs2010; Garcia et al., Reference Garcia, Landeck, Martínez and Carbonell2013; but see Madurell-Malapeira et al., Reference Madurell-Malapeira, Alba, Minwer-Barakat, Aurell-Garrido and Moyà-Solà2012). If the middle unit from Vallparadís is not considered, the threshold value for human presence would be GCI=0.62, as in Imola (Greece). However, Homo was present in some LFAs during the three time intervals, and our results suggest that there were not significant differences in competition intensity among LFAs with human presence from different time intervals, nor between faunas from the same period with and without hominins (analyses not shown). Therefore, human populations occurred in places under similar conditions during the three periods.
On the other hand, differences in the competition indexes (GCI and GCIB) between sites without human presence from the first versus the second interval were nonsignificant, considering either the TAB-m or the TAB-M, although in the first interval competition intensity was very high and in the second interval it was lower (Table 4, Figs. 5 and 6). Our results show that competition tends to be higher in assemblages from the first interval than in the assemblages from the two younger periods (Figs. 5 and 6). According to these results, Homo was present in ecosystems with a wide range of competition intensity values during the three intervals, but it was not present in conditions of extreme competition intensity. Nevertheless, those extreme conditions have been detected only in two sites, one from the first interval (the latter part of the early Pleistocene Vallparadís middle unit from Spain) and the other from the third interval (the latter part of the middle Pleistocene Hundsheim site from Germany; Table 3). Thus, Homo was able to successfully compete with other secondary consumers in most conditions, exhibiting a similar and remarkable adaptation capacity during the three periods studied. However, humans were not in communities where competition was extremely high. Moreover, the latter part of the middle Pleistocene Hundsheim site was excavated at the beginning of the twentieth century, and it no longer exists (Frank and Rabeder, Reference Frank and Rabeder1998). The probability that this assemblage was affected by a time-averaging phenomenon cannot be ruled out. Hundsheim is a faunal assemblage rich in biodiversity with different kinds of secondary consumer (omnivores, scavengers, and hypercarnivorous like canids, felids, hyenids, and ursids). In our sample, Hundsheim appears as an outlier for its time interval regarding competition intensity. If Hundsheim were removed from the sample, there would not be any middle Pleistocene site with competition intensity comparable to the latter part of the early Pleistocene sites, and this would suggest opposite tendencies for the early and middle Pleistocene LFAs.
In light of the results presented here, the discontinuity in the human occupation of Europe observed by Mosquera et al. (Reference Mosquera, Ollé and Rodríguez2013) on the basis of the scarcity of evidence of occupation and the apparent gap between the Mode 1 and Mode 2 technocomplexes in Europe cannot be explained by high competition inside the paleocommunities from 0.8 to 0.5 Ma. In the previous section, we commented on the relative frequency of human presence in the LFAs selected for this study. Wider analyses with the whole sample of 98 LFAs initially compiled, which includes LFAs that did not meet the selection criteria (see “Methods”), provide a more complete view. Thirty out of the 98 LFAs belong to the first interval, 12 to the second interval, and 56 to the third interval. Evidence of human presence is recorded in this sample in 43% of the LFAs from the first interval, 25% of the LFAs from the second interval, and 34% of the LFAs from the third interval. Taking these data into consideration, the lowest relative frequency of human presence corresponds to interval 2, being 18% and 9% lower than intervals 1 and 2, respectively. These figures show a reduction in the number of sites and in human presence in the early Middle Pleistocene.
If a reduction in human abundance during the middle Galerian is taken for granted, a priori, both the local extinction and/or the demographic decrease of humans might be caused by adverse conditions. From a paleoecological point of view focused on macromammals and resource availability, two scenarios might explain human local extinction and/or demographic decrease by adverse conditions, where climatic changes are considered intrinsically. In the first one, the competition among secondary consumers reached a high level during 0.8–0.5 Ma across Europe. The human populations were unable to address this competition, and they became extinct. However, according to the results presented here, competition intensity was lower during the middle Pleistocene than in the latter part of the early Pleistocene, except for Hundsheim, and this hypothesis is not supported. In an alternative scenario, changes in the configuration of paleocommunities might displace human populations from many faunal assemblages, although they were able to survive in areas where the structure of the mammalian communities allowed them to successfully compete for resources.
With respect to changes in community configuration or food web structure, the limit between the first and second intervals at 0.8 Ma roughly coincides with the Matuyama-Brunhes boundary and the MPR (Maslin and Ridgwell, Reference Maslin and Ridgwell2005). Carrión et al. (Reference Carrión, Rose and Stringer2011) suggest a cyclic replacement of forested landscapes with open landscapes between 0.9 and 0.4 Ma coincident with the changes in climatic periodicity from 41 ka to 100 ka cycles. A faunal turnover also occurred in Europe at this time with the appearance of new carnivores and increased herbivore richness that modified the structure of mammalian communities (Turner, Reference Turner1992; Palombo, Reference Palombo2007, Reference Palombo2010, Reference Palombo2013, Reference Palombo2014; Rodríguez et al., Reference Rodríguez, Rodríguez-Gómez, Martín-González, Goikoetxea and Mateo2012). However, Meloro (Reference Meloro2011a) suggests a similar exploitation of the ungulates by the guild of secondary consumers throughout the entire Plio-Pleistocene in Italy based on the relative stability of the disparity in the shape of the mandibular corpus. Palombo (Reference Palombo2013, Reference Palombo2016) concludes that during this transition, each mammal species varied its distribution and abundance according to its own environmental tolerances and ecological flexibility because the factors that trigged its dispersal varied from one species to another. Palombo (Reference Palombo2013, Reference Palombo2016) suggests that climate change and environmental instability facilitated the settlement of the European ecosystems by new species with wider ecological niches, such as humans. According to the results presented here, the mid-Pleistocene turnover produced conditions more favorable to Homo, from the point of view of meat resources, with lower competition intensity and better access to carcasses than in the former period. Surprisingly, these improved environmental conditions did not coincide with an increased presence of Homo. Human populations only increased and expanded much later, at ~0.5 Ma. It is also conceivable that other aspects of the human-fauna interactions not included in our analyses limited the viability of human population in this period.
A high biodiversity, with many complex ecological interactions, induces speciation, reduces extinction rates, and may facilitate the support of biodiversity in an ecosystem (Bascompte et al., Reference Bascompte, Jordano and Olesen2006; Ricklefs, Reference Ricklefs2010). The number of secondary consumer species was higher in the latter part of the early Pleistocene than in the middle Pleistocene, and perhaps this higher biodiversity induced specialization, made paleocommunities more stable, and allowed paleocommunities to support higher levels of competition intensity. In contrast, the faunal turnover associated with the MPR induced the breakup of the existing relationships and perhaps made it more difficult for secondary consumers to enter and survive inside a Galerian paleocommunity. Moreover, the Galerian secondary consumers were more generalist than the Epivilllafranchian species (Croitor and Brugal, Reference Croitor and Brugal2010), making their niches more prone to overlap and promoting territorial competition and spatial exclusion, as has been observed for recent cheetahs and wild dogs (Laurenson et al., Reference Laurenson, Wielebnowski and Caro1995; Mills and Gorman, Reference Mills and Gorman1997). In any case, human populations overcame these difficulties between 0.5 and 0.2 Ma when they increased in numbers. Niche overlap may be analyzed by looking at the species that co-occur with Homo in the LFAs. Only six species did not show co-occurrences with Homo at any LFAs, but if analyzed at the genus scale, Homo coincided with all genera (see Supplementary Table 2) in at least one LFA. For this reason, we assume that the human niche did not overlap completely with the niche of any other genus. The flexibility and diversity of the omnivorous human diet and the lack of exclusion with other secondary consumers hampers a better understanding of the role played by humans in these paleoecosystems.
There are not enough middle Pleistocene LFAs with high competition intensity to discuss in detail the relevance of the improvements in lithic technology in this period (Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011; Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013). However, our results show that, considering ecological competition intensity, the Acheulean technology spread in Europe when environmental conditions were more favorable for secondary consumers (reduced competition). Moreover, human subsistence strategies were successful with both Oldowan and Acheulean technologies (Blasco et al., Reference Blasco, Rosell, Fernández Peris, Arsuaga, Bermúdez de Castro and Carbonell2013; Huguet et al., Reference Huguet, Saladié, Cáceres, Díez, Rosell, Bennàsar and Blasco2013), but the latter coincided with the demographic increase and the range expansion (Roebroeks, Reference Roebroeks2001). Perhaps these technological improvements allowed humans to attain a higher performance with the same resources. Kahlke et al. (Reference Kahlke, García, Kostopoulos, Lacombat, Lister, Mazza, Spassov and Titov2011) suggest that improvements in hunting, gathering, food-processing techniques, and other cognitive capacities supplied opportunities for subsistence and dispersal during latter part of the early to early middle Pleistocene interglacials. They suggest that technological innovation occurred along with unfavorable conditions, high seasonality, and low levels of habitat variability because in stable environmental conditions hominins could rely on traditional subsistence strategies rather than develop technological innovations, as Moncel (Reference Moncel2010) also defends. Our results show more favorable conditions for secondary consumers during 0.8–0.5 Ma than 1.1–0.8 Ma. The earliest European Mode 2 appears in Barranc de la Boella (Spain) at approximately 1.0 Ma (Vallverdú et al., Reference Vallverdú, Saladié, Rosas, Huguet, Cáceres, Mosquera and Garcia-Tabernero2014), but Mode 1 remained in Europe at least until around 0.61 Ma at Isernia La Pineta (Italy) (Coltorti et al., Reference Coltorti, Feraud, Marzoli, Peretto, Ton-That, Voinchet, Bahain, Minelli and Thun-Hohenstein2005; Peretto, Reference Peretto2006; Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013). Under the premise that hostile environments promote technological innovation, and assuming that the European Acheulean could be a local innovation, it would be expected to appear during a period of high intraguild competition. Interestingly, intraguild competition was high during the latter part of the early Pleistocene, when the late Mode 1 or early Mode 2 industries appeared (Scott and Gibert, Reference Scott and Gibert2009; Barsky and de Lumley, Reference Barsky and de Lumley2010; Santonja and Pérez-González, Reference Santonja and Pérez-González2010; Barsky et al., Reference Barsky, Garcia, Martínez, Sala, Zaidner, Carbonell and Toro-Moyano2013; Moncel et al., Reference Moncel, Despriée, Voinchet, Tissoux, Moreno, Bahain, Courcimault and Falguères2013; Mosquera et al., Reference Mosquera, Ollé and Rodríguez2013; Walker et al., Reference Walker, López-Martínez, Carrión-García, Rodríguez-Estrella, San-Nicolás del-Toro, Schwenninger and López-Jiménez2013; Vallverdú et al., Reference Vallverdú, Saladié, Rosas, Huguet, Cáceres, Mosquera and Garcia-Tabernero2014). If the human occupation of Europe was continuous from 1.0 to 0.5 Ma, these pre–Mode 2 technologies would be progressively developed into a full Mode 2 from 0.8 to 0.5 Ma, a period of moderate intraguild competition. The lithic complex from Barranc de la Boella might be a good candidate to represent an initial stage in the progressive development of the European Mode 2, although Mosquera et al. (Reference Mosquera, Ollé, Saladié, Cáceres, Huguet, Rosas and Villalaín2016) suggest that it was actually a dead line, based on the apparent archaeological gap that occurred in Europe between 0.9 and 0.6 Ma. According to that interpretation, those first stages of the European Acheulean would be unsuccessful in providing significant survival advantages to those populations, and only the appearance of the full Acheulean around 0.5 Ma would represent a real step forward. Otherwise, if the Acheulean evolved locally during the Middle Pleistocene, it was not triggered by harsh environmental conditions, at least with reference to intraguild competition. Finally, if the Acheulean arrived in Europe from overseas, as claimed by several scholars (e.g., Carbonell et al., Reference Carbonell, Mosquera, Rodríguez, Sala and Made1999, Reference Carbonell, Sala Ramos, Rodríguez, Mosquera, Ollé, Vergès, Martínez-Navarro and Bermúdez de Castro2010; Santonja and Pérez-González, Reference Santonja and Pérez-González2010; Jiménez-Arenas et al., Reference Jiménez-Arenas, Santonja, Botella and Palmqvist2011), its appearance had nothing to do with the variations in intraguild competition. As an example, Carbonell et al. (Reference Carbonell, Sala Ramos, Rodríguez, Mosquera, Ollé, Vergès, Martínez-Navarro and Bermúdez de Castro2010) defend that the expansion of the Acheulean across Europe was because of a demographic increase in the Near East and Africa. To date, current knowledge of the European bifacial technology does not clarify the arguments in favor of a phenomenon from an African origin or in favor of a local substratum (Moncel et al., Reference Moncel, Arzarello, Boëda, Bonilauri, Chevrier, Gaillard, Forestier, Yinghua, Sémah and Zeitoun2016b). The debate is still open. In any case, it is beyond the scope of this article to support any of the three possible scenarios; our intention here is only to provide an environmental background to them.
The results in the present study are in disagreement with those presented in Rodríguez-Gómez et al. (Reference Rodríguez-Gómez, Mateos, Martín-González, Blasco, Rosell and Rodríguez2014b). Applying the same methodology, it was shown that at the local scale, the depopulation of the Sierra de Atapuerca area by ~0.6 Ma coincided with a relatively high intraguild competition (Rodríguez-Gómez et al., Reference Rodríguez-Gómez, Mateos, Martín-González, Blasco, Rosell and Rodríguez2014b). Studies at a local scale may provide information about the evolution of faunal assemblages in the long term, but analyses at the continental scale focus on average paleocommunities. An average is a simplification, so there is a loss of information when it is applied to obtain a global perspective. From the continental point of view, competition does not appear to be a relevant factor to limit human presence in the early middle Pleistocene. Combined analyses are desirable to help track the changes in intraguild competition both at the continental and local scales, although unfortunately, there are few local sequences like Atapuerca that are long enough to apply this approach.
This study addresses the constraints imposed by the communities of large mammals to the human presence in the European Pleistocene from a paleoecological point of view. We aimed to describe with precision the role of humans in the paleocommunities of the studied period. However, to understand human behavior during the MPR, it is not enough to know the limitations imposed on them by the availability of trophic resources. To obtain a more accurate view of the human niche in this period, it will be necessary to improve our knowledge of the interactions between species and/or food web structure in the paleocommunities.
CONCLUSION
Measurement of the intensity of intraguild competition as a limiting factor in human expansion across Europe tests whether the apparently scarce human presence at the beginning of the middle Pleistocene could be attributable to this factor. Our results show that competition intensity was higher in the latter part of the early Pleistocene LFAs than in the middle Pleistocene LFAs. This decrease in the intensity of competition inside the guild of secondary consumers parallels a decrease in human presence across Europe; the more abundant evidence of human presence is observed in our sample for the latter part of the early Pleistocene, and it is particularly scarce for the eraly part of the middle Pleistocene. Surprisingly, this decrease in human presence coincided, at the European scale, with a period of more favorable conditions from the point of view of competition for meat resources. Thus, it could be concluded that competition with carnivores was not the main cause of the apparent crisis suffered by human populations at the beginning of the middle Pleistocene, although the ecological interactions and the intensity of competition show a marked variation from the latter part of the early Pleistocene to the latter part of the middle Pleistocene. Humans were not present where competition intensities were the highest but showed a great adaptation capacity, being present in paleoecosystems with widely different levels of intraguild competition intensities. Remarkably, the niche of Homo in the early Pleistocene paleoecosystems did not completely overlap with that of any other genus, because the complete exclusion between Homo and another genus of secondary consumer was not observed. The analyses presented here, focused at a continental scale, suggest different conclusions than the analyses carried out at a local scale. Thus, future research should combine both scales of analyses to obtain a more comprehensive view of what seems to be a complex pattern. Moreover, a better knowledge of the role of Homo in the Pleistocene ecosystems could be obtained by delving into the structure of the paleocommunities and the relationships among the species included in them.
ACKNOWLEDGMENTS
We thank Professor Adrian M. Lister and the Natural History Museum of London for providing permitted GRG access to the Pleistocene macromammal material. The short stay to revise this material was funded with an Estancias Breves Grant (Ref. EEBB-I-12-03857) by the Spanish Ministerio de Economía, Industria y Competitividad (MINECO). This research was funded by the MINECO project CGL2012-38434-C03-02. GRG was the beneficiary of a predoctoral Formación de Personal Investigador Grant from the Spanish Ministerio de Ciencia e Innovación (Ref. BES-2010-033410). An anonymous editor from Elsevier’s Language Services improved the English of the original manuscript. GRG acknowledges the suggestions provided by Laura Martín-Francés, Estefanía Pérez-Fernández, and Mathieu Duval. This research was carried out in the framework of the International Union for Quaternary Science project 1403 “Modeling human settlement, fauna and flora dynamics in Europe during the Mid-Pleistocene Revolution (1.2 to 0.4 Ma).” Finally, we acknowledge the suggestions and constructive remarks provided by editor Lewis Owen and two anonymous reviewers.
Supplementary materials
To view supplementary material for this article, please visit https://doi.org/10.1017/qua.2017.20
