“….if true that the juxtapositions of ideas and data from largely independent studies can inspire new hypotheses, then mammalogy is surely a marvelous substrate for discoveries within the evolutionary paradigm. In total range and strength, mammalian research is unrivalled in biology.” (Vrba, 1992, p. 2)

The multifaceted and intriguing evolutionary history of mammals, which led to their current biodiversity and biogeographical setting, is tightly linked with palaeogeographic, climatic and environmental changes. The complex synergistic action of physical and biological factors shaped faunal evolution, species originations and extinctions, and the timing and mode of species dispersal through time and across continents.

The ecological and evolutionary responses of mammalian faunas to stimuli perturbing the internal dynamic equilibrium of palaeocommunities varies with the temporal and spatial scales at which such factors acted. Climate changes have been considered by some scholars as the most influencing causal factor in triggering species evolution and faunal diversity (FDy) at ecological and geological temporal scale, hypothesising that evolutionary change (e.g. speciation events) may be less frequent in times of ecosystem stability rather than during phases of variability triggered by physical environmental events (e.g. Alroy et al., 2000, Barnosky, 2001, Barnosky, 2005, Bennett, 1997, Berteaux et al., 2006, Eldredge, 1999, Gienapp et al., 2008, Hua and Wiens, 2013, Jansson and Dynesius, 2002, Smith, 2012, Vrba, 1992, Vrba, 1995a, Vrba, 2005, Webb, 1995 and Webb and Barnosky, 1989).

Some scholars believed extrinsic factors to have a minimal effect on species evolution, and that intrinsic biological factors must be the most important. Accordingly, changes in ecosystem equilibria and faunal turnovers may be due to the internal dynamics of competitive relationships, without necessarily indicating a strict interdependence between major climatic changes and evolutionary events (e.g. Jaeger and Hartenberger, 1989, Prothero, 1999, Prothero, 2004 and Tsubamoto et al., 2004). The statement may be actually true especially as regards to macroevolutionary events at a high taxonomic level in deep time, but some evidence has demonstrated that climatic changes related to geological/Quaternary time may have profound effects on patterns of mammalian FDy even at a continental scale (e.g. Blois and Hadly, 2009, Figueirido et al., 2012, Gingerich, 2003, Gingerich, 2006, Saarinen et al., 2014 and Webb et al., 1995 Woodburne et al., 2009). During times of environmental normality, indeed, interactions among ecosystem species evolve in a regime of dynamic equilibrium. Perturbations of physical parameters produce disequilibrium by causing the extinction of the most specialized species. The resulting unbalanced structure of the community stimulates new individual responses of other species, de facto causing ecosystems to significantly restructure (Graham and Lundelius, 1984).

Climate and biotic interactions, however, likely contribute to faunal evolution in successive phases: climate changes and physical-environmental disturbances, altering the ecosystem structure and functioning, acted as a trigger factor in promoting functional and taxonomic turnovers, while internal dynamics of competitive relationships constrained the faunal reorganization leading to a new equilibrium (e.g. Faith and Behrensmeyer, 2013, Palombo, 2007 and Palombo, 2014). Some researchers, conversely, assumed that although global climate changes may have influenced longer-term evolutionary trends, species interactions and local environmental changes were the most influential factors at shorter time scales (Bibi and Kiessling, 2015).

The biotic response of individuals, species and communities to climatic warming and cooling events is, however, a highly complex phenomenon, making it sometimes difficult to accurately disentangle causal factors behind faunal dynamics. Some organisms, for instance, seem to have persisted over thousands to perhaps millions of years in the face of environmental perturbations and species invasions; others evolved, and others changed their range. Bioevents related to secular dispersal were particularly important during phases of marked climatic and environmental instability, e.g., during the Quaternary period, which represents a crucial transition between geological and ecological time. Any ecosystem, indeed, provides finite resources that species have to partition, optimizing the expenditure cost by reducing competition and achieving a dynamic equilibrium during periods of environmental instability. As a result, the extent of FDy mainly depends on resource shifts, in turn related to changes in physical parameters (e.g., temperature, humidity).

During the Pleistocene, particularly during the mid-Pleistocene Revolution (MPR) (Clark et al., 2006, Maslin and Brierley, 2015 and Maslin and Ridgwell, 2005), climate forcing induced deep but gradual alterations and latitudinal displacements in terrestrial biomes and greatly influenced dispersal and dispersion of species. A number of species reacted to ecosystem disturbances by varying the geographic range in keeping with the displacement of biomes, acting as invasive species in the new territory they were dispersing across. Each species changed its range depending on the suitability of environmental conditions with respect to its own environmental tolerances and ecological flexibility. Such evidence is partially consistent with the Habitat Theory (Vrba, 1992), which hypothesizes that low latitude and higher latitude mammals move respectively toward higher and lower latitudes when climates become warmer and cooler, tracking the changing distributions of vegetation zones.

Whatever the extent and trajectory of dispersals should have been, discrete dispersal bioevents, merging alien species into previously existing large mammal communities, changed the faunal structure, giving rise to new internal dynamics that led to a progressive reconstruction of the mammalian fauna, although different processes predominated in different places. Dispersal hence is one among the fundamental processes in biogeography (crucial for understanding the evolutionary dynamics of organism distribution throughout time and across space) and a factor to carefully consider in analysing faunal dynamics, making correlations among distant stratigraphic sequences, and chronologically ordering faunal assemblages.

The Pleistocene fossil record of the Mediterranean region is particularly suitable for studying effects of climate change and comparing FDy at local and regional levels due to the complex physiographic and climatic heterogeneity of the region, the presence of important geographical/ecological barriers, and the long history of invasions of species of varying geographical origin and provenance.

This research aims to highlight the main aspect of FDy in selected North Mediterranean territories (Iberian Peninsula, France, Italy and Greece), each showing current peculiar physiographic configuration and climate regime. The geological evolution of the region during the early Middle Pleistocene suggests that the main differences from one territory to another have roughly been maintained throughout the Pleistocene, differently influencing the time of dispersal and distribution patterns of taxa. The mammal FDy (shifts in biodiversity, change in taxonomic composition and ecological structure) of each territory is analyzed to verify whether or not the major modifications match climatic and environmental changes. Biogeographic insights and chronological issues are briefly discussed in the light of diachronous/asynchronous dispersal events.

The database consists of taxonomically revised lists of large mammal species from selected Western (Iberian Peninsula, France and Italy) and Eastern (Greece) Mediterranean local faunal assemblages (LFAs) (a list of the species identified from the same stratigraphical horizon at a given fossiliferous site) ranging in age from about 2.6 to 0.62 Ma (from about MIS 104 to MIS 16, from the middle Villafranchian to “Epivillafranchian”/early Galerian European Land Mammal Ages, ELMAs).

Three main aspects were considered here for exploring the FDy in each studied territory: variations of diversity/richness (Estimated mean Standing Diversity, emSD, and Standing Richness, SR), changes in the taxonomic composition (Turnover index, TI; origination, OR, and extinction, ER, rates) helpful to reveal the three major components of FDy (i.e. evolution, extinction, dispersal), and modifications of the ecological structure of Faunal Complexes, FCs.

The sampling adequacy was estimated by means of two completeness indices (CI, CI_bda), based on the proportion of Lazarus taxa (or range-through taxa, RT).

Details about material, data sources, and methods are provided in Supplementary Information.

Trends in TI (Fig. 1) and in emSD and SR (Fig. 2) basically result from the same combined effect of appearances of new taxa (newcomers dispersing into the focal territory, and taxa originated by anagenetic/sympatric evolution in situ within phyletic lineages already inhabiting the area) and local disappearances/extinctions. Therefore patterns are expected to be similar throughout time, although they may vary from one territory to another, and at regional and local geographical scales.

Turnover trends in the West Mediterranean region, for instance, roughly mirror only the trend shown by the fauna from the Iberian Peninsula. Both show a high value of the three indices (global, g-TI, per in-loco-origination, o-TI, and per dispersal, d-TI, turnover indices) during nearly the entire Early Pleistocene (with a maximum during the post-Olduvai/pre-Jaramillo period), even if the value d-TI always exceeded that of o-TI. In the Iberian Peninsula, however, this aspect was more evident during the Gelasian.

The asynchrony/diachrony of local appearances due to dispersal (cf. Palombo, 2014 and Palombo, 2017) caused turnovers to differ from one territory to another in extent and significance. By the end of the Early Pleistocene, for instance, g-TI and d-TI turnover indices reached the maximum value in Italy, while the low amount of dispersal events accounts for the low increase and the slight decrease of the g-TI value during the post-Olduvai/pre-Jaramillo period in France and Italy respectively. Divergences among trends also depend on disparities in the amount of dispersal events, and some asynchrony in the last local historical appearances (LlHA). The assumption is supported by some differences in ER, in particular during the Gelasian and during the post-Jaramillo Early Pleistocene. In the West Mediterranean region, Italy shows the most accentuated departures, with a nearly constant ER during the Early Pleistocene. The relatively low ER value during the post-Jaramillo Early Pleistocene is consistent with a longer duration of some Villafranchian species in Italy than in the Iberian Peninsula and France. OR, conversely, shows only minor divergences.

Dissimilarities increase comparing the trend of TI, OR and ER shown by territories of the West Mediterranean region with those of eastern Mediterranean, e.g. Greece. The Greek fauna shows pronounced fluctuations: renewal phases alternated with periods of relative stability. TI were high in the Gelasian and during the late early Pleistocene, while in the post-Olduvai/pre-Jaramillo period changes in taxonomic composition were moderate in spite of the First local Historical appearances due to dispersal (d-FlHA) of some species (Fig. 1). The taxonomic changes were mainly related to extinctions and, to some extent, to in situ originations as supported by the ER trend, showing peaks in ER rates during the Gelasian and in the latest Early Pleistocene (the ER value are much higher than in the western Mediterranean region). During the transition to the Middle Pleistocene, ER was definitely lower and originations prevailed (Fig. 2). As a result, the α-diversity (emSD and SR) decreased on average throughout the Early Pleistocene (Fig. 2). Conversely, in the West Mediterranean diversity increased during the Gelasian and the post-Jaramillo Early Pleistocene, showing no or moderate fluctuations from about 1.6 to 1.1 Ma. The West Mediterranean trend mainly related to the emSd and SR decreases in the French and Italian FCs, while the drop in diversity in the Italian FCs in the latest Early Pleistocene was balanced by the increase in the French ones (Fig. 2).

An increase/decrease in diversity does not necessarily entail an analogous increase/decrease in the biomass (BM), which is more related to the average body mass and density of the species present in a given FC rather than to their number. In the western Mediterranean region, for instance, an increase of BM occurred during most of the Early Pleistocene, peaking during the post-Olduvai/pre-Jaramillo period (as the BM of fauna from the Iberian Peninsula and Greece did), but successively dwindled, notwithstanding the opposite trend shown by the French and Italian faunas. According to the results obtained, neither trends in diversity and BM, nor trends at local and regional geographical scales are consistent with each other. Regarding the BM, for instance, a moderate decrease occurred in Italy during the Early Pleistocene followed by a noticeable increase during the early middle Pleistocene. Conversely, at the western and eastern edges of the studied area, faunas from the Iberian Peninsula and Greece show some increase in the large mammal BM since Olduvai time, a quasi-stationary phase in the latest Early Pleistocene followed by a marked decrease after the transition to the middle Pleistocene. Despite such disparities, the differences in the total amount of BM were not so important during the late Gelasian and the successive Early Pleistocene, except for the French fauna, which shows a low value throughout the studied period, with a minimum during the post-Olduvai/pre-Jaramillo time and a moderate increase in the following Early Pleistocene (Fig. 2).

The differences in the BM trends would suggest some differences in the ecological structure of FCs, and in turn in the habitats, at a local geographical scale across the studied region. The hypothesis, however, is hardly supported by the somehow similar BM trends of the ecological groups related to the vegetational cover (Fig. 3). The BM of forest dwellers, for instance, decreased during most of the Early Pleistocene in all the studied territories, although at different degrees and paces, reaching the minimum value during the post-Olduvai/pre-Jaramillo period, as it occurred to the BM of more ecologically flexible large mammals, and on average of browsers. During this time span, the BM of taxa inhabiting open environments increased, as that of grazers did in the West Mediterranean area, except in Greece. The fluctuation of mixed-feeder BM differs at local geographical scale, although since Olduvai time analogous trends characterised France and Italy. These trends are opposed to that shown by both the Iberian Peninsula and Greece, suggesting one again some environmental similarities between these two distant territories.

All in all, the increased difference in the trends of BM shown by ecological groups related to vegetational cover and to the dietary behaviour of primary and non-carnivorous secondary consumers in the latest early and the beginning of the Middle Pleistocene (Fig. 3) suggests an increasing environmental heterogeneity across the North Mediterranean region in a period of climate instability, roughly coinciding with the most pronounced phases of the MPR. In that time, temperature and humidity on average decreased and grassland and savannah-like environments spread especially in Spain, Greece and southern Italy (Magri and Palombo, 2013 and references therein).

The presence of carnivores in a given territory, and consequently the diversity of the carnivore guild, has generally been regarded as affected, on average, more by the resources (prey) availability/abundance and food-web functioning rather than by abiotic (e.g. climate) factors, although some complex, rarely direct relationships exist between environmental changes and shifts in the diversity of Carnivora (i.e. forest dweller species and the most specialized ones) (see among several others Carbone and Gittleman, 2002, Carbone et al., 2011, Carbone et al., 2007, Chesson, 2000, Palombo et al., 2009 and Pettorelli et al., 2011). The trends shown by the BM of carnivores in ecological groups related to their feeding behaviour (Fig. 4) suggest that the major local changes, as well as the dissimilarities from a region to another, mainly relate to the appearance/disappearance (sometimes diachronous, sometimes asynchronous) of selected species (cf. Palombo, 2017 and references therein). The pick in BM of French hypercarnivorous species in the post-Olduvai/Pre-Jaramillo Early Pleistocene depends, for instance, on the late appearance in this area of large felids (e.g. Panthera gombaszoegensis). During the same period, a nearly contemporaneous decrease of bone-cracker BM occurred the Mediterranean region, which was likely due to the disappearance of Pliocrocuta perrieri. In the following Early Pleistocene, the contemporaneous presence of Crocuta in the North of the Iberian Peninsula and P. brevirostris in the South, accounts for the increase of the bone-cracker BM.

The predator-prey interaction is an important component of FDy and evolution of large mammal community structure. The diversity of predators and prey within the system, the relative proportion of body size groups of both prey and the predators, the availability of the most suitable prey for which predators are competing, and the presence of alternative prey are among the factors that mostly influence the predator-prey interaction (e.g. Cohen, 1977, Croft, 2006, Sinclair et al., 2003 and Van Valkenburgh and Janis, 1993).

The BM trends of predators and common prey (i.e. carnivores and primary and secondary consumers whose interaction is within the common rule), suggest the tendency of a progressive reduction of the predation pressure during the whole Early Pleistocene in the North Mediterranean region, except for the Iberian Peninsula where the BM of predators increased in the latest Early Pleistocene more than that of prey. As underlined above, at this time, the spotted hyaena is recorded for the first time in northern Spain (TD4W level, Atapuerca) (Garcia and Arsuaga, 2001), while P. brevirostris was still present in Southern Spain (Vallparadís EVT7, Catalonia) (Madurell-Malapeira et al., 2010). The ecological traits of the Pleistocene European Crocuta suggest that the species cannot have been successful in competing with the short-faced hyaena. Therefore, the BM of Spanish predators in this period may be regarded as overestimated if Pachycrocuta had already disappeared from the Cantabrian territory by the latest Early Pleistocene, and the two species never coexisted in any Spanish palaeocommunity.

All things considered, the analysis of the FDy in the North Mediterranean mammalian FCs and the comparison among trends in taxonomical turnovers, biodiversity and changes in ecological structure indicate significant modifications at regional geographical scale of the mammalian fauna during the Early Pleistocene, particularly during the MPR. The faunal change, however, developed at different pace and with different timings and modes at local geographical scale, underlining the important role played by the dynamics of single dispersal events in changing the fauna composition and structure in each studied territory. The peculiarity of FDy at local geographical scale mainly depends on palaeogeographical factors that may have prevent or slowed the appearances of some taxa in some territories, the different biological response of species to the local manifestation of the effect of global climate changes, and the resilience of local mammalian palaeocommunities in face of the arrival of invasive species and environmental perturbations. In addition, the nature of the data may have contributed to differentiate the FDy trends during time and across space. Large mammal LFAs, indeed, mostly come from deposits sparse in space and time. Therefore, a low number of LFAs included in some FCs and chronological insufficient controls may result in large uncertainties in the age of the appearance and disappearance of taxa and detection of geographical patterns and diachrony/asynchrony of biological events. The hypothesis that the differences at local geographical scale may depend on the disproportion among fossil records finds, however, only a little support in the quality of fossil record as estimated by means of the completeness indices (Fig. 5). The quality of the fossil data is crucial for validating results and conclusions, because the fluctuating frequencies of coexisting species may depend simply on sampling bias. In the analysed sample, the value of the completeness index CI is consistently high through the Early Pleistocene, except the Greek earliest Pleistocene fauna, as expected due to the dramatic paucity of LFAs known at that time. The value of the more conservative CI_bda index is generally higher than the minimum value (> 70) regarded as suitable for accepting estimates of originations, extinctions, turnover, and diversity/richness (Maas et al., 1995). In two cases (Italy and France in the late Gelasian and the post-Olduvai/pre-Jaramillo time respectively) the value is slightly lower, but close to the minimum. The low values of the CI_bda index shown at the beginning of the Middle Pleistocene would be expected due to the relatively large ratio of LlHA/FlHA with respect to the number of Early Pleistocene taxa that were present even in the following early Middle Pleistocene.

During the late Early Pleistocene, the climate forcing known as mid-Pleistocene Revolution (MPR) induced deep, more or less gradual alterations and latitudinal displacements in European terrestrial biomes and exerted great influence on dispersal and dispersion of mammalian species. The discrete dispersal bioevents that progressively contributed to the changes in taxonomic composition and ecological structure of the large mammal fauna of North Mediterranean mainly concern species of Palearctic origin, especially bovids and cervids, and some Nearctic groups, such as equids of the genus Equus. The arrival of a few species of African origin (baboon, elephants, hippos, and large carnivores) is comparatively rare and sometimes contentious. The Levantine Corridor has generally been accepted as the preferential route of dispersals from Africa. The region possibly acted as a sort of a “filter” corridor because a few African large mammals recorded in the Levant never dispersed further into the Eurasian continent.

Mammals did not generally move in multi-species waves of dispersal; rather each species changed its range depending on the suitability of environmental conditions with respect to its own environmental tolerances and ecological flexibility. As a result, the remodelling of the geographic range of a taxon, and the time and mode of its dispersal and diffusion into SW Europe, differed from species to species as well as from one territory to another, leading to diachronicity/asynchronicity in FlHA/LlHA. This and other factors (e.g. heterogeneous consistency of the fossil record in space and time, weak or unavailable chronological constraints of LFAs, exclusive presence of some species in confined areas, confusing taxonomic treatment of some taxa, and effects of depositional context, preservation and taphonomical biases in determining the chronological reliability of FlHA/LlHA and patterns of presence/absence especially of rare species) make it difficult to order LFAs in correct chronological sequences and gathered them in short-time chronological units.