Figure 3 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Local extinction rate ( e ) affects total cultural diversity as measured by Shannon’s H 9 under equilibrium (A) and non- equilibrium (B) conditions. Each data point provides the mean 6 1 standard deviation of 20 unique simulated populations. Black symbols represent data collected from populations with unbiased cultural transmission and white symbols data collected from populations with conformist cultural transmission. doi:10.1371/journal.pone.0015582.g003
Source publication
The persistence of early stone tool technologies has puzzled archaeologists for decades. Cognitively based explanations, which presume either lack of ability to innovate or extreme conformism, do not account for the totality of the empirical patterns. Following recent research, this study explores the effects of demographic factors on rates of cult...
Context in source publication
Context 1
... is straightforward: Note that F ST makes little sense when H T = 0. We found this to be the case for many of the simulated populations characterized by a relatively low copying error rate ( m # 0.0001) and a relatively high frequency of local extinctions ( e $ 0.01). For this reason, we present the F results of the two higher copying error rates only. To investigate the consequences of local extinctions on the rate of cumulative cultural change for a given m , we compare the number of copying errors that the metapopulation accumulates under different values of e . Recall that every non-equilibrium simulation is initialized with a homogeneous metapopulation in which all individuals display the value ‘‘0’’ as their cultural variant. Also recall that copying errors can only increase or decrease the value of a cultural variant by 1 (i.e., a copying error cannot result in a value of ‘‘5’’ unless the target value was either ‘‘4’’ or ‘‘6’’), and that we hold the number of cultural transmission events constant. Thus, a larger accumulation of copying errors in a metapopulation signifies a faster rate of cumulative culture change per transmission event. Let us consider the dynamics of neutral culture change in a structured population with unbiased cultural transmission and a bidirectional model of innovation. In the absence of local extinctions ( e = 0), the metapopulation accumulates copying errors at a rate proportional to m . The actual rate of cumulative culture change is less than m because of ‘‘back innovations’’ and the fact that many variants are lost to the sampling effects associated with randomly choosing teachers within groups. Stochastic local extinctions may also serve to remove cultural variants from the metapopulation. While it is apparent that local extinctions may further inhibit the accumulation of copying errors in a structured population, understanding the magnitude of this effect requires systematic investigation. The dynamics of cumulative culture change in a population with conformist transmission are quite different. In this case, the variants introduced via copying errors are lost immediately so long as copying errors are not commonplace ( m , 0.5). This is not because of drift, but because the frequency-dependent mechanism of conformist cultural transmission actively selects against all non- modal variants. Local extinctions are unlikely to affect the rate of cumulative change in the presence of conformist transmission because drift is weak relative to the bias introduced by copying the most common variant in the group. The number of copying errors that accumulate in any group during the course of a non-equilibrium simulation can be assessed by the absolute value of the group’s modal cultural variant. Because all individuals display a cultural variant of ‘‘0’’ at the start of each non-equilibrium simulation, we refer to the value of a group’s modal cultural variant as its distance from ancestral ( d A ). We use the absolute value of the group’s modal variant rather than the variant with the maximum absolute value in order to conform to the normative way archaeologists most often perceive and describe the material record. Perceptions of variation within and among Paleolithic assemblages are strongly biased in favor of the most common or most ‘‘important’’ technological variants. Rare artifact classes or unique technological procedures may be systematically reported but are seldom accounted for in large-scale syntheses and regional comparisons. The maximum rate of cumulative change per cultural transmission event in a structured metapopulation is represented by the maximum | d A | value found among its subpopulations ( d A max ). Note that d A max does not provide a suitable proxy for the rate of cumulative cultural change if metapopulations are initialized with maximum heterogeneity. The results of the model show clearly that rates of local extinction could influence total diversity, group differentiation, and rates of long-term cumulative change. These findings are summarized below. One of the simplest measures of total diversity is richness (Figure 2). As one would expect, richness increases with m , although the magnitude of this effect decreases as e increases. More importantly, e has a significant effect on richness for all values of m tested under equilibrium (Kruskal-Wallis H-test results: m = 0.00001: x 2 = 72.54, P , 0.001; m = 0.0001: x 2 = 73.26, P , 0.001; m = 0.001: x 2 = 74.06, P , 0.001; m = 0.01: x 2 = 74.12, P , 0.001) and non-equilibrium ( m = 0.00001: x 2 = 67.87, P , 0.001; m = 0.0001: x 2 = 69.82, P , 0.001; m = 0.001: x 2 = 72.29, P , 0.001; m = 0.01: x 2 = 73.91, P , 0.001) conditions. Richness decreases as e increases. Our results also suggest that the combination of unbiased transmission and frequent local extinctions can maintain a similar number of unique cultural variants as conformist biased cultural transmission in the absence of local extinctions (Figure 2b). Figure 3 summarizes the H 9 results of our experiments. First, note that H 9 increases with m . Second, e has a significant effect on H 9 under equilibrium ( m = 0.00001: x = 71.96, P , 0.001; m = 0.0001: x 2 = 73.80, P , 0.001; m = 0.001: x 2 = 73.56, P , 0.001; m = 0.01: x 2 = 74.07, P , 0.001) and non-equilibrium ( m = 0.00001: x 2 = 66.53, P , 0.001; m = 0.0001: x 2 = 70.76, P , 0.001; m = 0.001: x 2 = 73.06, P , 0.001; m = 0.01: x 2 = 73.64, P , 0.001) conditions. Holding m constant, frequent local extinction ( e = 0.1) yields metapopulations that display substantially lower H 9 than cases where there is no local extinction ( e = 0). Conformist cultural transmission also yields low values of H 9 , even when there is no local extinction (Figure 3b). Although H T accounts for population structure while richness and H 9 do not, all three measures provide similar pictures of how copying errors and local extinctions affect total cultural diversity. H increases with m , and e has a significant effect on H under equilibrium ( m = 0.00001: x = 71.68, P , 0.001; m = 0.0001: x 2 = 73.80, P , 0.001; m = 0.001: x 2 = 73.35, P , 0.001; m = 0.01: x 2 = 74.07, P , 0.001) and non-equilibrium ( m = 0.00001: x 2 = 64.62, P , 0.001; m = 0.0001: x 2 = 69.84, P , 0.001; m = 0.001: x 2 = 72.63, P , 0.001; m = 0.01: x 2 = 73.93, P , 0.001) conditions. H T decreases monotonically as e increases (Figure 4). And, as was the case for the other two measures of total diversity, frequent group extinctions ( e = 0.1) and conformist transmission have similar consequences for H (Figure 4b). Researchers are divided about the magnitude of geographic variation in Middle Paleolithic technological behavior. Some remark at the high level of regional diversity in stone artifacts (e.g., [8,48]), whereas others emphasize the similarity of evidence across Eurasia (e.g., [49,50]). Nonetheless, it is worth considering how the frequency of localized extinctions might act on geographic differentiation, or variation among spatially defined groups. The F ST results are summarized in Figure 5. Recall that F ST makes use of the relative frequencies of all cultural variants, not just the modal variant of each group. More importantly, F ST also takes into account the effect of e on H T . Two points are worthy of note. First, higher copying error rates yield lower F ST for all e . Second, e shows a significant effect on group differentiation under equilibrium ( m = 0.001: x 2 = 25.22, P , 0.001 and m = 0.01: x 2 = 12.77, P = 0.005) and non-equilibrium ( m = 0.001: x 2 = 24.37, P , 0.001 and m = 0.01: x = 19.28, P , 0.001) conditions. In our model, higher rates of local extinction constrain differentiation among groups as measured by F ST . Figure 6 summarizes the d G results for metapopulations simulated over different values of m and e . In general, d G increases with m . In addition, e has a significant effect on d G for all levels of m tested ( m = 0.00001: x 2 = 65.67, P , 0.001; m = 0.0001: x 2 = 62.12, P , 0.001; m = 0.001: x 2 = 66.19, P , 0.001; m = 0.01: x 2 = 66.69, P , 0.001). As was the case with F ST , the d G results show that there is less differentiation between groups in metapopulations plagued by a higher frequency of local extinctions (Figure 6). In short, the model predicts that group differentiation as measured by the mean distance between modal variants would decrease as the frequency of local extinction increases. It should be emphasized that this result assumes that cultural variants are neutral and have no effect on the probability of a local group going extinct. Frequent local extinctions can decrease d G to levels similar to those that result from conformist cultural transmission with no local extinctions (Figure 6b). This is not the case with the other measure of group differentiation calculated here. With unbiased cultural transmission, increasing e does not bring the F ST values closer in line with those that result from conformist cultural transmission (Figure 5b). Given that metapopulations are initialized as perfectly homogeneous in the non-equilibrium version of our model, conformist transmission within groups yields a very low level of group differentiation even in the absence of local extinction. On this point, F ST and d G agree. One might reasonably predict that conformist transmission within groups should yield a high level of group differentiation regardless of how it is measured, but this prediction is likely to be met only when starting metapopulations are highly polymorphic and local extinctions are extremely rare ( e < 0). The results concerning the effect of e on d A max —and, by extension, the effect of local extinction and recolonization on the rate of neutral cumulative change per cultural transmission event—are summarized in Figure 7. As one would expect, there is a positive relationship between d A max and m : the number of copying errors that accumulate in a metapopulation increases with the rate of ...
Citations
... Tourloukis 2010), significantly restricting the understanding of Middle Palaeolithic site burial, preservation and exposure. This, in turn, limits the degree to which the lack of sites can be discussed in terms of different factors, i.e. demographics (Premo & Kuhn 2010) or ecology (Karavanić et al. 2022), while also constraining reconstructions of the variability of land use in a regional setting. An exception is represented by the more intensively researched topic of marine transgression during the Late Glacial-Holocene, which has been recognized as a significant factor in altering the archaeological record of the Adriatic region (Benjamin et al. 2017;Karavanić & Barbir 2020). ...
In recent decades, the body of evidence from Croatian sites contributing to the understanding of Middle Palaeolithic behaviours has been significant. However, the data has been biased towards cave sites. Until recently open-air sites have exclusively been identified on the basis of surface finds, which often raise questions regarding assemblage integrity. Rescue excavations in the Istrian peninsula have recently brought to light the open-air site of Campanož and a substantial amount of new data. The site is a large and densely packed lithic scatter found stratified between two horizons of typical Mediterranean terra rossa soil. Among the lithic finds there is a large presence of nodular chert fragments and a smaller proportion of classifiable chert artefacts, which have been recognized as Middle Palaeolithic based on both typological and technological characteristics. A preliminary analysis shows that the blank production methods are coherent at the site. There are few flaking methods in the sample, with most being related to different modes of discoid reduction. Middle Palaeolithic toolmakers repeatedly procured raw materials and produced blanks on-site. Evidence points to the production of small tools, and also indicates recycling of previously discarded artefacts. Although these data are preliminary, the evidence seems to suggest an expedient and flexible technology may have been present in the Middle Palaeolithic of the Northeastern Adriatic. Despite the limited data on age and site formation processes, the site represents a valuable source of information in our understanding of Middle Palaeolithic technological behaviour and land use in the region.
... It may be that, while experiencing a megadrought, populations in Nevada experienced higher selection pressure to get the technology right, which led to less individual experimentation and more selective copying. In their own simulation, Premo and Kuhn (2010) also see reduced variation in culturally learned behavior with higher local group extinction rates when individuals rely on social learning. In any case, what is clear is that the model provides archaeologists an opportunity to test assumptions about the drivers of technological change and to generate new hypotheses about that change based on model outcomes. ...
Archaeology and cultural evolution theory both predict that environmental variation and population size drive the likelihood of inventions (via individual learning) and their conversion to population-wide innovations (via social uptake). We use the case study of the adoption of the bow and arrow in the Great Basin to infer how patterns of cultural variation, invention, and innovation affect investment in new technologies over time and the conditions under which we could predict cultural innovation to occur. Using an agent-based simulation to investigate the conditions that manifest in the innovation of technology, we find the following: (1) increasing ecological variation results in a greater reliance on individual learning, even when this decreases average fitness due to the costs of learning; (2) decreasing population size increases variability in the types of learning strategies that individuals use; among smaller populations drift-like processes may contribute to randomization in interpopulation cultural diffusion; (3) increasing the mutation rate affects the variability in learning patterns at different rates of environmental variation; and (4) increasing selection pressure increases the reliance on social learning. We provide an open-source R script for the model and encourage others to use it to test additional hypotheses.
... There is significant information-inspired archaeological literature effectively invoking entropy that is a precursor to our approach (e.g. Bevan et al., 2013;Barjamovic et al., 2019;Crema, 2015;Diachenko et al., 2020Diachenko et al., , 2022Dickens and Fraser, 1984;Drost & Vander Linden, 2018;Furholt, 2012;Gjesfjeld et al., 2020aGjesfjeld et al., , 2020bGronenborn et al., 2014Gronenborn et al., , 2018Gronenborn et al., , 2020Kandler & Crema, 2019;Neiman, 1995;Paige & Perreault, 2022;Premo & Kuhn, 2010;Shott, 2010;Hegmon et al., 2016;Wiśniewski et al., 2022), but in some cases, internal consistency is not clear. ...
The main objective of this paper is to develop quantitative measures for describing the diversity, homogeneity, and similarity of archaeological data. It presents new approaches to characterize the relationship between archaeological assemblages by utilizing entropy and its related attributes, primarily diversity, and by drawing inspiration from ecology. Our starting premise is that diachronic changes in our data provide a distorted reflection of social processes and that spatial differences in data indicate cultural distancing. To investigate this premise, we adopt a parsimonious approach for comparing assemblage profiles employing and comparing a range of (Hill) diversities, which enable us to exploit different aspects of the data. The modelling is tested on two seemingly large datasets: a Late Bronze Age Cretan dataset with circa 13,700 entries (compiled by PG); and a 4th millennium Western Tripolye dataset with circa 25,000 entries (compiled by AD). The contrast between the strongly geographically and culturally heterogeneous Bronze Age Crete and the strongly homogeneous Western Tripolye culture in the Southern Bug and Dnieper interfluve show the successes and limitations of our approach. Despite the seemingly large size of our datasets, these data highlight limitations that confine their utility to non-semantic analysis. This requires us to consider different ways of treating and aggregating assemblages, either as censuses or samples, contingent upon the degree of representativeness of the data. While our premise, that changes in data reflect societal changes, is supported, it is not definitively confirmed. Consequently, this paper also exemplifies the limitations of large archaeological datasets for such analyses.
... The process in either case might have increased the production of symbolic artifacts as they played a role in communicating "we're not you" or alternatively, "we are you." Finally, population size also plays a role, since the frequency of symbolic artifacts in a region may only reflect the density of the hominin population (Premo and Kuhn, 2010;Mellars and French, 2011). Evaluating the effects of climate and population will require more precise analysis of the geographic and temporal distribution of symbolic artifacts (e.g., Mackay et al., 2014;Will et al., 2019) and a future project might be to not only correct the data for taphonomic loss and research bias, but also for estimates of population size across Europe, the Near East, and Africa. ...
Researchers have long been aware of an apparently rapid increase ~40-45,000 BP in the frequency of "symbolic" artifacts in the Old World paleolithic record. However, some hypothesize that if not for taphonomic loss the data would instead show a gradual increase in such artifacts' frequency during the Middle Stone Age/Middle Paleolithic. We test this hypothesis by correcting the record for taphonomic bias. We find that even after correction, the ~40-45,000 BP peak remains with no prior gradual increase. However, analysis also suggests this peak may be a product of research bias. We note small peaks in symbolic artifact production at ~65,000, 75,000, and 115-120,000 BP, although these too might be a product of research bias. We end with a discussion of how symbolic artifact production might be expected to wax and wane as a function of adaptive pressures. This points to the importance of understanding the effects of taphonomic and research bias.
... That will be our focus, too, but these are probably not sufficient. Skills become distributed over the community through the division of cognitive labor and through various mechanisms of mutual support, so a large and/or networked population is likewise necessary (Henrich 2004;Powell et al. 2009;Premo and Kuhn 2010;Sterelny 2020). Likewise, selective pressures imposed by the increasing climatic variability of the Pleistocene were probably important but not our current focus. ...
Sterelny and Hiscock (S&H) argue against the centrality of high-fidelity copying in cumulative culture. I address one key strand of their case, the decoupling of expertise from precise imitation. This advances understanding of hominin skill acquisition, and underlines a puzzle about domain-specificity.
... Also, cultural complexity and the impact of population size and connectedness thereupon, as well as on cultural evolution (e.g. Kline and Boyd, 2010;Muthukrishna et al., 2014;Premo and Kuhn, 2010;Vaesen et al., 2016), is treated. ...
Bibliometrics offers powerful means of visualising and understanding trends within research domains. We here present a first exploratory bibliometric analysis of cultural evolutionary theory and attendant methods as applied specifically within archaeology across the last four decades (1981–2021). Bibliographic coupling network analysis shows that there exists a broadly successive series of author clusters making up the core of this research domain. A broader vernacular version of cultural evolution is also commonly used in thematic or regional research traditions that fall outside of cultural evolutionary studies in the strict sense. Our bibliometric networks trace the development of evolutionary archaeology over the last four decades and while they demonstrate the centrality of computational models, they also suggest a stagnation in the application of precisely that suite of methods—phylogenetics—that is central to evolutionary archaeology’s biological counterpart palaeontology. Recent methodological innovations in palaeobiology are, however, offering new ways of integrating artefact shape data directly with phylogenetic applications. This development may usher in a renaissance in artefact phylogenetics and appropriately marco-scale applications of cultural evolutionary theory in archaeology.
... Additionally, contact between groups may allow for the combination of cultural features (as explored in [8]), which may also increase the cultural repertoire. Since the 1960s [19], a number of findings have supported this hypothesis [5,10,[20][21][22][23], some have provided conflicting evidence [9,13,14,24,25], and others suggest it might have more predictive power in certain contexts (e.g. in food-producing societies more than in hunter-gatherer groups) [26]. In this article, we use an agent-based model of cultural evolution to illustrate a simple scenario that could cause cultural repertoire sizes to deviate from the predictions of the population-size hypothesis even when the underlying assumptions hold: when there is inter-group contact, even when it occurs rarely, the cultural repertoire of a focal population could be influenced by those of the populations connected to it. ...
Human populations rely on cultural artifacts for their survival. Populations vary dramatically in the size of their tool repertoires, and the determinants of these cultural repertoire sizes have been the focus of extensive study. A prominent hypothesis, supported by computational models of cultural evolution, asserts that tool repertoire size increases with population size. However, not all empirical studies have found such a correlation, leading to a contentious and ongoing debate. As a possible resolution to this longstanding controversy, we suggest that accounting for even rare cultural migration events that allow sharing of knowledge between different-sized populations may help explain why a population’s size might not always predict its cultural repertoire size. Using an agent-based model to test assumptions about the effects of population size and connectivity on tool repertoires, we find that cultural exchange between a focal population and others, particularly with large populations, may significantly boost its tool repertoire size. Thus, two populations of identical size may have drastically different tool repertoire sizes, hinging upon their access to other groups’ knowledge. Intermittent contact between populations boosts cultural repertoire size and still allows for the development of unique tool repertoires that have limited overlap between populations.
... Additionally, contact between groups may allow for combination of cultural features (as explored in [8]), which may also increase the cultural repertoire. Since the 1960s [19], a number of findings have supported this hypothesis [5,10,[20][21][22][23], some have provided conflicting evidence [9,13,[24][25][26], and others suggest it might have more predictive power in certain contexts (i.e., in food-producing societies more than in hunter-gatherer groups) [27]. In this article, we use an agent-based model of cultural evolution to illustrate a simple scenario that could cause cultural repertoire sizes to deviate from the predictions of the population-size Draft hypothesis even when the underlying assumptions hold: when there is inter-group contact, even when it occurs rarely, the cultural repertoire of a focal population could be influenced by those of the populations connected to it. ...
Human populations rely on cultural artifacts and complex cumulative culture for their survival. Populations vary dramatically in the size of their tool repertoires, and the determinants of these cultural repertoire sizes have been the focus of extensive study in recent years. A prominent hypothesis, supported by computational models of cultural evolution, asserts that tool repertoire size increases with population size. However, not all empirical studies seeking to test this hypothesis have found such a correlation; this has led to a contentious and ongoing debate. As a possible resolution to this longstanding controversy, we suggest that accounting for even rare cultural migration events that allow sharing of knowledge between different-sized populations may help explain why a population's size might not always predict its cultural repertoire size. Using an agent-based model to explore different assumptions about the effects of population size and migration on tool repertoires, we find that connectivity of one population to others, particularly to large populations, may significantly boost its tool repertoire size when population interactions lead to cultural exchange. Thus, two populations of identical size may have drastically different tool repertoire sizes, hinging upon their access to other groups' knowledge. Intermittent contact between populations boosts cultural repertoire size and still allows for the development of unique tool repertoires that have limited overlap between populations.
... While developed in information theory, the concept of entropy is widely applied in various natural and social sciences to analyze systems' internal diversity. Archaeological applications mainly consider Shannon's (1948) entropy, also known as Shannon's diversity index (Bevan et al., 2013;Bobrowsky & Ball, 1989;Crema, 2015;Dickens Jr. & Fraser, 1984;Drost & Vander Linden, 2018;Fedorov-Davydov, 1987;Furholt, 2012;Gjesfjeld et al., 2020a, b;Gronenborn et al., 2014Gronenborn et al., , 2018Gronenborn et al., , 2020Justeson, 1973;Kandler & Crema, 2019;Neiman, 1995;Nolan, 2020;Premo & Kuhn, 2010;Shott, 2010). According to this approach, "unification" is not opposed to "diversity"; instead, both categories are considered to gradually replace each other. ...
Contributing to the issue of complex relationship between social and cultural evolution, this paper aims to analyze repetitive patterns, or cycles, in the development of material culture. Our analysis focuses on culture change associated with sociopolitical and economic stasis. The proposed toy model describes the cyclical character of the quantitative and qualitative composition of archaeological assemblages, which include hierarchically organized cultural traits. Cycles sequentially process the stages of unification, diversity, and return to unification. This complex dynamic behavior is caused by the ratio between cultural traits’ replication rate and the proportion of traits of the higher taxonomic order’s related unit. Our approach identifies a shift from conformist to anti-conformist transmission, corresponding with open and closed phases in cultural evolution in respect to the introduction of innovations. The model also describes the dependence of a probability for horizontal transmission upon orders of taxonomic hierarchy during open phases. The obtained results are indicative for gradual cultural evolution at the low orders of taxonomic hierarchy and punctuated evolution at its high orders. The similarity of the model outcomes to the patters of material culture change reflecting societal transformations enables discussions around the uncertainty of explanation in archaeology and anthropology.
... New environments did not entail changes in behaviour and we cannot suggest that local extinctions of human groups in unfavourable environments led to decreased cultural diversity or changes in the structure of groups, as we observe the same features among different cultural traditions (cf. Premo and Kuhn, 2010;Kolodny et al., 2015). Who occupied Moulin Quignon and made the bifaes at Barranc de la Boella, la Noira and Notarchirico between 900 and 700 ka? ...
Current data seem to suggest that the earliest hominins only occupied the Northwest of Europe during favourable climatic periods, and left the area when the climate was too cold and dry, in the same way as Neandertal and even Homo sapiens. However, several sites in England and the North of France indicate that the earliest hominins, possibly Homo antecessor and/or Homo heidelbergensis, could adapt to cool environments and open grasslands without the use of fire. Recent discoveries of Acheulean lithic assemblages in early glacial fluvial deposits at Moulin Quignon in the Somme Valley in the Northwest of France reveal new knowledge on the earliest occupations in north-western territories and indicate hominins’ capacity to live above the c. 45th N. under a cold climate. The site shows evidence of occupations at the beginning of MIS 16 at around 650–670 ka. These findings bring to the forefront the possible ability, flexibility and resilience of Acheulean hominins at around 700 ka to extend to northern territories during transitional climatic periods (interglacial/glacial events), even if the climate was not fully favourable. Recent fieldwork has changed our interpretation of the timing and characteristics of the earliest Acheulean techno-complexes in Western Europe over a large geographical area, from Northwest Europe to the Mediterranean coast. In Western Europe, the earliest evidence, Moulin Quignon, is now dated to a narrow timeframe, between 700–650 ka, and is the northernmost evidence of biface production. This latter is earlier than British Acheulean records. Based on new findings at Moulin Quignon, we explore whether Acheulean traditions and associated new technological abilities could have facilitated the dispersal of hominins in Western Europe over large territories, regardless of climatic conditions. Changes in behavioural flexibility, and not only phenotypic changes in Homo groups, have to be investigated. Here, we examine the behavioural and technological abilities of hominins in north-western Europe in light of the available environmental data and compare them to those in southern areas between 700 and 600 ka. This event occurred at the end of the “Middle Pleistocene Transition” (MPT), a period marked by cyclical climate changes and vegetation and faunal turnovers (less competition with big carnivores). The extension of the grassland habitat into higher latitudes could have led to the opening and/or closing of migration corridors in these regions, probably favouring hominin expansion depending on tolerance to climate variability.
