The ‘mosaic habitat’ concept in human evolution: past and present

The habitats preferred by hominins and other species are an important theme in palaeoanthropology, and the ‘mosaic habitat’ (also referred to as habitat heterogeneity) has been a central concept in this regard for the last four decades. Here we explore the development of this concept – loosely defined as a range of different habitat types, such as woodlands, riverine forest and savannah within a limited spatial area – in studies of human evolution over the last 60 years or so. We outline the key developments that took place before and around the time when the term ‘mosaic’ came to wider palaeoanthropological attention. To achieve this, we used an analysis of the published literature, a study of illustrations of hominin evolution from 1925 onwards and an email survey of senior researchers in palaeoanthropology and related fields. We found that the term ‘mosaic’ starts to be applied in palaeoanthropological thinking during the 1970s due to the work of a number of researchers, including Karl Butzer and Glynn Isaac, with the earliest usage we have found of ‘mosaic’ in specific reference to hominin habitats being by Adriaan Kortlandt (1972). While we observe a steady increase in the numbers of publications reporting mosaic palaeohabitats, in keeping with the growing interest and specialisation in various methods of palaeoenvironmental reconstruction, we also note that there is a lack of critical studies that define this habitat, or examine the temporal and spatial scales associated with it. The general consensus within the field is that the concept now requires more detailed definition and study to evaluate its role in human evolution.


INTRODUCTION
In 1965 the distinguished ecologist G. Evelyn Hutchinson published a book of his collected essays entitled "The Ecological Theater and the Evolutionary Play" (Hutchinson, 1965). His title neatly captures the idea that the environment (ecology) sets the context for evolutionary change: an idea that was completely mainstream in mid-20th century science. Such interactions have had a prominent role in studies of human evolution in the mid to late 20th centurya well-known example being the Savannah Hypothesis (Bender et al., 2012;Domínguez-Rodrigo, 2014). However the role of the environment, and particularly adaptation to the environment, played a lesser role in theories of human evolution in the 19th and first half of the 20th century, when there was often an emphasis on inbuilt evolutionary drives towards certain goals and an assumption that human features, such as bipedalism, were so obviously advantageous that they required little explanation (Bowler, 1986). As Bowler (1986) pointed out, it was the increased emphasis on adaptations that arose from the 'new ', or 'modern' evolutionary synthesis (mainly between 1936', or 'modern' evolutionary synthesis (mainly between and 1947', or 'modern' evolutionary synthesis (mainly between (Mayr, 1980), that started to make the narratives of human evolution more mainstream, which put significant emphasis on environmental adaptations. Although an increased interest in such questions started around the 1930s, as so often in science it is possible to find a number of interesting precursors. For example, as far back as Lamarck (1809) and Darwin (1871), naturalists have made at least passing references to the importance of the type of habitats our early ancestors lived in and the role that this may have played in our evolution. Lamarck is a particularly interesting example, although his comments on the selection pressures leading to bipedalism are restricted to less than a page he produced what Mayr (1982: 352) described as a "startlingly modern" suggestion about the role of potential habitat change in the evolution of a bipedal stance. More recently, placing hominins within a distinct habitat type, against the backdrop of shifting African Pliocene and Pleistocene climates and environments, has been an important goal of palaeoanthropology (Butzer, 1977;deMenocal 1995deMenocal , 2004deMenocal , 2011Reed, 1997Reed, , 2013Potts, 1998Potts, , 2013Bobe & Eck, 2001;Bobe, 2006;. Several different habitat scenarios exist, including the Savannah Hypothesis, the Forest Hypothesis, the Turnover Pulse Hypothesis, the Variability Selection Hypothesis and the Tectonic Landscape Model (Dart, 1925;Vrba, 1980;Potts, 1998Potts, , 2013Retallack, 2001;Kingston, 2007;Trauth et al., 2010;Reynolds et al., 2011;Bender et al., 2012;Domínguez-Rodrigo, 2014). One of the reasons for this multiplicity of hypotheses (in addition to it being unlikely that there is one 'correct' answer) is that the environmental evidence is being viewed in very different ways. A recent review of the place of the Savannah Hypothesis in human evolution studies (Domínguez-Rodrigo, 2014) clearly shows that part of the problem lies with imprecise definitions of vegetation states that lie on a continuum, where the term 'savannah' could equally denote pure grasslands or mosaic habitats where a significant amount of trees is present in grasslands ( Figure 1). Therefore, there is a degree of overlap between the vegetation referred to here as 'mosaic habitats' and some of the habitats described in the Savannah Hypothesis.
The 'mosaic habitat' is deserving of particular attention because over the last few decades it has been reconstructed at many hominin sites across Africa from various types of palaeoenvironmental evidence (Table 1) and the use of the term in published literature is increasing, as we demonstrate later. To our knowledge, a formal palaeoenvironmental definition of 'mosaic habitat' does not exist, but it may well have developed out of the older, more established, Savannah Hypothesis, which often emphasised extensive grassland habitats (Domínguez-Rodrigo, 2014). A typical working definition of a mosaic habitat is as follows: "a range of different habitat types, scattered across and interspersed within a given area" (Elton, 2008: 381). Perhaps surprisingly, given its modern ubiquity, this term only appears in the palaeoanthropological literature in the 1970s (see below). In the four decades since then, the term has become more widely used. So much so, that recent literature has started to expand the application of mosaics in human evolution from Africa to more recent hominin taxasuch as those in Eurasia (Finlayson, 2009;Finlayson et al., 2011).
An early usage is seen in Glynn Isaac's (1976: 501) chapter in "Human Origins: Louis Leakey and the East African Evidence", in which he described the attractive nature of the "diversity of resources" offered by the "mosaic of diverse habitats" at Omo and Koobi Fora (in Ethiopia and Kenya), and it still remains central to our perceptions of human evolution, as illustrated by the following recent quotation: "The cradle of humanity was … in the savanna forest, favoured by its complex mosaic of different local habitats" (Wilson, 2012: 29). However, certain researchers have begun to voice the opinion that as a habitat concept, the mosaic is too general to be informative (Kingston, 2007).
Here we review the origin and development of the 'mosaic habitat' in human evolution studies. As the word 'mosaic' suggests, this habitat concept implies several different vegetation types arranged in patches within a delimited area. Focusing on a particular wordin this case 'mosaic'could be seen as less than ideal as it is the idea (the importance of habitat heterogeneity) rather than the word itself that is important. However, as Gould (2002: 875) pointed out, words matter as "phenomena without names, and theories marking them as worthy of notice, will probably not be recognised at all." While mosaic is the most frequently encountered term, we here consider the words heterogeneous, mixed and patchy as synonyms for similar vegetation distributions. Habitat heterogeneity is clearly important in palaeoanthropology, but this concept requires more in-depth research than it has currently received. Calls for more explicit definitions of habitats are not new: indeed Irwin Bernstein (1967: 178) specifically mentioned the lack of spatial and temporal scales in definitions of natural habitats in the 1960s.
For the purposes of our review, we outline the key developments that took place before and around the time when the term 'mosaic' came to wider palaeoanthropological attention. To achieve this we use several strands of evidence, including published research (online and print-based, summarised in Table 2), artistic representations of early hominin reconstructions, and focused questions to senior academics in palaeoanthropology and primatology, many of whom were students during the time period in which mosaics became incorporated into human evolution studies (Table 3). Such survey-based data potentially add important historical information about the use of the concept outside of published literature (see Glen, 1994;Sapp, 2009, for the use of this approach in other areas in the history of science), as there may be time lags between an idea being informally discussed and it appearing in publications. In addition to which, scientific papers are often uninformative when it comes to articulating the original genesis of the ideas they discuss (Medawar, 1963) making such an oral history approach potentially very valuable.

MOSAICS IN A WIDER CONTEXT -A BRIEF HISTORY OF MOSAICS IN THE ECOLOGY LITERATURE
Our intention with this paper is not to review the use of the concept of habitat mosaics across all of evolutionary biology and ecology, but to focus particularly on the area of human evolution. However, because the term has been widely used in plant ecology we first briefly review the more conventional ecological usage before focusing on human evolution in African environments. The term 'mosaic vegetation' was first used in plant ecology over 100 years ago (e.g. Pound & Clements, 1897;Cavers, 1914), but it was not until the 1940s that mosaic became a commonly applied idea (e.g. Watt, 1947;McIntosh, 1999). For example, vegetation mosaics were given particular prominence in the work on plant community ecology by R.H. Whitaker from the 1950s onwards (Kent, 2011). The term is still widely used in ecology today (e.g. Campioni et al., 2013;Fonderflick et al., 2013;Soininen et al., 2013;Marimon et al., 2014), and the current importance given to ideas of metapopulations and spatial dynamics in ecology makes patchy habitats of particular interest (Nee, 2007). In addition, the rise in landscape ecology as a discipline over the past 25 years has also given a particular prominence to mosaics, as landscapes are often viewed as composed of a variety of patches of different vegetation types (Kent, 2011).
Several scientists with field naturalist backgrounds who also have an interest in human evolution (especially Jonathan Kingdon and Edward O. Wilson, personal communications, 2012) pointed out to us that mosaics are a fairly obvious concept for any naturalist who works in a part of the world where vegetation is not found in large uniform blocks. We are aware of no formal definition of 'mosaic' in the ecological literature, but the generally accepted ecological usage suggests at least two types of vegetation that are distributed across a landscape. There are exceptions to this definition: if two types of vegetation meet along a gradient (e.g. a solid block of woodland meeting a solid patch of grassland), then this is usually referred to as an ecotone, rather than a mosaic (Calow, 1999).
Early 20th century descriptive studies of African vegetation, in particular in eastern Africa, make frequent mention of the mosaic-like nature of the vegetation (Michelmore, 1934(Michelmore, , 1939; Table 1. A sample of hominin habitat reconstructions from sites in Chad, Ethiopia, Kenya, Tanzania and South Africa. Additional reconstructions given in Wood and Strait (2004: 134-136) and Reed (2013). In most cases, the term 'mosaic' signals a variety or diversity of habitat and vegetation types.

Country and site
Reconstruction Data type and references Chad: Toros-Menalla "a mosaic of environments from gallery forest at the edge of a lake area to a dominance of large savannah and grassland" Macromammalian taxa (Vignaud et al., 2002: 155). Chad: Koro Toro "a lakeside environment, with both perennial and permanent streams, and a vegetational mosaic of gallery forest and wooded savannah with open grassy patches" Macromammalian taxa (Brunet et al., 1995: 273).
(b) Herbivore enamel stable isotopes (Kingston & Harrison, 2007: 272) South Africa: Sterkfontein (Member 4 and later) "overall change from tropical to sub-tropical gallery forest, forest fringe and woodland conditions in Member 4 to more open woodland and grassland habitats in the later units" (b), based partly on (c) (a) Fossil wood (Bamford, 1999); (b) Macromammalian fossils (Kuman & Clarke, 2000: 827); (c) Vrba, 1974Vrba, , 1975 South Africa: Swartkrans "a mosaic of grassland and tree cover which was probably denser alongside the ancient Blaaubank stream" Stable isotopes of mammalian enamel (Sponheimer and Lee-Thorp, 1999: 724) South Africa: Makapansgat "a woodland-savannah mosaic with a greater proportion of woodland than savannah" Multiproxy analysis of microwear and stable isotopes (Hopley et al., 2006: 248). -Fitzgerald, 1963); later ecological studies focused on the mechanisms that create and stabilise mosaics, such as grazing by herbivores and fire (Belsky, 1986;Bond & Keeley, 2005;Sinclair, 2012). There are interesting parallels between the term 'mosaic' and 'savannah' in the ecological literature. As with mosaic, savannah is a term widely used by plant ecologists that also lacks a widely-agreed formal definition (Torello-Raventos et al., 2013).

DEVELOPMENT OF THE MOSAIC CONCEPT WITHIN HUMAN EVOLUTION STUDIES
Firstly we briefly review the development of this idea before describing the results of different lines of historical evidence (visual reconstructions, focused questions and textbased sources) to study the history of the mosaic concept in palaeoanthropology. We then draw these lines of evidence together into a more detailed history of the development of the concept during the late 20th century.
Although Wells et al. (1931) in a popular encyclopaedia of biologybriefly raised the possible importance of forest/open country ecotones in the context of human evolution, most early palaeoanthropological studies largely focused on the hominin fossils, as well as the comparative anatomy of primates, in order to clarify taxonomic affinities of the hominins (Elliot Smith, 1924;Dart, 1925;Le Gros Clark, 1959a). Other important foci included dating the early sites and their faunas and establishing relative chronologies (e.g. Clark, 1957;Bishop & Miller, 1972). In the context of Wells et al.'s (1931) suggestion it is probably relevant that one of the authors was Julian Huxleya key player in the evolutionary synthesis of the late 1930s and early 1940s and actively thinking about the role of selection in relation to the environment (Bowler 1986). Reviews of progress in the field of palaeoanthropology by Le Gros Clark (1967), Leakey and Goodall (1970), Cartmill et al. (1986), Tattersall (2000) and Goodrum (2013) clearly highlight these early goals, but make no detailed mention of the environmental context for hominin evolution. Indeed, Karl Butzer in the preface to his book Environment and Archaeology (1971: vii) wrote "In preparing the original version of this book [published in 1964] I made an explicit plea for a Pleistocene geography concerned with man and the land. This revised edition reflects my concern that dating techniques, geomorphological theory, and regional stratigraphy continue to be overemphasized in the literature and in the classroom, with insufficient stress on the total environment." The fact that Butzer was trained as a geographer is obviously relevant to understanding why he came to consider this aspect of human evolution important and understudied.
Notable early exceptions to this lack of an environmental perspective are the works of Basil Cooke, Lawrence Herbert Wells and Ronald Singer who, in the 1950s, were using fossil fauna to attempt to reconstruct the past environments at several South African sites. Singer (1957: 178) wrote of Elandsfontein that the "site represents the desiccated basin of a prehistoric marshy vlei (lake or pool) surrounded by grassy plains and hills in which fairly dense scrub bushes and leafy trees must have been prominent features." While for the site of Makapansgat, Wells and Cooke (1956: 48) made a link between the "great wealth of species" observed for the fossil   American Anthropologist  Search for specific terms gives indication of years in which terms feature in published articles. Determines a likely period in which term becomes more widely used in scientific literature. Frequently cited papers indicate key texts. Citation data from Web of Science taken as an indicator of the influence of these publications.

E-mail survey:
Sent to 90+ senior academics, artists involved in hominin reconstructions and 'popular', science authors working in Africa and Asia during the timeframe in which the concept emerged.
Requested information from respondents about their recollections of the context in which they first encountered the term mosaic habitats.
Provided key sources, people and localities to be further investigated.

Book index search:
Consulted over 200 books covering a range of topics (ecology, palaeoenvironments and human evolution) published between approx. 1916 and 2013. Indexes were checked for a predefined range of terms, relating to habitats and environments. These were: environment; habitat; mosaic, heterogeneity, heterogeneous and patchiness, as well as combinations of these.
Key texts, in particular early textbooks and conference proceedings were examined for associated references in an effort to identify the texts most influential in popularising the term 'mosaic'. Table 3. Questions e-mailed sent to senior academics to obtain first-hand information about the development of the 'mosaic habitat' concept.
Question 1 What is the first paper that you remember discussing mosaic habitats in the context of the hominin record (rather than in the ecological literature)? Please give bibliographic details, if possible. Question 2 When do you first remember hearing about mosaic habitats in relation to hominin evolution? Was it at a conference, seminar or meeting? If so, can you remember the speaker, and the date and place of the meeting? Question 3 As pictures have been key in promoting human evolution in books and museum displays, is there a picture or diorama that you think encapsulates the idea of a mosaic habitat? Please could you tell us where or what it is? And is there an earlier example?
antelopes and the "physiographic setting of the site in a scrub-covered mountain mass, with extensive open or bushcovered plains at no great distance." Both descriptions would probably be termed mosaic today. Thus, until the mid-1970s the term 'mosaic habitat' occurred infrequently in the palaeoanthropological literature, but the appreciation of the role of mixed, heterogeneous, habitats was being established from the 1950s onwards. Gradually, the perspective altered to encompass detailed environmental and climatic reconstructions and evaluation of the role played by environments in shaping the course of human evolution; a trend still very much in evidence today (e.g. Cooke, 1978;Peters, 1979;deMenocal, 1995deMenocal, , 2004Potts, 1998Potts, , 2013. Bernard Campbell's textbook Human Evolution, an introduction to man's adaptations, aptly illustrates this shift. A subsection entitled 'leaving the forest' in the 1967 edition makes almost no mention of habitats except as a substrate for hominin movement, while the 1974 second edition has a paragraph inserted at the start of the same subsection that explicitly mentions an 'ecotone' (Campbell, 1974: 376).

ARTISTIC RECONSTRUCTIONS
Illustrations can be highly influential in communicating ideas in science (e.g. Rudwick, 1976). Therefore artistic attempts to portray hominins provide additional insights into changing perceptions of past environments, as an illustrator is forced to make a deliberate choice about what to show while a verbal description can more easily miss out whole aspects of a scene. Although it may be tempting to see such images as objective illustrations of the science of the time, they can clearly be influenced by a wide range of other cultural, artistic and commercial influences (Privateer, 2005), so they need interpreting with caution and in the context of other lines of evidence.
Darwin's early speculations on human evolution included an African origin and upright stance predating later increases in intelligence and brain size. He suggested a possible environmental context for the switch to a bipedal stancenamely increasing terrestrial locomotion and use of habitats away from dense forest, although he conceded that this could just have plausibly led to improvements in quadrupedal locomotion too (Darwin, 1871;Bowler, 1986Bowler, , 2003. However, until well into the 20th century Africa was largely seen by most workers as irrelevant to the key developments in human evolution. Instead, much attention focused on Europe and later Asia following the discovery of the first Pithecanthropus (later renamed Homo erectus) fossils (Bowler, 2003). Indeed many of the earlier versions of the savannah hypothesis envisaged our ancestors evolving in the grasslands of central Asia (Bowler, 1986;Bender et al., 2012). Only some years after the discovery of Australopithecus africanus (Dart, 1925) did the potential importance of African savannah habitats start to become apparent.
Many of the key early 20th century texts, such as Keith (1925), focus almost exclusively on the anatomy of the fossils and with little discussion of the potential ecological context. A similar trend is mirrored in the artistic representations of the period. Once Australopithecus africanus had been announced to the world, reconstructions began to appear in popular publications. These early reconstructions tended to be vignettes of just the head of the hominin, with no environment (Figure 2), or had the reconstructed hominin dominating the picture with only very limited landscape visible (Figure 3).
Classic early 20th century examples of this style include the work of artist Charles R. Knight (Milner, 2012).
These illustrations reveal the focus on anatomy, but also reflect a concomitant lack of palaeoecological context for the early excavations at African sites, such as Olduvai Gorge, Tanzania. Louis Leakey's remarks in Olduvai Gorge 1951-61 Volume 1 (1965) reveal a reluctance to determine habitats based on the data available to him at the time: "While we may justifiably conclude that the presence of fish, hippopotamus, crocodiles and aquatic birds indicates the presence of a relatively stable body of water, it is necessary to use great caution when attempting to assess climatic conditions on the basis of most mammals" (Leakey, 1965: 74). In contrast, the presence of fruits and seeds in the Miocene deposits at Rusinga Island (Kenya) meant that habitat reconstructions for this earlier period were being created and discussed much earlier than those of Australopithecus and early Homo. For example, the Miocene taxa Proconsul and Limnopithecus (now known as Dendropithecus) were described as being "found in association with fossil faunal and floral assemblages that showed these creatures had once lived in open grasslands, between forest galleries, along the rivers which once flowed into an ancient Miocene lake" (Leakey, 1960: 18). Therefore the relative lack of non-faunal habitat indicators (such as plant remains) may have hindered palaeoecological research at Pliocene and Pleistocene sites for some time.
Despite this lack of palaeoecological data, by the 1950s colour illustrations were being produced for the British Museum (Natural History) by Maurice Wilson that are still being used today in a non-historical context (e.g. Andrews & Stringer, 1989;Stringer & Andrews, 2005). Figure 4 shows a reconstruction of Australopithecus africanus in a rocky landscape, but with distinctively African Euphorbias in the backgroundan example of modern African ecology standing in for missing palaeoecological information (taken to its extreme in Edey [1973] where drawings of hominin reconstructions are inserted into photographs of modern African landscapes).
By the later 1950s, Neave Parker had created an 'active' reconstruction for hominins at Olduvai Gorge, using information supplied by Louis Leakey on the recently discovered Chellean (Early Stone Age) living floor and fauna (Leakey, 1957(Leakey, , 1958; and see de la Torre, 2011). This 1958 illustration ( Figure 5) is a mosaic in all but name, showing bushland, grassland and water. By the mid-1960s Jay Matternes' hominin dioramas, as seen in Howell (1970), show hominins in highly variable environments, although mosaics are not explicitly mentioned in Howell's text. In E.O. Wilson's (1975) influential text Sociobiology, the new synthesis his fig 27-5 shows a reconstruction of a group of Homo habilis in a mosaic landscapedrawn by Sarah Landy (the text credits F. Clark Howell with providing advice on the reconstruction). However the text makes no explicit reference to the mosaic landscape and Wilson (personal communication, 2012) says he only gave serious thought to the role of mosaics in human evolution while researching his much later book 'The Social Conquest of Earth' (Wilson, 2012).
It is possible that the illustrations of variable or mosaic habitats in popular science books and the press played a role in communicating the idea of the mosaic habitat in scientific palaeoecological publications some 15-20 years later.

SENIOR ACADEMIC RECOLLECTIONS
We approached a total of 95 senior academics, asking for their recollections of when they first heard the term mosaics applied to hominin environments, whether in Africa or in Asia (Table 3). These academics were chosenafter our literature search and discussions with a range of colleaguesas likely to have been at key conferences and other more informal meetings during the time when these ideas were starting to emerge. A total of 29 responded, and of these, 8 respondents indicated that ecology books and books on human evolution had been highly influential, coupled with lectures at undergraduate level, and 12 said they were highly influenced by their own early fieldwork, or by discussions with colleagues who had done fieldwork in Africa. The balance of the respondents could not recall exactly the context in which the term appeared, but mentioned that by the time the term became current in publications, it had already been in use for some time at meetings and in conversation. One of these meetings may have been the IXth International Congress of Anthropological and Ethnological Sciences in Chicago, 1973, which will be discussed in more detail below.  The key point that came out of these responses was that the idea appears to have emerged slowly, rather than appearing as a new concept in any particular publication and that it is clear that it was being informally discussed in advance of its appearance in the formal literature. The response by the primatologist William (Bill) McGrew is typical: "My immediate response is that mosaic habitats just sort of crept in to the conversation." In addition, most of the people suggested as key in early discussions and publications workedor had been trainedin areas such as zoology, geography and forestry and only published on human evolution later in their careers.

PRINTED SOURCES
Our analysis of the published literature was based on extensive reading of paleoanthropological articles and books and those on related topics. In addition, much of the scientific literature now exists in searchable digital archives, and this allows a more quantitative approach to tracking the spread of an ideaor to be more exact, the terms (such as 'mosaic') associated with that idea (McIntosh, 1999). We have examined the first 40 years of the archive issues of the Journal of Human Evolution (1972Evolution ( -2011 to quantify the number of times in which the term 'mosaic' was used in reference to habitats (distinct from other uses such as 'mosaic evolution' 1 , or 'photographic mosaic', which were disregarded). Of a total of 3680 articles published over this 40 year span, only 88 make reference to mosaic habitats. The use of the term has become more common in recent years ( Figure 6). For the period 1972-1976, only three papers (i.e. less than 1%) of all papers published (n=473) used the term mosaic habitat, while for the later period 2007-2011, 37 articles of a total of 652 articles mentioned this habitat type (almost 6%). This increasing trend is discernible for the case study example for Journal of Human Evolution, but is also seen in several other publications listed in Table 2, reflecting the growth in detailed and specialised studies of palaeoenvironments of the last 40 years. Clearly the use of such quantitative analysis of keywords in digital archives has drawbacksa concept could be discussed but without the use of the focal keyword. Because of this, our extensive reading of the appropriate literature was especially importantboth to spot the use of the concept without the associated word 'mosaic' and to better understand the context in which mosaic was being used in any particular source.

HISTORIC DEVELOPMENT OF THE MOSAIC CONCEPT
Years preceding the appearance of the mosaic -1960s to 1975 In the years immediately preceding the widespread appearance of the mosaic habitat in print, the conception of hominin habitats seems to have been a fairly simple dichotomy of early hominins moving from forest to open habitats (e.g. Sahni, 1952;Washburn & Howell, 1960;Schultz, 1961). The volume ' African Ecology and Human Evolution', edited by Howell & Bouliére (1963) has a number of papers and discussion transcripts that discuss vegetation patchiness and a forestgrassland ecotone, although this latter term and the term mosaics was not used. By the end of the 1960s there was a clear shift in thinking, with people referring to the complexities of the savannah habitat. For example, John Napier (1967: 44) wrote "an environment neglected by scholars but one far better suited for the origin of man is the woodland-savanna, which is neither high forest nor open grassland". Glynn Isaac (1969: 8) was also well aware of the shortcomings of the open grassland version of the savannah model, stating "Available data including the distribution of fossils make it appear that savannah and open grassland was the ecological setting for early cultural developments, but African savannahs are extremely diverse in their biology, climate and physiography. We are wholly ignorant of the extent to which geographical or chronological variation amongst them or within them has influenced hominid evolution." In this transitional period between the dominance of the classic savannah hypothesis and the arrival of the mosaic habitat, we have noted four key trends within the literature Figure 5. An illustration of 'Chellean' hominins at Olduvai Gorge, published in the Illustrated London News on 28 June 1958. The reconstruction is by Neave Parker, following guidance from L.S.B. Leakey, and the accompanying text states "The illustration above shows Chellean hunters on their campsite near a shallow marshy stream." © Illustrated London News Ltd/Mary Evans. Figure 6. Review of the archive issues of the Journal of Human Evolution since its inception in 1972 indicates a steady rise in the use of the term 'mosaic habitat'. The data were calculated as follows: each mention of mosaic habitat was recorded (each article was only counted once) and the total number of articles published (n=3680) were summed into five-year bins and expressed as a percentage. and the wider field: (1) a proliferation of primate studies, particularly in the wild; (2) the increase of large-scale African palaeoanthropological field projects with multiple specialists; (3) detailed research into the palaeoenvironments associated with Miocene localities (and fossils of primate genera like "Ramapithecus" sensu lato) such as Fort Ternan, Kenya (Tattersall, 1969;Andrews & Van Couvering, 1975;Andrews & Nesbit Evans, 1979;Shipman et al., 1981); and (4) an increase in interdisciplinary conferences and the publication of their proceedings on specific topics such as chronology (Bishop & Miller, 1972), the Middle Pleistocene (Butzer & Isaac, 1975) and savannahs (Harris, 1980). Together, these four areas of specialisation appear to have led to a deeper awareness of the role of ecology and African environments within palaeoanthropology, and they are reviewed below.

Primate studies
An increasing number of very detailed studies of African primate behaviour and ecology, on a range of taxa including vervets (Chlorocebus pygerythrus), geladas (Theropithecus gelada), baboons (Papio spp.), and chimpanzees, (Pan spp.), provided the first detailed information about the ecology and behaviour of living primates (Schaller, 1963;Goodall, 1965;Reynolds & Reynolds, 1965;Kortlandt, 1967;Rahm, 1967;Struhsaker, 1967;Altmann & Altmann, 1970;Clutton-Brock, 1977). An important subset of these studies sought to link insights gleaned from primate behavioural studies to those of human evolution, and these were very influential in this respect (Washburn & DeVore, 1961;Jolly, 1970;Kortlandt, 1972;Teleki, 1975;Washburn & McCown, 1978;Bernstein & Smith, 1979;Kortlandt, 1972Kortlandt, , 1980aPeters & O'Brien, 1981;Kinzey, 1987). Indeed Washburn and Howell (1960: 47) explicitly stated that "Extraordinarily little is known about the necessary background, including environmental situations and the particular structures, physiological mechanisms, or behaviour patterns, which formed the basis of the transition of some proto-hominid group to an early (australopithecine) grade. Certainly much more might be learned of these basic, important behavioural and structural preadaptations from field studies of African-ape behaviour in particular ecological situations." As part of the focus on the free-living primate populations, there was necessary attention paid to the ecological aspects of their habitats, ranging behaviours and diets. Importantly, the emphasis widened from the gorillas and chimpanzees to examining other primate species as ecological models for hominins. Specifically, DeVore and Washburn (1963), DeVore and Hall (1965), and Hall and DeVore (1965) argued for the relevance of baboon ecology to the understanding of human evolution, because these terrestrial primates offered "some insights into the problems which confronted early man" (DeVore & Washburn, 1963: 335). In addition, during the late 1960s and early 1970s several biologists also argued that field studies of African social carnivores might be highly relevant to understanding hominin behaviour and ecology (e.g. Schaller & Lowther, 1969;Tinbergen, 1972).
Of the primate ecology studies, two stand out as being particularly important. Clifford Jolly's 1970 interpretation of seed-eating by geladas as a model for early hominin diets and habitats seems to have been highly influential. Jolly envisioned the transition to seed-eating taking place "… in a dambo-like environment, later shifting to wider floodplains" (Jolly, 1970: 21). Dutch ethologist, Adriaan Kortlandt, who wrote several papers on chimpanzees and human evolution, was the author of the second influential study. He was the first to use the term 'mosaic habitat' in his 1972 book "New Perspectives on Ape and Human Evolution" which was obscurely published, but later reviewed across 22 pages of Current Anthropology in 1974, which may have helped bring the idea of the mosaics to a wider audience.

International field expeditions
Early field pioneers, such as Raymond Dart, Robert Broom, and Louis and Mary Leakey, tended to work alone or in small research groups, inviting specialists to become involved at appropriate stages in the research and publication. In the 1960s and early 1970s, a new model developed where large international and interdisciplinary field teams were put together and major field projects begun, such as the International Omo Research Expedition, the Koobi Fora Research Project, the Afar expedition and the Middle Awash Project (Boaz, 1981;Goodrum, 2013). These expeditions provided the opportunity for many young scientists to gain first-hand experience of the complexity of African habitats.
Detailed geological research led to a greater appreciation of the information inherent in the sedimentological contexts of hominin fossils (Bishop & Miller, 1972;Coppens et al., 1976;Hay, 1976;Bishop, 1978). These expeditions were also the impetus for the beginnings of some long-term actualistic research projects, such as those of Anna K. Behrensmeyer on taphonomy, in the Amboseli Basin (Kenya) and which have provided many new insights into African palaeoenvironments and site formation processes (e.g. Behrensmeyer, 1975Behrensmeyer, , 1978Western & Behrensmeyer, 2009).

Miocene site reconstructions
Analyses of primate-bearing deposits from the Siwaliks (Pakistan), Songor and Fort Ternan (Kenya) indicated that Miocene apes were most likely associated with forest habitats. Therefore, at some point during the course of human evolution, a shift to terrestrial savannah habitats had occurred, to which Pat Shipman and colleagues (1981: 49) ascribe the "appearance of Ramapithecus and other new species (that) may have been related to the ecological shift towards more open country." This habitat shift was considered a critical step in human evolution, which was linked to key hominin adaptations, such as the appearance of hunting behaviours (e.g. Cachel, 1976;Andrews & Nesbit Evans, 1979;Shipman et al., 1981). During this time, the term 'mosaic habitat' appeared in connection with the Siwalik palaeoenvironment, which was reconstructed as "a mosaic of more and less open habitats, rather than riparian forest" (Pilbeam et al., 1977: 691). Andrews and Van Couvering (1975) used a detailed perspective of modern East African vegetation types, geology, geomorphology and faunas to reconstruct the Miocene faunal environments for sites such as Rusinga and Songor in Kenya. These authors described modern habitats as mosaics (in the ecological sense), but they did not apply this term to the Miocene palaeohabitats.

Conferences and conference proceedings
A number of major conferences took place in the early 1970s. Two were sponsored by the Wenner-Gren Foundation at Burg Wartenstein, Austria. The first was on dating of fossil sites in 1971 and the proceedings were published as Calibration of Hominoid Evolution (Bishop & Miller, 1972). The second was on the Middle Pleistocene in 1973 and published as After the Australopithecines (Butzer & Isaac, 1975). While the papers from the latter meeting were revised before publication, it is notable that three authors used the term mosaic in their chapters -Karl Butzer, Hilary Deacon and Glynn Isaac. In the concluding chapter, Glynn Isaac also noted that during the conference both he and Hilary Deacon had been emphasising the role of the "complex mosaics of biotopes in Africa" acting as a buffer against climatic change (Isaac, 1975: 877). This meeting took place immediately before the 9th International Congress of Anthropological and Ethnological Sciences in Chicago and the results were reported there (Tax, 1975). The Chicago Congress was also attended by Adriaan Kortlandt who is recorded as using the term 'mosaic' in a discussion session (Anon, 1976) and by Susan Cachel, who presented her paper on Australopithecus behaviour that explicitly used the word 'mosaic' (Cachel, 1976). Many of the senior figures in palaeoanthropology were present at the Chicago meeting, and it would appear, given the time lags that occur between writing papers and book publication, that these 1973 meetings may have been the catalyst for the rise of the term 'mosaic' from the mid-1970s onwards.

The mosaic spreads: 1976 onwards
In Glynn Isaac's 1976 chapter in "Human Origins: Louis Leakey and the East African Evidence", published in California, he presented the evidence for hominin preferred habitats as follows: In summary, the palaeoenvironmental reconstructions of the basins (i.e. Omo and Koobi Fora) that preserve Plio-Pleistocene archaeological traces suggest a mosaic of diverse habitatsbeaches, reed beds, swamps, edaphic grassland, savanna, riverine woodland and bush, some gallery forest. … I strongly suggest that it was the diversity of resources that may have been attractive. (Isaac, 1976: 501, original emphasis) This chapter was republished verbatim two years later in the volume edited by Clifford Jolly (1978) entitled: "Early Hominids of Africa" and published in London. It is possible that Isaac's chapter enjoyed a wide readership on both sides of the Atlantic and that this visibility helped his proposition of the importance of the diversity offered by the mosaics of Africa gain wider acceptance. Harris and Isaac (1976) also mentioned mosaics in their reconstruction of the Karari escarpment (Kenya) palaeohabitat in the influential journal Nature.
High latitude studies of Pleistocene vegetation changes may also have played a role. Valerius Geist (1978: 214) emphasised the role of transitional areas in human evolution "The widely accepted notion …, that during the Miocene and Pliocene a drying trend forced arboreal apes to the ground to assume a terrestrial life on the savannah, is simply not tenable. It ignores the great opportunities of the forest-steppe [here steppe = savannah] ecotone". He also refers to the periglacial vegetation of northern latitudes as a mosaic (Geist, 1978: 202). Alan Gentry also cites high latitude Pleistocene vegetation changes as where he first met concepts of mosaic vegetation in his response to our e-mail survey.
By the late 1970s, several authors make use of the term in different contexts. In addition to the work in the Siwaliks (e.g. Pickford, 1977), authors such as Susan Cachel (1976: 193) stated that "hominids lived in environmental mosaics, and not exclusively in a single, completely open-country environment" and Raymonde Bonnefille's palynological reconstructions of the lower Omo valley confirmed the "mosaic nature of the vegetation" (1976: 428). In 1977, Karl Butzer placed hominins firmly within mosaics in a review paper that makes multiple reference to the importance of this habitat, and he states that "(t)he location of all early hominid sites (were) in mosaic environments along ecotones of the seasonally dry African 'savanna'" (Butzer, 1977: 577). The mosaic had arrived in palaeoanthropology.
The mosaic takes off : 1980-present In 1980, Adriaan Kortlandt published two papers on primate behaviour and habitats, with implications for human evolution, and both of these referred explicitly to the mosaic habitat (Kortlandt 1980a, b). J. Desmond Clark (1980) also mentioned both 'mosaic' and 'ecotone' in the first two pages of a chapter on the 'Early human occupation of African savanna environments' (part of a conference proceedings following a Wenner-Gren meeting on the topic of savannahs in 1978), demonstrating that both mosaics and ecotones were becoming established in discussions. Then, in 1981, two key papers were published: firstly, Bill McGrew and colleagues concluded that "vegetation of the East African Plio-Pleistocene was of a mosaic nature, a patchwork of riverine forest, open and closed woodland, scrub and dry and wet grassland" based on their chimpanzee studies in the dry savannah of Mount Assirik, Senegal (McGrew et al., 1981: 241); secondly Owen Lovejoy's highly cited Science publication, entitled "The origin of man" linked omnivorous Miocene hominoids with "variable and mosaic conditions" (Lovejoy, 1981: 344), and was probably instrumental in bringing the mosaic habitat term to the wider attention of palaeoanthropologists and students.

The present state of the mosaic habitat
Since then, the use of the mosaic has steadily increased in published articles, as shown in our summary of site reconstructions across Africa (Table 1) and our case study of the Journal of Human Evolution (Table 2 and Figure 6). However, there remains a lack of focus on the drivers of habitat heterogeneity as well as temporal and spatial scales. The challenge for the future will be to define and refine the temporal and spatial scale of the fossil mosaics, rather than simply noting that heterogeneous landscapes and vegetation were present at a site. There is now a desire on the part of palaeoecologists and palaeoanthropologists to focus on the mosaic and clearly define its role in human evolution. The American Association of Physical Anthropology meeting in Calgary, Canada in 2014 hosted a themed session on mosaic habitats and the concept of scale emerged as being of central importance.

CONCLUSIONS
The term 'mosaic habitat' originated in plant ecology and started to be applied in palaeoanthropological thinking during the 1970s. The earliest usage we have found of 'mosaic' in specific reference to hominin habitats is by Adriaan Kortlandt (1972). Glynn Isaac related the term 'mosaic habitats' to the diversity of resources they offered to hominins (Isaac, 1976(Isaac, , 1978. From a history of science perspective this provides a good example of a concept that emerged over a period of time and became influential without ever becoming associated with one or two influential 'named' scientists. It also illustrates the importance of conferences and informal discussionsrather than just formal academic publicationsin the development of scientific concepts.
The mosaic concept in terms of hominin landscapes and evolution remains loosely defined as a mixture of arboreal and savannah environments, typically around a water source at a fossil site. Defined in these general terms, it is easy to believe that these types of habitats are widespread, that they are stable and that they function in similar ways across regions. Importantly, this static view of mosaic habitats is at odds with what ecologists know about modern mosaics: that they are often a dynamic and unstable state in a vegetation succession between closed woodland and open grassland.
One of the problems with the present state of the mosaic concept is that it lacks a clear statement of what the appropriate temporal and spatial scales should be. Because a mosaic for a small mammal or insect is unlikely to be a mosaic for a hominin, scale is an important missing part of the concept as a whole. We are in need of several organismspecific definitions, instead of just a single definition. We may then be able to reassess the proper place of the mosaic habitat within the framework of human evolution studies.