Wolfe [14] and Olson [42] have produced generalized vegetation maps for this period .sx Figure 9.4 is based on these maps , and shows clearly the vegetation changes that are presumed to have taken place since the Middle Eocene some 30 Ma earlier ( Figure ) .sx The wide-ranging , large-leaved , evergreen , tropical and paratropical rain - forests of the Eocene had contracted through about 35 degree of latitude , and been replaced at higher latitudes by smaller-leaved seasonal forests and woodlands with a variable ( but often high ) proportion of deciduous taxa .sx The greater diversity of vegetation types in the Middle Miocene ( nine , as against five Eocene types ) was largely the result of adaptation to strongly seasonal climates at all latitudes :sx deciduous woodlands in the tropics and subtropics of Africa and India , sclerophyll woodlands between latitudes 20 degree and 40 degree in both hemispheres , and mixed coniferous forest at higher latitudes and altitudes in the northern hemisphere .sx Nevertheless , the tenuousness of the evidence on which this Miocene map is based must not be overlooked .sx The vegetation reconstructions for almost the whole of the southern hemisphere rely on a very small number of fossil sites , and rather different reconstructions , based on the same very limited data , have been proposed for South America [43] , Africa [9] and Australia [44] .sx The extent to which non-forest types ( savanna , grassland , or even desert ) , in particular , had developed in continental interiors in the subtropics is quite unclear .sx The best documented feature of the Middle Miocene vegetation is the wide-ranging occurrence of temperate broadleaved forests in the northern hemisphere .sx These forests contained taxa now characteristic of the present-day forest vegetation of western and eastern North America , non-Mediterranean Europe and East Asia , but were more diverse than any single modern forest-type in these regions .sx Parallels have been drawn with the modern " mixed mesophytic forests " of China [45] , but a better analogy might be to regard them as an amalgam of at least four of Wolfe's forest-types [16] ( types 6-9 in Figure ) .sx Although there must have been regional variation in these Miocene forests , many taxa apparently occurred over almost the entire range .sx Among these taxa were :sx cold-temperate hardwoods such as Alnus , Betula , Corylus , Nyssa , Salix and Ulmus ; cold-tolerant conifers such as Picea , Pinus and Tsuga ; warm-temperate trees such as Carya and Liquidambar ( still widespread ) , Ceridiphyllum and Glyptostobus ( now restricted to eastern Asia ) , and Sequoia ( now restricted to western North America ) ; and subtropical taxa such as Engelhardtia .sx All these ( and many other ) taxa had also been widespread earlier on , in the Middle Eocene Epoch , but in rather different forest-types .sx Consequently , to regard the middle-latitude Miocene forests as derived solely from southward displacement of an Eocene Arctotertiary Geoflora would be erroneous .sx The middle-latitude Miocene flora certainly contained many temperately - adapted taxa of that geoflora that had migrated southward with deteriorating climate , but it had also drawn in taxa from the Early Tertiary paratropical forests - taxa that had been 'left behind' , and had evolved in situ to produce new lineages adapted to colder conditions than those favoured by their Eocene ancestors [45] .sx Similar considerations must also apply to the flora of the sclerophyll woodlands at lower latitudes ( the Madrean-Tethyan Geoflora ) , associated particularly with the subtropical high-pressure belt and its margins in both hemispheres .sx That belt , with its seasonal or more prolonged drought conditions , remained much more nearly stationary as vegetation zones were displaced progressively equatorwards through the Tertiary Period .sx Consequently , new lineages were continually being drawn in as temperate taxa migrated equatorwards .sx Some adapted to the drier climates , and persisted , so that recruitment to the Madrean-Tethyan Geoflora was continually occurring .sx The concept of geofloras as originally proposed by Chaney ( section 1.3 ) can thus have only limited signficance in the context of Tertiary vegetation history [45,46] .sx 9.3.3 The spread of non-forest communities .sx Following the thermal optimum of the Middle Miocene Epoch ( section 4.4 ) , renewed orogenesis affected most parts of the globe , and continued , often with increasing intensity , into the subsequent Pliocene Epoch ( section ) .sx As a result , the range of climatic and topographic conditions over the earth's surface was enhanced , and global ecospace was further diversified .sx Ocean temperatures ( and presumably land temperatures also ) fell sharply between about 15 and 13 Ma ago , and again between 7 and 5 Ma ago , leading first to enlargement of the East Antarctic Ice Sheet and then to icing-up of West Antarctica [47] .sx The ensuing fall in global sea levels , coupled with intensification of high pressure in the subtropics and increasing rainshadow effects in the lee of newly-uplifted mountain ranges , combined to make the Late Miocene the driest part of the Tertiary Period .sx This arid episode was manifested most spectacularly in the almost complete drying-up of the Mediterranean Sea [48] .sx In northern Israel as a result [49] , pollen spectra with 50% tree pollen in the Middle Miocene ( and including such mesophytic taxa as Alnus , Corylus , Engelhardtia , Juglans , Platycarya and Pterocarya ) were replaced in the Late Miocene by spectra with almost no tree pollen , and with 'arid' indicators such as Artemisia and Chenopodiaceae prominent .sx There have been speculations that a wide range of dryland communities had developed earlier , in the drier conditions of the Oligocene 25-35 Ma ago [9,50] , with centres of aridity in South America in the rainshadow of the rising Andean Cordillera , in the low-latitude Sahara and Kalahari Deserts , and in the remote interior of Middle Asia [51] .sx However , the major evidence of diversifying dryland vegetation comes from the Late Miocene and Pliocene .sx Certainly , the environmental conditions necessary for the creation of some present-day desert regions did not exist before the Miocene :sx the Middle Eastern desert region was largely submerged below the Tethys Sea [52] , the American Southwest was still open to the influence of rain-bearing westerly winds , and parts of Middle Asia were not yet isolated from the south-west monsoons by the mountain barrier of the Himalayas .sx All that changed with the frenzied tectonic activity of the Miocene and Pliocene , and during the last 5-10 Ma of the Tertiary Period non-forest communities became widely established over the same general areas occupied today by desert , steppe and savanna vegetation .sx However , although floras containing non-woody taxa similar to those predominant today in desert , steppe and savanna areas were undoubtedly present in the Late Tertiary , the component taxa might have been assembled into communities very different from modern dry-climate ones .sx Thus pollen floras of Late Miocene and Pliocene age , even from present-day desert areas , almost invariably contain tree pollen , often of mesophytic taxa .sx For example , pollen of Betulaceae , Fagaceae , Juglandaceae and Ulmaceae accompanies that of Pinus , Artemisia and Chenopodiaceae in Miocene pollen spectra from the Gobi Desert region of north-west China [53] .sx Whether this pollen was transported from faraway deciduous forest sources to the south , or whether it was derived from local woodland patches growing in damper situations within the region , is not clear .sx Most palaeoecologists would probably favour an explanation based on local sources of pollen in arid regions - either from continuous scrubby woodland with an understorey of desert plants , or from local woodland patches intermixed with more open vegetation .sx Nevertheless , the macrofossil , pollen and faunal evidence for the Late Tertiary supports the existence of a wide range of dryland communities .sx True desert conditions may have been present in the Sahara and the Namib [54] , and semi-desert in southern Africa and the Near East [9,49] .sx The extensive saline deposits that formed as the remnants of the Tethys Sea evaporated could have triggered the evolution of many desert halophytes [15] .sx Some sort of sclerophyllous woodland was almost certainly present in the American Southwest , whilst fossil floras of Pliocene age from the Great Plains region of the United States attest to a gradual restriction of forest to river valleys there as grassland communities developed on the drier interfluves .sx The existence of extensive areas of open grassland is suggested also by the presence of grazing and browsing faunas ( horses , camels and osteodonts ) in Kansas and Oklahoma after 10 Ma ago [55] .sx The presence of grazing faunas has similarly been used as evidence of grassland conditions in South America , but here aridification occurred rather earlier - by 15 Ma ago at least [13] - and by the Pliocene thorn scurb and grassland probably occupied much of southern South America east of the Andes [13,43] .sx More open conditions probably developed within the Amazonian forest too in the Pliocene , to form a major 'savanna avenue' extending north from the dry woodlands of the Brazilian Chaco , through the Mato Grosso of Brazil , to the highlands of Guyana and Venezuela [13,56] .sx In East Africa , grazing faunas diversified later , about 2.5 Ma ago , at a time when pollen evidence indicates a contraction of rainforest [57] .sx Savanna and grassland may have spread sufficiently then to separate the formerly continuous rainforest into eastern and western segments [9] .sx In India the dominant rainforest of the Middle Tertiary retreated gradually eastwards and southwards in the Pliocene , as arid conditions developed in Pakistan and to the south of the rapidly-rising Himalayas [8,58] .sx Arid conditions almost certainly developed too in the interior of the Australian continent , but adequate fossil documentation is lacking [2,59,60] .sx In areas outside the tropics where seasonally dry climates developed in the Late Tertiary , the major dry period initially occurred during the cool season , and 'mediterranean' climates with summer drought did not develop until the accumulation of polar ice allowed the formation of cold ocean currents affecting the western sides of the major land masses [61] .sx For some areas , full summer drought conditions may not have been realized until the glacial stages of the Pleistocene [9] , but in the northwest Mediterranean Basin at least there is some evidence that dry summers developed in the Late Pliocene .sx About 2.8 Ma ago here , pollen of woody taxa requiring year-round moisture ( e.g. , Engelhardtia , Hamamelis , Nyssa and Symplocos ) declined to near-extinction , and pollen of taxa characteristic of present-day shrublands around the Mediterranean Basin ( Cistus , Olea , Phillyrea , Pistacia , Quercis ilex -type ) became a constant component of the sediments [62] .sx However , it must not be forgotten that adaptation to summer drought is a secondary evolutionary feature that developed in plants already adapted to seasonal drought ; the adaptation necessitated a shift in the time of seedling germination and establishment from late spring/early summer to late winter/early spring , together with concomitant adjustments in the timing of flowering and seed maturation [8] .sx Hence the presence of 'mediterranean' genera in the fossil record does not necessarily imply the existence of a mediterranean climate .sx Arbutus , Chamaerops , Juniperus , Pistacia and Quercus coccifera , for example , characteristic members of present-day shrublands in southern Europe , are recorded from European sites back to the early Oligocene , when they were part of the developing Madrean-Tethyan Geoflora [63] .sx Spread of non-forest plant communities late in the Tertiary Period occurred also at high altitudes , as mountain ranges were thrust up above the attainable treeline .sx Documentation of this spread of herbaceous plants relies largely on the pollen record , and is consequently rather non-specific .sx However , it is clear both from the pollen record and from present-day distribution patterns that the colonizing species of these newly-created habitats were drawn both from local sources and from more distant ones .sx Thus pollen analyses from the high northern Andes [43] show developing montane and alpine floras in the Pliocene Epoch that were derived mainly from neotropical elements , but drew in some North American taxa as well ( Alnus , Aster , Cruciferae , Ericaceae , Erigeron , Sambucus and Viburnum ) and at least one Antarctic one ( Lagenophora) .sx Present distribution patterns of alpine plants on the East African mountains [9] similarly indicate an input of exotic genera in the past - from southern Africa and from temperate regions of the northern hemisphere .sx Adaptation to conditions of extreme cold almost certainly occurred in these habitats and not in high-latitude ( arctic ) ones .sx Tundra vegetation dominated by herbaceous plants was almost non-existent before the Pleistocene Epoch , with boreal forest and spruce parkland reaching north to the shores of the Arctic Ocean [64] .sx At a few sites only is there pollen evidence of dwarf-shrub tundra in the Late Pliocene :sx on Seward Peninsula , Alaska [39] , and in the Kolyma lowlands of north-east Siberia [65] .sx 9.4 THE TERTIARY VEGETATION OF WESTERN NORTH AMERICA .sx