Ordovician palaeogeography and climate change

New palaeogeographical reconstructions for the earlier Ordovician (480 Ma), and later Ordovician (450 Ma) integrate revised longitude-calibrated palaeomagnetic reconstructions and the inclusion of synthetic plate margins within the now-vanished oceanic areas. There are substantial published differences from the previous placing of some of the continents and terranes in Asia; for example, Siberia and Gondwana have previously been placed at varied distances and relative positions in relation to the Kazakh terranes, South and North China, and Tarim. But there are only minor changes for most of the world, particularly in the North American and European areas. The global distributions of benthic trilobites and brachiopods within faunal provinces and their changes through the Ordovician are plotted, including the new term Cathay-Tasman Province for some pan-equatorial brachiopod faunas from China and Australia, and key sites and provinces are shown on the revised maps. The 30 Myrs from 470 to 440 Ma (mid Ordovician to early Silurian) saw some of the most varied and changeable climates of the whole Phanerozoic culminating in the ‘Hirnantian’ ice age. Those changes in turn much affected the rates of evolution of many benthic and pelagic animal groups which were driven by both biological and environmental causes. Global cooling during the Ordovician was a prime factor by reducing sea surface temperatures which challenged life to evolve faster and more substantially than before. That cooling was driven by decreasing atmospheric CO2, for reasons that are not fully resolved, but probably included reduced sourcing (reduced continental arc activity) combined with increased silicate weathering due to the advent of land plants and perhaps the progressive exhumation of low-latitude collisional arcs. Since long-term CO2 sinks are largely controlled by palaeogeography, the general increase in the GR Centennial volume 2021 (accepted) 2 concentration of continents in the tropics during the Ordovician increased the overall global weathering.


Introduction
The Ordovician has long attracted the interest of palaeogeographers because there was a particularly large number of diverse continents and terranes in those times which were well dispersed in equatorial and southerly latitudes (Figs. 1 and 2). However, the majority of the previously published research has been concerned with today's Europe and North American areas, and the remaining three-quarters of the world have been less intensively studied. No in situ Ordovician oceanic crust is preserved, and thus the principal quantitative tool for establishing the ancient geography has been through palaeomagnetic studies, but those only indicate the palaeolatitudes and rotations of the former continents and terranes, not their longitudes, and the original primary magnetic signature has in many places been overprinted (reset) by subsequent tectonics. The margins of all the terranes can also be defined by analysis of their separating sutures, which indicate subduction, obduction, or strike-slip activity at the time of closure, but they do not place the terranes' positions on the globe in relation to their neighbours until that suturing occurred. Although more qualitatively than quantitatively based, the distributions of the differing benthic faunal assemblages, for example of brachiopods and trilobites, have often proved effective indicators of relative palaeolongitudes as well as latitudes. Fortey and Cocks (2003) reviewed the shallower-water benthic faunas helpful in the placing of continents in the Ordovician and Silurian, but that account was written before the technical advances mentioned in the following paragraph, and thus those provinces and their changing distributions from the beginning to the end of the Ordovician are revised here. All the benthic and planktonic animal phyla underwent a massive evolutionary radiation during the Ordovician, known as the Great Ordovician Biodiversification Event (GOBE), which was reviewed by Webby et al. (2004) and is briefly discussed in Section 4 below.
Fresh insights have transformed our understanding of pre-Mesozoic palaeogeography since 2008, when the plume generation zone reconstruction method was identified as useful in placing some of the major continents in longitude through the occurrences of large igneous provinces (LIPs) and kimberlites within them (Torsvik, 2019;Torsvik and Cocks, 2017;Torsvik et al., 2008;. In addition, the generation of mathematically objective and kinematically coherent successive synthetic ocean models has provided a base for conjecturing where the plate boundaries lay under many of the oceans, thus making much more realistic Palaeozoic maps for much of the Earth (Domeier and Torsvik, 2014;Domeier, 2016;. We have constructed new global maps within a revised GPlates (Boyden et al., 2011) framework for the earlier Ordovician (near the end of the Tremadocian Stage) at about 480 Ma ( Fig. 1), and the later Ordovician at 450 Ma during the Katian Stage (Fig. 2), shortly before the advent of the end-Ordovician Hirnantian global cooling. During those thirty million years the Iapetus Ocean narrowed and the Rheic Ocean widened in today's western hemisphere; and the many terranes mostly now in Asia and southern Europe performed a complex dance round the northern border of the immense Gondwanan continent and to the then east of Siberia. Figs 1 and 2 also show representative sites of the shallower-water brachiopod benthic faunal provinces.
Those provinces and their positions on the globe are revised in broad terms and discussed below in Section 5.
The Ordovician saw some of the most varied climates and sea level variations of the whole Phanerozoic (Fig. 3). At the beginning of the Ordovician the Earth was very warm with tropical sea surface temperatures of around 45°C or more, and by ~450 Ma sea levels were so high ( Fig. 3d) that they have only since been exceeded in mid-Cretaceous times. At the Ordovician's end at 444 Ma there occurred one of only three glacial episodes known in the most recent half billion years but it is noteworthy that the 'Hirnantian' glaciation lasted only for perhaps less than a million years: a sharp contrast to the 40 Myrs Late Palaeozoic ice age (Carboniferous-Permian) and the End Cenozoic ice age, first recorded by glaciations in Antarctica from about 34 Ma and peaking from ~2.7 Ma in the current Plio-Pleistocene glaciation (which has not yet ended). The 'Hirnantian' glaciation is shown in quotation marks throughout this paper since it is probable that the glaciation commenced at some time in the latter half of the preceding Katian Stage in a few places, as well as icecaps continuing on into the early part of the Silurian, certainly in Brazil. The description and possible causes of those Ordovician climate fluctuations are discussed in Sections 6 and 7.

Outline
There is only a brief summary heremuch more can be found in Torsvik and Cocks (2017).

The North Atlantic area, Laurentia, and Baltica
The Iapetus Ocean, which lay largely between the continents of Laurentia,Baltica,and Gondwana (Figs. 1,2), was over 4,000 km wide at the start of the Ordovician at 487 Ma. The ocean extended northward over the Equator between north-east Laurentia  Laurentia was a large continent, comprising most of the USA, Canada, Greenland, much of Mexico, Scotland, and part of Ireland, a high proportion of whose craton was flooded by shallow seas during most of the Early Palaeozoic. Its equatorial position made it warm enough to host many carbonate sediments and bioherms, with thicker, deeper-water origin sediments including turbidites in the surrounding and less stable tectonically-active continental shelves.
Because of the passive margins which surrounded it on all sides, Baltica underwent little tectonism during the entire Ordovician and also contained many carbonates, although those limestones were initially of cooler-origin (Jaanusson, 1973), when the continent was at a much higher southern latitude ( Fig. 1) than later, when it neared the Equator (Fig. 2). In contrast, the much more tectonically active microcontinent of Avalonia included substantial volcanism, which peaked at around 455 Ma, and the substantial Welsh Basin within Avalonia deepened throughout the period and was largely filled with many thick turbidites often interbedded with volcanics (reviewed in Brenchley and Rawson, 2006).
The Iapetus contained many oceanic island arcs, but by the end of the Ordovician at 444 Ma it was much narrowed and most of those volcanic islands previously within the western part of the ocean had become accreted to Laurentia within the Appalachians of North America and those in the eastern sector to the Caledonides of north-western Europe to the north-west of Avalonia and Baltica. Many of those arc terranes were reviewed by Domeier (e.g. 2016) whose papers we have used in the changes to our previous published maps. Offshore from Avalonia and separated from it by the Tetagouche-Exploits back-arc basin, was the composite microcontinent of Ganderia (sometimes termed Western Avalonia), which included Maine, Tetagouche (Miramichi), and various Newfoundland units today on the American side of the Atlantic; as well as Tramore, Bellewstown, Anglesey, and others which are now parts of north-west Europe (Pothier et al., 2015).

Gondwana and its periphery
As reviewed by Torsvik and Cocks (2013)

Background
Apart from the large continental plate of Siberia, whose Palaeozoic positions and faunas were summarised by Cocks and Torsvik (2007) and whose palaeomagnetic inversion and progressively changing palaeolatitudes as it drifted northwards have been widely accepted for many years, there is little unanimity on the divisions and development of the many terranes and continents which made up most of the rest of central and eastern Asia during the Ordovician. As well as the Siberian craton itself, there are numerous terranes, many with Cambrian or Precambrian cores, which were independent entities during the Ordovician. Some were substantial continents during the Ordovician, including the closely-related South China and Annamia (Indochina), North China (including the Sino-Korean Peninsula), and Tarim (today within south-west China), and others which were smaller  which not only occupy much of Kazakhstan, but also Kyrgyzstan, Uzbekistan, Tajikistan, and adjacent countries, and which extend eastwards for a relatively short distance into south-western China.
Because so many different definitions and names have been used for the numerous Kazakh terranes, we have constructed Fig. 4, each part of which has a modern base-map, to help orientate the reader. Fig. 4a is essentially the same as Torsvik and Cocks (2017, fig. 3.7) but modified to denote those terranes, such as South Tien Shan, whose rocks are entirely of post-Ordovician (largely Devonian) ages and which can thus be discounted in the present discussion.  fig. 5), which was largely unchanged from Şengor et al. (1993). Popov and Cocks (2017), discussed further below, largely followed the modern-day boundaries of the Şengor et al.
As well as those three relatively recent reconstructions discussed further below, numerous other authors, including Golonka (2007), Metcalfe (e.g. 2011), Wilhem et al. (2012), Li et al. (2018), and those in the book edited by Kröner (2015), have also presented different and variable versions of how the Altaid fragments were distributed during the Lower Palaeozoic.
Unfortunately, particularly when attempting to synthesise the palaeogeography of the whole Ordovician globe, yet many more papers simply omit the Kazakh terranes altogether or depict them as a single entity. In addition, various authors have postulated that the continent of Kazakhstania, which did eventually incorporate the majority of the Kazakh terranes, existed as an entity during the Ordovician. For example, Zhao et al. (2018, fig. 9) show only 'Kazakhstan' in their Cambrian to Ordovician (500-460 Ma) reconstruction and linked it to Siberia with little discussion, as well as placing a substantial Paleo-Asian Ocean between those two continents and Gondwana, which they considered as including North and South China as well as Alex, Qaidam, and Tarim: an interpretation which can only be wrong. After detailed analysis of the many different local rocks throughout the region, Popov and Cocks (2017, fig. 15) demonstrated that Kazakhstania was only united to become a substantial continent from Devonian times onwards, and thus the name should not be used for periods before then.  (Isozaki et al., 2010). The whole region was described and reviewed by Cocks and Torsvik (2013), but here we further analyse and compare the three differing palaeogeographies of Şengor et al. (2018), Popov and Cocks (2017), and Domeier (2018).

Şengor et al. (2018)
A milestone paper was by Şengor et al. (1993), and its essentials were repeated with only minor changes for the Kazakh region in several subsequent papers including Şengor et al. (2018). Their scenario (summarised here in Fig. 5a) shows a single enormous Kipchak Arc curving all the way from Baltica to Siberia; an island arc which included many of the relatively small microcontinents and terranes now making up the Kazakh Terranes and adjacent areas of central Asia. The palaeolatitudes which they showed for Baltica and Siberia at either end of the arc were originally derived (within errors) from those of Torsvik et al. (1992). The Şengor et al.
interpretation was heavily influenced by their assessment of the history of the large and obvious orocline in Kazakhstan which had been noted for many years; however, that orocline was more fully investigated by Abrajevitch et al. (2007), who, by using a substantial amount of then new palaeomagnetic data from the widespread and abundant Devonian volcanics, established that the orocline was entirely formed during the Devonian. Also after analysis, Fortey and Cocks (2003) found that the Ordovician benthic trilobite (particularly the unmistakeable Taklamakania) and brachiopod faunas show a close relationship of the Kazakh terranes to Gondwana and South China and had no links with either Siberia or Baltica. Thus it can be reaffirmed here that the Kipchak Arc scenario can definitely be discounted. In contrast, the reality of the extensive Tuva-Mongol Arc with its many terranes today to the south-west and in the Ordovician to the north of Siberia (Fig. 5a), also originally postulated by Şengor et al. (1993), is a probable scenario for those areas.

Our compromise conclusions
In contrast to , we conclude here from all the geophysical and faunal evidence that the Boshchekul-Chingiz terrane group and the Siberian Plate must have been quite separate from each other, and our new reconstructions (Figs 1 and 2) show oceanic separation of over 1000 km between Siberia and the other terranes and continents to its east. However, we have followed Domeier (2018) rather than Popov and Cocks (2017) in placing the Kazakh terranes to the west of North China rather than to the east, since the faunal data previously used by Popov and Cocks is suggestive rather than objectively definite, and Domeier's continental placings have more kinematically consistent geophysical evidence to support them.

Ordovician evolutionary radiations of the biota
The term GOBE was introduced in the book edited by Webby et al. (2004), to which many individual experts in all the early Palaeozoic major animal and plant groups contributed: its conclusions have since been used and amplified by many subsequent workers (e.g. Servais and Harper, 2018). It is noticeable from that volume that, although every organic group did indeed diversify greatly during the Ordovician, most of those evolutionary peaks were not simultaneous for all the different groups, both benthic and planktonic, during the period or even within the various subgroups (Orders etc) within each phylum.
The definition of GOBE has varied between authors. Biodiversity patterns differ between various continents, various animal groups and the kind of statistical analysis used. Stigall et al.

The changing distributions of benthic faunal provinces
Since the pioneering works of Whittington (1966) on trilobites and Williams (1969) on brachiopods, as reviewed by Fortey and Cocks (2003), it has been known that the distribution of the various benthic provinces (Fig. 6) often reflect the contemporary Ordovician palaeogeography, particularly at the margins of the larger oceans. The opportunity is taken here to plot some of the key brachiopod provincial sites on our new maps (Figs 1, 2). However, each benthic phylum has different faunal provinces, as upgraded in the book edited by Harper and Servais (2013), particularly the analysis of Harper et al. (2013) for the brachiopods. As a generality the distribution of both benthic and planktonic faunal provinces are subparallel to the palaeolatitudes, but by no means entirely, probably indicating that the distribution of many of the provinces were much influenced by the prevalent colder or warmer ocean currents and gyres of Ordovician times. The distances between some of the major continents, particularly near the old Equator, were sufficiently wide enough so that separate shallower-water benthic provinces were hosted on some of the different continental margins at the same lower latitude. The development of the various provinces is now reviewed at successive times.

Mediterranean Province
In the earlier Ordovician, Gondwana was so large (Fig. 1) that the Mediterranean brachiopod Province (which largely coincided with the calymenoidean-dalmanitoidean, sometimes termed Neseuretus, trilobite Province) was widely developed on the fringes of its margins at higher and intermediate latitudes (Fig. 6a). The province has two principal aspects, whose sites are separately distinguished on Fig. 1; firstly, the higher-latitude faunas with their distinctive large inarticulated lingulide brachiopods, notably Lingulobolus brimonti, Lingulobolus hawkeii, Pseudobolus, salteri, Ectenoglossa leseueuri, and Lingulepis crassipixis, whose distributions primarily lay within northern Africa, and the then adjacent polar parts of Gondwana, notably Iberia, Armorica, and Bohemia (Torsvik and Cocks, 2011, fig. 6).
The second aspect of the Mediterranean Province, which covered a more widespread area, (Fig. 6a), is less eye-catching but still includes distinctive brachiopod faunas including assemblages initially dominated by Prantlina, Ranorthis, Nocturniella, and others during the Floian and Dapingian, and in slightly later (Darriwilian) times by Tissintia and Tafilaltia. These faunas occur in South America and the Middle East as well as in slightly lower latitudes in Avalonia (see 'Celtic' Province below), and some parts of the Armorican Terrane Assemblage such as France and Bohemia. In addition to these and over a similar wide latitudinal belt largely between 30° and 60°S, there were also communities of varied composition and diversity.
Variably diverse Ordovician shallow-marine benthic faunas, such as the brachiopods and trilobites described in the volume edited by Benedetto (2003), occur in the island arcs now in Argentina which fringed south-western Gondwana. Also abundant in some mid-shelf assemblages was the large porambonitacean brachiopod Yangzteella, which was originally described from South China, but is also known from southern Turkey, Karakorum, and Iran: however, those Yangzteella Community sites can only loosely be included within the Mediterranean Province: they also lived in rather lower latitudes.
Included as well in those Mediterranean Province more diverse regions are those with assemblages of the 'Celtic' Province. That province was originally defined by Williams (1969) to include two Early Ordovician faunas in Anglesey and south-eastern Ireland, which are both very comparable to the fauna now revised from south-west Wales (Cocks and Popov, 2019 although there seem to be few genera in common and those few appear to reflect latitudinal rather than geographical similarities. Thus Celtic is recognised here as a sub-province, perhaps comparable in status to the Yangzteella faunas mentioned above, particularly since there are no distinctive particular brachiopod genera found in all of its sites which would normally characterise a province, although some endemic genera are known from many of the individual Celtic terranes.

Baltic (or Baltica) Province
This province, based on the distinctive benthic faunas of many phyla identified by numerous authors in classic monographs since Linnaeus in the eighteenth century nineteenth century, was largely centred around today's Baltic Sea area ( Fig. 6a) but extended eastwards to the Urals and north-eastwards to Nova Zemlya, and was bounded to its modern south by the Trans-European Suture Zone (apart from the Holy Cross Mountains of Poland, which is to the south-west of that zone and contains characteristic Baltic trilobites and brachiopods). As summarised by many authors, for example Öpik (1935). Whittington (1966), Williams (1969), and Fortey and Cocks (2003), the province was particularly distinctive in the Early Ordovician, notably in its megistaspinid trilobites and the brachiopods Lycophoria (which is occasionally abundant to the extent of being rock-forming and is also the only genus known within its family) and many endemic clitambonitoideans. That distinctiveness reflected the relative isolation of the Baltica continent during the later Cambrian and earlier Ordovician ( Fig. 1) before its northward migration to lower palaeolatitudes (Fig. 2).

Laurentian Province
In the Early Ordovician of the USA and Canada, many of the abundant and diverse previously termed Ozarkian and Canadian (now Tremadocian) brachiopod faunas were largely endemic.
These faunas were originally described by Ulrich and Cooper (1938), although they are in some need of modern revision, and were succeeded by equally diverse brachiopod and trilobite faunas locally known as Whiterock, for example by Ross (1970). In addition, the widespread bathyurid trilobite Province dominated both Siberia and Laurentia in shallower waters, but the deeper-water trilobite faunas in the same regions have been termed the olenid Province, which is less significant when analysing the continental palaeogeography since the ambient sea temperature is always globally more equable in the deeper parts of the oceans. Both the bathyurid and the dikelokephalinid provinces lay in similar tropical latitudes but were far enough apart to have hosted largely different benthic faunas, even though there is a bathyurid record from a single site in North China (Fortey and Cocks, 2003). The Laurentian Province (Fig. 6a) hosted the largest benthic diversity of the period, not just trilobites and brachiopods, but many representatives of several other phyla, including echinoderms and corals (Harper and Servais, 2013).
We include within the Laurentian Province the so-called Toquima- Table Head Province, originally erected by Ross and Ingham (1970) and modified by Neuman and Harper (1992).
However, that consists essentially of the deeper-water marginal faunas found offshore from some of the main shallower-water benthic Laurentian Province, and is therefore recognised as only having subprovincial status here.
The Cuyania (or Precordillera) Terrane of north-western Argentina (Fig. 1) also hosted trilobites and brachiopods of largely North American aspect at that time (Benedetto, 2003), very different from the medium-latitude peri-Gondwanan faunas of the rest of South America adjacent to Cuyania today. That indicates its lower Early Ordovician palaeolatitude as well as its relative proximity to the south-western USA.
Although Siberia is grouped here as within the main Laurentian Province at this time, its trilobite and brachiopod benthic faunas may only be differentiated as a Siberian Subprovince in the Early Ordovician, partly because they were so much less diverse there than in Laurentia itself.
Nevertheless, we show that subprovince from the rest of the Laurentian Province separately in Fig. 6a. Although there were some distinctive shallow-water Siberian endemic trilobites (Fortey and Cocks, 2003), heralding the even more distinctively different faunas seen there in the subsequent parts of the Ordovician, there were no endemic earlier Ordovician brachiopods.
Siberia also shared the olenid trilobite 'Province' with Laurentia, but again that merely reflected the deeper-shelf biota.

Cathay-Tasman Province
The most characteristic trilobites of the Kazakh terranes, North China, South China, and the Australian sector of Gondwana are all representatives of the dikelokephalinid trilobite Province.
In contrast to the fauna within the Laurentian Province, the Australian pan-tropical shores of Gondwana hosted both the dikelokephalinid trilobite Province and the Cathay-Tasman brachiopod Province in the same equatorial palaeolatitudes. The latter is a new term, coined here to recognise the brachiopod faunas ( Fig. 6) found in both the Gondwanan margin, for example in Tasmania (Laurie, 1991) and in China (which was historically named Cathaysia). However, earlier Ordovician brachiopod diversity was often locally very low; for example, at some shallow-water localities in Australia and western Malaysia only numerous specimens of the plectambonitoidean Spanodonta occur in a virtually monospecific community , although the species diversity of both brachiopods and trilobites rose soon afterwards to reflect the GOBE radiation (Webby et al., 2004). A few Ordovician benthic faunas have been found in the continent of Amuria, particularly in the substantial Khinggan area (then part of the Khanka-Jiamusu-Bureya sector of Amuria) which today spans the junction between Siberia, China, and Mongolia. However, the described faunas from Khinggan are rather sparse and, since they were identified with few rather poor illustrations over a hundred years ago, are much in need of modern systematic revision before their provincial affinities can be confidently assessed.

Later Ordovician
Particularly during the approximately 30 Myr between the end of the Tremadocian at 478 and the middle of the Katian at about 450 Ma, not just the brachiopods and trilobites but all the benthic faunas evolved and expanded in diversity very dramatically as parts of the GOBE radiation (Webby et al., 2004). For example, the bivalve molluscs were insignificant during the Cambrian, but radiated sharply, firstly in the temperate latitudes of Gondwana soon after the start of the Ordovician before spreading from there to lower latitudes, where they later became more generally diverse (J.C.W. Cope and J. Kříž in Harper and Servais, 2013).

Mediterranean Province
Gondwana remained so large that the Mediterranean brachiopod Province was still developed on the fringes of its southern margins (Fig. 6b), but now without the previously-distinctive large inarticulated brachiopods or widespread quartzites. However, the trilobites remained grouped within the calymenoidean-dalamanitoidean Province, albeit with different genera (Fortey and Cocks, 2003). In general, only clastic rocks continued to be deposited at those higher latitudes, which was a major factor in keeping the diversities of most shelf benthic communities relatively low. However, for example, in Bohemia (then still an integral sector of Gondwana) Havlíček et al. (1994) identified 38 brachiopod genera in their later local Beroun stage (late Katian), of which over half were endemic to the province. During the later Katian, the Boda warming Event was reflected even in the higher-latitude parts of Gondwana, such as Morocco, which were host to relatively small bioherms of cooler-water origin largely formed of bryozoans: which are locally very conspicuous since they are the only carbonates seen in the whole thick Ordovician succession there (Fortey and Cocks, 2005).

Baltic-Anglo-Welsh Province
With the dwindling of the Tornquist Ocean, Avalonia became progressively closer to Baltica at its eastern end, and the two continents merged obliquely very close to Ordovician-Silurian boundary time at 444 Ma  the Baltic assemblages as within the Anglo-Welsh-Baltic Province, and we follow them here (Fig. 6). The changing distributions of the trilobites in these areas were also comparable to the brachiopods (Fortey and Cocks, 2003).

Siberian Province
By comparison with the other large continents, Siberia became progressively more isolated as the Ordovician progressed and as it drifted steadily northwards to straddle the Equator, and thus its benthic faunas had become progressively more endemic by the late Ordovician so as to merit full provincial status, especially for the trilobites. From the Sandbian to the end of the Ordovician, endemic trilobites were particularly distinctive, particularly emphasised by the Subfamily Monorakinae, eight out of ten of whose genera are confined to Siberia, and which formed a large part of the shallower-water monorakine-cheirurid-illaenid Association (Ebbestad and Fortey, 2019). Their distribution confirms that the New Siberian Islands in today's Arctic Ocean were part of Siberia, as concluded on other grounds by Cocks and Torsvik (2007), and their presence in the Khinggan (or Xing'an) area also supports the integration of that block within Siberia as suggested by . In addition, Ebbestad and Fortey (2019) confirmed that the trilobites in the Arctic-Alaska Chukotka terrane were typically monorakine and therefore also within the Siberian Province, which is why we have included a queried line on Fig. 2 to indicate that Arctic-Alaska Chukotka may have been nearer Siberia than previously shown by Cocks and Torsvik (2011), although the brachiopods from that terrane require revision. As also noted by Cocks and Torsvik (2011), monorakine trilobites have been found in the Farewell terrane of Alaska, although since the Ordovician location of that enigmatic and relatively small terrane is poorly constrained, it is not shown on Figs 1 and 2.
Ebbestad and Fortey (2019) also pointed out that the deeper-water trilobites in the Siberian area were much more cosmopolitan, and are comparable with the Scoto-Appalachian 'Province' originally identified by Williams (1969). However, the shallower-water brachiopods and trilobites in the Siberian area still had a much lower diversity than in most other parts of the world and the relatively few brachiopods are notably represented by largely cosmopolitan genera, which also emphasises the difficulties that their spat must have had in successfully crossing the very wide oceans which surrounded the Siberian continent (Fig. 2).

Cathay-Tasman Province
South China, North China, the peri-Australian pan-tropical shores of Gondwana (including Sibumasu, which extended into today's south-western China), and nearby areas (Fig. 2) continued to host largely equatorial brachiopod and trilobite shallow-water benthic provinces with only a small minority of genera in common with the Laurentian Province. This included the  (Fortey, 1997). The brachiopod faunas from the various Kazakh terranes analysed by Popov and Cocks (2017) also include some Sandbian to Katian genera otherwise endemic to North and South China, but those faunas are very different from the sparse faunas known from Siberia, which also straddled the Equator by that time (Fig. 2). Tarim, which is not firmly positioned palaeomagnetically, hosted typical Cathay-Tasman brachiopods, including the Altaethyrella-Schachriomonia Assemblage described by Sproat and Zhan (2019). Percival (1991) monographed the extensive Late Ordovician brachiopods found in New South Wales which lived within the volcanic arcs immediately offshore of Australia in those times, many of which genera are also only found elsewhere in the Kazakh terranes.

Laurentian Province
By the later Ordovician (Fig. 2), the Iapetus had become much smaller than before, with the Rheic Ocean much wider and thus the Avalonian faunas underwent progressive interchange with those in both Baltica and Laurentia. During the latest Sandbian to earlier Katian there was a major marine transgression after which Sproat and Jin (2017) identified the dominance of the relatively diverse Scoto-Appalachian Fauna (not here recognised as a benthic province) in the deeper waters off the eastern margin. There is also a widespread but less diverse epicontinental fauna (Fig. 6b) in much of shallower waters covering the craton, often known as the Richmondian brachiopod and coral faunas, which have sometimes been termed a province. Those Richmondian faunas remained distinctively different from their European contemporaries until almost the end of the Ordovician, when the majority of them became extinct, although from Katian times onwards they had been joined by an increasing proportion of immigrants from Baltica.
Before the end-Ordovician ice age, during the later Katian at around 447.5 Ma (Fig. 3a), there was relatively brief global warming termed the Boda Event (Fortey and Cocks, 2005), in which extensive bioherms developed in the low latitudes of Laurentia (particularly north-eastern USA and south-eastern Canada), as well as in Baltica (Boda itself and elsewhere in Sweden, Norway, and Estonia), and Avalonia (Kildare, Keisley, Portrane, and others), both of which had drifted northwards into much lower latitudes comparable to Laurentia (Fig. 2). Numerous endemic brachiopods, trilobites, and other phyla are known from within each individual bioherm and its immediate surroundings.

The Hirnantian finale
Near the close of the Ordovician at about 445 Ma, the 'Hirnantian' Ice Age (Fig. 3a) much affected the planet but for substantially less than a million years, although there may have been some minor glaciation at the highest latitudes of Gondwana in the late Katian which certainly continued into the earliest Silurian in Brazil (Grahn and Caputo, 1992) (Fig. 7). But within that relatively short time two different extinction phases showed a massive faunal turnover, notably including the demise of the distinctive Richmondian brachiopod faunas in Laurentia (Harper et al., 2013). However, the palaeogeography did not undergo any substantial changes and, in contrast to the earlier half of the Ordovician, there were few active orogenic areas.
A direct result of the 'Hirnantian' glaciations was better oceanic circulation which led in turn to increased oxygenation to lower depths within the oceans, thus enabling some more enterprising pioneer benthic forms, for example the brachiopod Hirnantia and its community associates, to colonise deeper depths than hitherto on the continental shelves over much greater areas of the globe. The diversity of that Hirnantia Fauna varied considerably from just a very few brachiopod genera including Hirnantia itself as well as Eostropheodonta and Cryptothyrella and the trilobite Mucronaspis, to over twenty genera, which were mostly brachiopods, but there were also sometimes a few representatives of other phyla (Rong and Harper, 1988). Although some authors have divided the Hirnantia Fauna into two separate provinces, we do not follow them and conclude that the variable diversity and detailed composition of the fauna depended on local factors, including variable depths and palaeolatitudes rather than relative palaeogeography. Thus the Hirnantia Fauna seems best considered as cosmopolitan rather than provincial.

Ordovician Climates and Biodiversity
Much has been published on the possible causes of the Ordovician climatic and evolutionary changes, some of which have been included in Fig. 3, including variations in biodiversity, sea surface temperatures (SST), atmospheric CO2 and O2, continental arc-activity, global sea level and continental connectivity expressed in terms of the width of some important oceans.
Plate tectonics plays an intricate role in shaping the long-term climate by controlling the distribution of continents and oceans (palaeogeography), mountain building, arc-volcanism, topography and weathering. Plate tectonics also has a major effect on the hydrosphere because the variation in seafloor spreading is the most important driver of sea-level rise and falls. In turn, that also affects the biosphere and biodiversity, which is strongly influenced by continental flooding, intercontinental connectivity, the link between latitudes of the various continents and their overall average temperature, habitat, and oceanic-atmospheric circulation. It is perhaps naïve to assume that radiations in biodiversity can be linked to any single specific cause, but for some time there has been much agreement (e.g. Trotter et al., 2008) that changes in the marine biosphere during the Ordovician (Fig. 3a), can be at least partly related to a steady falling in SSTs, from ~45 o C in the Early Ordovician to near modern equatorial SSTs (27-32 o C) by the mid-Ordovician (Fig. 3b). Cooler oceans would store more dissolved oxygen and Edwards et al. (2017) argued for a strong temporal link between GOBE and O2 concentrations. The O2 level in the early Ordovician was about 10-13%, but then there was a sharp rise from the early Darriwilian up to about 24% by the mid-Katian (red curve in Fig. 3c). That was followed by a major reduction in O2 immediately before and during the Hirnantian. Edwards et al. (2017) therefore argued that oxygen levels played an important role in regulating Ordovician biodiversity; we find a fair (negative) correlation between O2 levels and SSTs (r=-0.74) whilst there is a respectable positive correlation between atmospheric O2 and the diversity of global articulate brachiopods (r=0.86).

The link between palaeogeography and biodiversity.
It is perhaps also naïve to consider only one parameter (e.g. temperature) as the chief cause for diversity changes. Variations in the geographical distribution and diversity of species are also affected by tectonic and magmatic pressures, for example from LIPs. As discussed by Stigall (2018), plate tectonic divergence causes physical isolation and can drive natural selection via allotropic speciation whilst habitat reconnections during plate convergence leads to competition between species occupying similar niches. Continental 'block-fragmentation' was tentatively correlated with diversity by Zaffos et al. (2017); however, although they observed peak Palaeozoic block fragmentation in the Mid-Ordovician, they identified no changes at all during Ordovician to Early Silurian times, which is surprising since that was a period with much major reorganization of the palaeogeography. The Iapetus closed progressively during the Ordovician (Figs. 1-2) and from the mid-Silurian, Avalonia and Baltica collided with Laurentia to form Laurussia. The Rheic Ocean, at the expense of the Iapetus, widened considerably during the Ordovician and reached a width of about 7000 km by the Early Silurian (Fig. 3e). We notice that the width of the most important Ordovician oceans, such as the Iapetus, Rheic, but also AEgir (separating Baltica from Siberia) were all in the order of 3000-4000 km at around 465 Ma and at the main GOBE initiation (Fig. 3a,e).

What led to cooler Ordovician oceans?
When discussing Ordovician temperature fluctuations there are two main questions; firstly, what led to the initial and progressive cooling during the Early Ordovician, and secondly what caused the more abrupt cooling that is contemporaneous with mass extinctions during the 'Hirnantian'?
CO2, the most important greenhouse gas in the Earth's atmosphere, regulates planetary temperature, and on a geological time-scale its changes have been principally controlled by plate tectonics and the consequent developing palaeogeography. Long-term magmatic-related degassing occurs at mid-ocean ridges, island arcs, continental arcs, and continental rifts. Zircons in the sedimentary record provide a way to track the past presence of continental arc systems, and temporal changes in continental arc-activity has been argued to play an important role in regulating long-term climate changes (McKenzie et al., 2016); in other words, high continental arc activity is linked to warm climate (greenhouses) and reduced arc activity explains icehouse climates. Subduction fluxes derived from full-plate models provide a powerful means of estimating plate tectonic CO2 degassing (sourcing) through time , and correlate well with zircon age frequency distributions. No reliable global full-plate model (Domeier and Torsvik, 2014) is available before 410 Ma (Devonian) and therefore we have rescaled and normalized arc-related zircon frequencies to subduction flux for the past 410 Ma and extended the normalized (1 = present day) zircon frequency curve to pre-Devonian times. In the Early Ordovician (Fig. 3d), we note peak arc-activity of ~1.7 (i.e. 70% higher than modern activity), after which there was a marked drop during the Floian-early Dapingian. However, the Ordovician is essentially characterised by a 35% reduction in continental arc-activity, starting at 480 Ma. Reduced plate tectonic CO2 degassing is compatible with a generally cooler climate during the Ordovician but cannot explain the abrupt temperature/diversity shift during the 'Hirnantian'.
Silicate weatherability and cooling can be explored by calculating the fraction of the total continental area that is located in the tropics (currently ± 23.5 o ), i.e. subjected to potentially warm and wet climates with high weatherability. During the Cambrian at around 510 Ma, 35% of the landmasses were located within the tropics, but maximum tropical exposure (Fig. 3d) is seen in the Ordovician between 460 and 440 Ma (45-46%, Table 1). That was followed by a dramatic reduction during the Silurian resulting in about 30% exposure of land in the tropics by the Early Devonian. The Ordovician therefore had the highest exposure of landmasses within the tropics, which potentially cooled the climate through silicate wreathing. Estimates naturally depend on plate models and effective weatherability would be lowered by high sea levels. Climate gradients also appear much different during most of the Ordovician with evaporites commonly found at tropical latitudes, and thus pointing to an arid equator with low silicate weatherability. Globally, this was the case for the entire Ordovician (Fig. 7), except perhaps for the Katian and Hirnantian, when low-latitude evaporites were mostly confined to Laurentia (Fig. 8).
Low atmospheric CO2 is a principal variable in controlling continental-scale glaciations (Royer, 2006), but the short-lived Hirnantian cooling event (~445 Ma) is paradoxical because of its apparent association with high atmospheric CO2 levels. However, the sun was 3-5% fainter then and the CO2 threshold for nucleating ice sheets at that time could have been 4-8 times (1120-2240 ppm) higher than pre-industrial levels of ~280 ppm (Gibbs et al., 2000;Herrmann et al., 2003;Royer, 2006;Lowry et al., 2014). GEOCARBSULF models atmospheric CO2 levels of 2536 ppm (450 Ma) and 2600 ppm (440 Ma), and it is perhaps interesting to see how our revised estimates of 'plate tectonic degassing' (continental arc-activity in Fig. 3d) affects the GEOCARBSULF model. Changes in seafloor production rates is an important time-dependent parameter in long-term climate models and named fSR in GEOCARBSULF, but subduction flux must equal the seafloor production rate to a first order. We use normalized arc-related zircon frequencies as a measure of subduction flux and as a proxy for fSR. If we recalculate atmospheric CO2 in GEOCARBSULF (all other parameters similar to those in Royer et al. 2014) we notice that early Ordovician CO2 levels are slightly higher, but end-Ordovician/early Silurian levels are reduced by ~200 ppm (450 Ma) and 350 ppm (440 Ma)354 ppm at 440 Ma (stippled black curve in Fig. 3b). The two models are quite similar but the Pearson correlation between sea surface temperatures and our revised estimates for atmospheric CO2 is slightly improved (r=0.94).

The end-Ordovician mass extinction: links with cooling or a Large Igneous Province?
Peak weatherability (at least theoretically) during the Mid-Late Ordovician can be estimated by other proxies for silicate weathering, e.g. strontium (Saltzman et al., 2014) and neodymium argued that the end-Ordovician mass extinction was linked to intense volcanism during the mid-Late Ordovician, but citing a so-called LIP event in South Korea (Ogcheon Volcanics) with reference to Kravchinsky (2012) who had assigned a rather imprecise age (480-430 Ma) for that intrusion in his LIP compilation. But the South Korean rocks in question are in fact Neoproterozoic (~750 Ma; Kim et al., 2006), and the end-Ordovician mass extinction cannot be linked to any known continental LIP .
Might the end-Ordovician extinction also have been triggered by an oceanic LIP which has since vanished through subduction? That was mooted by Jones et al. (2017) who documented unusually high concentrations of mercury in the late Katian and mid-Hirnantian of South China and the Hirnantian of Nevada, then part of Laurentia (Fig. 8), which they suggested might be the products of an undiscovered or a now-vanished (subducted) LIP. This was reinforced by Bond and Grasby (2020), who analysed deeper-water sediments, including the type Ordovician-Silurian boundary section at Dob's Linn, Scotland, then offshore from Laurentia, and found comparable levels of mercury. However, oceanic LIPs are limited in terms of the production of volatiles through contact metamorphism and/or crustal contamination, and environmental impacts must be considered much less severe than their continental counterparts . Thus, even if there was an end-Ordovician LIP, it is unlikely to have been more than a minor factor in causing the 'Hirnantian' glaciation.

The development of Ordovician climates
7.1. The climate before 460 Ma.
There is general agreement that the latest Cambrian and earlier Ordovician were relatively very warm (Fig. 3b); a warmth reflected in the massive carbonates widespread in many continents, some even in high paleolatitudes such as Sardinia. That warmth was matched by high atmospheric CO2 levels, and that was echoed in turn by the great variety and increased speciation not only of benthic phyla, particularly trilobites, brachiopods, and echinoderms, but also swimmers including cephalopods and conodonts (Harper and Servais, 2013 . 3b).
Although Rasmussen et al. (2016) concluded that the GOBE of Webby et al. (2004) was triggered by an emerging mid-Ordovician icehouse, there are no known mid-Ordovician glacial deposits from anywhere to firmly support any icecaps anywhere: for example, those glacial pavements previously reported from Morocco by Hamoumi (e.g. 2001) from the Sandbian of Morocco have been falsified, as summarised in Torsvik and Cocks (2017). Saltzman & Young (2005) had also postulated that there was glaciation as early as late Sandbian or early Katian (locally termed the Chatfieldian Stage) in Laurentia as reflected in a δ 13 C excursion in the Monitor and Antelope ranges of Nevada, but, although there may well have been some cooling at that time, there seems no proof of any icecaps or glaciations either near there or anywhere else.

Climate from 460 Ma (late Darriwilian) to 450 Ma (Katian).
It was during the Middle Ordovician, between 460 Ma (late Darriwilian) and just after 450 Ma (early Katian), that there were the highest sea-level stands in the Palaeozoic (Fig. 3d). Many parts of the old cratons were flooded and the resultant warm-water shallow seas were enhanced centres of evolution for the benthic faunas, particularly those at low latitudes such as around and over the partially flooded craton of the continent of Laurentia (Fig. 2), as well as around some of the many contemporary microcontinents such as the Chu-Ili Terrane of Kazakhstan (Popov and Cocks, 2021). Bergström et al. (e.g. 2010) identified a substantial δ 13 C isotope excursion at 453 Ma ( Fig. 3c), approximately the Sandbian-Katian boundary, which they named the Guttenberg Excursion (GICE), and which they documented in detail in many sections in Baltica and Laurentia. Somewhat prior to the GICE, in the later Sandbian at 454 Ma, there occurred one of the largest known volcanic explosions ever, probably at a now-subducted site under the North Sea, of calc-alkaline magmatism beneath Avalonia, and which caused massive and very widespread bentonites to be laid down. Since Avalonia was nearing Baltica at that time : those bentonites are termed the Kinnekulle Bentonite in Baltica. It was once thought that the extensive Millbrig Bentonite in eastern Laurentia was a product of the same eruption, but it is now known that the two are not of precisely the same age and have different geochemical composition (Haines et al., 1995), but nevertheless Millbrig confirms the roughly contemporary very active arc volcanism. Bergström et al. (2010, fig. 17) accurately located both the GICE excursion as well as the Kinnekulle Bentonite some 4 m below it in the Fjäcka section and the nearby Smedsby Gård borehole in the Siljan area of Sweden. What precise effect those volcanic outpourings had on both the GICE and the longer-term climate change is uncertain.

Climate from 450 Ma (Katian) to 444 Ma (end Hirnantian).
A relatively short warm period was named the Boda Event by Fortey and Cocks (2005), which peaked in the early late Katian at around 447.5 Ma and saw numerous bioherms in several continents as plotted on Fig. 8. Those bioherms are of two very different types, the 'normal' and often very substantial bioherms up to several hundred metres in lateral extent and largely made up of crinoids, calcareous algae, and stromatoporoids, with subsidiary corals and bryozoans, as well as epifaunal benthic brachiopods, trilobites, and other phyla. These major rock-forming structures were in marginally subtropical areas, notably in Europe (as far east as the Ural Mountains) and North America. The second category are cooler-water bioherms, usually less than 20 m wide and which are chiefly dominated by bryozoans: those are vividly seen as relatively small white carbonate structures which stand out in stark contrast to the surrounding thick and substantial darker-coloured clastic deposits, notably in the high cliffs of the Anti-Atlas Mountains of Morocco, and other regions of North Africa. Before the Boda Event was named, Boucot et al. (2003) had previously identified some 'Warm-Water' faunas in various previously colder places outside Laurentia during the Katian. However, by the time of the Boda Event the overall biodiversity was past its Ordovician maximum   (Fig. 3a).
After the Boda Event, the sea level remained high until around the end of the Katian (Fig. 3c), which increased the endemism within, for example, the brachiopods (Jin et al., 2014). Villas et al. (2002)  poorly constrained, and thus the glacial reality of some have been controversial, both in the interpretation of their sedimentology as well as in their dates. However, an important paper, which incorporated the much earlier pioneer results of Beuf et al. (1971), is the compilation from northern Africa by Ghienne et al. (2007), which demonstrated clearly the reality of a late Ordovician ice cap there. That cap appears to have extended to cover a large part of central Africa as well as South America, Arabia and adjacent parts of south-eastern Asia. There are also contemporary glaciogenic sediments in South Africa, notably the Pakhuis Tillite (I.C. Rust in Holland, 1981), although it is unknown whether or not the icecap there was connected with the main polar icecap and what proportion of the whole Gondwanan continent was covered by a continuous sheet of ice.
Icebergs from the polar ice sheet are known to have travelled at least 3,500 km northwards before generating dropstone deposits on the subtropical shelf of Baltica (now within Poland) after melting (Porębski et al., 2019); however, that does not indicate that the icecap extended anything like as far towards the Equator as the low latitudes in which Baltica was then situated, and the icebergs are almost certainly comparable to those seen near the coast of Newfoundland today.

Climate immediately after the Ordovician.
Although a few icecaps persisted in the Brazilian sector of Gondwana for much of the early Llandovery (Fig. 7), they had largely vanished from most of that vast continent very soon after the start of the Silurian. The melting ice and the slowly rising average global temperature caused the volume of water in the oceans to expand. The consequent rise in sea level caused major transgressions during the early Silurian in a great number of places, many documented in the volume edited by Landing and Johnson (2003). Those transgressions continued in many regions at least until the end of the Llandovery at 433 Ma.

Conclusions
We have combined newly generated maps with calibrated longitude and plotted on them revised benthic faunal distributions to produce palaeogeographical and biogeographical maps for the earlier and later Ordovician, with the boundaries of lands, continental shelves, and oceans updated from those in Torsvik and Cocks (2017). Largely following Domeier (2016;, the maps show the plate boundaries within most of the oceans apart from the immense Panthalassic.
Those plate boundaries are synthetic since no in situ Ordovician oceanic crust is preserved today, although the postulated boundaries are kinematically consistent with each other through time.
The new maps presented here supersede the reconstructions in our various previous papers (e.g. Cocks and Torsvik, 2007;Torsvik and Cocks, 2013;. The benthic faunal provinces defined by the brachiopods in the earlier and later Ordovician are also shown (Fig. 8), after assessment of the assemblages from many sites across the world, some shown in Figs 1 and 2.
Other reconstructions are reviewed, including Popov and Cocks (2017), who, from analysis of later Ordovician brachiopods of the Kazakh terranes, concluded that those faunas demonstrate progressively closer links between contemporary faunas in South China and nearby Australia (Fig. 5b); and Domeier (2018)  The Ordovician is known for worldwide tectonic activity and volcanism, major plate tectonic reorganizations and changes in palaeogeography, and wide oceans separating many of the major continents, which must have contributed to distinctive faunal provinces found in the marine benthos of the continental shelves. The Ordovician is also very special since virtually all the continents were located in the southern hemisphere (although some spanned the Equator) and the northern hemisphere was dominated by a single huge ocean named the Panthalassic. The Ordovician was also an exceptional period for biological and climatic changes, including the advent of land plants (which might have begun as early as the Late Cambrian; Morris et al. 2018), as well as the Great Ordovician Biodiversification Event (GOBE), which progressively increased the biota, particularly from the early Darriwilian onwards (Fig. 3a). The Early Ordovician Earth was a greenhouse planet with high temperatures (around 45 o C) and high atmospheric CO2 levels, but a 40 Myrs cooling trend culminated in the 'Hirnantian' glaciations in the high latitude parts of Gondwana, and the subsequent lowering of global sea level and the associated first Phanerozoic mass extinction at the end of the Ordovician period at 444 Ma. This is the only Phanerozoic mass extinction that cannot be linked to or explained by a contemporaneous LIP event (or an asteroid impact) and all other mass extinctions or anoxia events occurred during greenhouse conditions . A yet unidentified LIP or an oceanic LIP now vanished through subduction has been suggested to explain the end-Ordovician mass extinction but there is no evidence for that. Such a LIP would have triggered an abrupt global warming event superimposed on the cold climate at that time for which there is no evidence.
Although the GOBE was driven by a combination of biological and environmental factors, global cooling during the Ordovician must have been the prime factor, by reducing SSTs to temperatures that challenged life to evolve faster and more substantially than before. Cooling probably also caused the modelled increase in atmospheric O2. Temperature and oxygen levels are anti-correlated and both echo the diversity of global articulate brachiopods. Global cooling was probably driven by decreasing atmospheric CO2, which is most noticeably recognized from the earliest Ordovician to the Late Darriwilian in the GEOCARBSULF model. Although the reason for reduced CO2 levels is not fully resolved, it probably reflects a combination of causes in this extraordinarily dynamic period in Earth evolution. These include reduced sourcing (reduced continental arc activity in Fig. 3d) and increased silicate weathering due to the advent of land plants (Lenton et al., 2012;2016;Porada et al., 2016;Morris et al., 2018) as well as the progressive exhumation of low-latitude collisional arcs. Ultimately, long-term CO2 sinks are largely controlled by palaeogeography, and the general increase in the concentration of continents in the tropics during the Early Ordovician (Fig. 3d) appears to have increased the overall global weatherability.    (2018), and (c) Popov and Cocks (2017). Af, Afghan terranes. For discussion see text.    Figure 3a-d.