This special issue of Geological Magazine is dedicated to the memory of Dr Alan Gilbert Smith, Fellow of St John's College and Emeritus Reader in Geology at the University of Cambridge, who passed away on 13 August 2017 at the age of 80. I first met Alan at the 5th International Symposium on Eastern Mediterranean Geology in Thessaloniki, Greece, in spring 2004 and later on several occasions when I was working on the Cambridge Arctic Shelf Programme (CASP) in Cambridge. The palaeotectonic evolution of Greece was one of our common interests. Alan was one of the pathfinders in palaeogeographic research in the 20th century. Together with Sir Edward Bullard (1907–1980) and Jim E. Everett, he published the first computational approach in palaeogeography in their famous paper ‘The fit of the continents around the Atlantic’ (Bullard, Everett & Smith, 1965), which shows a very accurate geometrical fit of the circum-Atlantic continents using the early Cambridge University EDSAC 2 computer. Later, in a contribution in Nature entitled ‘The fit of the southern continents’, Smith & Hallam (1970) presented the first computer fit of the contour of the southern continents forming Gondwanaland. Worth mentioning also are his detailed palaeogeographical maps of the entire Earth, down to epoch level (e.g. Smith, Briden & Drewry 1973; Smith, Hurley & Briden 1981) and his work on the first three editions of A Geologic Time Scale (Harland et al.1982, 1990; Gradstein, Ogg & Smith 2005). Alan's great achievements in the Earth sciences have stimulated new ideas and had a huge impact on geological research, including palaeogeography.

Palaeogeography is the cartographic representation of the past distribution of geographic features such as deep oceans, shallow seas, lowlands, rivers, lakes and mountain belts on palinspastically restored plate tectonic base maps. It is closely connected with plate tectonics which grew from an earlier theory of continental drift and is largely responsible for creating and structuring the Earth's lithosphere. Today, palaeogeography is an integral part of the Earth sciences curriculum. Commonly, with some exceptions, only the most recent state of research is presented; the historical aspects of how we actually came to the insights which we take for granted are rarely discussed, if at all. It is remarkable how much was already known about the changing face of the Earth more than three centuries before the theory of plate tectonics, despite the fact that most of our present analytical tools or our models were unavailable then. Here, we aim to present a general conspectus from the dawn of ‘palaeogeography’ in the 16th century onwards. Special emphasis is given to innovative ideas and scientific milestones, supplemented by memorable anecdotes, which helped to advance the theories of continental drift and plate tectonics, and finally led to the establishment of palaeogeography as a recognized discipline of the Earth sciences.

Since the 1970s, numerous global plate tectonic models have been proposed to reconstruct the Earth's evolution through deep time. The reconstructions have proven immensely useful for the scientific community. However, we are now at a time when plate tectonic models must take a new step forward. There are two types of reconstructions: those using a ‘single control’ approach and those with a ‘dual control’ approach. Models using the ‘single control’ approach compile quantitative and/or semi-quantitative data from the present-day world and transfer them to the chosen time slices back in time. The reconstructions focus therefore on the position of tectonic elements but may ignore (partially or entirely) tectonic plates and in particular closed tectonic plate boundaries. For the readers, continents seem to float on the Earth's surface. Hence, the resulting maps look closer to what Alfred Wegener did in the early twentieth century and confuse many people, particularly the general public. With the ‘dual control’ approach, not only are data from the present-day world transferred back to the chosen time slices, but closed plate tectonic boundaries are defined iteratively from one reconstruction to the next. Thus, reconstructions benefit from the wealth of the plate tectonic theory. They are physically coherent and are suited to the new frontier of global reconstruction: the coupling of plate tectonic models with other global models. A joint effort of the whole community of geosciences will surely be necessary to develop the next generation of plate tectonic models.

Constructing palaeogeographical maps is best achieved through the integration of data from hotspotting (since the Cretaceous), palaeomagnetism (including ocean-floor magnetic anomalies since the Jurassic), and the analysis of fossils and identification of their faunal and floral provinces; as well as a host of other geological information, not least the characters of the rocks themselves. Recently developed techniques now also allow us to determine more objectively the palaeolongitude of continents from the time of Pangaea onwards, which palaeomagnetism alone does not reveal. This together with new methods to estimate true polar wander have led to hybrid mantle plate motion frames that demonstrate that TUZO and JASON, two antipodal thermochemical piles in the deep mantle, have been stable for at least 300 Ma, and where deep plumes sourcing large igneous provinces and kimberlites are mostly derived from their margins. This remarkable observation has led to the plume generation zone reconstruction method which exploits the fundamental link between surface and deep mantle processes to allow determination of palaeolongitudes, unlocking a way forward in modelling absolute plate motions prior to the assembly of Pangaea. The plume generation zone method is a novel way to derive ‘absolute’ plate motions in a mantle reference frame before Pangaea, but the technique assumes that the margins of TUZO and JASON did not move much and that Earth was a degree-2 planet, as today.

A half-century has passed since the dawning of the plate tectonic revolution, and yet, with rare exception, palaeogeographic models of pre-Jurassic time are still constructed in a way more akin to Wegener's paradigm of continental drift. Historically, this was due to a series of problems – the near-complete absence of in situ oceanic lithosphere older than 200 Ma, a fragmentary history of the latitudinal drift of continents, unconstrained longitudes, unsettled geodynamic concepts and a lack of efficient plate modelling tools – which together precluded the construction of plate tectonic models. But over the course of the last five decades strategies have been developed to overcome these problems, and the first plate model for pre-Jurassic time was presented in 2002. Following on that pioneering work, but with a number of significant improvements (most notably longitude control), we here provide a recipe for the construction of full-plate models (including oceanic lithosphere) for pre-Jurassic time. In brief, our workflow begins with the erection of a traditional (or ‘Wegenerian’) continental rotation model, but then employs basic plate tectonic principles and continental geology to enable reconstruction of former plate boundaries, and thus the resurrection of lost oceanic lithosphere. Full-plate models can yield a range of testable predictions that can be used to critically evaluate them, but also novel information regarding long-term processes that we have few (or no) alternative means of investigating, thus providing exceptionally fertile ground for new exploration and discovery.

We present a statistical approach to data mining and quantitatively evaluating detrital age spectra for sedimentary provenance analyses and palaeogeographic reconstructions. Multidimensional scaling coupled with density-based clustering allows the objective identification of provenance end-member populations and sedimentary mixing processes for a composite crust. We compiled 58 601 detrital zircon U–Pb ages from 770 Precambrian to Lower Palaeozoic shelf sedimentary rocks from 160 publications and applied statistical provenance analysis for the Peri-Gondwanan crust north of Africa and the adjacent areas. We have filtered the dataset to reduce the age spectra to the provenance signal, and compared the signal with age patterns of potential source regions. In terms of provenance, our results reveal three distinct areas, namely the Avalonian, West African and East African–Arabian zircon provinces. Except for the Rheic Ocean separating the Avalonian Zircon Province from Gondwana, the statistical analysis provides no evidence for the existence of additional oceanic lithosphere. This implies a vast and contiguous Peri-Gondwanan shelf south of the Rheic Ocean that is supplied by two contrasting super-fan systems, reflected in the zircon provinces of West Africa and East Africa–Arabia.

Dynamic topography is a well-established consequence of global geodynamic models of mantle convection with horizontal dimensions of >1000 km and amplitudes up to 2 km. Such physical models guide the interpretation of geological records on equal dimensions. Continent-scale geological maps therefore serve as reference frames of choice to visualize erosion/non-deposition as a proxy for long-wavelength, low-amplitude vertical surface motion. At a resolution of systems or series, such maps display conformable and unconformable time boundaries traceable over hundreds to thousands of kilometres. Unconformable contact surfaces define the shape and size of time gap (hiatus) in millions of years based on the duration of time represented by the missing systems or series. Hiatus for a single system or series base datum diminishes laterally to locations (anchor points) where it is conformable at the mapped resolution; it is highly dependent upon scale. A comparison of hiatus area between two successive system or series boundaries yields changes in location, shape, size and duration, indicative of the transient nature of vertical surface motion. As a single-step technique, it serves as a quantitative proxy for palaeotopography that can be calibrated using other geological data. The tool magnifies the need for geological mapping at the temporal resolution of stages, matching process rates. The method has no resolving power within conformable regions (basins) but connects around them. When applied to marine seismic sections that relate to rock record, not to time, biostratigraphic and radiometric data from deep wells are needed before hiatus areas – that relate to time – can be mapped.

Palaeogeographic reconstructions have been proposed for years. The technique employed, however, is more or less always the same: it consists of determining the palaeoenvironment at the local scale and extending it to the regional scale. Such work is carried out in a maximum number of locations all over the planet and the global palaeogeography is the result of interpolation of those reconstructions. Advances in palaeogeography can be made via an alternative way, which consists of integrating and then coupling various global models. It results in the proposal of synthetic palaeogeographies that can be compared a posteriori to local or regional data. The advantage is twofold: (1) the view is really global and it avoids gaps (in particular in the oceanic realm) in the reconstructions, and it is very much less focused on the coastline; (2) it takes advantages from almost all the fields of geosciences, so that reconstructions can be constrained from a large variety of data. The two techniques – the ‘classic’ and the ‘alternative’ – are not contradictory but complementary, and it is desirable that one feeds the other and the study of palaeogeography be revived.

Whether the latitudinal distribution of climate-sensitive lithologies is stable through greenhouse and icehouse regimes remains unclear. Previous studies suggest that the palaeolatitudinal distribution of palaeoclimate indicators, including coals, evaporites, reefs and carbonates, has remained broadly similar since the Permian period, leading to the conclusion that atmospheric and oceanic circulation control their distribution rather than the latitudinal temperature gradient. Here we revisit a global-scale compilation of lithologic indicators of climate, including coals, evaporites and glacial deposits, back to the Devonian period. We test the sensitivity of their latitudinal distributions to the uneven distribution of continental areas through time and to global tectonic models, correct the latitudinal distributions of lithologies for sampling- and continental area-bias, and use statistical methods to fit these distributions with probability density functions and estimate their high-density latitudinal ranges with 50% and 95% confidence intervals. The results suggest that the palaeolatitudinal distributions of lithologies have changed through deep geological time, notably a pronounced poleward shift in the distribution of coals at the beginning of the Permian. The distribution of evaporites indicates a clearly bimodal distribution over the past ~400 Ma, except for Early Devonian, Early Carboniferous, the earliest Permian and Middle and Late Jurassic times. We discuss how the patterns indicated by these lithologies change through time in response to plate motion, orography, evolution and greenhouse/icehouse conditions. This study highlights that combining tectonic reconstructions with a comprehensive lithologic database and novel data analysis approaches provide insights into the nature and causes of shifting climatic zones through deep time.

The Phanerozoic evolution of the atmospheric CO2 level is controlled by the fluxes entering or leaving the exospheric system. In this contribution, we focus on the role played by the palaeogeographic configuration on the efficiency of the CO2 sink by continental silicate weathering, and on the impact of the magmatic degassing of CO2. We use the spatially resolved numerical model GEOCLIM to compute the response of the silicate weathering and atmospheric CO2 to continental drift for 22 time slices of the Phanerozoic. Regarding the CO2 released by the magmatic activity, we reconstruct several Phanerozoic histories of this flux, based on published indices. Again using the GEOCLIM model, we calculate the CO2 evolution for each degassing scenario. We show that the palaeogeographic setting is a main driver of the climate from 540 Ma to about the beginning of the Jurassic, with the noticeable exception of the Late Palaeozoic ice age. Regarding the role of the magmatic degassing, the various reconstructions do not converge towards a single signal, and thus introduce large uncertainties in the calculated CO2 level over time. Nevertheless, the continental dispersion, which prevails since the Jurassic, promotes CO2 consumption by weathering and forces atmospheric CO2 to stay low. Warm climates of the ‘middle’ Cretaceous and early Cenozoic require enhanced CO2 degassing by magmatic activity.

Palaeogeography is the representation of the past surface of the Earth. It provides the spatial context for investigating how the Earth evolves through time, how complex processes interact and the juxtaposition of spatial information. In hydrocarbon exploration, palaeogeographies have been used to map and investigate the juxtaposition, distribution and quality of play elements (source, reservoir, seal and trap), as boundary conditions for source-to-sink analysis, climate modelling and lithofacies retrodiction, but most commonly as the backdrop for presentations and montages. This paper demonstrates how palaeogeography has been and can be used within an exploration workflow to help mitigate exploration risk. A comprehensive workflow for building palaeogeographies is described which is designed to provide a standard approach that can be applied to a range of tasks in exploration and academia. This is drawn from an analysis of the history of palaeogeography and how it has been applied to exploration in the past and why. Map applications, resolution and content depend on where in the exploration and production (E&P) cycle the map is used. This is illustrated here through three case studies, from the strategic decisions of global new ventures exploration to the more detailed basin and petroleum analyses of regional asset teams evaluating basins and plays. Through this, the paper also addresses three commonly asked questions: (1) How can I use palaeogeography in my workflow? (2) How reliable are the maps? (3) How do I build a palaeogeography?