Palaeovegetation modelling of the Cape during the Pleistocene

Project Abstract:

High-resolution regional scaled palaeovegetation maps derived from our theoretical understanding of vegetation (via modelling) can complement, contextualise and extend empirical palaeo-archival studies (such as speleothems, pollen in sediments etc.). Here I propose to model palaeovegetation under strong glacial (Last Glacial Maximum) and strong interglacial conditions (mid-Holocene) of the Pleistocene over the Cape and surrounding regions. Two divergent theoretical approaches will be utilised: correlative community distribution modelling and mechanistic dynamic global vegetation modelling. These maps (hypotheses of distribution) will be used to explore the history of the Cape biota during the Pleistocene, including early human evolution and the exposed continental shelf during glacials that would have given rise to an extensive Agulhas Plain. The palaeovegetation models will also be tested using the existing empirical records and with independent lines of study (e.g. phylogeography, ecological studies on biome boundaries) based on predictions made by the models.
Required funding period: 4 years.

Details of research: Problem statement

The south-western tip of Africa has long been recognised as a geographic singularity, most notably in terms of its biodiversity (Darwin, 1836, Cowling, 1992, Goldblatt & Manning, 2002). The causes for this biodiversity lie in the history of the Cape and this has been of intense research interest, yet we remain restricted to the somewhat vague, yet most likely accurate, reasons of complex environments and relative climate stability (Linder, 2003, Hopper, 2009, van der Niet & Johnson, 2009, Schnitzler et al., 2011). There have been major strides towards understanding the evolutionary past of the Cape system using molecular ecology (including phylogenies and phylogeography; e.g. Verboom et al., 2009) and palaeo-environmental records  (including speleothems, pollen sequences, etc.; reviewed in Chase & Meadows, 2007). Despite this, there remain temporal or spatial discontinuities that cloud any attempts to generate a regional synthesis. For example, the phylogenetic record often has exceptionally wide and vague dating estimates, but provides powerful regional biogeographic patterns, whereas palaeo-environmental records are usually temporally well-resolved but may only represent localised changes in the landscape that is out of sync with regional patterns. 

General pictures have emerged of the past history of the Cape as more and more of these empirical studies are completed, but the increasing number of studies does not necessarily result in increased resolution of the temporal or spatial picture; this is unsurprisingly given the nature of genetic signals or preservation environments (see Graur & Martin, 2004, Chase & Meadows, 2007, Lewis, 2008). The problem is a frustrating lack of any palaeo-records that are both highly resolved in terms of time and widely distributed in space. This limits our ability to infer or contextualise Cape evolutionary research – a brief example of this is how to interpret the effects of the Pleistocene climate fluctuations on biota, including floristic diversity and early human evolution, both of which have rich records during this time (Marean, 2010c, Marean, 2010a). 

A further limitation to using the molecular or deposit approach is that these are restricted to the current Cape landscape, and give very limited insights into the nature of the large Agulhas Plain that existed off the current coastline during the majority of the Pleistocene (Fisher et al., 2010). Lowered sea levels and an extremely shallow off-shore continental shelf led to an area up to the size of Ireland being exposed during the lowered sea-levels of Pleistocene glacials. This likely had a dramatic, and currently undervalued, impact on the fauna and flora of the Cape (e.g. Compton, 2011).  

In this project I aim to generate theory-based palaeovegetation maps that span the levels of species, community and functional types of two specific past climate states that represent extremes of the Pleistocene fluctuations (strong glacial and strong interglacial conditions).  These will be based on a combination of correlative and mechanistic modelling methods and will include any additions to the Cape by exposure of the currently submerged Agulhas Plain.  This will enable us to test old, and generate new, hypotheses of palaeo-environmental change including such things as the placements of refugia, potential distributional shifts, and localised extinctions of plant communities and species. Such maps will characterise the extremes of the Pleistocene and provide a common hypothesis on which all Pleistocene-related Cape research from a range of disciplines can be contextualised and tested.

Details of Research : Literature Review and Conceptual Framework

In the empirical domain it remains essential to retrieve and study a wider selection of palaeo-archives that are well resolved both spatially and temporally. However, such archives are usually temporally or spatially discontinuous (Chase & Meadows, 2007, Lewis, 2008). This has frustratingly handicapped attempts to synthesise the palaeorecord into a regional understanding of environmental changes and this directly impacts on our ability to interpret the history or evolution of biota (Chase & Meadows, 2007). An extreme, but critically important example already mentioned above is the glacial-period terrestrial environment of the current offshore Agulhas shelf. This shelf was exposed during the lower sea-levels of glacial periods (Fisher et al., 2010); for example during the Last Glacial Maximum the landmass exposed off the coast of South Africa effectively increased the size of the Cape Floristic Region by a factor of 1.4 (Fisher et al., 2010, Compton, 2011). This exposed shelf likely provided a now extinct ecosystem for exploitation for a range of fauna and flora (including early humans). This has already been proposed as a driver for a now extinct migration-based ecosystem (Compton, 2011) and also as a possibly decisive region for early human survival (Marean, 2010b, Compton, 2011) that occurs in a region evident to be a hotspot of novel human innovations (Marean et al., 2007, Brown et al., 2009, Henshilwood et al., 2011) and may, indeed, be the locus for the evolution of cultural and behavioural innovations associated with contemporary humans (Marean, 2010c). 

In the theoretical domain, a model-based palaeovegetation synthesis in a regional context is required to contextualise the Cape palaeorecord, and that this can be provided by the application of a range of vegetation models coupled with simulations of palaeoclimates. This will produce detailed maps of the potential distribution of plant communities of past climate states. Without such high-resolution palaeovegetation maps of the Cape, the inferences based on the current palaeorecord sites (which are by nature spatially-limited) will remain largely localised enigmas and any regional extrapolations may extend beyond available evidence. Here I propose to develop high-resolution vegetation maps for two extreme periods of the Quaternary cycling, the warmer-than-present mid-Holocene (MH: ~6000yr BP) and Last Glacial Maximum (LGM: ~21,000yr BP) using a two-pronged vegetation modelling approach. The reasons for selecting these two periods specifically are discussed later.

Vegetation modelling builds on first principle understanding of how plants and the environment are correlated and/or interact (Franklin, 1995, Scheiter & Higgins, 2009).  There are two largely disparate fields of vegetation modelling: community distribution modelling and dynamic global vegetation modelling (DGVM). The first is an offshoot of species distribution modelling (Franklin, 2010) where the distribution of vegetation units (identified based on floristics) is modelled. This approach has been used to model distributional changes, in response to altered climates, of tropical forests (Hilbert et al., 2007, Carnaval & Moritz, 2008, VanDerWal et al., 2009), subtropical thicket (Potts et al., 2013c), and fynbos and succulent karoo (Midgley & Roberts, 2003). The second utilises our fundamental knowledge of plant ecophysiology to predict the presence of plant functional types, and hence vegetation (Bond et al., 2003, Scheiter & Higgins, 2009, Prentice et al., 2011). 

Both modelling approaches have limitations and advantages. There are important aspects of plant ecology that cannot easily be incorporated into community (and species) distribution models, such as biotic interactions (e.g. competition, Franklin, 2010), and the effects of fire or atmospheric CO2 (Scheiter & Higgins, 2009, Scheiter et al., 2012). Nonetheless, such methods have successfully predicted refugia across a range of species in the southern African context that have been supported by genetic data (e.g. Potts et al., 2013a, Potts et al., 2013b, Potts et al., 2013c) , strongly suggesting that the correlative distribution modelling approach will provide valuable insights into the past changes of plant communities and species of interest (Svenning et al., 2011). In contrast, the DGVM approach can include complex ecological components such as interactions, fire and atmospheric CO2 (Higgins & Scheiter, 2012, Scheiter et al., 2012) but is limited to plant functional types (such as trees, shrubs and grasses) for vegetation predictions and cannot make some community distinctions (e.g. between subtropical thicket and forest). Investigating the congruence between these modelling approaches will enable us to determine the strengths and limitations of our current theoretical understanding of vegetation distributions. 

Details of Research : Research Question and Hypothesis

The cycling between glacial and interglacial periods drove shifts in species distributions globally (Dynesius & Jansson, 2000). Understanding the effects of these dramatic shifts on vegetation across the Cape landscape is critical for understanding the stresses and selective forces operating on the region’s remarkable biota (including early humans). This project seeks to answer the primary question of how the vegetation of the Cape and surrounding regions responded to the changes in climate during specific periods of the Pleistocene representing extreme climatic (strong glacial and interglacial) states. This includes projections onto the exposed Agulhas Bank, which would have been an Agulhas Plain, during glacial periods (Fisher et al., 2010). This will provide a theoretical framework onto which empirical palaeorecords can be connected. In order to do this, I will generate vegetation maps based on state-of-the-art modelling and simulation methods. These maps will include levels of model uncertainty both within and between the two vegetation modelling approaches. 

Above and beyond the primary question outlined above, these maps will also be used to assess a wide variety of existing questions regarding the evolutionary history of the Cape. Below is a brief list of examples:
  • How have biomes shifted in response to changes in climate and what variables are responsible for driving biome boundaries?
  • Is the longitudinal gradient of species richness identified by Levyns (Cowling & Lombard, 2002) due to greater distributional stability of fynbos in the west relative to the east through the Pleistocene?  
  • Albany subtropical thicket is predicted to have fragmented and contracted into the drainage basins during glacial periods (Potts et al., 2013c) – what vegetation replaced the thicket? 
  • How did vegetation vary through the Pleistocene in terms of resource availability for early humans? For example, the earliest record of heat treatment of stone tools is found in the Cape that was a significant and marvellous technological innovation (Brown et al., 2009), yet its use is sporadic through the archaeological record. Is this due to the fluctuations of fuel-wood availability in response to shifting vegetation as hypothesised by Brown and Marean (2010)?
Simply testing the map predictions will also provide further avenues of research. For example, the community distribution modelling projections onto the Last Glacial Maximum for subtropical thicket vegetation showed dramatic contraction and fragmentation (Potts et al., 2013c); analysis of the models suggested that this was driven by declining minimum temperatures. We have explored this as a limiting environmental factor and determined that thicket vegetation in general is highly intolerant to frost and this is responsible for driving at least one biome boundary (with the Nama Karoo) (Duker et al., in review). Another example was the testing and confirmation of phylogeographic predictions made by the community distribution model of the subtropical thicket (Potts et al., 2013b, Potts et al., 2013d). 

Details of Research : Proposed Research Plan 


I, in collaboration with others, will develop detailed palaeovegetation models and maps of the extreme climate states of the dramatic glacial-interglacial cycling of the Quaternary, specifically a strong interglacial (mid-Holocene) and a strong glacial (Last Glacial Maximum). I have selected these two states as they are manageable and cover the extreme range of climate variation over the temporal boundary conditions (EPICA community members, 2006, Barbante et al., 2010, Masson-Delmotte et al., 2011).  This research also aims is to include a range of Masters and Doctorate dissertations in order to test the vegetation models using independent lines of investigation (e.g. phylogeography and ecological studies on biome boundaries). 


The primary objectives of this project are split into four major categories, with main collaborators highlighted.
- Regional downscaled climate modelling of the Cape (Project led by Dr. F Engelbrecht of the CSIR and Ramapulana Nkoana [PhD student]).
- Testing climate models output with available palaeoclimate archives (Dr. AJ Potts, Dr. F Engelbrecth, R. Nkoana).

- community distribution modelling (Dr. AJ Potts and Prof. J Franklin).
- species-ensemble distribution modelling (Dr. AJ Potts, Prof. RM Cowling and Prof. J Franklin).
- adaptive dynamic global vegetation modelling (Dr. AJ Potts, Dr. S Scheiter and Prof. SI Higgins).

- Comparing outputs from the different models, and detailing model congruence and incongruence (Dr. AJ Potts, Prof. J Franklin, Dr. S Scheiter and Prof. SI Higgins). 
- Testing palaeovegetation maps against the available palaeovegetation archives (Dr. AJ Potts and others). 
- Using the maps to make predictions regarding population history of Cape biota, and testing these using ecological and genetic studies (AJ Potts and others). This objective (along with objective 4) has the greatest and most exciting scope for including postgraduate research dissertations (MSc and PhD), where the vegetation models will inform the questions we ask yet the research remains stand-alone. For example, Robbert Duker (MSc upgrading to PhD) is exploring the effects of minimum temperature (including frost events) on the subtropical thicket boundary. This research was directly initiated by examining drivers of palaeovegetation models of the subtropical thicket (Potts et al., 2013c) and feeds back into the model testing, but is also providing stand-alone publications (e.g. Duker et al., in review). 

These maps will be evaluated in relation to a range of palaeoscience and ecological questions, including, but not restricted, to:
- Cape floral evolution (Dr. AJ Potts, Prof. K Esler and Prof. RM Cowling).
- Biome boundaries and shifting climates (Dr. AJ Potts, Prof. RM Cowling and Prof. W Bond). 
- Human evolution along the south coast (Prof. CW Marean, Prof. RM Cowling, Prof. K Hill and Dr. AJ Potts).
This objective has also wide scope for training students. Student projects that are aimed at testing the map predictions (outlined above in objective 3) will also answer some Cape-related questions, but a number of student projects will investigate such questions directly solely based on the palaeovegetation maps. 


The expected timeline for the project in relation to the objectives stated above are given below. Note that there are overlaps as much can be done in parallel once a previous objective has started producing results.

This section of the overall project is already well underway as part of an NRF African Origins Platform grant and is led by F Engelbrecht of the CSIR; the first downscaled results expected in Mar 2014. 

The bulk of the required distributional datasets have been collected for the community and species-ensemble distribution models. The aDGVM is currently being refined for the southern African context (led by S Scheiter and SI Higgins) and will be available by Aug 2014. The current continentally-calibrated aDGVM will still provide useful insights and be used until the calibrated version is available. By and large, I will run and analyse all model simulations. This is a lengthy process given the quantity of data produced.
Comparing and synthesising the outputs of the different vegetation modelling approaches. This step includes publishing the map projections for each vegetation modelling method as well as a synthesis across the two methods. 

This will include reviewing the palaeovegetation maps in terms of the current knowledge and questions regarding the Cape biota. 

The models will provide predictions regarding vegetation shifts and the potential drivers of those shifts. These will be tested using independent studies as outlined above. 

Expected outputs

This research aims to produce high resolution maps of vegetation of the Cape and surrounding regions and will give rise to a wide array of further outputs. These palaeovegetation maps will have broad impacts on a wide range of palaeosciences. I expect no less than 10 papers published in international journals to arise from this research. These  range from publishing the palaeovegetation maps generated from the different methods (including the overall synthesis), exploring the palaeoenvironment of the Agulhas Plain during glacial periods, investigating the predictions made from the maps regarding population history of Cape biota, and detailing the effects of vegetation dynamics in terms of human occupation and evolution in the Cape. 

Furthermore, I expect a minimum of three MSc and two PhD degrees arising from objectives 3 and 4 listed above. Also, this research will be presented at a range of local (e.g. Fynbos Forum; Arid Zone Forum; Thicket Forum) and international conferences and meetings (e.g. Human Behaviour and Evolution Society Conference; European Society for Human Evolution Meeting; International Biogeography Society Meeting).  

Details of Research : Methods 

The spatial domain of this research is largely restricted to the south-western tip of Africa, the Cape and surrounds. I use the term “Cape” to refer to the region completely encompassing the Cape Floral Kingdom or Cape Floristic Region. However, the study region also includes surrounding regions namely the succulent karoo of the Greater Cape Floristic Region (Born et al., 2007) and Albany subtropical thicket (Vlok et al., 2003). The Cape is a natural bounded laboratory for the study of the evolution of fauna and flora through time because it differs dramatically from the surrounding Paleotropics, and it is a manageable space within which we can construct good models of climate and vegetation.  

The temporal domain is limited to the two extremes of the Pleistocene fluctuations which are representative of strong glacial and strong inter-glacial conditions, specifically the Last Glacial Maximum (LGM: ~21,000yr BP) and the warmer-than-present mid-Holocene (MH: ~6000yr BP). The reason for this is that they are both focal periods for the Paleoclimate Modelling Intercomparison Project 3 (PMIP3; and have been extensively simulated by a wide a range of groups and global climate models. Thus the simulations necessary for regional climate downscaling are already available (discussed further below). 
As stated previously, the vegetation modelling will involve a two-pronged approach using distribution modelling and global dynamic vegetation modelling. Community distribution modelling will be performed on vegetation types (e.g. fynbos, succulent karoo, subtropical thicket etc.) identified using floristics, structure and ecology (e.g. Vlok et al., 2003, Mucina & Rutherford, 2006). Such methods build on the well-established field of species distribution modelling (Franklin, 2010) and have been used to predict distributional changes due to altered climate for a wide range of vegetation types (Midgley & Roberts, 2003, Hilbert et al., 2007, Carnaval & Moritz, 2008, VanDerWal et al., 2009, Potts et al., 2013c). I will use a range of algorithms that span the major classes of algorithms (profile, regression and machine learning;  Franklin, 2010) : Bioclim, Domain, generalized linear models, generalized additive models, random forests and Maxent. This will form part of an “ensemble” approach as suggested by Araújo and New (2007) whereby projection uncertainty due to model choices (e.g. the algorithm used or the predictor variables selected) is incorporated into the results (e.g. Potts et al., 2013c). 

The application of community distribution models to predefined assemblages of species assumes that species swarm together and have a shared distributional history and future. However, species may not share trajectories with their currently associated vegetation type, and may be re-organised into different communities rather than act as coherent units (Midgley et al., 2002). Thus, I will also utilise a species-ensemble distribution modelling (SEDM) approach (which is an addition to the ensemble approach mentioned above) whereby dominant and characteristic species of a vegetation type are modelled individually. These are then combined to assess the congruence or discord of distributional changes within the species assemblage, as well as with the relevant community distribution model (min. of 50 characteristic species per vegetation type). The necessary locality datasets are readily available from the South African National Biodiversity Institute, and any additional data can be collected via additional georeferencing and/or fieldwork.

The aDGVM integrates plant physiological models with novel sub-models that allow plants to dynamically adjust carbon allocation and leaf phenology to environmental conditions (Scheiter & Higgins, 2009), and can incorporate aspects of competition, fire and even some degree of evolution (Scheiter et al., 2013). This is an individual-based model which keeps track of state variables such as height, leaf area index, phenological status and above- and below-ground biomass for individual plants. The model considers a range of functional types such as savanna or forest trees, and C3 or C4 grasses. The range of functional types within a cell is then used to classify its vegetation type. 

Both vegetation modelling approaches require high-resolution environmental layers, most of important of which for this research is climate. The other variables, such as topography, geology, and soil are readily available in a digital format – this includes the now submerged Agulhas plain   (e.g. Siesser & Dingle, 1977; H. Cawthra [Council for Geoscience], unpublished data). Sediment transported from shale geologies from the interior and deposited onto the exposed Agulhas Plain may have been an important factor for localised increases in the nutrient status on the Plain. We are investigating this nuanced angle (led by Prof. F. Ellery, Rhodes University) using landscape evolution modelling (Tucker & Hancock, 2010). Where environmental variables are mapped at resolutions coarser than those required for models, these will be downscaled using a semi-random-specification algorithm (which produces many different versions of the downscaled outputs; A.J. Potts, unpublished), which will allow the assessment of the level of bias in the palaeovegetation maps that may be introduced by this step. 

The aDGVM approach requires monthly and daily environmental estimates such as daily temperature range, mean daily temperature, daily relative humidity and monthly precipitation. Such estimates for altered climate can only be derived from dynamical downscaling (not statistical downscaling). Thus, in order to obtain high resolution and realistic climate over the Cape, we are downscaling the global climate models using regional climate modelling. This downscaling is performed in collaboration with the Climate Studies and Modelling and Environmental Health group at the Council for Scientific and Industrial Research (funded by the NRF/RISA African Origins Platform, Grant #: AOP1207264432). The aim of that project is to provide regional downscaling from multiple global climate model simulations generated by the PMIP for the two periods of interest. From this, the necessary input variables for both vegetation modelling approaches will be available. 

Projections of vegetation models are influenced by many potential sources of uncertainty (e.g. Araújo & New, 2007, Scheiter & Higgins, 2009). There is uncertainty in the equations used to represent the plant-environment relationship, and there is uncertainty in the data used for parameterization and validation. Further sources of uncertainty involve the environmental input data (e.g. climate or soil data). However, this research specifically involves two approaches that are on opposite ends of the modelling continuum - correlative and mechanistic - and thus uncertainty across general modelling approaches can be established. Furthermore, uncertainty within approaches can be quantified, and these explicitly included in the palaeovegetation maps to ensure that the projected outputs are not misconstrued or misinterpreted (for example, see Fig. 4 in Potts et al., 2013c, pg. 317).  This research is the most ambitious use of our theoretical understanding of plant dynamics to produce palaeovegetation maps (which can be seen as distributional hypotheses) against which the empirical record can be compared. As such it represents a meeting point of the empirical and theoretical domains and should provide a holistic hypothesis of vegetation change in the Cape during the periods of interest.

Details of Research : Alignment to National and Institutional Strategies

This proposed research aligns with the “Biodiversity” and “Climate Change” sub-themes of Nelson Mandela Metropolitan University’s institutional research theme of “Biodiversity Conservation and Restoration”. The palaeovegetation maps will give direct insights into the evolution of the Cape biodiversity, and understanding changes in response to past climate change is necessary to accurately predict potential changes under altered climates of the future. As a primary focus of the interpretation of the palaeovegetation maps will also be to contextualise human evolution in the Cape, this research also touches on the sub-theme of “Narratives and Meaning of the Past” within the “Earth Stewardship Science” thrust of NMMU’s research themes. All current evidence suggests that the transition from anatomically-modern to behaviourally-modern humans took place in the Cape (e.g. Marean et al., 2007, Brown et al., 2009, Henshilwood et al., 2011) and this was potentially a last refuge for our lineage during a period of drastic and rapid climate change (Marean, 2010c). From here, this key innovation allowed humanity to radiate across the world. This is an exceptionally powerful metaphor in the light of the racially- and culturally-driven divisions that remain in the new South Africa: the lineage of all of humanity stems from our shores and any returning lineage, irrespective of when they returned, can be seen as a homecoming. 

This research is also strongly aligned to the African Origins research strategy for the palaeosciences which forms a part of the National Research and Development Strategy coordinated by the Department of Science and Technology. This research also feeds into the “Science and technology in response to global change” grand challenge declared by the Department of Science and Technology. The methods utilised to develop and test the palaeovegetation maps will have significant implications regarding the reliability of these vegetation models projected onto the altered climates of the future. This project can be seen as a necessary forerunner and proof-of-concept for vegetation modelling focussed on future changes. 


Araújo, M. B. & New, M. (2007) Ensemble forecasting of species distributions. Trends in Ecology & Evolution, 22, 42-47.
Barbante, C., Fischer, H., Masson-Delmotte, V., Waelbroeck, C. & Wolff, E. W. (2010) Climate of the last million years: new insights from EPICA and other records. Quaternary Science Reviews, 29, 1-7.
Bond, W. J., Midgley, G. F. & Woodward, F. I. (2003) The importance of low atmospheric CO2 and fire in promoting the spread of grasslands and savannas. Global Change Biology, 9, 973-982.
Born, J., Linder, H. P. & Desmet, P. (2007) The Greater Cape Floristic Region. Journal of Biogeography, 34, 147-162.
Brown, K. S. & Marean, C. W. (2010) Wood fuel availability for heat treatment drives the rise and fall of silcrete as a raw material in the Middle Stone Age of South Africa. 2010 Paleoanthropology Society Meetings. St. Louis.
Brown, K. S., Marean, C. W., Herries, A. I. R., Jacobs, Z., Tribolo, C., Braun, D., Roberts, D. L., Meyer, M. C. & Bernatchez, J. (2009) Fire as an engineering tool of early modern humans. Science, 325, 859-862.
Carnaval, A. C. & Moritz, C. (2008) Historical climate modelling predicts patterns of current biodiversity in the Brazilian Atlantic forest. Journal of Biogeography, 35, 1187-1201.
Chase, B. M. & Meadows, M. E. (2007) Late Quaternary dynamics of southern Africa's winter rainfall zone. Earth-Science Reviews, 84, 103-138.
Compton, J. S. (2011) Pleistocene sea-level fluctuations and human evolution on the southern coastal plain of South Africa. Quaternary Science Reviews, 30, 506-527.
Cowling, R. M. (1992) The ecology of fynbos: nutrients, fire and diversity. Oxford University Press, Cape Town.
Cowling, R. M. & Lombard, A. T. (2002) Heterogeneity, speciation/extinction history and climate: explaining regional plant diversity patterns in the Cape Floristic Region. Diversity and Distributions, 8, 163–179.
Darwin, C. (1836) Darwin’s Beagle Diary. Online at:
Duker, R., Cowling, R. M., Preez, D. R. D., Vyver, M. L. V. D., Weatherall-Thomas, C. R. & Potts, A. J. (in review) Community-level assessment of freezing tolerance: frost dictates the boundary between subtropical thicket and Nama-Karoo shrubland in South Africa. Journal of Biogeography.
Dynesius, M. & Jansson, R. (2000) Evolutionary consequences of changes in species' geographical distributions driven by Milankovitch climate oscillations. Proceedings of the National Academy of Sciences, U.S.A., 97, 9115-9120.
Epica Community Members (2006) One-to-one coupling of glacial climate variability in Greenland and Antarctica. Nature, 444, 195-198.
Fisher, E. C., Bar-Matthews, M., Jerardino, A. & Marean, C. W. (2010) Middle and Late Pleistocene paleoscape modeling along the southern coast of South Africa. Quaternary Science Reviews, 29, 1382-1398.
Franklin, J. (1995) Predictive vegetation mapping: geographic modelling of biospatial patterns in relation to environmental gradients. Progress in Physical Geography, 19, 474-499.
Franklin, J. (2010) Mapping Species Distributions: Spatial Inference and Prediction. Cambridge University Press, Cambridge, UK.
Goldblatt, P. & Manning, J. C. (2002) Plant diversity of the Cape region of southern Africa. Annals of the Missouri Botanical Garden, 89, 281-302.
Graur, D. & Martin, W. (2004) Reading the entrails of chickens: molecular timescales of evolution and the illusion of precision. Trends in Genetics, 20, 80-86.
Henshilwood, C. S., D'errico, F., Van Niekerk, K. L., Coquinot, Y., Jacobs, Z., Lauritzen, S.-E., Menu, M. & Garcia-Moreno, R. (2011) A 100,000-year-old ochre-processing workshop at Blombos Cave, South Africa. Science, 334, 219-222.
Higgins, S. I. & Scheiter, S. (2012) Atmospheric CO2 forces abrupt vegetation shifts locally, but not globally. Nature, 488, 209-212.
Hilbert, D. W., Graham, A. & Hopkins, M. S. (2007) Glacial and interglacial refugia within a long-term rainforest refugium: the Wet Tropics Bioregion of NE Queensland, Australia. Palaeogeography, Palaeoclimatology, Palaeoecology, 251, 104-118.
Hopper, S. (2009) OCBIL theory: towards an integrated understanding of the evolution, ecology and conservation of biodiversity on old, climatically buffered, infertile landscapes. Plant and Soil, 322, 49-86.
Lewis, C. A. (2008) Late Quaternary climatic changes, and associated human responses, during the last ~45000 yr in the Eastern and adjoining Western Cape, South Africa. Earth-Science Reviews, 88, 167-187.
Linder, H. P. (2003) The radiation of the Cape flora, southern Africa. Biological Reviews, 78, 597-638.
Marean, C. W. (2010a) Pinnacle Point Cave 13B (Western Cape Province, South Africa) in context: the Cape Floral kingdom, shellfish, and modern human origins. Journal of Human Evolution, 59, 425-443.
Marean, C. W. (2010b) Pinnacle Point Cave 13B (Western Cape Province, South Africa) in context: the Cape Floral kingdom, shellfish, and modern human origins. Journal of Human Evolution, 59, 425-443.
Marean, C. W. (2010c) When the sea saved humanity. Scientific American, 303, 54-61.
Marean, C. W., Bar-Matthews, M., Bernatchez, J., Fisher, E., Goldberg, P., Herries, A. I. R., Jacobs, Z., Jerardino, A., Karkanas, P., Minichillo, T., Nilssen, P. J., Thompson, E., Watts, I. & Williams, H. M. (2007) Early human use of marine resources and pigment in South Africa during the Middle Pleistocene. Nature, 449, 905-908.
Masson-Delmotte, V., Stenni, B., Pol, K., Braconnot, P., Cattani, O., Falourd, S., Kageyama, M., Jouzel, J., Landais, A., Minster, B., Barnola, J. M., Chappellaz, J., Krinner, G., Johnsen, S., Röthlisbergere, R., Hansen, J., Mikolajewicz, U. & Otto-Bliesner, B. (2011) EPICA Dome C record of glacial and interglacial intensities. Quaternary Science Reviews, 29, 113-128.
Midgley, G. & Roberts, R. (2003) Past climate change and the generation and persistence of species richness in a biodiversity hotspot, the Cape flora of South Africa. Global Change and Protected Areas. (ed. by G. Visconti, M. Beniston, E. Iannorelli & D. Barba), pp.393-402. Kluwer Academic Publishers, Boston.
Midgley, G. F., Hannah, L., Millar, D., Rutherford, M. C. & Powrie, L. W. (2002) Assessing the vulnerability of species richness to anthropogenic climate change in a biodiversity hotspot. Global Ecology & Biogeography, 11, 445-451.
Mucina, L. & Rutherford, M. C. (2006) Vegetation map of South Africa, Lesotho and Swaziland. South African National Biodiversity Institute, Pretoria.
Potts, A., Hedderson, T., Vlok, J. H. J. & Cowling, R. M. (2013a) Pleistocene range dynamics in the eastern Greater Cape Floristic Region: a case study of the Little Karoo endemic Berkheya cuneata (Asteraceae). South African Journal of Botany, 88, 401-413.
Potts, A. J., Hedderson, T. A. & Cowling, R. M. (2013b) Testing large-scale conservation corridors designed for patterns and processes: comparative phylogeography of three tree species. Diversity and Distributions.
Potts, A. J., Hedderson, T. A., Franklin, J. & Cowling, R. M. C. (2013c) The Last Glacial Maximum distribution of Albany Subtropical Thicket inferred from community distribution modelling. Journal of Biogeography, 40, 310-322.
Potts, A. J., Hedderson, T. A. & Grimm, G. W. (2013d) Constructing phylogenies in the presence of intra-individual site polymorphisms (2ISPs) with a focus on the nuclear ribosomal cistron. Systematic Biology.
Prentice, I. C., Harrison, S. P. & Bartlein, P. J. (2011) Global vegetation and terrestrial carbon cycle changes after the last ice age. New Phytologist, 189, 988-998.
Scheiter, S. & Higgins, S. I. (2009) Impacts of climate change on the vegetation of Africa: an adaptive dynamic vegetation modelling approach. Global Change Biology, 15, 2224-2246.
Scheiter, S., Higgins, S. I., Osborne, C. P., Bradshaw, C., Lunt, D., Ripley, B. S., Taylor, L. L. & Beerling, D. J. (2012) Fire and fire-adapted vegetation promoted C4 expansion in the late Miocene. New Phytologist, 195, 653-666.
Scheiter, S., Langan, L. & Higgins, S. I. (2013) Next-generation dynamic global vegetation models: learning from community ecology. New Phytologist, 198, 957-969.
Schnitzler, J., Barraclough, T. G., Boatwright, J. S., Goldblatt, P., Manning, J. C., Powell, M. P., Rebelo, T. & Savolainen, V. (2011) Causes of plant diversification in the Cape biodiversity hotspot of South Africa. Systematic Biology, 60, 343-357.
Siesser, W. G. & Dingle, R. (1977) Geology of the Continental Margin Between Walvis Bay and Ponta Do Ouro: Geologie Van Die Vastelandsrand Tussen Walvisbaai en Ponta Do Ouro. Geol. Surv. S, Afr., Mar. Geosci. Ser. 2 (map and explanatory note).
Svenning, J.-C., Fløjgaard, C., Marske, K. A., Nógues-Bravo, D. & Normand, S. (2011) Applications of species distribution modeling to paleobiology. Quaternary Science Reviews, 30, 2930-2947.
Tucker, G. E. & Hancock, G. R. (2010) Modelling landscape evolution. Earth Surface Processes and Landforms, 35, 28-50.
Van Der Niet, T. & Johnson, S. D. (2009) Patterns of plant speciation in the Cape Floristic Region. Molecular Phylogenetics and Evolution, 51, 85-93.
Vanderwal, J., Shoo, L. P. & Williams, S. E. (2009) New approaches to understanding late Quaternary climate fluctuations and refugial dynamics in Australian wet tropical rain forests. Journal of Biogeography, 36, 291-301.
Verboom, G. A., Archibald, J. K., Bakker, F. T., Bellstedt, D. U., Conrad, F., Dreyer, L. L., Forest, F., Galley, C., Goldblatt, P., Henning, J. F., Mummenhoff, K., Linder, H. P., Muasya, A. M., Oberlander, K. C., Savolainen, V., Snijman, D. A., Niet, T. V. D. & Nowell, T. L. (2009) Origin and diversification of the Greater Cape flora: Ancient species repository, hot-bed of recent radiation, or both? Molecular Phylogenetics and Evolution, 51, 44-53.
Vlok, J. H. J., Euston-Brown, D. I. W. & Cowling, R. M. (2003) Acocks' Valley Bushveld 50 years on: new perspectives on the delimitation, characterisation and origin of subtropical thicket vegetation. South African Journal of Botany, 69, 27-51.