Cryptic forest refugia on the ‘Roof of the World’


  • Arndt Hampe,

    1. Department of Integrative Ecology, Estación Biológica de Doñana (CSIC), Av. Américo Vespucio s/n, E-41092 Sevilla, Spain
    Search for more papers by this author
  • Rémy J. Petit

    1. INRA, UMR 1202 Biodiversité, Gènes & Communautés, 69, Route d’Arcachon, F-33610 Cestas, France
    2. Université de Bordeaux, UMR 1202 Biodiversité, Gènes & Communautés, 69, Route d’Arcachon, F-33610 Cestas, France
    Search for more papers by this author

(Author for correspondence: tel +34 954 232340; email

It has long been known that the Quaternary glaciations displaced temperate and boreal forests to lower latitudes or altitudes, where species persisted until warmer climates allowed them to expand again (Reid, 1899). Researchers have assumed for decades that most tree taxa and their associated organisms survived the Last Glacial Maximum ((LGM), 18–20 kyr before present (BP)) only in areas far from the continental ice sheets, such as the southern peninsulas of Europe and the southeastern regions of North America. This long-held belief is now increasingly challenged by palaeoecological and genetic surveys. A rapidly growing body of evidence indicates that numerous forest plant and animal species, in fact, maintained small populations at considerably higher latitudes and altitudes than previously thought (reviewed in Bhagwat & Willis, 2008; Provan & Bennett, 2008; Rull, 2009). A paper by Opgenoorth and co-workers in this issue of New Phytologist (pp. 332–342) illustrates this trend with a particularly spectacular and convincing example. Using genetic markers, the authors were able to infer the past population dynamics of a group of Juniperus species that are currently growing up to 4900 m above sea level (asl) on the Tibetan Plateau and the northern Himalayas, forming one of the highest known tree lines on earth (Miehe et al., 2007; Fig. 1). Their results represent sound evidence that junipers probably withstood the coldest phases of the Pleistocene at altitudes of > 3500 m asl.

Figure 1.

 Photograph of the highest known juniper forest stand growing at approx. 4900 m above sea level (asl) in southeastern Tibet (Xizang Autonomous Region, Baxoi County). This stand is also one of the two highest known forests in the world (following a definition widely used in treeline research that considers ‘forests’ groups of individuals > 3 m high; Körner, 2003). The rosaceous tree Polylepis tarapacana grows at up to 5100 m asl on the slopes of the volcano Salama in Bolivia. However, stands of this species that fulfill the above definition of a forest occur only up to 4810 m asl (Hoch & Körner, 2005).

‘… identification of cryptic refugia has direct consequences on our understanding of the impacts of modern climate change on biodiversity.’

At present the Tibetan Plateau is largely covered by alpine pastures and desert-steppes. Juniper forests are mostly restricted to its eastern and southern declivities, although some forest islands occur on the Plateau itself. Pollen records indicate that juniper forests were considerably more widespread in the past and declined as a consequence of human activities during the last few centuries (Miehe et al., 2008). Opgenoorth et al. analysed the chloroplast DNA (cpDNA) of almost 600 individuals from 102 Juniperus stands. They observed geographic patterns of genetic variation that cannot be explained by a scenario involving complete extinction and subsequent recolonization of the Tibetan Plateau by the investigated species. Hence, they suggest that tree populations must have persisted during the LGM in multiple small refugia situated on the southern slopes of several river gorges that cross the Plateau.

This represents an extreme case of so-called ‘cryptic refugia’ (Provan & Bennett, 2008) or ‘microrefugia’ (Rull, 2009), defined as areas outside the major refugia that allowed species to maintain small populations through the LGM as a result of their favourable local environment and microclimate. The term ‘cryptic’ pinpoints the difficulties in identifying and delimiting such refugia using fossil remains. Inferring the existence of local plant populations from trace amounts of fossil pollen is burdened with uncertainties, whereas macrofossils provide more conclusive evidence but are rare and difficult to find. As a consequence, phylogeographic surveys are increasingly used for this purpose. With adequate population sampling and molecular analyses, they can reveal the presence of otherwise undetectable cryptic forest refugia from which small-scale expansions took place following postglacial climate warming (Anderson et al., 2006; Provan & Bennett, 2008). The rapidly growing evidence for the existence of such refugia questions the generality of the established paradigm of postglacial recolonization that emphasizes wave-like expansions from distant refugia mediated by repeated long-distance dispersal (Hewitt, 2000). It also suggests that the distances over which species expanded after the LGM have frequently been overestimated, resulting in an overoptimistic appreciation of the ability of species to track future climate changes (Anderson et al., 2006; Svenning & Skov, 2007; Provan & Bennett, 2008). Clearly, the identification of cryptic refugia has direct consequences on our understanding of the impacts of modern climate change on biodiversity.

The study by Opgenoorth and colleagues provides not only convincing evidence for the existence of multiple cryptic refugia but also enables the authors to locate them precisely. Three particular circumstances contribute to this achievement. First, the study is based on extensive and geographically comprehensive population sampling. Second, the topographic structure of the study area is such that suitable sites for juniper survival through the LGM were restricted to the major river gorges that cross the Tibetan Plateau, many of them situated well above 3500 m asl. Third, the analysed junipers contain remarkably high levels of cpDNA diversity, including many haplotypes that are restricted to a single population. Drawing on the generally low mutation rate of cpDNA (especially in trees; Petit & Hampe, 2006), the authors argue that the patterns observed are unlikely to be of postglacial origin but should instead reflect population divisions older than the LGM and, in some cases, possibly even dating back to the late Tertiary (> 1.8 Myr BP). Although this interpretation involves some uncertainty (as cpDNA mutation rates might be unusually high because of strong UV irradiation in the area), it would be in line with other recent studies that found extant tree population structures to reflect surprisingly ancient events (Hampe & Petit, 2007). Interestingly, cases of deep phylogeographic structure have so far only been reported from regions with warm temperate or tropical climates, very different from the cold-dry conditions prevailing on the Tibetan Plateau.

The strong genetic structure observed indicates that Tibetan junipers underwent only localized postglacial range expansions and experienced little admixture between lineages originating from different refugia. This finding is especially remarkable because the paternally inherited cpDNA of junipers is disseminated through wind-dispersed pollen, a priori the most efficient vehicle for long-distance gene flow between populations. However, the Tibetan juniper populations did not actually need to be very mobile in order to track suitable climatic conditions, because the steep terrain permitted them to ascend several hundred metres by expanding only over short distances. In fact, these small refugia apparently experienced climatic conditions during the LGM that are similar to those prevailing today at the highest known juniper tree lines of approx. 4900 m (Miehe et al., 2007; Opgenoorth et al.), suggesting that postglacial population expansions reached areas near the environmentally set growth limits of the species. Such reconstructions extend in time ongoing analyses of tree line responses to climate warming, confirming that these transition zones are particularly temperature sensitive (Harsch et al., 2009). An interesting perspective would be to investigate such rapidly changing tree lines at different spatio-temporal resolutions using an interdisciplinary approach (including population genetic analysis) to characterize their dynamics in greater detail and extrapolate them through time (Holtmeier & Broll, 2005).

Finally, the study of Opgenoorth et al. highlights the potential of some tree species to survive over centuries, or even millennia, under extremely cold and dry conditions (Petit & Hampe, 2006). The great longevity of juniper trees probably played a central role for population persistence in view of presumably extremely low rates of reproduction and plant recruitment. Unfortunately, the results also imply that extant Tibetan juniper forests face bleak prospects. The ecological strategy of the species, characterized by great adult longevity and a low dispersal potential (probably exacerbated by low rates of successful plant establishment), renders junipers particularly vulnerable to the major anthropogenic disturbances in the region: uncontrolled logging for firewood, overgrazing, and drought stress resulting from climate change (Miehe et al., 2007, 2008). Water supply on the Tibetan Plateau depends largely on glaciers, which are experiencing a significant melt down (Yao et al., 2007). As a consequence, more water is currently available, but a significant increase in aridity is to be expected soon (Cruz et al., 2007). This increase is likely to exacerbate the already severe effects of logging and overgrazing, leading to continued fragmentation and destruction of the remaining Tibetan juniper forests and the ecological communities which they are sustaining.