Portal de Periódicos da CAPES

Carta Acesso aberto Produção Nacional Revisado por pares

BIBTEX RIS

Climate refugees going underground – a response to Maurin et al . (2014)

2016; Wiley; Volume: 209; Issue: 3 Linguagem: Inglês

10.1111/nph.13567

1469-8137

Manfred Finckh, Rasmus Revermann, Marcos Pereira Marinho Aidar,

African Botany and Ecology Studies

The paper of Maurin et al. (2014) has a lot of merit, bringing the fascinating and widely understudied subject of the 'underground forests of Africa' (White, 1976) back onto the scientific agenda. Based on a large sample of geoxyles and their tree counterparts the paper places these species in a comprehensive dated phylogeny of the southern African flora. From there, the authors delineate very convincingly the phylogenetic relationship and evolutionary origin of the geoxylic life form, dating it back in most cases to the Pliocene. In their final analysis, however, the authors formulate the hypothesis that geoxyles have evolved in response to the interactive effects of high precipitation and frequent fires, and rather nonchalantly negate the possibilities of other environmental factors such as frost driving the evolution of suffrutices. In our response we would like to challenge this interpretation and point out that the fire and precipitation hypothesis of Maurin et al. (2014) does not explain the coexistence of the geoxylic life form with closed forests in the immediate proximity. We base our argument on our own observations and measurements from the Angolan Plateau in the provinces of Bié and Moxico. This region is situated in the core of the Zambezian phytoregion as delineated by White (1983) and is rich in geoxyle diversity (White, 1976). The natural vegetation of the Bié Plateau is composed of closed Miombo forests, geoxyle-rich 'grasslands', and wetlands. Our field data on vegetation, soil temperature and microclimate show that night frosts occur frequently during the dry season, shaping the landscape boundaries between forests on topslopes and hills and geoxylic grasslands on midslopes and footslopes. In the first part of our response, we will critically revisit the points in support of fire being an evolutionary driver in the Zambezian region. In the second part we will discuss the authors' points against frost playing an important role ecologically, and in the third part we will present arguments suggesting the contrary. Maurin et al. (2014) state that geoxyles in Africa are restricted to savanna habitats or upland grasslands and that they occur almost exclusively in higher rainfall savannas that are prone to fires. From our own data and observations we would like to question fire as the predominant evolutionary driver of geoxyle evolution using three arguments. The savannas of southern Africa undergo frequent burning (Barbosa et al., 1999; Bond & Keeley, 2005). Archibald et al. (2012) stated that in southern Africa most fires occur in the dry season when ignition by thunderstorms is largely absent. Similarly, for the region of the Angolan Plateau it has been shown that most grassland fires occur early in the dry season (Stellmes et al., 2013). This indicates an anthropic rather than natural origin to fire. The available literature on paleofires makes no reference to fire frequency. However, it can be assumed that the fire return rates were much lower in the absence of human induced fires, which are the number one cause of wildfires today. According to Archibald et al. (2012) humans only learnt to ignite their own fires c. 200–400 thousand yr ago. Moreover, the landscape of the Angolan Plateau has a forest cover (considered inflammable) beyond the threshold of 40% where fire cannot spread (Hennenberg et al., 2006; Archibald et al., 2009). However, only a short fire return rate of < 5 yr captures tree regeneration in the sapling stage. This process is commonly referred to as 'regeneration bottleneck' or 'fire trap' (Sankaran et al., 2004; Bond & Keeley, 2005). Therefore, we question whether fire could have been the main evolutionary driver during the Pliocene, the evolutionary starting point of the geoxylic life form according to the study of Maurin et al. (2014). The current vegetation of the southern slopes of the Angolan Plateau shows an inverse pattern as one would expect from a fire driven landscape pattern. By contrast, the natural landscape follows a very regular pattern. While the hill tracts are covered by forests, slopes are dominated by geoxylic grasslands and only narrow linear strips on the valley bottoms by wetlands (Fig. 1; Supporting Information Fig. S1). As wildfires normally run uphill due to chimney effects, a fire driven landscape pattern would be expected to show an inverse forest–open land distribution, with forest remnants in the valleys and creeks and grassland on the burned slopes. Vegetation inversion as found on the Bié Plateau is, however, a frequently cited phenomenon under topographic conditions which allow for the accumulation of cold air (see later). Maurin et al. (2014) hypothesize that, notwithstanding the fire adaptation of many savanna trees (thick bark, fire-resistant shoots), geoxyles may have escaped fire by developing their woody component belowground, thus minimizing their resource input into annual vegetative growth to the benefit of flower and fruit production. The authors neither explain nor quantify how geoxyles optimize flower and fruit production with their dwarf stature and how this may lead to outcompete large trees. Thus, this argument only holds true if we assume short fire return periods of up to 5 yr (see comment earlier). Furthermore, the authors do not provide any evidence that the geoxyles optimize their resource allocation to the benefit of flower and fruit production as already pointed out by Pennington & Hughes (2014) in their commentary on Maurin et al. (2014). A valid point emphasized by Pennington & Hughes (2014) is the synchronicity of the evolution of geoxyles in South America and southern Africa, which calls for a global explanation. Their argument for a 'complex set of shared climate–fire–vegetation feedback mechanisms' is partly flawed by the fact that in climatically comparable regions, for example the Sudanian center of endemism, only a very small number of geoxylic species occurs, as already pointed out in White's groundbreaking paper (White, 1976). Maurin et al. (2014) admit that, in the South African context, cold temperatures may well have been a contributing factor in the evolution of underground trees. However, they claim that the region of greatest geoxyle diversity, the northern areas of the Zambezian Domain (White, 1983), receives little or no frost. Maurin et al. (2014) further state that they did not find a general relationship between the geoxylic life form and mean annual temperature or elevation. Here we would like to contradict Maurin et al. (2014) based on our own measurements. Data from our network of 21 microclimatic temperature loggers installed in the upper Cubango catchment, Angola, show that frost nights occur recurrently in valleys and depressions of the Zambezian Domain (Notes S1; Fig. S2). The predominant weather conditions for frost occurrence in the Zambezian region are cloudless nights in the dry season with thermal radiation and subsequent accumulation of cold air in the valleys (Fig. 2a). Between the end of May and mid-September 2012 to 2014 we measured up to 44 frost nights yr−1, reaching an absolute minimum of −7.5°C. However, we recorded frost mainly on the grasslands and peatlands on slopes and in valley bottoms, with only a few light frost events in the forest-covered hill tracts (Fig. 3; Table S1). These measurements support the bioclimatic assessment by Le Houérou (2009) who shows that the Angolan Plateau is under risk of frost hazard for as much as 30 to 60 d yr−1. The gridded climate data set CRU TS v3.22 for the period 1971–2000 (Harris et al., 2014) also shows recurrent frost events in the Zambesian region. At the half degree grid resolution of the data set mean annual frost occurrence in our study area is about four frost nights (Fig. 4); however, our data depict nicely that local topography can strongly accentuate frost occurrences. The use by Maurin et al. (2014) of mean elevation of the distributional range of a species as an explanatory variable has little value in this regard as topographic position of the site is more important. In our case study on the Angolan Plateau we found that in many cases the tree partner of geoxyles occurs on the hills while the suffrutices grow in close proximity on the footslopes. Furthermore, the authors make use of mean annual temperature (MAT) derived from the Worldclim database (Hijmans et al., 2005) for their analysis. MAT does not reflect frost occurrence in subtropical and tropical regions with a diurnal instead of a seasonal temperature regime and hence has hardly any explanatory power at all. By contrast, frost events are related to cloudless days in the dry season (May–September) with high temperatures during daytime. The registered frost events are always coupled to sunny days with temperature amplitudes reaching > 35°C within 8 h (Fig. 5). Impact of frost on southern African savanna ecosystems has been documented by several authors (Holdo, 2007; Chafota & Owen-Smith, 2009; Whitecross et al., 2012). Whitecross et al. (2012) in their case study on the impact of frost on Colophospermum mopane in South Africa proved that frost resembles a disturbance regime maintaining saplings in a 'freeze trap' similar to the often cited 'fire trap' and 'browse trap'. Furthermore, they found evidence that topographic position in the landscape has an influence on the severity of frost damage. Similarly, Brando & Durigan (2004) studied changes in the Brazilian Cerrado vegetation after disturbance by frost. Of 57 species studied, 15% were unaffected, 19% had only their leaves damaged, 25% had some of their leaves and branches damaged, and 41% had all their aerial parts killed. They conclude that the frequency and intensity of frosts can maintain open forms of Cerrado vegetation even in sites where both water and nutrient availability could support denser vegetation. Several other studies report frost events in the Cerrado core area (Hamilton & Tarifa, 1978), and that Cerrado plants are subject to the 'freeze trap' phenomenon killing saplings and seedlings, but not the adult trees (De Vuono et al., 1982; Delitti, 1984; Filgueiras & Peña, 1989). In our study area, we frequently found frost burns on tree species in the ecotones between forests and grasslands (Fig. 2b). Most suffrutices in the low-lying grassland die back due to frost burns at the beginning of the dry season (Fig. 2c) and start resprouting from their buried and frost protected buds at the end of the frost season. In general, buds of suffrutices are placed in the upper 10 cm of the soils. As nocturnal frost episodes are always of short duration, this soil depth is sufficient to prevent any frost damage. On frost prone sites soil temperature at 10 cm depth never drops below 9°C (Fig. 5; Revermann & Finckh, 2013). In spite of their short duration, these frost events have a high impact on the flora of the Zambezian phytoregion. Sakai & Larcher (1987) indicate that most tropical plants are sensitive to freezing and killed by temperatures of −1 to −4°C, and some of them are even severely damaged by low temperatures above freezing level. Most geoxyle species listed by Maurin et al. (2014) belong to woody tropical and subtropical families such as Apocynaceae, Moraceae, Myrtaceae, or Rubiaceae, being notorious for lacking ecophysiological adaptations to frost (Griffin & Antikienen, 1996). Werneck (2011) and Giehl & Jarenkow (2012) consider intolerance to low temperatures to be one of the major factors preventing the tropical flora of South America from occurring in subtropical regions, including the Cerrado. Maurin et al. (2014) argue that a shift in general climatic conditions has favored the expansion of savannas. We concur with the fact that the climate has changed, and triggered evolutionary development of the geoxylic life form. However, we see the main significance of such change being the increasing seasonality of the climate starting in the late Miocene with a marked dry season and corresponding enhanced risk of frost nights via thermal radiation (Keeley & Rundel, 2005). Frost as an evolutionary driver would therefore explain the evolutionary onset of the geoxylic life form long before men started burning. The geoxylic life form – evolved to escape thermic stress – is, however, a perfect preadaptation to man-made fires as frost burns and fire burns have similar effects on plants and necessitate immediate re-sprouting. The strong increase in fire frequency and burnt area since people learned to use fire as a tool (Archibald et al., 2012) probably has allowed geoxyles to leave their localized frost-prone niche and expand into secondary grasslands maintained by frequent fires. In conclusion, there are good reasons to regard the evolution of the geoxylic life form as an adaptation to low temperatures by frost avoidance rather than purely as a response to fire. We therefore could look at the Zambezian geoxyles as climate refugees gone underground. In general, the role of frost should be reconsidered as an evolutionary driver for the vegetation of south-central Africa and tropical biomes beyond. Research was funded by the German Federal Ministry of Education and Research (BMBF) in the context of The Future Okavango (TFO) project, grant number 01LL0912A. The authors thank Torsten Weber, Climate Service Center, Hamburg, for processing the CRU TS data. Furthermore, the authors acknowledge valuable comments from three anonymous referees. Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Fig. S1 Cross-sections of the Sovi and the Cusseque Valley indicating vegetation types and the lower tree line. Fig. S2 Spatial layout of the logger network in the Cusseque Valley, Bié Province, Angola. Table S1 Synoptic table of annual minimum air temperatures, number of frost days and length of frost period for the years 2012–2014 Notes S1 Technical note on the climate logger network in the Cusseque Valley, Bié Province, Angola. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.