干旱区生态环境知识资源中心

Aim Grasslands and savannas, which make up > 75% of Madagascar’s land area, have long been viewed as anthropogenically derived after people settled on the island c. 2 ka. We investigated this hypothesis and an alternative - that the grasslands are an insular example of the post-Miocene spread of C(4) grassy biomes world-wide.

Location Madagascar, southern Africa, East Africa.

Methods We compared the number of C(4) grass genera in Madagascar with that in southern and south-central African floras. If the grasslands are recent we would expect to find fewer species and genera in Madagascar relative to Africa and for these species and genera to have very wide distribution ranges in Madagascar. Secondly, we searched Madagascan floras for the presence of endemic plant species or genera restricted to grasslands. We also searched for evidence of a grassland specialist fauna with species endemic to Madagascar. Plant and animal species endemic to C(4) grassy biomes would not be expected if these are of recent origin.

Results Madagascar has c. 88 C(4) grass genera, including six endemic genera. Excluding African genera with only one or two species, Madagascar has 86.6% of southern Africa’s and 89.4% of south-central Africa’s grass genera. C(4) grass species make up c. 4% of the flora of both Madagascar and southern Africa and species : genus ratios are similar (4.3 and 5.1, respectively). Turnover of grasses along geographical gradients follows similar patterns to those in South Africa, with Andropogoneae dominating in mesic biomes and Chlorideae in semi-arid grassy biomes. At least 16 monocot genera have grassland members, many of which are endemic to Madagascar. Woody species in frequently burnt savannas include both Madagascan endemics and African species. A different woody flora, mostly endemic, occurs in less frequently burnt grasslands in the central highlands, filling a similar successional niche to montane C(4) grasslands in Africa. Diverse vertebrate and invertebrate lineages have grassland specialists, including many endemic to Madagascar (e.g. termites, ants, lizards, snakes, birds and mammals). Grassland use of the extinct fauna is poorly known but carbon isotope analysis indicates that a hippo, two giant tortoises and one extinct lemur ate C(4) or CAM (crassulacean acid metabolism) plants.

Main conclusions The diversity of C(4) grass lineages in Madagascar relative to that in Africa, and the presence of plant and animal species endemic to Madagascan grassy biomes, does not fit the view that these grasslands are anthropogenically derived. We suggest that grasslands invaded Madagascar after the late Miocene, part of the world-wide expansion of C(4) grassy biomes. Madagascar provides an interesting test case for biogeographical analysis of how these novel biomes assembled, and the sources of the flora and fauna that now occupy them. A necessary part of such an analysis would be to establish the pre-settlement extent of the C(4) grassy biomes. Carbon isotope analysis of soil organic matter would be a feasible method for doing this.