Proceedings of the Royal Society B: Biological Sciences
You have accessInvited reply

Beyond ancient versus anthropogenic for Madagascar's grassy ecosystems. A Reply to: Crowley et al. (2021)

A fundamental challenge for the natural sciences is understanding the distribution of ecosystems. Nowhere has discourse been more contested than in Madagascar around perceived ancient versus anthropogenic grassy ecosystems. Data collected across different disciplines and at varied spatio-temporal scales have produced conflicting inferences about the mosaic of Madagascar's vegetation before the human settlement of the island. Crowley et al. [1] describe new palaeo-records demonstrating extinct hippos were associated with grassy ecosystems and with different abundances of grassy ecosystems containing a C4 component associated with different bands of elevation across the Central Highlands. These and other recently published data from the Central Highlands [1,2] support the findings of Solofondranohatra et al. [3] that grazing associated with grassy ecosystems existed in some forms prior to human settlement of the island, but open questions remain around the extent. Here, we propose a research agenda and ecological considerations to help advance understanding of the distribution and dynamics of ecosystems across Madagascar.

Persistent questions remain around ecosystem definitions, limits and their dynamism in Madagascar, and Crowley et al. [1] note the complicated nature of resolving vegetation mosaics and their dynamism through time. Within this frame, it is important to note that Madagascar is a case study in evolving global discourse around ecosystem definitions and their application with respect to grassy ecosystems [46] and a false dichotomy between ancient versus anthropogenic ecosystems. Not only in Madagascar, but the world over, such as in India and Southeast Asia, grassy ecosystems have been interpreted through colonial lenses and in utilitarian terms, that have impacted ecological research and understanding [7]. Questions of ecosystem delimitation and dynamism are not esoteric discovery science but central to conservation priorities [8], land-use policies and nature-based mechanisms for climate change mitigation involving tree planting [5]. The debate in Madagascar, which can only be resolved through interdisciplinary collaboration, has consequences for landscapes across the tropics of the Global South where ecosystems and livelihoods are threatened by compounded misunderstandings and global change [9,10].

The extent of grassy ecosystems prior to human settlement of Madagascar remains a persistent unknown. Reconstructions or delimitations of pre-human settlement ecosystems are inherently complicated by ecosystems and landscapes being dynamic entities. Ecological and palaeo-evidence demonstrates the capacity for substantial ecosystem and species shifts over decadal, centennial and millennial timescales in the absence of direct human influence. Change in vegetation structure and limits may appear slow over a human lifetime leading to shifting perceptions and unawareness of how natural and anthropogenic drivers modify ecosystem limits through time [11]. Meaningful quantification of grassy ecosystem extent prior to human settlement is complicated by contemporary accelerating and extreme rates of land transformation. Accelerating rates of land transformation are not unique to Madagascar and its human occupation history but reflect global acceleration across the tropics over the last century [12,13]. Crowley et al. [1] recognize the complicated nature of vegetation reconstructions. We suggest that, along the lines of the new palaeo-research findings by Crowley et al. [1], Samonds et al. [2] and the seminal work of Gasse & Van Campo [14], next steps are to move beyond antiquated dichotomies. Collaborative inter-disciplinary research spanning the physical, natural and social sciences is needed to address the following questions. (i) To what extent were grassy ecosystems a component of the island's pre-history? And, what processes drove grassy ecosystem assembly and dynamics over the wide environmental gradients of the island? (ii) Can the spatial distribution of grassy ecosystems prior to human settlement be separated from the more recent trends and locations in land transformation since French colonization and over the last century to gain a robust understanding of the dynamism of ecosystems over time? (iii) What are the discipline-specific biases to understanding grassy ecosystems and mosaic vegetation? And, how can these biases and limitations be resolved or minimized via inter-disciplinary collaboration? Between disciplines, discourse can confound centennial versus millennial vegetation shifts and terms such as ‘mosaic vegetation’ or ‘woodland’ are widely applied but can be understood at cross-purposes among disciplines. Stable carbon isotopes of bone and tooth enamel have been instrumental to inferring past vegetation, but nuanced inference would benefit from the examination of how diverse C3 plant life-history strategies relate to both livestock and animal diets and soils. C3 plants have a wide diversity of life-history strategies related to both open and closed ecosystems, meaning investigation of C3–C4 isotope records from an ecological point is important especially as many animals are highly selective in their feeding habits. Functional ecology meshed with biodiversity analyses can help understand how the diversity of plant life-history strategies relate to community assembly, usefully applied by Solofondranohatra et al. [3] to infer plant–animal relationships even though, recognition of assemblages cannot delineate past ecosystem extent. These ecological methods should be applied to a wider set of ecosystems and regions in Madagascar related to both fire and herbivory and could help target palaeo-research. A nuanced understanding of how site-level dynamism through the Pleistocene and Holocene reconciles with centennial and decadal variation observed by written records, remote sensing, landscape ecology and ecological experimentation is much needed. (iv) How do modern Malagasy grassy ecosystems support and contribute to livelihoods? And, what are the threats to livelihoods in grassy ecosystem regions? Given the precarious livelihoods and widespread malnutrition of the region, this last question is paramount. Work on the political ecology of these systems is limited [15], and on climate change impacts it is further limited. Given global patterns and trends, there are highly likely existing impacts on drought, fire regimes, ecosystem productivity and key plant and animal species central to livelihoods [16,17]. Invasive plants and the propensity for the planting of Eucalyptus and Pinus for fuel wood will have lasting impacts as ecosystem transformers [18] such that future ecosystems may little resemble the present. Although, different experiences of invasive species can equate to different perspectives and values of them [19].

With respect to understanding vegetation distributions, ecological research has developed the theory to consider how vegetation–disturbance interactions and feedbacks in both woody plants [20] and grasses [21] structure ecosystem mosaics at local scales over years [22] to modifying biome limits at a global scale over millennia [23,24]. The biogeography of the southern continents demonstrates that closed-canopy vegetation of shrublands, thickets and forests can be found in almost any climate across the tropics [25]. Interactions between the likelihood of fire/herbivore disturbance and plant growth strategies see that where seasonal drought, mean annual rainfall from 500 to 2000 mm and disturbance combine the likelihood of grassy ecosystems is approximately 80% based on analyses of existing vegetation maps across Australia, Africa and South America [20,25]. Comparing the modern limits of grassy ecosystems across these continents relative to Madagascar demonstrates modern Malagasy grassy ecosystem limits are like Australia and with substantially overlapping distributions to Africa and South America (figure 1). Grassy ecosystems are readily considered ancient in these other regions, each with unique flora and fauna [10]. Although, across all continents, fundamental assessment of floral and faunal diversity of grassy ecosystems needs urgent investigation given the current rates of grassy ecosystem transformation and to address an often-implicit assumption that low diversity equals recent assembly or less value even where different processes drive ecosystem assembly [28].

Figure 1.

Figure 1. Distribution of grassy biomes occurrence in Africa, Australia and South America along continental rainfall gradients relative to Madagascar based on existing vegetation maps (reproduced from [26]). Data for Australia, Africa and South America are originally from [25] and Madagascar from [27]. (Online version in colour.)

African biogeography has much to offer Madagascar with respect to predicting potential pre-human settlement location and limits of grazing associated grassy ecosystems (figure 1). The southern continents have distinct evolutionary histories in their vegetation and megafauna [25,29]. Megafauna in South America, Australia and Madagascar became at least selectively extinct in association with human colonization. Africa is the odd one out, a continent where megafauna and humans long coexisted in contrast with the comparatively short timescale of megafauna–human interactions elsewhere. In Africa, mammalian grazing megafauna and the grassy ecosystems they structure are predominantly found in the semi-arid parts of the continent [30]. The higher rainfall zones of continental Africa share similar climates and highly weathered soils with Madagascar's Central Highlands [25,27]. Where grazer-dominated grasslands do occur in higher rainfall continental Africa, they tend to be restricted to nutrient hotspots or associated with rivers and water bodies, enabling congregation and intensive concentrated grazing (e.g. [31]). Based on these data, and the feedbacks that structure grazer determined grassy ecosystems along gradients in productivity [21,30], we can infer that in the Central Highlands, such grazer-associated grasslands likely were restricted. In comparison, most prevalent in this environmental space are large expanses of fire-determined grassy ecosystems with a variable woody component [10]. Although the apparent absence of any large-bodied predator in Madagascar and therefore the potential absence of a landscape of fear [32] likely impacted vegetation limits and their dynamism in ways yet to be considered. Consequently, vegetation in Madagascar would possibly have been more homogeneous over wider spatial extents in comparison to the semi-arid regions of Africa characterized by contingent and heterogeneous patterns of woody cover.

Closed versus open ecosystem occupancies of the drier environmental spaces across the southern continents (figure 1) also provides important pointers to the potential dynamism of biomes over millennial time scales with the loss of megafauna. In South America, Australia and Madagascar, these arid and semi-arid regions have extensive areas characterized by shrublands, thickets and dry forests. In Africa, this same environmental space is more widely characterized by herbivore-dominated grassy ecosystems but by lineages of animals never present in Madagascar [33]. Where hippos are important, it is around and in proximity to water bodies, but hippos are just one species critical to grassy ecosystems [34]. Palaeo-data have provided limited support for grassy ecosystems in this region of Madgascar [35] despite the presence of animals physiologically likely to be able to have consumed grasses (e.g. elephant birds, tortoises) and an extant and diverse array of endemic grass species [26]. Understanding this arid transition in Madagascar and how closed and open ecosystem limits have reorganized in the face of megafaunal loss in Australia, South America and Madagascar is an important unresolved question hampered by limited ecological investigation, a limited palaeo-record and the unique faunal assemblages of each region.

We support the ongoing important research by Crowley et al. and look forward to helping solve critical unknowns for informed and appropriate management and conservation. Binary classification of modern ecosystems as either ancient or anthropogenic is an outdated and harmful approach to ecology, conservation and land management that risks discriminating against dynamic ecosystems. It also risks blaming people for ecosystem changes even when these changes are natural or neutral. The antiquated colonial vision of a pristine and exotic pre-human Madagascar paradise needs to give way to a nuanced realistic understanding of socio-ecological systems led by Malagasy scientists.

Authors' contributions

C.E.R.L. wrote the manuscript with input and comments on the draft from C.L.S. and M.S.V.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.


This study was funded by the Office of the Royal Society (grant no. IC170015).


The accompanying comment can be viewed at

Published by the Royal Society. All rights reserved.