ANALYSISSustainable management of extensively managed savanna rangelands
Introduction
Savannas are defined as tropical ecosystems where grasses and trees co-dominate (Huntley and Walker, 1982). They cover large proportions of the tropical continents, that is 65% of Africa, 60% of Australia and 40% of South America (Huntley and Walker, 1982). The large area covered by savannas means that their sustainable management is of regional and global concern. One of the main land-use activities practised in savannas is livestock production, yet consensus on what might constitute sustainable livestock production systems remains elusive (Vetter, 2005).
Defining sustainable land-use systems requires careful consideration of what one means by sustainability. Following the Brundtland Report (WECD, 1987), sustainable development aims to guarantee inter- and intragenerational fairness concerning the use of natural resources. Hence a thorough evaluation of sustainable development requires the consideration of economic, ecological and ethical factors in an integrated framework. Economic sustainability typically means that resources should be managed in such a way that the utility does not decline over time (Perman et al., 2003). Ecological sustainability may mean preserving ecological resilience over time or ensuring that the flow of some ecological service does not decline over time (Daily, 1997). For instance, in a grazing system maintaining ecological resilience might involve preserving the soil layer or preventing the system from moving into a tree dominated state; while sustaining ecological services may involve maintaining large trees that act as keystone species. The interpretation of sustainability also varies among actors, because different actors have different risk aversions, discount rates and constraints. For instance, a farmer might maximise utility by maximising profit while discounting future earnings strongly; whereas a section of society might maximise utility by maximising the stability of production.
The focus of this study is the sustainable management of extensive rangeland systems in savanna regions. That is, we consider systems in which supplementary feeding of livestock is not economically viable and where grass and livestock production are primarily limited by rainfall. Rainfall in savanna regions varies with season; most rain falls in the summer months, very little rain falls in the winter months. Rainfall also varies stochastically between years. As a consequence, livestock farmers in savanna regions must develop strategies for dealing with variability and uncertainty in resource supply. Theory based on the harvesting of a stock that has density dependent and deterministic growth has strongly influenced the management of savanna rangelands (Mentis, 1984). However, more recent studies (Ellis and Swift, 1988, Behnke and Scoones, 1993) have argued that the stochastic nature of rainfall, which is the primary driver of resource supply in savannas, make such theories inappropriate for savannas.
This tension between “classic” and “new” rangeland theories (Cowling, 2000) has often been cast as a conflict between equilibrium and non-equilibrium paradigms, or between stochastic and deterministic paradigms. However, such polarisation is counter productive, because in any dynamical system both non-equilibrium or transient dynamics and equilibrium dynamics are important. Similarly, many dynamical systems are influenced by a mixture of both stochastic and deterministic processes. In savannas, the equilibrium–non-equilibrium dichotomy is particularly inappropriate because even when a savanna is not at equilibrium, its trajectory in state space is determined by the system's equilibrium points. The stochastic–deterministic dichotomy is also inappropriate in savannas because while the stochastic effects are apparent (e.g. rainfall on grass production), deterministic linkages are equally undeniable (e.g. grass consumption on animal production).
The broad aim of this study is to propose a model of savanna rangelands that combines a mix of the deterministic and stochastic, equilibrium and non-equilibrium processes that form the keystones of the “classic” and “new” rangeland paradigms. We use this model to explore guidelines for the sustainable management of livestock production systems in savannas. Our intent is to explore the extent to which different aspects of sustainability are compatible with one another. Specifically, we aim to define strategies that optimise the sustainability from the perspective of agents that are motivated by economic, ecological and production factors under deterministic and stochastic environmental conditions.
Section snippets
Ecological sub-model
We create a discrete time model, inspired by the models of Perrings and Walker (1997) and Janssen et al. (2004). The most important difference between our model and previous models is that we model, for both grasses and trees, two biomass compartments, roots and shoots. This allows us to simulate the fact that herbivores and fire cannot consume roots, and allows a separation of below ground and above ground competition (Fig. 1). The model of Perrings and Walker (1997) does not separate above
Utility
Given the different definitions of sustainability outlined in the introduction the agent may either aim to increase income, increase production or increase the value of some ecological indicator. A farmer may be interested in the economic definition, the state may be interested in the production definition and the ecological definition. In the paragraphs that follow we propose functions for the utility of these different agents.
The economic utility is determined by income and costs, which are a
Optimisation
The model is non-linear, stochastic and has multiple domains of attraction; these attributes make it poorly suited to classic optimisation methods (e.g. optimal control or dynamic programming). As an alternative, we use a simulation–optimisation approach, that is we combine our process orientated simulation model with a numerical optimisation algorithm. Simulation–optimisation methods have been successfully used to optimise hydrological (Aly and Peralta, 1999, Koutsoyiannis and Economou, 2003),
Model behaviour
The system has three attractors that are of ecological interest, grass dominated, tree dominated and savanna (mixed tree–grass) states. Scheiter and Higgins (submitted) provide a more formal discussion of the mathematical conditions necessary for the savanna state. These analyses show that the savanna state is possible when either fire is present or when grass–tree competitive interactions are weak. The weight of empirical evidence for savannas, however, supports the notion that competition
Discussion
The new rangeland ecology criticises classic rangeland ecology's recommendation of constant stocking on the basis that this represents a lost opportunity to rangeland farmers, preventing them from taking advantage of bumper grass production years (Behnke and Scoones, 1993). One alternative to constant stocking is to attempt to force animal numbers to track variation in grass production. Although intuitively appealing, the feasibility of the tracking strategies suggested by the new rangeland
Conclusions
In this paper we analyse a dynamic model that enables us to analyse a rangeland system under stochastic conditions. The model illustrates how both deterministic and stochastic processes influence rangeland dynamics. The results of our model analysis are that under stochastic conditions livestock farmers should be more conservative than under deterministic conditions. This result is perhaps surprising for those (e.g. Ellis and Swift, 1988, Westoby et al., 1989, Behnke and Scoones, 1993, Sullivan
Acknowledgements
We thank the Robert Bosch Foundation for financial support and the reviewers for valuable comments.
References (37)
Challenges to the ‘new’ rangeland science
Trends in Ecology and Evolution
(2000)- et al.
Evaluation of strategies for tracking climatic variation in semi-arid grazing systems
Agricultural Systems
(1998) - et al.
Robust strategies for managing rangelands with multiple strange attractors
Journal of Environmental Economics and Management
(2004) - et al.
Biodiversity, resilience and the control of ecological economic systems: the case of fire-driven rangelands
Ecological Economics
(1997) - et al.
Interdependence of ecological risk and economic profit in the exploitation of renewable resources
Ecological Modelling
(1985) Rangelands at equilibrium and non-equilibrium: recent developments in the debate
Journal of Arid Environments
(2005)- et al.
Comparison of a genetic algorithm and mathematical programming to the design of groundwater cleanup systems
Water Resources Research
(1999) DEoptim: differential evolution optimization
R Package Version 1.1–2
(2005)- et al.
Rethinking range ecology: implications for range management in Africa
- et al.
The global distribution of ecosystems in a world without fire
New Phytologist
(2005)
A limited memory algorithm for bound constrained optimization
SIAM Journal of Scientific Computing
Cattle population dynamics in the southern Ethiopian rangelands, 1980–97
Journal of Range Management
Stability of African pastoral ecosystems — alternate paradigms and implications for development
Journal of Range Management
Fire, resprouting and variability: a recipe for grass–tree coexistence in savanna
Journal of Ecology
Ecology of Tropical Savannas
On the relevance of nonequilibrium concepts to arid and semiarid grazing systems
Ecological Applications
Evaluation of the parameterization–simulation optimization approach for the control of reservoir systems
Water Resources Research
Cited by (45)
Modeling rangelands as spatially-explicit complex adaptive systems
2020, Journal of Environmental ManagementSustainable grazing management in rangelands: Over a century searching for a silver bullet
2019, Agriculture, Ecosystems and EnvironmentHow Does Climate Change Influence the Economic Value of Ecosystem Services in Savanna Rangelands?
2019, Ecological EconomicsCitation Excerpt :Scheiter and Higgins (2009) and Scheiter et al. (2012) show that the aDGVM can simulate the current distribution of vegetation in Africa better than alternative dynamic vegetation models. Scheiter and Higgins (2009) show that the aDGVM can reproduce biomass measurements obtained from a long-term fire manipulation experiment in the Kruger National Park (Experimental Burn Plots, Higgins et al., 2007). In Scheiter et al. (2018) we investigated the risk of biome shifts in the Limpopo Province, South Africa, induced by various climate change scenarios.
Global effectiveness of group decision-making strategies in coping with forage and price variabilities in commercial rangelands: A modelling assessment
2018, Journal of Environmental ManagementHow vulnerable are ecosystems in the Limpopo province to climate change?
2018, South African Journal of BotanyGrazing ecology and the conservation of the Caldenal rangelands, Argentina
2016, Journal of Arid Environments