There have been significant declines in the perennial grass content in native and sown pastures across temperate Australia. Not only has this reduced agricultural productivity, but has contributed to a range of external costs associated with serious degradation such as loss of soil and biodiversity, decreasing water quality, and dryland salinity caused by rising watertables. This paper presents an economic framework for examining a range of on-farm production and environmental issues in a perennial grazing system in the New South Wales temperate perennial pasture zone. This involves a combination of simulation and dynamic programming models, with the state of the system represented by variables for the perennial grass composition and soil fertility. The paper considers a range of management strategies that increase the perennial grass composition in terms of net income from grazing, and the impact upon externalities. The study concludes that long-term economic returns are improved by strategies that lead to an increase in perennial grass composition over time. The study also determined that environmental factors such as deep drainage, runoff and soil loss are reduced as perenniality is increased. However, the study suggests that it not appropriate to claim whether the grazing systems are actually ‘sustainable’. The concept of a sustainable agricultural system can only be considered at an industry or a catchment level, not at paddock or farm scale studies as reported here. Consequently, we suggest that there are positive economic and environmental effects from the adoption of grazing systems that result in greater perenniality, however we are unable to discern the overall impact upon the catchment processes from such adoption and whether agriculture is more sustainable as a result.

The temperate perennial pasture zone (TPZ) in New South Wales, Australia (>600 mm annual rainfall) is a major sheep and cattle production area. This region represents about 5 million hectares, sustains 23 million sheep and 3.5 million cattle, and contributes significantly to the New South Wales economy (Vere et al., 2002). There are a number of serious land degradation issues now facing agricultural systems in this region, in particular the decline in the perennial grass component and a range of external costs through changed water use patterns.

Pasture degradation due to a decline in the perennial grass content may be characterised by increased variability in production over time, poorer summer forage supply, weed invasions, more rapid soil acidification and the increased development of salinity through poor management of the watertable (Kemp et al., 2000). Vere (1998) found that there has been a decline in the productivity of improved pastures in the TPZ. This was largely attributable to the decline in perennial grass and legume pasture productivity due to invasion by annual grass weeds.

Soil related problems pose a significant sustainability challenge for the temperate perennial grazing industries with significant impacts already evident. For example, earlier studies found that soil acidity was limiting crop and pasture production on an area exceeding 7 million hectares (Helyar et al., 1990). The TPZ is also a major source of watertable accessions within the Murray-Darling Basin, the largest and most developed river system in Australia and accounting for 13% of the continent. Tree clearing to develop native and sown pastures for livestock grazing is suspected of increasing the threat of salinisation lower in the catchment through contributions to rising ground water levels (Gates and Williams, 1988).

The replacement of shallow rooted annual species with perennials can substantially reduce soil acidification and water losses from deep drainage in pasture systems (Ridley et al., 2001; Ward et al., 2001). Pasture systems with a high proportion of perennial species have also been demonstrated to provide significantly higher financial returns (Dowling et al., 2001; Dowling and Jones, 2002). Consequently, developing and maintaining a strong perennial grass component in pastures is a key objective to achieving a sustainable grazing system.

There are a number of management options that can be adopted by landholders to achieve this objective. These include reducing stocking rates of sheep and cattle so that grazing pressure on desirable perennial grass species is minimised, tactical rests to rehabilitate the perennial component, resowing of the pasture with perennial species when the composition gets to a low level, and herbicides to control undesirable weed species such as annual grasses (e.g. Vulpia spp., Bromus spp.) and broadleaf weeds (e.g. Echium spp.). Also, soil fertility levels can be manipulated through the application of phosphorus and nitrogen fertilisers, as well as lime, so as to manipulate the competitive relationships between the desirable and undesirable species. Finally, at a catchment level a policy of withdrawing parts of the landscape from agriculture and revegetation with trees for a better watertable balance may be considered.

Michalk et al. (2003) presented the results of a multi-disciplinary research program conducted in central New South Wales over the period 1997-2001. The aim of this research was to develop more profitable and sustainable management systems for perennial grass based pastures by imposing tactical management strategies. The treatments were based upon the application of phosphorus fertiliser and strategic grazing rests where paddocks were destocked over summer.

The concept of strategic grazing rests for improving the composition of perennial grasses in a pasture system was developed from a number of earlier studies (Dowling et al., 1993; Dowling et al., 1996; Kemp et al., 1996). In a broader context, modifying botanical composition by imposing management changes such as fertiliser inputs or by varying grazing strategies (stocking rate, various forms of rotational grazing) has been known for some time in European agriculture (Jones, 1933; Voisin, 1960). Implementation of this approach in temperate Australia has been delayed largely because good husbandry was not required when desired botanical composition could be reimposed by simply replanting the pasture. A decline in the cost-effectiveness of replanting over the last 20 years (Vere et al., 1997), and the increasing need to retain perenniality in pastures (Ridley et al., 1997) based on either exotic or native grass species has forced a reassessment in the way we manage pastures, and the need to regard pastures as a long-term investment.

More recent studies have confirmed the utility of grazing rests as a means of varying pasture composition, and increasing and maintaining long-term perenniality (Graham et al., 2000; Virgona et al., 2000; Michalk et al., 2003), though universality of response by species and season would appear to be dependent on local conditions (Kemp et al., 2000). The response by perennial grasses is closely associated with the timing of the grazing rest in relation to reproductive activity of the target species, and is enhanced by longer rests and more favourable conditions at the time of the grazing rest (Virgona and Bowcher, 2000).

The objective of this paper is to present an economic framework suitable for evaluating the economic benefits of technologies that increase perenniality and to examine the sustainability issues associated with grazing systems in the TPZ. A number of policy options for achieving increased perenniality are considered. These are a polluter-pays and beneficiary-pays approaches, and direct regulation.

A case study approach is used to determine the economic benefits and sustainability impacts of fertiliser and grazing management tactics in the TPZ. The case study region is the Central Tablelands of New South Wales, Australia. This region was selected because of the availability of suitable soil data and long-term daily weather data as well as the results from a number of locally based experimental studies.

Methods

Defining sustainability

There are many definitions of sustainability or sustainable economic development. Also, the term sustainability has a wide range of interpretations with Graham-Tomasi (1991) pointing out ‘just about everyone is on the sustainability bandwagon, and sustainability has come to mean all things to all the riders on this bandwagon’.

The most widely used definition of sustainable development is that put forward by the World Commission on Environment and Development, which is ‘development that meets the needs of the present without compromising the ability of future generations to meet their own needs’ (WCED, 1987, pp. 43). Many more definitions exist, with Pezzey (1992) listing 27 definitions of sustainable development. From an Australian perspective the National Strategy for Ecologically Sustainable Development defines ecologically sustainable development as ‘using, conserving and enhancing the community’s resources so that ecological processes, on which life depends, are maintained, and the total quality of life, now and in the future, can be increased’ (Commonwealth of Australia, 1992, pp. 6).

There are two potential versions of sustainability; strong sustainability and weak sustainability (Turner et al., 1994, pp. 55; Stoneham et al., 2003). The main difference between these two versions is the ability to make trade-offs between the various forms of capital (natural, human and social). In the case of strong sustainability the stocks of natural capital must not be depreciated over time, i.e. the same stock of resources available to current generations will be available to future generations. The concept of weak sustainability allows for substitution between the different forms of capital stock, thereby allowing an overall economic system to be sustainable by trading-in some forms of capital for others.

Pannell and Schilizzi (1999) in referring to the multitude of definitions in the literature conclude that sustainability is not a useful criterion in itself for planning or decision making. However, the term sustainability is useful as an emblem for other and clear criteria which capture the relevant issues pertaining to resource and environmental management.

Rather than be overly concerned with exact definitions of sustainability or sustainable development, a better approach may be to accept the goal of sustainability in a general sense to be an improvement in the productive performance of a system without depleting the natural resource base upon which future performance depends (Pearce and Turner, 1990, pp. 24). This is consistent with the views of Graham-Tomasi (1991) and Pannell and Schilizzi (1999) that sustainability should be viewed as a goal rather than as a set of actions or technologies. Sustainability generally involves several separate issues such as protection of ecological systems, inter-generational equity and efficiency of resource use (Pannell and Schilizzi 1999), valuation of environmental assets and recognition of constraints implied by the dynamics of environmental systems (Heal 1998, p48). Given the diversity of issues, sustainability is difficult to measure and it is not possible to identify a few key variables at the farm level which adequately gauge sustainability (Graham-Tomasi 1991).

Heal (1998, pp. 48) suggests that the essence of sustainability is defined by the following three axioms.
(i) A treatment of the present and the future that places a positive value on the very long run.
(ii) Recognition of all the ways in which environmental assets contribute to economic well-being.
(iii) Recognition of the constraints implied by the dynamics of environmental assets.

A bioeconomic model of sustainable grazing

Understanding the long-term implications of managing a resource or environmental asset is one of the key components of the sustainability problem. Pandey and Hardaker (1995) present a framework that incorporates the inter-temporal tradeoffs for evaluating the sustainability problem into a bioeconomic model.

(1)

subject to:

Where J is the discounted sum of the performance measure over the planning horizon T, t is an index for year, ∏ is a measure of farm performance, x is the stock of natural resources (state variables), u is the set of management decisions (control variable), δ is the discount factor (δ = 1/(1+r), r is the discount rate), and g is the measure of the change in the stock of the natural resources over time, which depends on the stock size and the management decisions. The final constraint (equation 4) reflects the requirement of a strong sustainability definition, where the stock of the resource is no less than some specified stock of the resource (X_T). In the case of a weak sustainability system the value of X_T in equation 4 is set to zero, with emphasis being on valuing the trade-offs between the sets of resource capital.

Heal (1998, p36) argues that society may experience economic value from the existence of a resource or environmental good in addition to the benefits derived from consumption of the stock of the resource over time. An example is a forest which is conventionally valued from the flow of wood for consumption. However, the forest can also be valuable as a source of biodiversity, carbon sequestration services, and as recreational facilities. Consequently, it may be important to include the preservation benefits of a resource along with any consumption benefits. In the case of a perennial grass based grazing system, environmental (i.e. preservation) benefits associated with the stock of a perennial grass resource can be attributable to reduced deep drainage to the watertable (and thereby less catchment salinity), improved quality of streamflow runoff, and greater biodiversity.

A dynamic programming model was developed for the sustainable grazing management problem and is outlined as follows.

(5)

subject to 
 

Where PG is the proportional composition of perennial grasses in the pasture, P is the level of soil phosphorus, SR is the livestock stocking rate (head ha^-1), and F is the amount of phosphorus fertiliser applied (kg ha^-1).

Three policy approaches are used to investigate the sustainability problem in perennial grazing systems.
(i) The outputs of deep drainage, runoff and soil loss are considered as external costs in the objective function. This treats the issue as a polluter-pays approach to the problem. Although deep drainage and poor quality runoff are both diffuse forms of pollution and in practice it would be difficult to identify and impose pollution taxes, this analysis is useful in terms of determining the potential change in farmer behaviour from such a policy action.
(ii) Following the approach of Heal (1998) a preservation benefit associated with the variable PG is included in the objective function. This represents a beneficiary-pays approach to the problem if compensation or direct subsidies were used as the preservation benefit payment.
(iii) A strong sustainability constraint (equation 4) is imposed, which restricts PG to be greater than some specified value. Direct regulation in environmental issues is often of this type where limits or targets on resource use are imposed by government.

The interactions between the species composition and soil fertility models, which represent the differential equations for the state variables, with the water balance, grazing systems and dynamic programming models are illustrated in Figure 1. The water balance and grazing systems models are calculated on a daily time step whereas the species composition and soil fertility models are calculated annually. All these biological models combine to provide the necessary data for the state transitions and objective function (π) of the dynamic programming model.

The objective function of the dynamic programming model included both private and public benefit components. The model results in a socially optimal solution of resource use when the environmental costs resulting from grazing management are included. 
 

(10)

Where LR is livestock revenue, LC is livestock production costs, SFC are supplementary feed costs, PVC are pasture variable costs, and FC are fertiliser application costs. The external impacts upon the environment are captured through the variables on the right hand side of equation 10. P_RO are unit costs associated with runoff ($ mm^-1), RO is runoff amount (mm ha^-1), P_DD are unit costs associated with deep drainage to the watertable ($mm-1), DD is the amount of deep drainage or water lost to the watertable (mm ha^-1), P_SL are unit costs associated with soil loss ($ t^-1), SL is the amount of soil loss (t ha^-1), and B_P is the preservation benefit associated with PG ($ ha^-1). A privately optimal objective function results from setting the values of P_RO, P_DD, P_SL and B_P to zero. The polluter-pays approach to the sustainability problem involves setting non-zero values for P_RO, P_DD and P_SL. For the beneficiary-pays approach these values are maintained at zero and a positive value for B_P is included. 

Figure 1. The sustainable grazing modelling system

The use of an external cost approach to the sustainability question is appropriate in this study for a number of reasons. Given the small scale of the analysis (i.e. hectare, paddock, farm), the environmental effects of management are mostly off-site and cannot be represented adequately through a state variable in the dynamic programming model. This problem could be resolved by setting the decision problem to be at a regional or catchment scale and including a state variable for the watertable or area/proportion of the catchment that is salinised. However, the processes between on-farm water management and regional watertables in the study region are poorly understood so there is little to be gained from increasing the level of complexity in this manner at this time.

The management decisions included in the model involve a range of variables that influence both species composition and economic returns. These are: livestock stocking rates, pasture establishment, tactical grazing rests and fertiliser options. A description of the decisions is given in Table 1. In this study the livestock system is represented by a merino wether (i.e. sheep) enterprise.

Table 1. Description of grazing management decisions

The species composition model

The state of the pasture system is defined by three species functional groups; perennial grasses (PG), legumes (LG) and annual weeds (AW). The PG and LG comprise the set of desirable species in the grazing system, while AW represents the set of undesirable species. The PG functional group includes all C3 and C4 (native and introduced) perennial grass species, while AW represents annual grasses (e.g. Vulpia spp., Bromus spp.) and broadleaf weeds (e.g. Echium). Combining the annual grasses and broadleaf weeds simplifies the modelling process and is biologically justifiable given the similar responses by both annual grasses and broadleaf weeds to the management options evaluated (e.g. Dear et al., 1998).

A pasture state variable is defined to represent the composition of the desirable perennial grass species. This variable takes values between zero and one, and represents the proportional composition of the pasture biomass in early spring. Given that all species are at a similar stage of growth at this period it provides the best representation of the space, or ecological field, occupied by the different functional groups.

The remaining space is assumed to be occupied by legumes and annual grass and broadleaf weeds. The legume composition was assumed to be in proportion to the composition of perennial grass. The following polynomial equation was used to calculate the legume composition (LG) and was based upon field data collected in the Central Tablelands. The composition of annual weeds (AW) was simply the difference between one and the summation of PG and LG.
 

(11)

A growth equation approach (Fitzpatrick and Nix, 1970) was adopted to measure the annual rate of change in PG (ΔPG). Two separate logistic equations were used to calculate ΔPG, depending on whether the value of PG is above or below a pre-specified asymptote. If composition was below the asymptote a growth equation was used, whereas if composition was above the asymptote a decay function was adopted to calculate ΔPG.

(12)

Where FI is a fertility index, a is an asymptote while b1, b2 and b3 are shape parameters for the logistic equations. The annual change in the value of the species state variables is derived from the following differential equation:

(13)

The soil fertility model

The fertility index was based upon the approach of Holford (1980) for estimating relative yield as a function of soil phosphorus (Bray1). Holford defined relative yield as the proportion of maximum (unconstrained by phosphorus) yield. The FI was calculated as follows:

(14)

Where a₁ is a slope coefficient, a₂ is the asymptote and P is the level of soil phosphorus measured in parts per million (mg kg^-1). The FI variable is used in the calculations of the daily growth of specific pasture species and the calculation of the rate of change in the composition of perennial species. Following the findings of Holford and Crocker (1988) separate FI were calculated for ΔPG as well as for the pasture growth rate equations for perennial species, legumes and annual weeds.

The effect of applied fertiliser is to raise the amount of phosphorus in the soil, while there is an ongoing annual loss in phosphorus due to grazing. The calculation of the annual transition in soil P follows that of Kriticos (unpublished), which is partly based upon the findings of McCaskill and Cayley (2000). The amount of phosphorus at the start of a year (P_t) is a function of phosphorus in the previous year (P_t-1), the annual loss in phosphorus (P_loss) and fertiliser applied (P_fert).

(15)

Where P_c is the proportion of P available in phosphorus fertiliser (P_c = 0.091), P_fert is measured in terms of kg ha^-1 and the remaining variables in mg kg^-1. The variable P_loss is represented by the following decay function.

(16)

Where a₃ is a decay coefficient (a₃ = 0.1852) and P_min is a minimum amount of phosphorus fertiliser available from non-expendable pools of phosphorus (P_min = 3).

The water balance model

The environmental parameters used in growth index calculations and the environmental effects from grazing management systems were derived from the PERFECT model (Littleboy et al., 1999). PERFECT (Productivity, Erosion and Runoff Functions to Evaluate Conservation Techniques) is a biophysical model that simulates the plant-soil-water-management dynamics in an agricultural system. In addition to calculating daily soil moisture the model was used to determine runoff, soil loss and deep drainage to the watertable for a range of perennial and annual vegetation systems.

PERFECT uses daily weather inputs for precipitation, maximum temperature, minimum temperature, pan evaporation and solar radiation. Weather data were obtained for the period 1930 to 2002 for Orange, Central Tablelands, (Lat -33.28, Long 149.10 decimal degrees) from the Silo dataset (http://silo) for use in the model.

Given the diversity of soil types within the study region, the PERFECT model was solved for two distinct soil types; ferrosol and kurosol. Ferrosol soil types are derived from basalt, are highly fertile and have high water holding capacity. Kurosol soil types are granite based and have greater drainage potential than ferrosols. Both soil types are widespread within the TPZ.

The grazing systems model

The interaction between the growth of individual pasture species and livestock feed requirements was determined using a daily time step grazing systems simulation model. This model system also determined the financial returns from the livestock enterprise, including the cost of supplementary feeding when pasture feed supply was less than livestock daily demand. This model is described in detail by Jones and Dowling (submitted).

Results

Deep drainage, runoff and soil loss

The effect of the level of perennial grass composition upon deep drainage, runoff and soil loss was simulated with the PERFECT model. Two pasture scenarios were considered; (1) a pasture system that was comprised completely of perennial species, and (2) a pasture system that was comprised completely of annual species. These two scenarios represent the extremes of the composition of a pasture in the TPZ, and the actual values used in the bioeconomic model for pasture with a mixture of annuals and perennials were obtained by linearly interpolating these results.

The PERFECT model was solved for a 72-year period 1930-2002, with the summary statistics reported in Table 2. The amount of excess water (the sum of runoff and deep drainage) is largely determined by the vegetation system, being considerably less under a perennial pasture than an annual pasture. There is a soil type effect on excess water, with the ferrosol resulting in less excess water than the kurosol soil for an equivalent pasture scenario. The partitioning of the excess water is largely determined by the soil type, with a ferrosol soil having less deep drainage and more runoff than a kurosol.

Soil loss is heavily influenced by both the pasture and soil type. Significantly greater soil loss occurred with annual pasture, particularly on the ferrosol soil (mean 4.6 t/ha).

The 5th, 50th and 95th percentiles of the results are included in Table 2. The divergence between the 5th and 95th percentile values indicates that there is significant annual variability in the results for runoff, deep drainage and soil loss.

Table 2. Summary statistics of PERFECT model simulations

Private benefits of perennial grazing systems

Optimal decisions

The impact of ferrosol and kurosol soils was not included in the private benefit analysis as that the private objective function was not influenced by soil type. A summary of the optimal decision rules derived from the dynamic programming model for combinations of the two state variables PG and P are given in Table 3. At low PG levels (PG ≤ 0.10) the optimal decision was to establish a new perennial grass pasture regardless of the prevailing level of soil P. Grazing rest tactics were adopted for levels of PG between 0.15 and 0.50, depending upon the value of P. At the higher levels of P, the grazing rest was selected at larger prevailing values of PG. This result indicates that there is an interaction between the levels of PG and P in determining the optimal grazing management decision.

The optimal fertiliser decision was also dependant upon the interaction of PG and P. At values of P between 5 and 10 mg kg^-1, a fertiliser application was selected for most values of PG considered. However, even at moderate to high P levels of 25 to 30 mg kg^-1 a fertiliser application was selected for some values of PG (PG = 0.20 to 0.60).

Table 3. Optimal private decision rules for combinations of the two state variables; perennial grass composition (PG) and soil phosphorus (P)