Farmland Biodiversity and the Footprint of Agriculture

+ See all authors and affiliations

Science  19 Jan 2007:
Vol. 315, Issue 5810, pp. 381-384
DOI: 10.1126/science.1136607


Sustainable development requires the reconciliation of demands for biodiversity conservation and increased agricultural production. Assessing the impact of novel farming practices on biodiversity and ecosystem services is fundamental to this process. Using farmland birds as a model system, we present a generic risk assessment framework that accurately predicts each species' current conservation status and population growth rate associated with past changes in agriculture. We demonstrate its value by assessing the potential impact on biodiversity of two controversial land uses, genetically modified herbicide-tolerant crops and agri-environment schemes. This framework can be used to guide policy and land management decisions and to assess progress toward sustainability targets.

Biodiversity and ecosystem function are inextricably linked. The case for biodiversity conservation can be argued on economic, sociocultural, and aesthetic grounds (13). Although biodiversity loss has occurred across all terrestrial ecosystems, many of its drivers are associated with the intensification of agriculture (4, 5). Agricultural production is set to double again by 2050 (6). Unless the foot-print of agriculture is carefully managed through sustainable development, both agricultural systems and remaining natural ecosystems will suffer further degradation, increasing the proportion of the world's species threatened with extinction and further limiting the ecosystem services they are capable of providing (7, 8).

Managing the environmental effects of agriculture requires an assessment of biodiversity risks and benefits for all new agricultural practices (4, 9). An appreciation of the ecological mechanisms that affect extinction risk is fundamental to the development of risk assessment protocols. One key factor appears to be the degree of specialization shownbyaspecies (10, 11). Specialists have narrower niche requirements and are disproportionately affected by reduced niche availability; the corollary is that generalist species are likely to be more resilient to environmental perturbation (12).

We have developed a trait-based risk assessment framework capable of predicting the impact of environmental change on biodiversity and ecosystem services. We used farmland birds as a model system to which to apply this framework. In the United Kingdom, birds have already been adopted as a focus for biodiversity conservation, with an index of wild bird population trends included as one of the government's 15 headline indicators of sustainable development. This indicator, presented as the overall proportional change since 1970, can be partitioned by habitat to reveal underlying trends (13). The farmland bird index (FBI) component shows that farmland bird populations have almost halved since 1970, and it is widely accepted that these declines have been driven by agricultural intensification (14, 15). The UK government has set a public service agreement (PSA) target to reverse this long-term decline in farmland bird populations by 2020.

Our framework draws on a matrix of species' ecological requirements covering components of diet, foraging habitat, and nesting habitat. The framework assumes that an agricultural change will affect a species if it leads to a change in food abundance and/or a change in nesting success. Food abundance can be altered by changes in foraging habitat availability and/or changes in food abundance in existing foraging habitats. Nesting success can be altered by changes in nesting habitat availability and/or changes in nest success in existing nesting habitats. Risk score calculation has its basis in the assumption that species with broader niches will be less vulnerable to the effects of agricultural change than species with narrower niches. Niche breadth is reflected in a species' risk score by calculating and summing the proportion of diet, foraging habitat, and nesting habitat components used by the species that are affected by an agricultural change. Higher scores are attributed to species demonstrating a greater proportion of affected requirements (16).

To validate our framework, we assessed the environmental effects of six key components of agricultural intensification in the United Kingdom over the past 4 decades: the switch from spring to autumn sowing, increased agrochemical inputs, loss of noncropped habitats (i.e., land not used for growing crops), increased land drainage, the switch from hay to silage, and the increased intensity of grassland management (14, 15). We determined from the available literature whether these components had led to a reduction in the abundance or availability of each diet, foraging habitat, and nesting habitat component included in our matrix of ecological requirements. By using this matrix, we identified every species likely to have been adversely affected by any such reduction and calculated a risk score for each species. When summed across all six changes, the overall risk score for each species reflects the degree to which agricultural intensification has affected the species' ecological requirements (16).

Each of these agricultural changes has occurred at a national scale, and any detrimental environmental effects are likely to have caused population-scale responses in vulnerable species. We predicted that the risk score for each species should be significantly related to its conservation status and population growth rate over the period of recent agricultural change. The conservation status of UK birds is listed as red (most threatened), amber, and green (least threatened) and is assigned according to a range of criteria covering breeding range and population trends (17). We found that risk score was significantly related to the probability of being listed in these conservation status categories (Fig. 1A) (mean score ± 1 SE for species on the red list was 6.6 ± 0.8; for the amber list, 4.9 ± 0.8; and for the green list, 2.2 ± 0.4; ordinal logistic regression, χ2 = 25.4, P <0.001). We also found that the risk score was significantly related to the annual rate of population growth (Fig. 1B) (16): Higher risk assessment scores were associated with species with negative population growth rates and therefore experiencing population decline [F(1, 49) = 11.3, P = 0.002]. The predicted FBI, based on population changes from 1970–2001 and calculated by using a bootstrapping procedure on population growth rates generated from jack-knife analyses (16), is 0.59 (0.42 to 0.85 are 95% confidence limits), compared to the actual FBI of 0.54.

Fig. 1.

Relationships between total risk score and (A) conservation status category and (B) annual population growth rate. (A) Probability of conservation status category classification derived from parameter estimates of the ordinal logistic regression model (16). Colors represent UK conservation status categories (17). Below a risk score of 4 (to the left of line X), the probability of being green-listed is highest. Between risk scores of 4 and 6.9 (between lines X and Y), the probability of being amber-listed is highest. Above a risk score of 6.9 (to the right of line Y), the probability of being red-listed is highest. Bar charts show the distribution of actual species status within these risk score boundaries. Cross-tabulation of predicted versus actual classification shows strong symmetry (Somer's d = 0.50, P < 0.001). (B) Annual population growth rate declines with increasing score from risk assessment of recent agricultural intensification (16). Data for species included in FBI are shown in blue. Solid black line shows fitted model for all species (y = 0.0079 – 0.0037x, r2 = 0.19); dashed lines show 95% confidence limits.

Our risk-scoring system assumes equal weighting for each source of risk in terms of its relation to conservation status and population growth and assumes that different risk sources have an additive effect. To critically assess these assumptions, we constructed a series of more-complex alternative models that decomposed the total risk score into various component parts, allowing the weighting of different sources of risk to vary. We also created a set of models that assumed multiplicative rather than additive effects (tables S6 and S7). This analysis showed that our assumptions were reasonable: The most parsimonious models of conservation status and population growth rate only included total risk score. However, two alternative models for predicting population growth rates, one specifying additive effects of risk score decomposed into diet-related and nest-related components and the other specifying multiplicative effects of these two variables, also received support.

Parameter estimates from these regression models can be used to predict the likely impact of new agricultural practices on farmland bird populations. We demonstrate this process by applying our risk assessment framework to two controversial land uses, genetically modified herbicide-tolerant (GMHT) crops and agri-environment schemes, which have generated debate over their possible contributions to sustainable development and their impacts on biodiversity (1822). Species scores from these risk assessments were combined with their scores from the validation process, which characterize responses to current landscape conditions, to predict population growth rates and conservation status in the resultant landscapes (16).

The Farm Scale Evaluation project investigated the effects of GMHT crop management on UK farmland wildlife. Its results suggested that introduction of GMHT sugar beet and oilseed rape is likely to cause a long-term reduction in above-ground invertebrates and weeds in the cropped area of fields (23). Thirty-nine farmland bird species have ecological requirements that make them susceptible to such changes. Each of these species would therefore be expected to experience reduced population growth rates following nationwide GMHT crop introduction. However, we predict that just one species, meadow pipit (Anthus pratensis), would be reclassified to a less-favorable conservation status as a consequence (changing from amber- to red-listed). Overall, it appears that replacing equivalent conventional crops in the current agricultural landscape with GMHT crops would only have a limited effect on the FBI (Table 1 and Fig. 2).

Fig. 2.

Predicted proportional changes in the populations of the 19 FBI species between 1970 and 2020 under three land management scenarios (16). Line colors represent current UK conservation status categories (17). Population growth rates estimated from parameter estimates were derived from the model of total risk score used to illustrate predicted responses.

Table 1.

Predicted FBI in 2020 derived from risk scores associated with continued current management, the introduction of GMHT crops in 2005, or the introduction of the ELS scheme in 2005. Mean predicted FBI values and 95% confidence limits (in parentheses) were generated from three alternative models of population growth rate (16). Predicted FBI for 2001 generated from the three alternative models are also shown. For comparison, the FBI in 2001 calculated from actual population growth rates was 0.54.

Predicted FBI in 2020
Model structurePredicted FBI in 2001Current management continuedGMHT crops introduced in 2005ELS introduced in 2005
Total risk 0.59 (0.42-0.85) 0.42 (0.26-0.65) 0.40 (0.25-0.62) 0.47 (0.31-0.71)
Diet-related risk plus nest-related risk 0.59 (0.38-0.86) 0.42 (0.27-0.62) 0.41 (0.26-0.62) 0.47 (0.30-0.70)
Diet-related risk multiplied by nest-related risk 0.62 (0.41-0.94) 0.47 (0.30-0.71) 0.45 (0.28-0.67) 0.52 (0.34-0.77)

Agri-environment schemes are designed to mitigate the detrimental impacts of agriculture and increase the value of the landscape to biodiversity. Clearly these schemes will be most effective if they target the main drivers of biodiversity decline. Our validation results suggest these key drivers have been the loss of food and nesting habitats in the cropped areas of the agricultural landscape (table S8).

An example scheme, entry-level stewardship (ELS), was launched in England in 2005, offering a range of management options for all farming types. Over 13,000 agreements, covering 1.5 million hectares, have already been implemented, with payments totalling £47 million in the first year (24). However, analyses of option objectives and initial uptake rates (16, 25, 26) show that the main emphasis of current agreements is on hedgerow and margin management rather than improving the environmental value of the cropped area (table S8). This disparity between the causes of farmland bird population decline and the uptake of mitigation measures, rather than scheme design per se, suggests that the ELS may not deliver its biodiversity objectives. Even if all causes of decline associated with margin and hedgerow habitats in the current agricultural landscape are countered by management agreements under ELS, our analyses suggest that the FBI will continue to decline, driven by the detrimental conditions persisting in the cropped area (Table 1 and Fig. 2). More importantly, the three models predict that the percentage of FBI species with annual population growth rates of zero or above will lie between only 37% and 53% [8/19, 10/19, and 7/19 of FBI species (table S9)]. Furthermore, several red-list species, such as skylark (Alauda arvensis) and corn bunting (Miliaria calandra), which rely solely on the cropped area of fields, are likely to continue declining at their current rate (Fig. 2). Unless greater emphasis is placed on improving the value of the cropped area for biodiversity, progress toward reversing the long-term declines in farmland birds is liable to fall short of the UK government's PSA target.

This trait-based approach could have a broad range of applications in agricultural ecosystems and beyond. For example, it could be used to assess the risk of agricultural changes to pollinating insect populations and therefore pollination services. The necessary data for developing an ecological requirements matrix for many of these species, particularly in the United Kingdom and Western Europe, are readily attainable (27). Pollinator diversity is essential for sustaining this highly valued service, estimated to be worth $14 ha–1 year–1 (28), but agricultural intensification has reduced both the diversity and the abundance of native insect pollinators (27). By assessing the impact on key ecological requirements, our framework could be used to predict the response of pollinating insect populations to any proposed change and therefore facilitate the effective management of pollination services in the agricultural landscape. Our framework provides a robust basis for assessing risk, and its application to GMHT crops and agri-environmental management has important implications for policy- and decision-makers. We believe our framework can also contribute greatly to the economic evaluation of proposed agricultural changes that alter the functioning of ecosystem services through their impact on biodiversity.

Supporting Online Material

Materials and Methods

Tables S1 to S9


References and Notes

View Abstract

Related Content


Navigate This Article