Future agricultural policies need to integrate biodiversity targets into smart farming

The benefits of SF for biodiversity can come in several ways¹¹. Here, we give several examples of such proposed benefits. First, the development of autonomous robots can break the “one man – one machine” paradigm, allowing one person to simultaneously operate several smaller machines for field management and harvesting, reducing labor costs. Labor costs have been the main driver of mechanization in agriculture, resulting in large and heavy machinery that is best operated in large, obstacle-free, monocultural fields. This has promoted the homogenization of agricultural landscapes to accommodate this machinery, causing strong disruption and compaction of soil with detrimental consequences for its associated biodiversity¹². As SF and autonomous robots may reduce the need for large machinery, maneuverability issues may be overcome. This will reduce soil compaction, allow for smaller fields and a larger diversity of crops per area, and hence a more diverse agricultural landscape^2,13.

Second, non-destructive remote-sensing technology can guide targeted use of chemical inputs⁶. Sensors can estimate the nutrient status of soils and crop plants, and fertilize according to need, reducing excessive use of fertilizers. Weeding robots can detect and mechanically remove undesired weeds, partly replacing herbicide use. Such robots could also be programmed to avoid weeding out or spraying entire fields, but instead focus on removing only the most noxious weeds, while leaving the less detrimental ones on the field¹⁴. This would increase within-field plant diversity, with concomitant effects on the diversity of associated insects and other animals above- and belowground¹⁵. Similarly, image detection algorithms may also identify the presence of pest infestations on plants and either mechanically remove infested plants or combat outbreaks using locally applied insecticides on the affected plants, leading to a drastic reduction in plant protection inputs. Unfortunately, there are currently few alternatives to fungicide applications, although smaller-sized robots could nevertheless help to minimize inputs by targeted spraying on plants instead of large field applications, preventing run-off and lowering inputs.

Third, through the Internet of Things¹¹, robots can be connected to weather information and forecasts, using this information to autonomously time their operations to appropriate weather and crop-growing conditions. For instance, programs could avoid fertilization or pesticide spraying during adverse weather conditions (enduring heat, rain, or wind) to reduce off-site effects. By making farm management decisions independent of the time constraints of the farmer, SF can use resources more efficiently.

Understanding under which conditions SF will foster effective biodiversity conservation requires reflection upon which negative effects of intensive land use are most likely to be alleviated by SF, which can be on-site, off-site or landscape level (Fig. 1). We argue that it is plausible that SF will alleviate some of the negative off-site effects of intensive agriculture on biodiversity, by more targeted application of inputs, thereby reducing spillover to other habitats^4,5,6,8,15 (Fig. 1). However, whether the proposed environmental benefits are large enough to impact biodiversity still needs to be investigated and quantified. The expected influence of SF on on-site effects is less clear. If future robot designs resemble the large machinery of the present, the effects of SF on biodiversity will be limited. In fact, agricultural intensification may also increase as workforce limitations dissolve, machinery use expands, and smaller, more maneuverable machines enable the cultivation of currently unused land. This threatens landscape elements that provide refuge for biodiversity, microclimatic heterogeneity, and buffer pollution. In addition, even with SF practices, a monoculture field will still promote limited biodiversity, and could become even more devoid of non-target species if biodiversity is not incorporated as a target in the decision-making by SF technology. Without appropriate guidance, robotic farming could easily suppress non-target organisms and overall biodiversity even more rigorously than current farming practices, stressing the importance of making biodiversity an integral target for SF. Development of SF in its current form fails to alter this fundamental aspect. Where SF could offer promising potential regarding on-site effects is by allowing more biodiversity within fields compared to current practices. However, positive change will require willingness and space to allow biodiversity into the confines of the field. It will ultimately depend on the programming of decision-making algorithms whether SF shall have biodiversity benefits: fields could become more diverse, but only if not all weedy plants or insects are destroyed or removed by its machinery¹¹. How the trade-offs among productivity, biodiversity, and economic sustainability within a field can be alleviated requires further collaborative research between agronomists and ecologists.

As SF operates mainly at the individual field or farm level, it is likely that the effects are only observed locally. However, the ecological functions necessary to sustain agriculture generally operate at the landscape level, which is the higher level of integration for agricultural systems. While including a landscape perspective in SF is mostly a topic of setting aims and targets that include biodiversity and ecological indicators^16,17, technological innovations of SF can make such a target more achievable. Smaller and more versatile machinery in SF can allow for increasing landscape heterogeneity by breaking large fields down into smaller units, optimized by yield potential¹⁶, and by planning for biodiversity-optimized field margins and semi-natural elements within the field. Thereby, SF could help to increase the connectivity of the agricultural landscape, facilitating the dispersal of species and enhancing biodiversity at the landscape scale^2,17. However, if SF only leads to the increased use of land that is currently unused due to topography or geometry incompatible with large machines, this would disadvantage biodiversity rather than promote it.

Preserving biodiversity at the local and landscape scale requires making biodiversity an integrated target of SF. However, to date, ecological principles have been insufficiently incorporated into the development, policies, and implementation of these new technologies, highlighting the need for a new framework that integrates agricultural (production) and ecological (conservation) targets. This could also include automated biodiversity assessments for decision-making. Given the recent developments in machine learning and AI-based species recognition from freely available applications, there is hardly a reason not to do so, and do it in a ‘smart’ way.

In order to achieve the proposed full potential of SF for both productivity and biodiversity outputs, a radical shift in vision is required that incorporates both crop production and biodiversity targets as integrated goals. For this, funding agencies should prioritize research that develops optimized SF technology suited to meet and promote both targets. Scientists from different disciplines, including information scientists, engineers, agronomists, and ecologists, need to work together to incorporate insights from agroecology and more biodiversity-friendly farming practices, such as intercropping, multi-cropping, or agroforestry¹⁸, to instruct SF software to make decisions balancing productivity and biodiversity. This knowledge will also guide the development of improved hardware optimized to operate in future landscapes. Information-technological advances will not improve agricultural practices if not taken up by both industry and farmers. Ways to co-create technology that is acceptable to farmers have already been laid out¹³ and need to be supported by appropriate legislation, solving questions of data ownership, which is a major obstacle to large-scale uptake by farmers¹⁹. Uptake by industry and development of best-practice examples should be supported by government incentives, through direct financial support and tax reductions.

It is important to recognize that uptake of SF entails significant risks, impacts, and challenges²⁰. For instance, their development and deployment rely on resource-intensive manufacturing, extractive supply chains, and data infrastructures²¹ requiring substantial energy and water use, potentially shifting environmental burdens rather than alleviating them²². Additionally, SF can create socio-economic challenges by introducing new dependencies or technological lock-ins for farmers, particularly where power asymmetries and data ownership issues prevail, thereby exacerbating farmers’ dependency on external actors^10,19. Acknowledging these challenges helps ensure that SF is embedded within broader sustainability strategies rather than becoming a stand-alone technological solution.

Biodiversity conservation and food security are global endeavors that require a globalized vision, that is combined with locally adapted approaches. SF technologies must be evaluated within current food system dynamics, where a substantial share of crop production goes to animal feed rather than human consumption²³, and in relation to emerging challenges such as improving nutritional security, resource use efficiency, and climate resilience, while minimizing negative environmental externalities²⁰. Thus, different regions of the world have different environmental, economic, and societal conditions that need to be acknowledged in the future development of SF. In the global North, agricultural yield is typically high, and biodiversity is already greatly impoverished. Here, SF could aid in maintaining yield while reducing external inputs and potentially restoring part of the lost biodiversity. In contrast, in the Global South, yields are typically lower, yet biodiversity is still higher in many places. Here, SF could help to close the yield gap, while preserving biodiversity²⁴. It is likely that implementation of SF in the Global South can have the most positive impact for yields and biodiversity if the destruction of nature caused by agricultural intensification and industrialization that was taken in the Global North can be avoided.

Despite the potential for global implementation, SF technologies are currently primarily adopted in the Global North, while in the Global South, its availability remains limited in many regions. The implementation will be challenging due to current trends in agricultural practices²⁴. Even in the Global North, uptake by farmers is constrained by factors such as market opportunities, labor availability, and government policies, which limit the wider transition toward more sustainable practices. Conversely, in many areas in the Global South, inequality in resource distribution is high. Policies should therefore be implemented by prioritizing SF adoption while regulating its use to ensure a fair distribution. That way, SF could contribute to achieving sustainable development goals, alleviating unfair global food distribution, and protecting biodiversity on-site, but particularly off-site and at the landscape level in the areas of the world that sustain most species.

We have pointed out how SF can lead to more biodiversity-friendly farming, on-site, off-site, and for the landscape as a whole. However, a new framework is necessary because the benefits of SF for biodiversity and sustainability do not come automatically with these technologies. We argue that healthy ecosystems should be an aim that is equally important as productivity, and hence biodiversity needs to be included in SF strategies and farming decisions. This implies that, in the future, negative effects of agriculture need to be addressed increasingly also at the landscape scale and that biodiversity assessments become an integral part of SF. Given a new framework to integrate agricultural (production) and ecological (conservation) targets into SF development, policies, and implementation, SF is a promising solution to agriculture’s environmental issues. However, without this framework, the biodiversity benefits of SF will not be guaranteed.