Local-field agrobiodiversity had a negative effect on pollination potential, while local-field agrobiodiversity, pollination and pest control potentials had a positive influence on cereal production . These effects were significant except for the effect of pest control potential, which was marginally significant . As a result, farming intensity had both a positive direct effect on cereal production and a negative indirect effect through detrimental effect on agrobiodiversity and ecological functions . Indirect negative effects were half as strong as direct positive effects. The total net effect of farming intensity and local-field agrobiodiversity on cereal production were of similar strength . In 2017, farming intensity did not have any effect on cereal production, local-field agrobiodiversity, and pollination and pest control potential, while the latter ones did not have any positive effect on cereal production . Only the negative relationship between local-field agrobiodiversity and pollination potential remained significant. In 2016, crop mosaic heterogeneity had, in general, a greater influence than SNC heterogeneity. Both had positive effects on local-field agrobiodiversity, although the latter only had a marginally significant effect . Crop mosaic heterogeneity also tend to influence positively pollination and pest control potential, although not significantly . Consequently, crop mosaic heterogeneity had an indirect and net positive effect on cereal production, while SNC heterogeneity had a slightly negative effect . In 2017, SNC heterogeneity and, to a lesser extent, crop mosaic heterogeneity had positive, significant effects on local-field agrobiodiversity. SNC heterogeneity also had a strong negative influence on cereal production and a marginally significant positive effect on pollination. In addition, crop mosaic heterogeneity had a marginally significant positive effect on pest control potential . Our results highlight the counter-productive effects of increased farming intensity within conventionally farmed crop fields. As expected, farming intensity had a positive direct influence on cereal production in 2016 , although this was not true in 2017, 4x8ft rolling benches which presented particular meteorological conditions. However, the negative effects of farming intensity on local-field agrobiodiversity and ecological functions indirectly reduced cereal production.

In turn, local-field agrobiodiversity and ecological functions positively influenced crop yield, and their combined contribution to cereal production was superior to the direct effect of farming intensity itself.The balance between direct and indirect effects of farming intensity on cereal production shows that the benefits may be halved. Similarly, a recent meta-analysis demonstrated that low-input farming systems promote pest control to a level able to compensate the absence of pesticides use, despite higher levels of pest infestations . Although the mechanisms underlying the direct and indirect effects of farming intensity on crop yield are very different, our results suggest that relaxing farming intensity in conventional farming systems could enhance the contribution of agrobiodiversity and related ecological functions to cereal production. These results are consistent with existing literature: while limiting the nutrient deficit and pest infestation of crops, mineral fertilisation and pesticides lessen carabid and plant diversity, as well as aphid and weed predation potential . Bees are also strongly impacted by farming intensity through direct lethal, and sub-lethal effects of insecticides , but also herbicides that reduce flower availability , with consequences on pollination . However, the positive effect of pollination potential on cereal production may be surprising here as we studied non-entomophilous crops.One potential reason is that some pollinators, such as hoverflies, are also natural predators of pests at larval stage.Our study follows the agroecological framework where ecological functions associated with mobile organisms depend on the dynamic of the entire landscape mosaic.Therefore, pollination may correlate with a generally high abundance of beneficial species and good agroecosystem functioning, which makes pollination measures indicators of other important ecological functions . This assumption is supported by the fact that in absence of these relationships , the model explained a lower proportion of observed variation. The contribution of local-field agrobiodiversity, and particularly of the carabid beetle community, to cereal production was crucial. Its net positive effect was as high as the net effect of farming intensity . However, we found negative and non-significant influence of our measures of local-field agrobiodiversity on pollination and pest control potential respectively . A potential reason for the absence of positive effect is that our measures were made 50 m away from field edges, where such relationships are expected to be highest as a result of spill-over.

Our result on the negative influence of local-field agrobiodiversity on pollination potential was unexpected. This result may reflect an antagonism between carabid diversity and pollinator communities, with, for example, contrasted responses to landscape context and/or interactions with weed communities . In that sense, our results highlight the difficulty to use comprehensive and representative indicators of agrobiodiversity and ecological functions in agroecological studies. For instance, frozen preys, glued on predation cards, are useful but imperfect surrogate for actual pest control.Ricketts et al. , in a meta-analysis, have shown that more than half of studies linking biodiversity and pest control have not demonstrated any significant relationship. However, opposite to our results, a recent global data synthesis showed that diversity of beneficial communities support pollination and pest control, which in turn increases crop production . Nevertheless, local-field agrobiodiversity had a direct positive effect on cereal production. This relationship may not be interpreted as a causal relation but could reflect a positive correlation between agrobiodiversity and ecological functions not, or only partially, measured here . For instance, carabid beetles are good indicators of soil characteristics, biodiversity, and related ecological functions, which were missing herein . Future research should include the soil components of agroecosystems, as they were found to interact with farming intensity and ecological functions . While farming intensity had negative effects on agrobiodiversity and ecological functions, landscape heterogeneity had positive effects in both years . In 2016, heterogeneity of the crop mosaic , had a stronger positive effect on local-field agrobiodiversity compared to semi-natural covers. Similarly, Sirami et al. found in an extensive study that increasing crop heterogeneity was more beneficial for multitrophic diversity of crop field communities than increasing semi-natural covers. While they found that smaller field size was more important compared to crop diversity, we observed the opposite trend, perhaps because of different study gradients or response variable. These relationships lead, in our study, to an indirect positive effect of crop mosaic heterogeneity on cereal production . Similarly to Martin et al. , we found that the positive effect of landscape heterogeneity on agrobiodiversity was similar or even larger than the negative effect of farming intensity .

Recent studies suggest that the effect of landscape heterogeneity on agrobiodiversity, ecological functions and cereal production may be even greater when local farming intensity is lower . For instance, Ricci et al. showed that an increased amount of semi-natural covers benefited biological pest control only at a low field-level pesticide use. Such interaction was also found at the landscape level, where the positive effect of landscape complexity on bee species richness occurred in landscapes with low nitrogen inputs . These results call for investigations that quantify the potential of agroecological systems, when low-input farming practices are applied at the landscape-scale. Our results strongly differ between the two years of measurements, with most relationships observed in 2016 collapsing in 2017 . A potential reason is that 2017 was marked by drought and local storm events in the spring, as compared to 2016 . Weather conditions have been shown to be the main drivers of variations in ecological processes such as pest outbreak risk . However, it is not possible to infer the respective influence of weather and stochastic variability based on a two-year study. Observed variations in yield, agrobiodiversity, and levels of ecological functions between 2016 and 2017 may have also resulted from other factors such as field history, as sampled fields differed between the two years. Non-measured local characteristics such as soil quality and soil biodiversity may have generated variations among study fields as well . Nevertheless, our result suggests that neither increased farming intensity nor ecological functions could compensate for the decline in crop production in 2017. However, persistence, or stability, across years of ecological communities and associated functions is critical for agroecological systems and stable crop production. Spillover of beneficial species into crop fields from semi-natural habitats is recognised to be critical for mitigating the negative effects of global environmental change on biodiversity patterns and ecological processes . Landscape heterogeneity, particularly the amount of semi-natural covers , significantly increased local-field agrobiodiversity in 2017 . This effect was still too weak to maintain ecological functions and cereal production. Stronger resilience to unfavourable factors, such as meteorological conditions, might however be provided by drastic modifications in farming systems such as less intense soil management , longer crop rotations, intercropping, remaining previous crop residues or crop mixing . For instance, organic farming was found to support spatiotemporal stability of bumble bee and butterfly communities . If such agroecological practices are implemented, in combination with increased landscape heterogeneity, it may be possible to increase the levels of biodiversity in crop fields and its contribution to cereal production, flood and drain table along with persistence of important ecological functions. In a national-scale study in the UK, Redhead et al. found that wheat crop yield was more stable and more resistant to extreme weather events in landscape with larger area and less fragmented semi-natural covers. Since the frequency of extreme meteorological events is likely to increase dramatically under climate change, research should urgently explore how landscape agroecological approaches can improve the resilience of agroecosystems.

Agricultural practices are often accompanied by a range of environmental costs including soil degradation, freshwater contamination, eutrophication, and biodiversity and habitat loss . Organic agriculture has been suggested as a more sustainable alternative with lower environmental costs than conventional high-input agriculture . Organic agriculture combines traditional farming methods such as natural pest management, rotating crops, and organic fertiliser application with modern technologies including biological control and reduced tillage . Today’s share of organic to total farmland is 1% globally, 7% in the EU, and 14% in Switzerland . Several long-term studies compared conventional and organic agricultural practices regarding yield and environmental impact . These studies revealed yield differences between organic and conventional agriculture that range from 5% to 35% less yield in organic agriculture, depending on the crop type and agroecological condition . At the same time, there is ample scientific evidence that organic agriculture improves soil quality and soil fertility, conserves biodiversity, increases energy and production efficiency, and reduces environmental pollution compared to conventional agriculture . To date, only a few studies have compared the impacts of organic versus conventional agriculture on the water relations of crops and/or entire cropping systems. This is surprising, given the importance of water for crop production and the fact that agriculture is the largest consumer of freshwater, leading industry and domestic use . The previously shown impact of organic agriculture on soil properties implies that organic agriculture could also improve ecosystem water relations. Organically managed soils have, for example, a greater soil organic matter content , which directly correlates with higher soil aggregate stability and reduced bulk density . Both contribute to the water holding capacity of soils . Some studies report higher resistance against surface runoff of organically as compared to conventionally managed soils because organic agriculture increases anectic earthworm biomass and diversity which in turn results in a higher abundance of vertical earthworm tunnels and improved water infiltration . In fact, Pimentel et al. as well as Kundel et al. reported higher ground-water recharge and soil moisture in organic compared to conventional farming approaches. Further indications for improved ecosystem water relations result from studies reporting higher crop yields in organic compared to conventional farming systems under drought conditions . Despite these indications, studies directly assessing the plant and ecosystem water relations of crops in organically versus conventionally managed agroecosystems are scarce and just about starting to roll . Whether organic agriculture can indeed reduce the water demand and improve the water use efficiency of crop production still remains unclear.