The study site was located in the western periphery of Rennes, Brittany, Western France. This site is a 50,000-ha farmland area that exhibits a “bocage” structure, with a dense hedgerow network, and is dominated by a polyculture-livestock system. Five locations were chosen in the study site, in which two paired fields in conservation and conventional soil management were selected (Figure 1). The distance between two locations was at least 1.9 km and at most 21.3 km. Locations were not supposed to be spatially auto-correlated, as previous studies in the same region have shown spatial independence (Puech et al., 2015; Djoudi et al., 2018). Conventional fields corresponded to crops where the soil management consisted in more than 25 cm depth tillage and inversion of the soil and conservation fields corresponded to crops where the soil management consisted in tillage of less than 10 cm depth, and non-inversed soil. Paired fields were less than 100 m apart and therefore assumed to share the same landscape environment. Reduced soil management was applied for at least 6 years in conservation fields except for one (Biodiversystem: 1 year). Fields surface varied from 0.25 ha to 2 ha. All fields were cultivated with winter wheat (Triticum aestivum), and previously cultivated with maize.

All statistical analyses were carried out with the software R (R Core Team, 2019), version 3.6.1.

Sampling session and its interaction with soil management had significant effects on aphidophagy, but only sampling session had a significant effect on parasitism service (Table 1). Aphidophagy significantly increased over time (March 21st, April 4th < May 9th, June 13th < July 11th, P < 0.0001 for all comparisons, Figure 8A), and on June 13th, aphidophagy service was greater in fields under conservation soil management than in fields under conventional soil management (P = 0.018). Reversely, parasitism service decreased over time (Figure 8B) with significant differences between April 5th and May 28th (P = 0.006), and between May 9th and June 13th (P = 0.0001).

Only sampling session had a significant effect on aphid damage (Table 1). Damage caused by aphids tended to be lower in fields under conservation soil management in Spring than in conventional fields but the trend was reversed from the end of May onward, however, the differences between soil management were not significant (SM5).

Soil management is an important factor in determining soil biotope properties and its biocoenosis (Kladivko, 2001). Ground dwelling arthropods, such as carabids, are especially sensitive to soil management. In accordance, conservation fields with reduced perturbations harbored more diverse and abundant carabid communities than conventional fields in this study. Regarding diversity indices similar findings have been consistently observed in previous studies (Baguette and Hance, 1997; Hatten et al., 2007; Menalled et al., 2007; Shearin et al., 2007) but abundance responses to conservation soil management varied. Most previous studies also showed a positive impact of conservation practices with higher abundances of carabids in conservation systems than in conventional systems (Kromp, 1999; Holland and Reynolds, 2003; Kosewska et al., 2014). However, in some cases higher abundances were found in conventional systems (Baguette and Hance, 1997; Hatten et al., 2007; Menalled et al., 2007), but this was usually due to a single dominant species that was either indifferent to disturbance or resilient to it thanks to good colonization capabilities (Baguette and Hance, 1997; Hatten et al., 2007). Allema et al. (2019) also hypothesized that in disturbed open habitat, such as conventional fields, carabids exhibit a more intense foraging activity than in complex habitat, resulting in increased trapping efficacy rather than increased abundance.

In general, conservation soil management is expected to have direct beneficial impacts on communities that overwinter in fields (Soane et al., 2012). Direct mechanical disturbances, such as deep tillage applied in conventional fields, result in high mortality of field-overwintering carabid individuals. In fact, emerging communities of carabids have been recorded to be more abundant in conservation fields or in less disturbed habitats than in conventional fields (Holland and Reynolds, 2003; Thorbek and Bilde, 2004; Mesmin et al., 2020). Similarly, the number of individuals that emerged in conservation fields in our study was 1.3 more abundant than in conventional fields (SM6). Carabid species overwintering as larvae are suspected to be more sensitive to tillage than overwintering adults (Purvis and Fadl, 1996; Mesmin et al., 2020) with the exception of those that overwinter below plough depth (Barney and Pass, 1986). In our case, almost all emerged carabid species overwintered as adults, but Trechus gr. quadristriatus, the most abundant species in the emerging community overwintered as both larvae and adults and was 2.45 more abundant in conservation fields than in conventional ones (SM6), which may indeed suggest a greater sensitivity of the larval stage to tillage. Emergence dynamics also depended on soil management: emerging communities were more abundant in conservation fields than in conventional ones in Spring (see Figure 3), but the trend was reversed in Summer. This temporal change in carabids response to tillage can be explained by two non-exclusive hypotheses: first, the soil is more loosened in conventional fields than in conservation ones due to tillage which facilitates carabid burial and Summer-emerging species overwinter below plough depth escaping direct mortality. For instance, Pseudoophonus rufipes, that starts emerging and being active in July in our study, have previously been found buried at 45 cm and below (Briggs, 1965). Second, carabid emergence in conservation fields is earlier because of changes in soil properties. Indeed, heat diffuses faster in conservation soils than in conventional soils (Potter et al., 1985), leading to faster development of carabid eggs and larvae (Tenailleau et al., 2011) which could explain earlier emergence.

One of the main benefits of conventional deep tillage is weed suppression, which leads to habitat homogenization, in contrast to the more weed-colonized conservation fields (which was actually observed in our fields pairs), and to indirect negative impacts on natural enemies. MacArthur (1972) hypothesized that the number of available ecological niches increased with habitat complexity, which is expected to have a positive effect on the co-existence of species (Tews et al., 2004; Stein et al., 2014). A high habitat heterogeneity supports a wide range of alternative preys (Langellotto and Denno, 2004; Finke and Snyder, 2008) and brings stability in natural enemies food webs (Staudacher et al., 2018) and biocontrol services (Langellotto and Denno, 2004; Birkhofer et al., 2008; Finke and Snyder, 2008) by lessening interspecific competition (Finke and Denno, 2002; Staudacher et al., 2018). In consequences, high habitat heterogeneity is expected to be beneficial for generalist predators like carabid beetles, by providing a wider range of prey and refuge, which is in accordance with the higher abundance and species richness observed in circulating communities of conservation systems in our study. Parasitoids are also expected to benefit from habitat heterogeneity. Conservation fields by providing a greater abundance and diversity of weeds (Soane et al., 2012) might provide more nutritional resources, overwintering sites and alternative preys (Rusch et al., 2010; Araj et al., 2011). The number of parasitoids was slightly higher in conservation fields than in conventional ones at peak aphid infestation, in late May and early June in our study. On the contrary, more parasitoids were present at the beginning of the season in conventional fields than in conservation ones, possibly due to a bottom-up control of parasitoid populations by aphid resources since aphids were more abundant in April in those fields. In any case, no significant differences in parasitoid abundance were found according to the type of soil management, suggesting that if an effect exists, it is of low magnitude.

Aphid predation by carabid beetles increased over time and was similar between soil managements, except in June when this service was greater in conservation fields than in conventional fields. The mechanisms governing temporal variations in aphids place in the carabid diet have not been studied but they have been for seed predation. Given the generalist diet of carabid beetles, the same mechanisms could be at play for aphid predation. According to Saska et al. (2008), temporal variations in seed predation result from three mechanisms. First, there is a turnover in carabid communities; even if carabids are generalist predators, there is a wide gradient of diet among species and a different food demand is expected depending on community composition. Second, food demand is expected to change with phenological stage, even within the same species. Third, changes in environment temperature lead to changes in activity and metabolic rates of carabids and in consequences, in predation rates. Aphidophagy service was linked to carabid abundance in a previous study that used aphids as sentinel prey (Boetzl et al., 2020) and, regardless of soil management we can assume that carabid community turnover, changes in phenological stage and in environment temperature, all contributed to changes in aphidophagy rates, with an increase over time in our study. In June, aphid predation rate was higher in conservation fields than in conventional ones while no significant difference in carabid abundance was observed at this time. However, in June, a significant difference in carabid community composition was noted between fields managed with conservation practices and fields managed with conventional practices, suggesting that species composition rather than total abundance determines aphid predation rate.

In our study, parasitism rates did not differ according to soil management. Contrarily, Tamburini et al. (2016) found higher parasitism rates in fields under conservation soil management than in fields under conventional management. However, in this study, parasitism rates also depended on the proportion of semi-natural habitat in the landscape, but only for conventional fields, suggesting that habitat heterogeneity within the field could compensate for poor landscape quality. Brittany’s agricultural landscape exhibits a “bocage” structure with semi-natural field-borders that may provide refuge and resources to arthropod communities. This type of structure could compensate for the poor habitat heterogeneity in conventional fields and mitigate the positive response of parasitoids to conservation practices. Parasitism rates decreased over time in our study. Until May, aphid infestation remained low resulting in a shortage of resources for the already present parasitoids and parasitism rates reaching over 50% (see Figure 8). At peak aphid infestation, parasitism rates decreased, probably due to a rapid increase in resources, and varied between 25% and 35% regardless of soil management, an order of magnitude similar to a previous study performed in Brittany (Derocles et al., 2014). Although differences were not significant, peak parasitism rate in conservation fields was earlier than in conventional fields, which makes us wonder about the potential benefits of conservation practices on early service of biocontrol.

Aphidophagy and parasitism have reversed temporal dynamics suggesting a temporal complementarity of carabid and parasitoid communities in the provision of aphid control service. An effective complementarity between flying and ground dwelling guilds of natural enemies had already been observed through lower aphid infestation when both guilds were present (Schmidt et al., 2004; Tamburini et al., 2016), despite the risk of negative interactions between natural enemies, such as consumption of mummified aphids by predators (Traugott et al., 2012; Ortiz-Martínez et al., 2020). These opposed dynamics between natural enemies might be due to the removal by predators of aphid resources which the parasitoids can no longer access. Similarly to the delay between complex and simple landscape context, conservation soil management, by favoring a precocious presence of predators compared to conventional soil management, might explain the delay in parasitism response in conventional system (Ortiz-Martínez and Lavandero, 2018). Regardless of the interaction between natural enemies, aphid population dynamics or their damage do not change with soil management. Indeed, despite higher carabid abundances and aphidophagy, especially in June, conservation fields displayed similar aphid infestation dynamics as conventional fields.

Services were measured with various methods, inducing some biases: parasitism rate was assessed directly on local aphid populations, while aphidophagy was measured using introduced sentinel prey, pea aphids. As a result, the measures of aphid control by parasitoids and by predators are not strictly identical, making it difficult to compare the two. Furthermore, sentinel prey have been criticized for their immobility and suspected of not truly reflecting natural trophic interactions (Birkhofer et al., 2017) but they have been shown to be an efficient proxy for assessing predation in an ecosystem (Boetzl et al., 2020) and immobility bias was removed in our study by using live aphids as sentinel prey.

Parasitoid community did not seem to be impacted by soil management: both assemblages were composed of Aphidiinae parasitoids. In contrast to parasitoid assemblages and in accordance with previous studies, soil management influenced carabid assemblages (Baguette and Hance, 1997; Holland and Reynolds, 2003; Thorbek and Bilde, 2004; Kosewska et al., 2014). One species was dominant in circulating communities regardless of soil management (Poecilus cupreus), suggesting that this species mainly overwinters outside the fields and recolonizes them afterward. Some others were specific to the circulating communities and to the emerging community in conservation soil management (Anchomenus dorsalis, Amara sp, Harpalus affinis), suggesting that these species overwinter both inside and outside the fields and are directly negatively impacted by autumn tillage. Similar results have already been observed for Amara and Harpalus genera (Tonhasca, 1993; Baguette and Hance, 1997; Thorbek and Bilde, 2004; Pretorius et al., 2018) and for A. dorsalis (Thorbek and Bilde, 2004 but Baguette and Hance, 1997) and corroborate the existence of direct mortality caused by tillage. It is worth noting that those three taxa were especially active in June, when aphidophagy was significantly higher in the conservation fields than in the conventional ones, and are recognized biocontrol agents. Indeed, A. dorsalis is a well-known aphid consumer (Chiverton, 1987, p. 0; Sunderland and Vickerman, 1980; Staudacher et al., 2018) and could explain the difference in aphidophagy between systems. Amara sp and H. affinis have also been observed consuming aphids but at lower predation rates (Sunderland and Vickerman, 1980; Staudacher et al., 2018). However, Amara and Harpalus genera are known for their predation service on weed seeds (Honek et al., 2007; Saska et al., 2008), highlighting that other services not assessed in our study could be affected by conservation practices.

When considering only circulating carabid communities, some species were also characteristic of the soil management mode. Demetrias atricapillus and Drypta dentata were more abundant in conventional fields, while Nebria brevicollis and Notiophilus biguttatus were more abundant in conservation fields. Spring breeders that overwinter as adults, such as D. atricapillus and D. dentata, should be resistant to soil disturbance. In contrast, species that breed in autumn and overwinter in their larval form are more sensitive to perturbation (Purvis et al., 2001; Ribera et al., 2001 but Kosewska et al., 2014), which is consistent with the observed higher abundance of N. brevicollis and N. biguttatus in conservation fields as they breed in both Autumn and Spring. In addition, the ability to fly of D. atricapillus and D. dentata, which had functional wings, gives them good dispersal capabilities, which is advantageous in conventional fields to escape disturbance and recolonize (Ribera et al., 2001). D. atricapillus has been shown to be a good aphid-consumer (Sunderland and Vickerman, 1980) but the diet of N. brevicollis includes a wide variety of pests in addition to aphids: lepidopterans, crane flies or even slugs (Sunderland and Vickerman, 1980; Seric Jelaska et al., 2014; Reich et al., 2020), confirming the presence of multiservice-providing species in conservation fields.