Vincent de Leijster

52 Chapter 3 rates under alternative management practices, that adding organic amendments is the most effective strategy to increase SOC. No differences in soil carbon stock were found between CT and the agroecological treatment, which combined no tillage with compost application, in a Spanish apricot study (Montanaro et al., 2010). The authors claim that Mediterranean soils with low organic matter content only show measurable changes in SOC levels after 7-10 years (Montanaro et al., 2012). As the build-up of SOC is slow, it is more likely that the higher levels of SOC in this study are the result of measuring the compost particles themselves, and these may still mineralise over time. Therefore, long-term monitoring of SOC after compost application is needed to demonstrate whether carbon stocks are also improved in the long term. Further, Luo et al. (2010) report that in a transition from CT towards NT management SOC increases in the top layer (0-10 cm), but decreases in deeper layers (10-40 cm). In the current study soil samples were taken from a depth between 0-20 cm, which includes both previous layers. Our results show that actual carbon storage is 12 to 44 times higher in this part of the soil than in understory vegetation, which implies that when targeting carbon stock rehabilitation in woody-crop systems it is important to focus on the soil. This makes CMmore effective in increasing carbon stock than the other agroecological management practices. Habitat provisioning, pest control and pollination We found 31–73% habitat provisioning improvements in the agroecological treatments by comparison with CM. This is strongly related to fraction of understory cover and plant species richness and not to arthropods abundance and richness. It is self-evident that the plant species richness and fractional cover are highest in the vegetation-related treatments (NT and GM) and is in line with previous research (Cucci et al., 2016). Understory vegetation in almond orchards has been shown to be positively related to pollinator abundance (Norfolk et al., 2016; Saunders et al., 2013), and parasitoid wasp abundance (Eilers and Klein (2009); however, we failed to find this relationship. One possible reason is that our study had a relatively small sample size and few sampling events over a short period of time (Westphal et al., 2008). A second possible reason is that pollination measurements were done in March, during the flowering season of the almonds, when the understory vegetation was not yet fully developed. Thirdly, the resolution of our taxonomic identification, which was only at the order level, might not be enough. Fourthly, for the analysis of natural enemies we may have missed important pest predators syrphid flies, parasitoid flies and wasps and lacewings as our method was not sensitive to these groups. Moreover, for the analysis of pollinators we missed butterflies (Lepidoptera) and pollinating wasps (Vespoidea), as the pan trap method was not sensitive to these groups. These groups, however, have a lower pollinator efficiency than bees (Apoidea) and flies (Diptera). A study in the southern Spain showed that butterflies,