1. It is widely recognized that landscape factors affect the biological control of weed seeds and insect pests in arable crops, but landscape effects have been found to be inconsistent between studies.
2. Here, we compare six different types of sentinels (surrogate prey that was either live insects or seeds) to measure the effects of semi-natural habitats at field to landscape scales on levels of biological control in winter wheat in the UK. Sentinels were located in fields adjacent to three boundary types: grassy margin, hedgerows or woodland to study local scale effects and in landscapes of varying heterogeneity in study areas of 1 km radius.
3. Overall mean levels of predation were higher for most insect prey (60.8%) located on the ground compared to the crop (12.2%) and was lower for seeds (5.8%). Predation of sentinels on the ground was attributed to generalist predators. Semi-natural habitats had both positive and negative effects at field and landscape scales, but the response varied with the sentinel type. Herbaceous linear semi-natural habitats had positive effects at local scales for Calliphora vomitoria and Sitobion avenae sentinels and provides evidence that farmers can introduce linear herbaceous features to benefit biological control. In contrast our distance weighted kernel models identified a positive relationship between woody habitats and the predation of Caliphora vomitoria and Chenopodium album. Natural aphid infestations were lower in landscapes with more semi-natural habitat.
4. Synthesis and applications. Sentinels may be sensitive enough to detect variation in levels of biological control influenced by semi-natural habitats, but this study confirms that landscape effects differ for different types of sentinel prey. This implies that it may not be possible to categorize landscapes as pest suppressive using a single sentinel type. Future studies should therefore consider using multiple sentinels to give a better perspective on predation intensity. The resulting recommendations for farm management include planting woodland adjacent wheat fields infested with seed predators and positioning herbaceous linear habitats adjacent wheat fields infested with Sitobion Avenae, particularly if fields are bordered by woody liner habitats due to their association with decreased Sitobion Avenae predation.

Study area

The study was conducted in 18 focal fields of winter wheat per year surrounded by a 1 km radius landscape circle in the South East and South West regions (counties Dorset and Hampshire) of the UK in 2014 and 2015 (Fig. 1). A 1 km radius was used because this was the distance over which many previous studies had detected an infuence on biocontrol in cereal crops (e.g. Rusch et al., 2016). The area has a temperate climate, summers are warm and humid (Kottek et al., 2006).

Due to crop rotation, focal fields sown with winter wheat in 2014 were put into a different crop in 2015 which prevented the use of the same focal fields and their surrounding landscape circles in 2015. Between years, landscape circles were, however, selected to be close together and are defined as landscape circle pairs. The central point of landscape circle pairs were seperated by a distance of 90 m to 3 km (mean 1.15 ± 0.05 km).

Sampling design

A standardised QuESSA protocol was followed that examined the impact of three different boundary types (local effect) on sentinel predation. From the 18 focal fields studied each year, six had a field boundary for each of three categories, 1) herbaceous strip between fields (control), 2) woody linear habitat (hedgerow) or 3) woody areal habitat (woodland) (Table A.1). Within each field, sentinels were placed at 2, 25, 48 and 71 m from the adjacent SNH, along two transects 10 m apart (Fig. A1). To minimize the influence of the other field boundaries on transects, the distance between the end of the transects and non-focal field boundaries had to be at least 1.25 times the length of the transect (i.e. 88.75 m).

All SNHs within a 1 km radius of the transect centres were digitised in Arc GIS V.10.5 (ESRI, 2011), using the Lambert Azimuthal Equal Area geo-referencing system. SNH present in landscape circles were mapped using the Customer Land Database (CLAD) and in-field observations, SNH were then categorised as either herbaceous linear, herbaceous areal, woody linear, woody areal or fallow (land which has been plowed and left unseeded for at least one season). Areal habitats were defined as having a minimum width of 25 m, whereas linear habitats were between 1.5 m and 25 m wide, both areal and linear habitats were defined as having a minimum length of 50 m and minimum surface area of 75 m². Herbaceous habitats comprised <30% shrub/tree cover and includes sown habitats, whilst woody habitats had >30% shrub/tree cover.  Total SNH was calculated as the sum of these habitats.

Sentinel Sampling

The sentinels were deployed on two occasions in June/July 2014 and 2015. They comprised Calliphora vomitoria larvae (Diptera: Calliphoridae), Ephestia kuehniella eggs (Lepidoptera: Pyralidae), Drosophila melanogaster pupae (Diptera: Drosophilidae, one round in 2014 only), Poa trivialis seeds (Poaceae) and Chenopodium album seeds (Amaranthaceae), which were all located on the ground, and Sitobion avenae adults which were artifically attached to the crop (see below). Calliphora vomitoria larvae were pinned live to strips of 6 mm thick plastazote (1 x 10 cm), 10 per strip. To mimic Lepidoptera egg-laying, E. kuehniella egg masses were exposed on four corners (0.25 x 0.25 cm each) of dry-stick paper (1 x 2 cm; supplied by Oecos Ltd.). The percentage of eggless surface on each corner of dry-stick was then estimated in the laboratory using microscopes). For the seed sentinels, fine sandpaper (5 x 10 cm) was attached to plastazote of the same size to provide some rigidity (Westerman et al., 2003). Twenty seeds of each species were then attached to the sandpaper using M3 spray mount in two blocks with a 5 x 4 arrangement. All other sentinels were attached to dry stick card which was then coated with fine sand to allow predatory insects to walk across the surface. Ten D. melanogaster pupae were attached to dry stick card and placed on the ground. Ten live adult wingless apterate cereal aphids S. avenae were attached to dry stick card and two cards were stapled onto flag leaves at each sampling point. Natural cereal infestation by aphids was assessed on 25 tillers at each distance along one of the transects, on one occassion in 2014 and two occasions in 2015 when populations would be peaking in both years. All sentinels placed on the ground were covered with a metal cage (1 cm mesh) to prevent access by birds and rodents. Numbers of all sentinels, except for E. kuehniella, were counted when deployed (in case any were lost in transit) and on collection in the field. Assessments were made after 24 h for animal prey and 7 days for seeds. Seed sentinels were left in the field for longer than animal prey because they are not as perishable as live prey and previous studies recommend that sampling takes place after 2 – 14 days (Westerman et al. 2003). Partially or totally consumed prey items were recoded as predated.