Understanding the habitat use and movement patterns of natural enemies in agricultural landscapes is essential for enhancing biological pest control. Since many natural enemies rely on floral resources, the distribution of these resources in combination with movement behaviour likely influence biological control in field crops. Nevertheless, few studies have assessed natural enemy movement at the landscape scale. Here we estimated minimal movement distances of the green lacewing Chrysoperla carnea based on consumed pollen and the spatial distribution of the respective plant species in 24 agricultural landscapes (500 m radius). Lacewings were sampled using sticky traps in the centre of each landscape, and the consumed pollen were identified. The location of the most important pollen-providing plants was mapped in each landscape. Distances to potential sources of pollen consumed by 346 lacewings were used to derive minimal movement distances. Lacewings consumed mostly pollen from insect-pollinated plants that were present within 500 m from the sampling location. The distance to the nearest source of consumed pollen exceeded 200 m in 31% of lacewing individuals, demonstrating the relevance of the landscape scale to understand their population dynamics. Distances were shorter to insect-pollinated than to wind-pollinated plants, and shorter early than late in the season. Mean and median distances to pollen sources were negatively associated to flower availability and edge density in the landscape, but this was not the case for minimal distances.

Synthesis and applications: Our findings suggest that the spatial distribution of flowering wild plants can inform movement patterns of lacewings and other flower-visiting insects in agricultural landscapes. In addition, the location of floral resources in the landscape is important for its capacity to sustain natural enemies. Given the strong reliance of lacewings on pollen from nearby insect pollinated woody plants, the promotion of native shrubs and trees, such as Prunus, Salix and Castanea, should be prioritised for natural enemy enhancement in agricultural landscapes. Doing this in the form of hedgerows or agroforestry systems would lead to only minimal reduction in production areas, and provide additional benefits such as biodiversity conservation.

Lacewing sampling and pollen diet

Adult lacewings were sampled in 2019 in the Upper Rhine Valley, Rhineland-Palatinate, Germany (49.07-49.29°N, 8.09-8.34°E), in 24 landscapes (radius 0.5 km) that span a gradient of landscape composition and configuration and floral resource availability (Fig 1; see Eckerter et al., 2020 and Supporting Information for details). Landscape centres were at least 800 m apart (average: 1993 ± 183 m). Land-use categories (areas of forest, arable land, semi-natural habitats, urban areas) at every landscape were mapped in 2019 during insect and plant samplings. Using QGIS, measures of landscape composition (proportion of cereals, proportion of semi-natural habitats, Shannon Index of landscape and crop diversity) and configuration (edge density, calculated as the total length of edges between landscape elements divided by the total area of the landscape; and mean field size, i.e., average size of arable fields on each landscape) were calculated for each landscape (Table S1).

In the centre of each landscape, two sticky traps were placed at the edge between a herbaceous semi-natural habitat and an arable field at 1.5 m height, separated by 20 meters. Traps had two collecting surfaces facing South and North and each surface had blue, white, and yellow sticky areas (see Supporting Information for more details). Traps were active for a total of eight weeks from May 1st to July 4th. For further analyses, the sampling period was divided into early season (from May 1st to June 4th, corresponding to approximately 100 to 200 Growing Degree Days -GDD-) and late season (from June 5th to July 4th, equivalent to 200 to 400 GDD). This classification was based on a previous study of C. carnea pollen diet, which showed a marked compositional shift between these phases (Bertrand et al., 2019). Each landscape was visited every two weeks and all the collected lacewings were removed from the traps using brushes with white spirit and tweezers. Each individual was stored in a separate Eppendorf tube with 70% ethanol. In the laboratory, all the lacewings belonging to the genus Chrysoperla were identified to the species level using morphological features when possible (San Martin, 2004; personal communication with P. Duelli).

The consumed pollen by lacewings (i.e., pollen grains in the gut) were retrieved using acetolysis following the steps of Jones (2012, 2014). During this procedure, insect tissues and all pollen material with the exception of the outer pollen wall, are removed (Southworth 1974), yielding clean pollen exines for identification. The method was previously used to study the pollen diet of lacewings and ladybeetles (Alcalá Herrera et al., 2021; Bertrand et al., 2019; Villenave et al., 2006). In short, external pollen on the bodies of lacewings were removed by rinsing them in ethanol. Each individual was placed in a centrifuge tube for acetolysis. After centrifugation, pollen pellets were diluted in 50% (v/v) glycerin and put on slides for pollen identification (Supporting Information). Pollen grains were identified using palynological keys (Beug, 2004), a photo atlas (Reille, 1992), and reference collections of the Ecosystem Analysis Group (University Kaiserslautern-Landau). Pollen were usually identified to plant species level, but for some cases to lower taxonomic levels. We further refer to “pollen records” to describe all the different pollen types identified across all collected lacewings.

Floral resource mapping

The major floral pollen resources for lacewings were assessed during two different sampling rounds (see Supporting Information for details). First, the cover (in m²) of all woody plants was mapped in 2017-2018 in sampling plots of 10 x 10 m in all hedgerows and forest edges (first ten meters) of each landscape (orange dots in Fig. 1D; Eckerter et al., 2020). The forest interior was not mapped since C. carnea is mostly active in open farmland during spring and summer (Duelli et al., 2002). The cover of each tree was estimated by measuring its crown projection area on the ground. Second, the cover (in m²) of herbaceous flowering plants was mapped in 2019 within each landscape element containing herbaceous flowering plants. In grassy field margins, hedgerows, and forest edges, the total cover of herbaceous flowering plants was recorded (red lines in Fig. 1D). In meadows and crops, the cover of flowering plants was upscaled from ten 1 m²-subplots to the total meadow and crop area (green and yellow polygons in Fig. 1D, respectively). Mapped plants were identified at species level using vegetation keys (Parolly et al., 2006; Schauer et al., 2012; Schulz, 2014). For herbaceous plants, only species that represented >5% of the pollen consumed by C. carnea in a previous study in the region (Bertrand et al., 2019) were sampled (Table S2). The locations of these flowering plant species were mapped and digitized to create floral resource maps in QGIS 3.18.3 (QGIS Development Team, 2021). Floral resource area (i.e., area of the landscape covered by plants consumed by lacewings), flower diversity (Shannon Index of diversity, considering the identity and area of each plant species), and early and late Flower Availability Indices (FAI) were calculated with this information (see Eckerter et al. 2020 and Supporting Information for details).

Floral resource distribution and movement distance estimation

To infer minimal movement distances and characterize the floral resource distribution, we measured all distances between the location of the sticky trap where a lacewing was captured and the locations of plants from which pollen detected in that lacewing could originate within that respective landscape using QGIS. Since lacewings usually ingested pollen from more than one plant species, this procedure was followed for each pollen type. This resulted in a list of distances for each pollen type that a specific lacewing individual had ingested. From this list, the minimal distance to each pollen source (distance to the nearest source of that pollen type, which represents the minimal distance a lacewing must have moved after consuming the pollen grains), and mean and median distances (average distances to all the resources of that type in the landscape, which is a measure for the resource distribution in the landscape) were calculated for each lacewing using the R package dplyr (Wickham et al., 2015). Furthermore, for lacewings that consumed more than one pollen type, we selected the plant species that had the highest minimal distance because this is the minimal distance that the lacewing individual must have moved. These calculations were done separately for insect- and wind-pollinated plants, but only data from insect-pollinated plants was used for addressing objectives (iii) and (iv), as we expected that these represent visits to insect-pollinated plants. For pollen from wind-pollinated plants, we cannot exclude that lacewings were taking up the pollen after being displaced by wind.