Aim

In this study, we assessed the importance of local to landscape-scale effects of land cover and land use on flying insect biomass. Our main prediction was that insect biomass would be lower within more intensely used land, especially in urban areas and farmland. Location Denmark and parts of Germany.

Methods

We used rooftop-mounted car nets in a citizen science project (‘InsectMobile’) to allow for large-scale geographic sampling of flying insects. Citizen scientists sampled insects along 278 five km routes in urban, farmland and semi-natural (grassland, wetland and forest) landscapes in the summer of 2018. The bulk insect samples were dried overnight to obtain the sample dry weight/biomass. We extracted proportional land use variables in buffers between 50 and 1000 m along the routes and compiled them into land cover categories to examine the effect of each land cover, and specific land use types, on insect biomass.

Results

We found a negative association between urban cover and insect biomass at a landscape-scale (1000 m buffer) in both countries. In Denmark, we also found positive effects of all semi-natural land cover types, i.e. protected grassland (largest at the landscape-scale, 1000 m) and forests (largest at intermediate scales, 250 m). Protected grassland cover had a more positive effect on insect biomass than forest cover. The positive association between insect biomass and farmland was not clearly modified by any variable associated with farmland use intensity. The negative association between insect biomass and urban land cover appeared to be lessened by increased urban green space.

Main conclusions

Our results show that land cover has an impact on flying insect biomass with the magnitude of this effect varying across spatial scales. However, the vast expanse of grey space in urbanised areas has a direct negative impact on flying insect biomass across all spatial scales examined.

Route design and flying insect sampling

Sampling was carried out along 211 routes from 1 - 30 June 2018 in Denmark, and along 67 routes between 25 June - 8 July 2018 in Germany. Across both countries, routes were created across five land cover types: farmland, grassland, wetland, forest and urban areas. No route could be designed to fit 100% to one land cover type, but each route was designed to target a specific land cover type. Each sampling event was either driven in one direction for 10 km or 5 km driven back and forth, to cover as much of the targeted land cover type as possible. The routes were constructed in ArcGIS and QGIS using information from Google Earth, Google Maps, OpenStreetMap (OSM), including data from Danish authorities on land cover types in Denmark, and also using the German ATKIS data (Amtliches Topographisch-Kartographisches Informationssystem) in Germany. Based on the target land cover, in Denmark, 64 urban, 59 farmland, 63 grassland, 62 wetland and 66 forest routes were designed. In Germany, 12 urban, 15 farmland, 12 grassland, 17 wetland and 14 forest routes were created.

Sampling of each route was carried out once during two time intervals on the same the day: between 12-15 h (midday) and again between 17-20 h (evening) with a maximum speed of 50 km/h and weather conditions of at least 15°C, an average wind speed of maximum 6 m/s and no rain. Insects were collected in individual sampling bags that were placed in 96% pure ethanol and stored in double-sealed plastic bags before the citizen scientists sent the samples back to the research institutions. 

Flying insect biomass

Insects were removed from the sampling bag with a squeeze bottle containing 96% EtOH and forceps. Empty 15 or 50 ml centrifuge tubes were weighed, and the insects were transferred to the tubes. The insects were dried overnight at 50 ̊C in an oven (>18hrs), and the tubes containing the dry insects were weighed (Mettler Toledo ME303 in Denmark, Quintix® Precision Balance 310 g x 1 mg in Germany) to obtain sample biomass (in total mg). 

Environmental variables

We extracted land use predictors for insect biomass from four buffer zones for each route: 50 m, 250 m, 500 m, and 1000 m in five categories; urban, farmland, grassland, wetland, and forest. The buffers were calculated as linear buffers around each route. Land use intensity data for Denmark were extracted for farmland and urban routes.

Other variables

We extracted potential car stop variables to account for sampling heterogeneity. We obtained the number of traffic lights or stops of any type (e.g. roundabouts, pedestrian crossings, stop signs, railroad crossings) within a 25-30 m buffer using OSM. For Danish routes, we obtained the number of roundabouts using data from the Danish Map Supply provided by SDFE (Agency for Data Supply and Efficiency) (GeoDenmark-data), since data on roundabouts in Denmark was limited to three records in OSM. Mean hourly temperature and wind was extracted for each route including date and time band from the nearest weather station using the rdwd R package for German routes. For Danish routes, temperature (within increments from 15-20, 20-25 and 25-30 ℃), average wind speed (within increments from light Breeze 1.6-3.3, gentle breeze: 3.4-5.5, and moderate breeze 5.5-7.9 metre/second), and sampling time (hh:mm) were registered by the citizen scientists.

Analyses must be run separately for each country.