Global warming is assumed to restructure mountain insect communities in space and time. Theory and observations along climate gradients predict that insect abundance and richness, especially of small-bodied species, will increase with increasing temperature. However, the specific responses of single species to rising temperatures, such as spatial range shifts, may weaken such prediction, asking for intensive monitoring of real-world communities over time. Here, we examined the temporal and spatial change in wild bee communities and its drivers along two largely well-protected elevational gradients (alpine grassland vs. prealpine forest), each resampled within the last decade. Along both gradients, we detected clear upward shifts in bee communities, with cold-adapted bumblebee species reacting particularly sensitive, demonstrating the speed with which mobile organisms can respond to climatic changes. Mean annual temperature (MAT) was identified as the main driver of species richness in both regions. Accordingly, and in large overlap with expectations under climate warming, we detected an increase in bee richness and abundance, and an increase of small-bodied species in low- and mid-elevations along the grassland gradient. While changes in upward shifts, species richness, abundance and body size in the prealpine forest gradient were partly consistent with the alpine grassland system, they were generally weaker, which could be due to a much less severe warming trend than in the alpine grassland system or due to the variable and possibly confounding effect of canopy cover. Our study highlights the use of accurate assessed abundance data revealing rapid changes in bee communities over only one decade. We conclude, that in well-protected temperate regions, small-bodied bees may initially profit from warming temperatures, by getting more abundant and diverse.

The study was conducted along two elevational gradients in two different mountain ranges both located in south-eastern Germany:

The first gradient was located in the National Park Berchtesgaden and vicinity, the so called Berchtesgadener Land (study region further referred to as BGL) and ranged from 641 m to 2032 m above sea level (m.a.s.l.).  In BGL, 33 grassland sites were selected (60 x 60 m) along 5 mountain transects in 2009, of which 18 were extensively managed (grazed by cattle or sheep or mowed once in late summer) and 15 were not managed (abandoned during the last century or above the treeline). Sites were resampled in 2019. In BGL, wild bees and honeybees were recorded from May to September thus encompassing approximately the full foraging season. Monitoring was conducted in standardized transect walks (10 evenly distributed sub-transects per site; duration of transect walk: 50 min (5 min per sub-transect)), six times in lower elevations (<1200 m.a.s.l.) and due to a shorter season five times in higher elevations. Monitoring took place from 9:30 to 18:00 when the weather was sunny or when temperature at 650 m.a.s.l. was above 17°C on cloudy days. We used sweep nets to catch the insects. 

The second prealpine forest gradient, ranging from 297 to 1368 m asl, was sampled along 4 transects within the Bavarian Forest National Park and in lowland plots (< 650 m.a.s.l.) outside the NP, which extended the elevational gradient down to the Danube River (further referred to BF). Here, 47 study sites were selected, which were first sampled in 2007 and 2008 and resampled in 2016. All sites were located within continuous forest or forest patches and varied in canopy cover as a measure of the summarized canopy cover of the tree layers and the shrub layer (range: 0- 180; with 0 = no forest; 300 = maximum possible density, details are described below). At elevations > 1150 m asl this was primarily caused by windthrow events or bark beetle damage. The study region is dominated by mixed mountain forests of Norway spruce Picea abies, European beech Fagus sylvatica and silver fir Abies alba in the montane zone and mixed beech forests in lower elevations.In BF, wild bees and honeybees were recorded each month during May and September by deploying one flight-interception trap and one malaise trap on each site. While flight interception traps were activated over 4 weeks, malaise traps sampled over a period of 2 weeks each month.

We further provide site information (elevation) including temperature data which was modeled from adjacent climate stations and data from the flowercover surveys for BGL and canopy cover per site data for BF.

Note that in BGL we recomment to summarise morphological very similar species Bombus lucorum and Bombus terrestris to Bombus sensu stricto complex due to field identifications.