Climatic variation shapes species interactions by filtering communities along environmental gradients, yet the role of fine-scale thermal heterogeneity in structuring mutualist assemblages along coastal-to-interior gradients remains underexplored. We tested the hypothesis that thermal niche differentiation promotes spatial, temporal, and phenological partitioning among insect pollinators in a Great Lakes landscape. Specifically, we examined how native and non-native pollinators responded to horizontal (coastal-to-interior) and vertical (ground-to-1 m) thermal gradients and how these responses aligned with the flowering phenology of Malus domestica. Over an 11-week period, we collected more than 14,000 insects representing 17 genera of flies and 10 genera of bees. Diptera were more abundant near the lake, earlier in the season, and at elevated trap heights, consistent with broader thermal tolerance. Native bees – particularly Lasioglossum and Agapostemon – showed stronger phenological synchrony with bloom than either Apis mellifera or syrphid flies, despite the presence of managed honeybee hives. These findings support the hypothesis that fine-scale thermal gradients drive niche segregation among mutualists, with consequences for the composition, timing, and potential resilience of pollinator communities. As climate change alters thermal regimes, understanding how pollinators partition microclimatic space and time will be essential for predicting shifts in plant–pollinator interactions.

Study System and Location

The research was conducted in commercial Malus domestica (apple) orchards owned by Bittner-Singer Farms, located along the southern shoreline of Lake Ontario in Appleton, New York (43.34° N, 78.67° W). The site was selected for its spatial proximity to Lake Ontario, creating a naturally occurring thermal gradient resulting from the lake’s thermal inertia. The orchards consist of mature M. domestica trees (~30 years old), planted in rows running north to south, allowing transects to be established perpendicular to the lake shoreline and aligned with the presumed inland thermal gradient.

Experimental Design

Four linear transects were established on April 5, 2024, to sample pollinator activity and tree phenology across a spatial thermal gradient. Three transects extended from the lake inland with approximate lengths of 525 m, 525 m, and 450 m. A fourth, shorter transect (75 m) was added to increase replication and distance from the lake. Each transect was subdivided into evenly spaced sampling plots at ~75 m intervals (n = 24 total plots).

Each plot was centered on three adjacent M. domestica trees – one focal center tree and one comparison tree each to the east and west. This yielded 72 trees in total. Plots were spaced to minimize spatial autocorrelation between pollinator communities. Pollinator communities were sampled using paired blue vane traps deployed at each plot. Each plot received two traps: (a) ground-level – 10 cm above the soil surface and (b) elevated to ~1 meter height on a standard shepherd’s hook. Blue vane traps are passive interception traps known for high efficacy in capturing a broad array of pollinating taxa, especially bees (Joshi et al. 2015, Hall 2018). The blue top and yellow funnel mimic floral visual cues, attracting pollinators that then fall into a detergent-water solution. Traps were deployed in the same position each week and left open for three consecutive days.

The 75 m spacing between plots with blue vane traps falls within the 100-300 m foraging radius of many native bees (Greenleaf et al. 2007, Zurbuchen et al. 2010), but the effective range of blue vane traps is relatively localized (25-50 m) as the passive traps do not attract bees from long distances but intercepts them as they forage (Cane et al. 2000, Kimoto et al. 2012). Sampling was conducted weekly from April 9 to June 17, 2024, encompassing pre-bloom to post-bloom phenological stages.

All insect specimens were sorted and identified to order. Hymenoptera and Diptera taxa were further identified to genus (Borror et al. 1989, Michener et al. 1994, Michener 2007, Wilson and Carril 2015, Skevington et al. 2019). Genus-level identification focused on known and potential pollinators, including bees such as Lasioglossum, Andrena, Agapostemon, Osmia, and Apis, and flies such as Toxomerus, Musca and Drosophila.

Malus domestica phenology

The flowering phenology of each center tree in all plots was scored twice per week during the sampling period using an M. domestica scale based on the BBCH system, which is a standardized coding system for the phenological development stages of plants that is widely used in agriculture and plant sciences for uniform documentation of plant growth stages (Meier, 2001). Our modified scale for M. domestica was: 0 = bud stage (pre-flowering); 1 = Initial bloom (<25% flowers open); 2 = peak bloom (>50% flowers open); 2.5 = bloom decline (<25% flowers remaining); 3 = senescence (no flowers present).

Temperature Monitoring

To assess spatial and microclimatic thermal gradients, temperature loggers (Onset HOBO Pendant MX2201) were deployed at each plot in two positions: buried ~10 cm below ground (which approximates nesting depth for solitary bees; Wilson and Carril 2015) at each trap (n = 24) and, to capture ambient temperatures, elevated at the ends of each transect on the north-facing side of trees on shepherd hooks (~1 m height ), shielded from solar radiation using custom white foam cups with ventilation slits. The loggers were programmed to record temperature at 12:00 AM and 12:00 PM daily.