Spoiled sweet: Effects of habitat and foraging conditions on sucrose responsiveness in honey bees

---------------------------------------------------------------------------------------------------------------------------------------------GENERAL INFORMATION--------------------------------------------------------------------------------------------------------------------------------------------- Dataset and code for: Balogh et al. (submitted). Spoiled sweet: Effects of habitat and foraging conditions on sucrose responsiveness in honey bees. Field data from six study sites in the Rhine-Main Metropolitan Region, Germany, collected June 2021 – August 2023. See README for details. ---------------------------------------------------------------------------------------------------------------------------------------------BRIEF ABSTRACT--------------------------------------------------------------------------------------------------------------------------------------------- Honeybee (Apis mellifera) foragers vary in how selectively they respond to nectar sugar content, yet the ecological drivers of this variation remain poorly understood. We investigated sucrose responsiveness (SR) — quantified via the Proboscis Extension Reflex (PER) — in relation to habitat-specific ecological parameters in a multi-site field study spanning three seasons (2021–2023) in the Rhine-Main Metropolitan Region, Germany. Colony growth was quantified using hive scales, recent weather conditions using data from nearby meteorological stations, and forage quality by analyzing the honey crop contents of foragers. Using generalized additive models with ordinal response, we found that SR was higher when conditions were poor: slow colony growth, unfavorable weather, and low forage quality, indicating increased acceptance of low-quality nectar when resources are scarce. These relationships were consistent across years. SR is a state-dependent trait that integrates ecological information across multiple biological scales. ---------------------------------------------------------------------------------------------------------------------------------------------CONTENTS--------------------------------------------------------------------------------------------------------------------------------------------- SpoiledSweet.zip/ Data/ (3 files): The bee-level dataset (3,899 rows, 31 columns), the hourly meteorological record from which the weather predictor is built (17,520 rows, 5 columns), and hive weight data (7,645 rows, 6 columns). Variable definitions are in README.txt. R/ (3 files): Figure constants, theme and export helper; model formulas, effect-size helpers and partial-effect extraction; binning and panel functions for the supplementary figure; the optimal foraging hours definition and the functions that rebuild it. Reports/ (3 files): The analysis, as an R Markdown source and its knitted HTML and Markdown output. Knitting the .Rmd regenerates every figure and table in this deposit; nothing else needs to be run. Figs/ (7 files): The four figures as submitted (Fig1, Fig2, FigS1, FIGs2) at 600 dpi, plus three screen-resolution images written automatically while knitting. Output/ (15 files): Derived tables written by the analysis. Model summaries, parametric and smooth terms, effect sizes, the seasonal model comparisons, the sensitivity checks, residual diagnostics, and the counts behind the supplementaryfigure. README.txt: Full variable definitions, site coordinates, data provenance, model specifications and software versions. Licenses.txt: Licence terms. Uploaded separately: README.txt and Licenses.txt, identical to the copies inside the archive, so they can be read without downloading it. ---------------------------------------------------------------------------------------------------------------------------------------------REPRODUCING THE ANALYSIS--------------------------------------------------------------------------------------------------------------------------------------------- Open Reports/Analysis_report.Rmd in R (4.5.1 or later) and knit it. Paths are relative to SpoiledSweet/, which the report sets as its working directory. No other input is required. A full knit takes about 8 minutes on a 2.3 GHz laptop, most of it in the residual diagnostics and the three sensitivity sweeps. Required packages: tidyverse, lubridate, mgcv, gratia, DHARMa, cowplot, ragg, viridisLite. ---------------------------------------------------------------------------------------------------------------------------------------------SOURCES AND LICENSES--------------------------------------------------------------------------------------------------------------------------------------------- Weather observations are derived from the German Meteorological Service (Deutscher Wetterdienst), stations Frankfurt/Main-Westend (01424), Kleiner Feldberg (02601) and Frankfurt Airport (01420), licensed CC BY 4.0 and publicly available at https://opendata.dwd.de/. Wind speed at the Messe site is a calibrated estimate; the calibration is derived in a companion study, archived separately (see Related identifiers). Bee Friendliness values are from Alves da Silva et al. (2021), B-GOOD deliverable D3.1, EU Horizon 2020 grant 817622. Data are released under CC BY 4.0. R code is released under the MIT licence. Both texts, and the full source attributions, are in Licenses.txt.