Caste-Specific Proboscis Extension Responses in Honey Bees to Sucrose and Royal Jelly Stimuli

33 Understanding the nutritional preferences of honey bees ( Apis mellifera ) is essential for 34 comprehending their behavioral ecology and the division of labor within a colony. While 35 gustatory sensitivity to sucrose is well-documented in workers, a significant research gap exists 36 regarding the sensory responses of queens and their reactions to caste-specific nutrition such as 37 royal jelly. This study utilized the proboscis extension response (PER) assay to compare the food 38 preferences of three distinct bee categories: foragers, 1-day-old workers, and queens. Subjects 39 were presented repeatedly, in a pseudorandom order, with water, sucrose, royal jelly, and a 40 sucrose–royal jelly mixture as gustatory stimuli. Foragers exhibited a high responsiveness to 41 sucrose and showed uniformly low responsiveness to other stimuli. Although 1-day-old workers 42 showed high responsiveness to sucrose, unlike foragers, they also responded to the sucrose–royal 43 jelly mixture. Queens displayed a unique response profile, with near-ceiling responsiveness to 44 both royal jelly and the mixture, followed by response to sucrose solution without habituation. 45 Additionally, responsiveness to the sucrose was higher in foragers than in 1-day-old workers. 46 These findings suggest that the honey bee gustatory and sensory system is tuned to the specific 47 nutritional requirements of caste and age. 48

Honey bee, Proboscis extension response, Caste, Queen bee, Royal jelly, Sucrose 49 50 1. INTRODUCTION 51 Understanding the food preferences of honey bees ( Apis mellifera) is crucial for unraveling the 52 behavioral ecology of this key pollinator species. Food selection not only affects individual bee 53 survival and colony-level resource allocation but also influences feeding dynamics, trophallactic 54 interactions, and division of labor (Farina & Núñez, 1991; de Brito Sanchez, 2011). These 55 preferences are driven by physiological requirements, sensory response thresholds, and the 56 specific tasks the bees perform. 57 The food sources of honey bees vary by their caste (i.e., queens versus workers), and their 58 temporal (i.e., foragers versus nurses) and behavioral division of labor (i.e., pollen versus nectar 59 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint foragers). Temporal polyethism is an age-related division of labor where younger bees carry out 60 in-hive tasks such as nursing, while older bees take on outside jobs such as nectar and pollen 61 foragers (Seeley, 1982). Pollen is the primary natural source of protein, and nurses consume 3.4 62 to 4.3 mg pollen per day (Crailsheim et al, 1992). Pollen provides essential amino acids, 63 including arginine, histidine, lysine, tryptophan, phenylalanine, methionine, threonine, leucine, 64 isoleucine, and valine (de Groot, 1952). This protein source is essential for the development of 65 their hypopharyngeal glands, to synthesize the protein-rich jelly used to feed larvae and the 66 queen (Crailsheim, 1990a). In addition to pollen consumption, young bees receive jelly via 67 trophallaxis from older nurse bees (Crailsheim, 1990a; Free, 1957; Crailsheim, 1990b). Nurses 68 also receive nectar collected by foragers. However, the sugar concentration in the crops of 5 to 6-69 day-old worker bees is typically lower than that found in the crops of active foragers (Pankiw et 70 al., 2004). Foragers primarily consume carbohydrates to fuel the intense metabolic cost of flight. 71 The transition to foraging causes physiological shifts. Hypopharyngeal gland atrophy occurs, and 72 proteolytic enzyme activity decreases significantly in foragers (Crailsheim & Stolberg, 1989). 73 Especially, pollen consumption decreases with age (Lotmar, 1938; Lindauer, 1952). Thus, the 74 diet shifts from protein accumulation to carbohydrate catabolism (Crailsheim, 1990a). The queen 75 is fed royal jelly. The queen receives protein that has already been processed and synthesized by 76 the nurse bees (Crailsheim, 1990a). This high-protein diet supports her highly developed 77 reproductive characteristics (Fèvre & Dearden, 2024). 78 The proboscis extension response (PER) is widely considered a standard and established assay 79 for quantifying both gustatory responsiveness and appetitive learning (Scheiner et al., 2003). The 80 PER can be utilized in a non-associative learning assay, which is especially used to test gustatory 81 responsiveness or non-associative forms of learning. This is measured by repeatedly stimulating 82 the antennae with a solution and observing the frequency or magnitude of the proboscis 83 extension over time (Scheiner et al., 2003). Honey bees give PER to sucrose when the solution 84 comes into contact with their antennae (Bitterman et al. 1983). Newly emerged bees typically 85 have low responsiveness to sucrose, and when they approach foraging age, their responsiveness 86 increases (Pankiw & Page, 1999). Sucrose responsiveness is also higher in foragers than in nurse 87 bees and is related to the expression of specific genes in the brain (Degirmenci et al., 2018). In 88 addition, young bees exhibit a higher habituation rate to sucrose than older bees (Guez et al., 89 2001; Scheiner et al., 2003). Thus, sucrose responsiveness generally increases as bees age; 90 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint younger bees are less responsive to sucrose. Another factor that influences responsiveness is the 91 social context: brood pheromone lowers the sucrose response thresholds of young bees (Pankiw 92 & Page, 2001). Foragers also vary in sucrose response. The sucrose sensitivity of foragers 93 correlates with their foraging specialization. Water and pollen foragers have the highest 94 responsiveness to sucrose, followed by nectar foragers (Scheiner et al., 2001). Queen bees are 95 also responsive to sucrose (Gong et al., 2018). In addition to sucrose, worker bees also give PER 96 to pollen (Nicholls & Hempel de Ibarra, 2013), and pollen foragers are more responsive to pollen 97 (Moreno & Arenas, 2023; Moreno & Arenas, 2024). 98 While taste sensitivity to sucrose and PER-based learning in honey bees have been extensively 99 characterized, particularly across age groups and nutritional specializations, there is a gap 100 regarding caste-comparative PER studies involving queens and queen-related nutritional stimuli 101 (royal jelly). Based on the caste-specific nutritional source of honey bees and the queen's royal 102 jelly-based diet, we hypothesize that PER will differ for caste-specific stimuli. We predict 103 foragers and young workers will show the highest sensitivity to sucrose, while queens will show 104 high sensitivity to royal jelly. We test this hypothesis using a PER assay in which foragers, 1-105 day-old workers, and queens are presented with water, sucrose, royal jelly, and a royal jelly–106 sucrose mixture across repeated, pseudorandomized trials. 107 108 2. METHOD 109 2.1. Collecting bees and handling 110 Proboscis extension response (PER) experiments were conducted with three bee categories: 111 foragers, 1-day-old workers, and queens. Returning foragers were collected at the hive entrance 112 using a plastic mesh (Giray et al., 2007) and placed in flight boxes, where they were food-113 deprived for 2 h. For the 1-day-old group, capped brood combs were removed from the hives 114 and placed in a box inside an incubator at 35 °C and 65% relative humidity overnight. Newly 115 emerged 1-day-old workers were collected 2 h before the experiment and placed in flight boxes 116 for food deprivation. Both foragers and 1-day-old workers were obtained from two source 117 colonies. After the food-deprivation period, bees were harnessed and kept ready for testing. 118 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint Queen bees were obtained from three different beekeepers. Mated queens were collected from 119 their mini-nucs, small colonies with ~2000 workers, where each queen was laying eggs, and was 120 attended by her worker offspring. Each queen was transferred to our lab in a queen cage with 5–121 10 attendant nurse bees and placed in the incubator until the experiment (Fahrbach et al. 1995). 122 Before testing, queens were separated from the nurses, harnessed, and prepared for PER assays 123 as described for the worker groups. 124 2.2. Test procedure 125 Four test solutions were prepared: water, pure royal jelly, 50% sucrose solution (w/v), and a 126 mixture consisting of 50% royal jelly, 25% sucrose, and 25% water (w/w/v). To assess PER, 127 separate cotton swabs were dipped into each solution and brought into contact with the antennae 128 of the bee for 3 s. Each bee received a total of 60 trials presented in a pseudorandom order (S, W, 129 RJ, M, S, M, W, RJ, M, S, W, RJ, M, W, S, RJ, M, W, RJ, S, W, S, RJ, M, S, M, W, RJ, M, S, 130 W, RJ, M, W, S, RJ, M, W, RJ, S, W, S, RJ, M, S, M, W, RJ, M, S, W, RJ, M, W, S, RJ, M, W, 131 RJ, S), where S = sucrose, W = water, RJ = royal jelly, and M = mixture, each stimulus appears 132 15 times overall with no consecutive repeats. The sequence prevents conditioning to a visual cue, 133 such as an upcoming cotton swab, and creates a similar sensitization or habituation effect for all 134 solutions. The inter-trial interval was 1 min for each bee. For each stimulation, the presence (1) 135 or absence (0) of PER was recorded. The forager and 1-day-old worker groups each consisted of 136 24 individuals, and the queen group consisted of 10 individuals. 137 2.3. Statistical analysis 138 All statistical analyses were conducted in RStudio using the R version 4.3 (R Core Team, 2020). 139 Data manipulation and reshaping were performed using the readxl , dplyr, tidyr , and stringr 140 packages. 141 While raw PER ratios (the total number of positive responses divided by 15 trials per bee) were 142 calculated to provide a descriptive overview of the data, these summary statistics do not account 143 for individual variation or temporal effects like habituation. Therefore, formal inference was 144 based on a binomial generalized linear mixed model (GLMM) with a logit link. 145 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint Because PER is a binary outcome (0 or 1) recorded repeatedly from the same individuals across 146 trials and stimuli, inference was based on a GLMM rather than independent-samples tests. A 147 binomial GLMM with a logit link was fitted using the g lmer function in the lme4 package, 148 including bee categories (forager, 1-day-old, queen) and stimulus (water, sucrose, royal jelly, 149 mixture) to test whether stimulus responsiveness differed among bee categories. The repeated-150 measures structure was accounted for by including bee identity as a random intercept, allowing 151 baseline response probability to vary among individuals. This modeling choice is appropriate 152 because it respects the non-independence of trials within bees, uses the correct distribution for 153 binary outcomes, and yields interpretable bee categories-by-stimulus effects while retaining trial-154 level information. We therefore analyzed PER using binomial logistic models, which estimate 155 effects on the log-odds scale and then convert them to predicted response probabilities via the 156 logistic link. Using probabilities provides model-based estimates that account for the discrete 157 nature of the data and the non-independence introduced by repeated measures within bees. 158 To capture systematic changes in responsiveness across repeated stimulation (e.g., habituation or 159 sensitization), trial order was included as a scaled linear term (trial) and a quadratic term (trial²). 160 Both trial-order terms were retained in the final model. 161 Model adequacy was evaluated using diagnostic functions from the performance package, 162 including checks for overdispersion and singularity. Predicted probabilities were obtained as 163 estimated marginal means (EMMs) on the response scale using the emmeans package, which 164 back-transforms log-odds to PER probabilities and provides confidence intervals. Multiple 165 comparisons among EMMs were conducted as Tukey-adjusted pairwise contrasts. 166 To test habituation within each stimulus, we evaluated whether PER decreased across the 15 167 repeated presentations of that stimulus. We fitted a binomial generalized linear mixed model 168 (GLMM) with a logit link, using the within-stimulus repetition index (1–15) as the focal 169 predictor and including bee identity as a random intercept to account for repeated measures 170 within individuals. The repetition index was z-standardized prior to model fitting to improve 171 numerical stability and allow comparable effect scaling across groups. Model fitting used 172 maximum likelihood with the Laplace approximation. Habituation was inferred when the 173 repetition coefficient was negative and statistically different from zero ( α = 0.05). Inference was 174 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint based on a one-sided test derived from the Wald z statistic, and p-values were adjusted across the 175 set of caste × stimulus tests using the Holm procedure to control the family-wise error rate. 176 177 3. RESULTS 178 3.1. PER ratios to gustatory stimuli 179 Foragers responded most strongly to sucrose, with comparatively low and similar responses to 180 royal jelly, the mixture, and water. In 1-day-old workers, response to sucrose was again the 181 highest, and the mixture elicited responses comparable to sucrose, whereas royal jelly and water 182 elicited lower responses. In contrast, queens showed near-ceiling PER to royal jelly and the 183 mixture, both stimuli exceeding sucrose and water. Foragers (n = 24) exhibited a high mean and 184 median PER ratio for sucrose (μ = 0.931, M = 1) but low ratios for water (μ = 0.403, M = 0.467), 185 royal jelly (μ = 0.386, M = 0.400), and the mixture (μ = 0.411, M = 0.400). In 1-day-old workers 186 (n = 24), sucrose elicited the strongest response ( μ = 0.792, M = 0.867), followed by the mixture 187 (μ = 0.619, M = 0.667), royal jelly ( μ = 0.494, M = 0.467), and water ( μ = 0.439, M = 0.467). 188 Queens (n = 10) exhibited near-maximal responsiveness to both the mixture ( μ = 0.993, M = 1) 189 and pure royal jelly ( μ = 0.993, M = 1). In contrast, sucrose yielded lower responses ( μ = 0.800, 190 M = 0.900), and water elicited the minimum response ratio (μ = 0.160, M = 0.100). 191 We then analyzed the proboscis extension responses (PER: 1, no PER: 0) (Figure 1A-C) at the 192 trial level using a binomial generalized linear mixed model (GLMM) with a logit link, with 193 stimulus (sucrose, water, royal jelly, and mixture), bee categories (forager, 1-day-old worker, 194 queen), and their interaction as fixed effects and a random intercept for individual bee. The 195 model indicated strong effects of trial order in scaled trial and its square (standardized trial: z = 196 −18.39, p < 0.001; trial²: z = 5.13, p < 0.001). Model diagnostics indicated no overdispersion and 197 non-singular random effects. Estimated marginal means (EMMs) were reported as predicted 198 probabilities with 95% confidence intervals (Table 1). 199 3.2. Response patterns of bee categories 200 Foragers showed a strong response to the sucrose stimulus. The predicted PER probability (prob) 201 for sucrose was high (prob = 0.965, 95% CI [0.931, 0.983]), whereas water (prob = 0.369, 202 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint [0.242, 0.517]), royal jelly (prob = 0.364, [0.238, 0.513]), and mixture (prob = 0.375, [0.246, 203 0.523]) were uniformly low (Table 1, Figure S1). 204 In 1-day-old workers, PER also peaked for sucrose but showed a graded pattern across stimuli. 205 Predicted PER was highest for sucrose (prob = 0.887, 95% CI [0.806, 0.937]), followed by 206 mixture (prob = 0.687, [0.542, 0.803]), then royal jelly (prob = 0.521, [0.371, 0.668]) and water 207 (prob = 0.415, [0.278, 0.567]) (Table 1, Figure S1). 208 Queens showed a qualitatively different profile. Predicted PER was near maximal for royal jelly 209 (prob = 0.999, 95% CI [0.991, 1.000]) and mixture (prob = 0.999, [0.990, 1.000]), high for 210 sucrose (prob = 0.907, [0.774, 0.965]), and low for water (prob = 0.064, [0.023, 0.163]) (Table 1, 211 Figure S1). 212 3.3. Comparison of responses to stimuli within bee categories 213 Tukey-adjusted pairwise comparisons confirmed these patterns. In foragers, sucrose exceeded 214 water, royal jelly, and mixture (all p < 0.001), whereas water, royal jelly, and mixture did not 215 differ (all p ≥ 0.995) (Figure 1D, Table S1). 216 In 1-day-old workers, sucrose exceeded all other stimuli (all p < 0.001). The mixture also 217 exceeded water (p < 0.001) and royal jelly ( p = 0.001), while the difference between water and 218 royal jelly was not statistically significant (p = 0.105) (Figure 1E, Table S1). 219 In queens, water was lower than each nutritive stimulus (all p < 0.001). Sucrose was lower than 220 royal jelly ( p < 0.001) and mixture ( p < 0.001), and royal jelly and mixture did not differ ( p = 221 1.000) (Figure 1F, Table S1). 222 3.4. Comparison of the responses of bee categories within each stimulus type 223 Bee categories differences depended strongly on the stimuli, which included water, sucrose, 224 royal jelly, and a mixture. Queens responded less than both 1-day-old bees ( p < 0.001) and 225 foragers (p = 0.001), whereas foragers and 1-day-old workers did not differ ( p = 0.897) (Table 226 S2) for the water stimulus. 227 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint With sucrose as stimulus, foragers responded more strongly than 1-day-old workers ( p = 0.027), 228 while neither foragers (p = 0.236) nor 1-day-old workers ( p = 0.938) differed significantly from 229 queens (Table S2). 230 Exposure of royal jelly and mixture stimuli, queens’ PER ratio exceeded both worker groups (all 231 p < .001). However, no difference was found between 1-day-old workers and foragers ( p = 232 0.312) for royal jelly (Table S2). 233 In addition, the 1-day-old workers’ PER ratio exceeded that of foragers for the mixture stimulus 234 (p = 0.009) (Table S2). 235 3.5. Habituation evaluation 236 Habituation was assessed as a decrease in PER probability across the 15 repeated presentations 237 of each stimulus (Figure 1A-C) using stimulus-specific binomial mixed-effects models and 238 defined with logit slopes (b), test statistic (z), and Holm-adjusted p values. 239 In foragers, PER decreased significantly across trials for water ( b = −1.31, z = −8.27, p < 240 0.0001), sucrose ( b = −0.572, z = −2.36, p = 0.046), royal jelly ( b = −1.08, z = −7.31, p < 241 0.0001), and mixture (b = −1.67, z = −8.90, p p = 0.046) 244 In 1-day-old workers, PER decreased significantly across trials for all stimuli: water (b = −0.653, 245 z = −4.71, p < 0.0001), sucrose ( b = −0.587, z = −3.47, p = 0.002), royal jelly ( b = −1.03, z = 246 −6.78, p < 0.0001), and mixture ( b = −1.45, z = −7.63, p < 0.0001) (Figure 1B). The slope for 247 sucrose is also less pronounced than for other stimuli. 248 Here, we observed that both foragers and 1-day-old workers showed evidence of habituation for 249 all stimuli, but the magnitude differed. Foragers exhibited a more pronounced decline in 250 response to water, royal jelly, and mixture stimuli compared to 1-day-old workers. However, the 251 PER ratio of foragers declined modestly (PER ratios from 1.00 on the first trial to 0.916 by the 252 15th trial), with a weaker negative slope for sucrose, whereas 1-day-old workers showed a 253 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint sharper decline (PER ratios from 1.00 at the first trial to 0.667 at the 15th trial) with a stronger 254 negative slope. 255 In queens, there was no statistically significant decrease across trials for water ( b = −0.231, z = 256 −0.943, p = 0.692) or sucrose ( b = −0.0489, z = −0.184, p = 1.00). Royal jelly ( b = 0.783, z = 257 0.659, p = 1.00) and the mixture ( b = 0.236, z = 0.231, p = 1.00) also showed no statistical 258 evidence of habituation in queens (Figure 1C). Notably, queen responses to royal jelly and 259 mixture were near-ceiling (mean PER = 0.993), which limits sensitivity to detect decreases over 260 trials. 261 262 Table 1. Descriptive and model-based summary statistics for PER by bee categories and stimuli. 263 For each category and stimulus condition, the table reports observed PER ratios and GLMM-264 derived predicted PER probabilities. PER ratio is computed as the total of positive PER numbers 265 divided by 15 trials for each stimulus per bee. It is summarized across individuals using sample 266 size (n), mean, standard deviation, standard error, median, and the 1st and 3rd quartiles. The 267 GLMM columns report the estimated marginal mean PER probability on the response scale 268 (prob), its 95% confidence interval (CI lower, CI upper), and the model-based standard error (SE 269 model), derived from a binomial GLMM that accounts for repeated trial-level observations 270 within individuals via a random intercept for bee identity. 271 Category Stimuli Descriptive Statistics for PER Ratios GLMM Sample Size Mean Standard deviation Standard Error Median 1 st Quartile 3rd Quartile prob CI lower CI upper SE model Forager Mixture 24 0.411 0.190 0.039 0.400 0.267 0.533 0.375 0.246 0.523 0.072 Royal Jelly 24 0.386 0.206 0.042 0.400 0.250 0.533 0.364 0.238 0.513 0.072 Sucrose 24 0.931 0.112 0.023 1.000 0.867 1.000 0.965 0.931 0.983 0.012 Water 24 0.403 0.183 0.037 0.467 0.317 0.483 0.369 0.242 0.517 0.072 1-day-old Mixture 24 0.619 0.269 0.055 0.667 0.467 0.817 0.687 0.542 0.803 0.068 Royal Jelly 24 0.494 0.276 0.056 0.467 0.317 0.683 0.521 0.371 0.667 0.078 Sucrose 24 0.792 0.241 0.049 0.867 0.600 1.000 0.887 0.806 0.937 0.033 Water 24 0.439 0.296 0.060 0.467 0.233 0.600 0.415 0.278 0.567 0.076 Queen Mixture 10 0.993 0.021 0.007 1.000 1.000 1.000 0.999 0.990 1.000 0.001 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint Royal Jelly 10 0.993 0.021 0.007 1.000 1.000 1.000 0.999 0.991 1.000 0.001 Sucrose 10 0.800 0.288 0.091 0.900 0.867 0.983 0.907 0.774 0.965 0.045 Water 10 0.160 0.170 0.054 0.100 0.017 0.250 0.064 0.023 0.163 0.032 272 4. DISCUSSION 273 This study demonstrates caste and age-related differences in PER to gustatory stimuli. A higher 274 PER and the absence of habituation (reduction in responsiveness to repeated stimulation) for 275 gustatory stimuli indicate a strong food preference. In the case of queen bees, these data indicate 276 a queen-specific tuning toward royal-jelly-based stimuli. We conclude that queen bees have a 277 very high preference for royal jelly. In contrast, foragers have a low preference for royal jelly 278 and a high preference for high-concentration sucrose reward. On the other hand, 1-day-old bees 279 are intermediate, with a high preference for sucrose and a lesser, but still higher, response to a 280 royal jelly mixture than foragers. These preferences are consistent with the known feeding habits 281 of these three bee categories, which are discussed below. 282 Sugar content of the royal jelly is on average 12% (w/w) (Ramanathan et al., 2011; Kolayli et al., 283 2025). Thus, the sugar concentration of our mixture solution should be higher than 30%, and if 284 this percentage of sugar is not present in the royal jelly mixture, it is used as a high stimulant that 285 breaks the habituation (Scheiner et al., 2004). However, the level of response to the mixture 286 foragers remains at the same level as the response to the water control. One characteristic of 287 royal jelly is its acidity, with a pH range between 3.6 and 4.2 (Ramanathan et al., 2011), which 288 may act as a gustatory deterrent and reduce the response to sugar for foragers. Additionally, 289 other components of royal jelly may alter feeding behavior. For instance, the addition of amino 290 acids, such as isoleucine, proline, phenylalanine, and methionine in sucrose solution decreases its 291 consumption rates by bees (Simcock et al., 2014). Also, proline, phenylalanine, and methionine 292 addition to the sucrose stimuli, reduce the responsiveness (Simcock et al., 2014; Nicholls et al., 293 2019). These amino acids are present in royal jelly (Howe et al., 1985; Bayram et al., 2021), 294 which may cause the reduction of PER to the mixture and royal jelly stimuli. Moreover, 295 typically, weaker stimuli may not even cause dishabituation, but high or strong stimuli do cause 296 dishabituation (Rankin et al., 2009). We observed habituation for mixture and royal jelly in 297 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint foragers and 1-day-old workers, even in the presence of intervening strong stimuli such as a high 298 concentration of sucrose solution. This also indicated that worker bees do not prefer royal jelly. 299 On the other hand, the proximity in PER ratios for the mixture and sucrose solution given by 1-300 day-old bees may be due to the protein affinity of young bees. Protein intake is crucial for the 301 development of the hypopharyngeal glands of 1-day-old workers, and they consume pollen and 302 receive protein-abundant jelly from older nurse bees (Crailsheim, 1990a; Free, 1957; Crailsheim, 303 1990b). The positive response of 1-day-old workers in this study towards the mixture solution 304 containing royal jelly could then be the typical feeding response of these bees due to the rich 305 protein content of royal jelly (Ramanathan et al., 2018). 306 Queens exhibited a near-maximal response to both royal jelly and the mixture throughout the 307 trials. Notably, the queens' response to royal jelly-based stimuli remained consistently high 308 throughout the trials without habituation (Figure 1C). This high and persistent response 309 underscores a highly specialized sensory tuning. Because the queen's reproductive output is 310 dependent on a royal jelly diet (Fèvre & Dearden, 2024). Queens can be fed nectar and pollen by 311 nurse bees, and they can feed individually on these sources (St Clair et al., 2024). Moreover, 312 when the queens were separated from the court bees, they could consume sugar. However, only 313 sugar as food decreases the queen’s longevity (Haydak, 1970). Additionally, there were only 314 trace amounts of sugar found in the gut of the queens during the winter period (Haydak, 1970). 315 Thus, queen bees are responsive to sucrose, but sugar is not an adequate or preferred food for 316 them. 317 Our findings on sucrose response are also consistent with the literature. Queen bees responded to 318 sugar in this study, as shown before (Gong et al., 2018). Older bees showed a higher affinity for 319 sugar than younger bees in this study, as shown by higher PER to sugar by foragers, as in other 320 studies (Pankiw & Page, 1999; Guez et al., 2001; Scheiner et al., 2003; Degirmenci et al., 2018). 321 We also found that the 1-day-old workers had more apparent habituation to sucrose than 322 foragers, in line with previous studies (Guez et al., 2001; Scheiner et al., 2003). 323 The novel finding of our study is that bees have caste and age-related food preferences. High 324 responsiveness to pure royal jelly is exclusive to the queen bee. Forager bees have a high level of 325 fidelity to sugar, and 1-day-old bees do show some preference for royal jelly in a mixture. One 326 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint practical implication of this finding is the ability to independently manipulate the nutritional 327 status of members of a hive to examine the nutritional ecology and physiology of the hive. 328 Future research may also unravel the molecular and neural basis of these caste-specific feeding 329 preferences. For example, investigating the differential expression of gustatory and olfactory 330 receptors across castes may explain how queens and workers are biochemically tuned to their 331 respective diets. Sensory neurons responsive to protein-rich stimuli like royal jelly remain poorly 332 characterized in queens, and their comparison with carbohydrate-sensitive pathways in foragers 333 could provide insight into the adaptive evolution of nutritional specialization in eusocial insects. 334 335 ACKNOWLEDGMENTS AND FUNDING INFORMATION 336 This work was supported by the EU grant RoboRoyale [grant number: 964492]; the Middle East 337 Technical University Research Fund [grant number: ADEP-302-2024-11468]; the U.S. National 338 Science Foundation grant [grant number: 2318597]. 339 340 DATA AVAILABILITY 341 The data are available in the Zenodo repository: https://doi.org/10.5281/zenodo.18347610 342 343 AUTHOR CONTRIBUTIONS 344 BE: Conceptualization, Methodology, Formal analysis, Investigation, Writing - Original Draft, 345 Visualization. SS: Investigation, Resources. OCA: Investigation, Writing – review and editing. 346 AGG: Wr iting - Original Draft. HA: Methodology, Writing – review and editing AET: 347 Methodology, Resources, Funding acquisition. TG: Conceptualization, Methodology, Formal 348 analysis, Writing - Original Draft. ES: Conceptualization, Writing - Original Draft, Resources, 349 Funding acquisition. 350 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint 351

352 Bayram, E. N., Çebi, N., Çelik, S., Gerçek, Y. C., Bayram, S., Tanu ğ ur Samancı, A. E., 353 Sağ dıç, O., & Özkök, A. (2021). Turkish royal jelly: amino acid, physicochemical, 354 antioxidant, multi-elemental, antibacterial and fingerprint profiles by analytical 355 techniques combined with chemometrics. Journal of Apicultural Research, 60(5), 751–356 764. https://doi.org/10.1080/00218839.2021.1889222 357 Bitterman, M.E., Menzel, R., Fietz, A. & Schäfer, S. (1983) Classical Conditioning of 358 Proboscis Extension in Honeybees (Apis mellifera). Journal of Comparative 359 Psychology, 97, 107. https://doi.org/10.1037/0735-7036.97.2.107 360 de Groot, A. P. (1952). Amino acid requirements for growth of the honeybee (Apis 361 mellifica L.). Experientia, 8(5), 192–194. https://doi.org/10.1007/BF02173740 362 Crailsheim, K. (1990a). The protein balance of the honey bee worker. Apidologie, 21(5), 363 417–429. https://doi.org/10.1051/apido:19900504 364 Crailsheim, K. (1990b). Protein synthesis in the honeybee (Apis mellifera L.) and 365 trophallactic distribution of jelly among imagos in laboratory experiments. Zoologische 366 Jahrbucher, Allgemeine Zoologie Und Physiologie Der Tiere, 94, 303–312. 367 https://www.cabidigitallibrary.org/doi/full/10.5555/19910230837 368 Crailsheim, K., Schneider, L. H. W., Hrassnigg, N., Bühlmann, G., Brosch, U., 369 Gmeinbauer, R., & Schöffmann, B. (1992). Pollen consumption and utilization in 370 worker honeybees (Apis mellifera carnica): Dependence on individual age and 371 function. Journal of Insect Physiology , 38(6), 409–419. https://doi.org/10.1016/0022-372 1910(92)90117-V 373 Crailsheim, K., & Stolberg, E. (1989). Influence of diet, age and colony condition upon 374 intestinal proteolytic activity and size of the hypopharyngeal glands in the honeybee 375 (Apis mellifera L.). Journal of Insect Physiology , 35(8), 595–602. 376 https://doi.org/10.1016/0022-1910(89)90121-2 377 de Brito Sanchez, M. G. (2011). Taste perception in honey bees. Chemical Senses, 36(8), 378 675–692. https://doi.org/10.1093/chemse/bjr040 379 Değ irmenci, L., Thamm, M., & Scheiner, R. (2018). Responses to sugar and sugar receptor 380 gene expression in different social roles of the honeybee (Apis mellifera). Journal of 381 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint Insect Physiology , 106(May 2017), 65–70. 382 https://doi.org/10.1016/j.jinsphys.2017.09.009 383 Fahrbach, S. E., Giray, T., & Robinson, G. E. (1995). Volume changes in the mushroom 384 bodies of adult honey bee queens. Neurobiology of Learning and Memory, 63(2), 181–385 191. https://doi.org/10.1006/nlme.1995.1019 386 Farina, W. M., & Núñez, J. A. (1991). Trophallaxis in the honeybee, Apis mellifera (L.) as 387 related to the profitability of food sources. Animal Behaviour , 42(3), 389–394. 388 https://doi.org/10.1016/S0003-3472(05)80037-5 389 Fèvre, D. P., & Dearden, P. K. (2024). Influence of nutrition on honeybee queen egg-laying. 390 Apidologie, 55(4), 1–16. https://doi.org/10.1007/s13592-024-01097-1 391 Free, J. B. (1957). The transmission of food between worker honeybees. The British Journal 392 of Animal Behaviour, 5(2), 41–47. https://doi.org/10.1016/S0950-5601(57)80023-9 393 Giray, T., Galindo-Cardona, A., & Oskay, D. (2007). Octopamine influences honey bee 394 foraging preference. Journal of Insect Physiology , 53(7), 691–698. 395 https://doi.org/10.1016/j.jinsphys.2007.03.016 396 Guez, D., Suchail, S., Gauthier, M., Maleszka, R., & Belzunces, L. P. (2001). Contrasting 397 effects of Imidacloprid on habituation in 7- and 8-day-old honeybees (Apis mellifera). 398 Neurobiology of Learning and Memory , 76(2), 183–191. 399 https://doi.org/10.1006/nlme.2000.3995 400 Gong, Z., Tan, K., & Nieh, J. C. (2018). First demonstration of olfactory learning and long-401 term memory in honey bee queens. Journal of Experimental Biology , 221(14). 402 https://doi.org/10.1242/jeb.177303 403 Haydak, M. H. (1970). Honey bee nutrition. Annual Review of Entomology, 15(1), 143–156. 404 https://doi.org/10.1146/annurev.en.15.010170.001043 405 Howe, S. R., Dimick, P. S., & Benton, A. W. (1985). Composition of freshly harvested and 406 commercial royal jelly. Journal of Apicultural Research , 24(1), 52–61. 407 https://doi.org/10.1080/00218839.1985.11100649 408 Kolayli, S., Sahin, H., Can, Z., Yildiz, O., Malkoc, M., & Asadov, A. (2016). A member of 409 complementary medicinal food: Anatolian royal jellies, their chemical compositions, 410 and antioxidant properties . Journal of Evidence-Based Complementary & Alternative 411 Medicine, 21(4), NP43–NP48. https://doi.org/10.1177/2156587215618832 412 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint Lindauer, M. (1952). Ein beitrag zur frage der arbeitsteilung im bienenstaat. Zeitschrift Für 413 Vergleichende Physiologie, 34(4), 299–345. https://doi.org/10.1007/BF00298048 414 Moreno, E., & Arenas, A. (2023). Changes in resource perception throughout the foraging 415 visit contribute to task specialization in the honey bee Apis mellifera. Scientific 416 Reports, 13(1), 1–10. https://doi.org/10.1038/s41598-023-35163-y 417 Moreno, E., & Arenas, A. (2024). Foraging task specialization in honey bees (Apis 418 mellifera): the contribution of floral rewards to the learning performance of pollen and 419 nectar foragers. Journal of Experimental Biology , 227(13). 420 https://doi.org/10.1242/jeb.246979 421 Nicholls, E., & Hempel de Ibarra, N. (2013). Pollen elicits proboscis extension but does not 422 reinforce PER learning in honeybees. Insects, 4(4), 542–557. 423 https://doi.org/10.3390/insects4040542 424 Nicholls, E., Krishna, S., Wright, O., Stabler, D., Krefft, A., Somanathan, H., & Hempel de 425 Ibarra, N. (2019). A matter of taste: the adverse effect of pollen compounds on the pre-426 ingestive gustatory experience of sugar solutions for honeybees. Journal of 427 Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral 428 Physiology, 205(3), 333–346. https://doi.org/10.1007/s00359-019-01347-z 429 Pain, J., & Maugenet, J. (1966). Physiologiques sur le pollen par les abeilles emmagasin par 430 les abeiles. Les Annales de l’Abeille , 9(3), 209–236. https://hal.science/hal-431 00890236v1 432 Pankiw, T., Nelson, M., Page, R. E., & Fondrk, M. K. (2004). The communal crop: 433 Modulation of sucrose response thresholds of pre-foraging honey bees with incoming 434 nectar quality. Behavioral Ecology and Sociobiology , 55(3), 286–292. 435 https://doi.org/10.1007/s00265-003-0714-0 436 Pankiw, T., & Page, R. E. (1999). The effect of genotype, age, sex, and caste on response 437 thresholds to sucrose and foraging behavior of honey bees (Apis mellifera L.). Journal 438 of Comparative Physiology - A Sensory, Neural, and Behavioral Physiology , 185(2), 439 207–213. https://doi.org/10.1007/s003590050379 440 Pankiw, T., & Page Jr, R. E. (2001). Brood pheromone modulates honeybee (Apis mellifera 441 L.) sucrose response thresholds . Behavioral Ecology and Sociobiology, 49 (2–3), 206–442 213. https://doi.org/10.1007/s002650000282 443 Ramanathan, A. N. K. G., Nair, A. J., & Sugunan, V. S. (2018). A review on royal jelly 444 proteins and peptides. Journal of Functional Foods , 44(December 2017), 255–264. 445 https://doi.org/10.1016/j.jff.2018.03.008 446 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint Rankin, C. H., Abrams, T., Barry, R. J., Bhatnagar, S., Clayton, D. F., Colombo, J., 447 Coppola, G., Geyer, M. A., Glanzman, D. L., Marsland, S., McSweeney, F. K., 448 Wilson, D. A., Wu, C. F., & Thompson, R. F. (2009). Habituation revisited: An 449 updated and revised description of the behavioral characteristics of habituation. 450 Neurobiology of Learning and Memory, 92 (2), 135–138. 451 https://doi.org/10.1016/j.nlm.2008.09.012 452 R Core Team (2020). R: a language and environment for statistical computing. 453 https://www.R-project.org/. 454 Scheiner, R., Barnert, M., & Erber, J. (2003). Variation in water and sucrose responsiveness 455 during the foraging season affects proboscis extension learning in honey bees. 456 Apidologie, 34(1), 67–72. https://doi.org/10.1051/apido:2002050 457 Scheiner, R., Page, R. E., & Erber, J. (2001). The effects of genotype, foraging role, and 458 sucrose responsiveness on the tactile learning performance of honey bees (Apis 459 mellifera L.). Neurobiology of Learning and Memory , 76(2), 138–150. 460 https://doi.org/10.1006/nlme.2000.3996 461 Scheiner, R., Page, R. E., & Erber, J. (2004). Sucrose responsiveness and behavioral 462 plasticity in honey bees (Apis mellifera). Apidologie, 35(2), 133–142. 463 https://doi.org/10.1051/apido:2004001 464 Seeley, T. D. (1982). Adaptive significance of the age polyethism schedule in honeybee 465 colonies. Behavioral Ecology and Sociobiology , 11(4), 287–293. 466 https://doi.org/10.1007/BF00299306 467 Simcock, N. K., Gray, H. E., & Wright, G. A. (2014). Single amino acids in sucrose 468 rewards modulate feeding and associative learning in the honeybee. Journal of Insect 469 Physiology, 69(C), 41–48. https://doi.org/10.1016/j.jinsphys.2014.05.004 470 St. Clair, A. L., Dwyer, B., Shapiro, M., & Dolezal, A. G. (2024). Adult honey bee queens 471 consume pollen and nectar. BioRxiv. https://doi.org/10.1101/2024.12.04.626851 472 473 Figures 474 Figure 1. Proboscis extension response (PER) to four stimuli across honey bee categories. Panels 475 A–C show trial-by-trial mean PER (number of positive PER/number of bees) for each stimulus 476 in foragers (A), 1-day-old workers (B), and queens (C). Panels D–F show boxplots of the ratio of 477 PER, which is computed as the total of positive PER numbers divided by 15 trials for each 478 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint stimulus, as for sucrose, water, royal jelly, and a sucrose–royal jelly mixture in foragers (D), 1-479 day-old workers (E), and queens (F). Boxes are centered on the median with interquartile range, 480 whiskers extend to 1.5×IQR. Letters above boxes are compact-letter groupings derived from the 481 GLMM (estimated marginal means on the response scale) and indicate Tukey-adjusted pairwise 482 differences among stimuli within each bee category, and shared letters indicate not significantly 483 different. 484 485 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted January 26, 2026. ; https://doi.org/10.64898/2026.01.23.701277doi: bioRxiv preprint