To test this hypothesis, we examined female reproduction in a total of five rural and five urban colonies over three consecutive years (Fig. 1A—spring of 2022, 2023, and 2024 and winter of 2022–2023, see Table 1 and Additional file 1: Table S1). In total, we sampled 561 mothers and 123 pups. We estimated the human population density at typical bat foraging ranges, i.e., a 5-km radius and a 15-km radius around each bat colony (Fig. 1B and Table 1 and see the “Methods” section). According to the estimated densities, we used a threshold of 3000 inhabitants per km2 to distinguish between rural and urban bat colonies. We also tested all models using a categorical division of the colonies into urban and rural. We then used generalized linear mixed effect models (GLMMs) to examine the effect of urbanization on the reproductive cycle. Finally, we employed a categorical urban/rural model that best fit the data (based on the AIC) but also provided results for the human population density at a 15-km radius, which fits better than the 5-km density and was not significantly worse than the categorical model. The 15-km density indicated an almost continuous variability in human population density for our sampled colonies (see Y-axis values for the light-blue points in Fig. 1B). Unless stated otherwise, we also used the sampling year and date (days from December 1st of the relevant year) as fixed factors in all models and the colony ID as a random effect (see the “Methods” section). To identify the best time slots for sampling, we first sampled two colonies (one urban and one rural) approximately once a month over a full year (Additional file 1: Table S2). This revealed a clear bi-annual reproductive cycle (Fig. 1C), which had already been documented for this species in the region [36]. Accordingly, in the following years, we conducted one sample in December to determine pregnancy probability in the colonies and another sample in March–April to assess parturition dates. We focused on the spring reproduction peak as it is the first after winter, which is the most difficult time for these bats [36]. All the results below are reported for adult females or pups of both sexes.

Fruit bat colonies in relation to human population densities. A A map depicting the sampled bat colonies’ locations (urban in brown and rural in green). Circles represent the 5-km radius (dark blue) and 15-km radius (light blue) around the bat colony. Red and brown areas represent urban cities and towns, respectively. Here, we define an urban colony as one with a human population of more than 3000 inhabitants per km2. B Human population density (per km2) for a 5-km and 15-km radius around the colonies. The dotted line shows the cutoff between urban and rural colonies. C Seasonal fluctuations in reproduction. During 2021/2022, one urban colony (Herzliya) and one rural colony (Tinshemet) were sampled approximately once a month for 1 year. Dark green (Tinshemet) and brown (Herzliya) lines and dots represent the percentage of pregnant females out of the total adult females caught in each sampling event. Light green (Tinshemet) and brown (Herzliya) lines represent the percentage of females with pups out of the adult females that were caught in each sampling event. Gray areas represent the dates chosen for sampling in the following years, based on peaks in pregnancies and pup occurrence

In winter, there was no significant difference in pregnancy proportion between rural and urban colonies (75% ± 19% [mean ± SD] of females in urban colonies were pregnant vs 84.5% ± 19.5% in rural colonies). The Null model with only the intercept as an explaining parameter had the best fit (binomial distribution GLMM null model, intercept p < 0.001, n = 197 adult female bats, see Table 2 in which we also provide the best non-null model). There was also no significant difference in the size (forearm; GLMM, p = 0.9, n = 197 bats) and fitness (represented by the body mass index (BMI), see the “Methods” section; GLMM, p = 0.77, n = 197 bats) of adult females in urban and rural colonies. Although there was no significant difference in the parasite load between urban and rural females, we did find a significant correlation between parasite load and reproductive state, with pregnant bats presenting a slightly higher parasite load than non-pregnant females (GLMM, p = 0.01 for reproductive state, n = 197 bats, Table 2). Winter BMI (a proxy for individual state) did not correlate with parasite infestation (GLMM, p = 0.3 for parasites, n = 197 bats, Table 2).

Pups in urban colonies were born earlier in spring than in rural colonies, as was evident from their significantly longer forearms and heavier weight (Fig. 2B, C, urban average forearm was 10 mm longer and average weight was 7 gr heavier; GLMM, forearm: p < 0.001, n = 123 pups; weight: p = 0.03, n = 123 pups, Table 2). Translating forearm to age (see the “Methods” section) suggested that urban pups were born on average 2.5 weeks earlier than rural pups (GLMM, pup age: p < 0.001, n = 123 pups, Table 2).

Pups are born earlier in urban colonies. For all parameters, data are presented on the left panel by colony type (A1–D1; lines represent the median and lower and upper quartiles. Circles represent individual data points for each colony type and a gray asterisk represents significance), and on the right panel by human population density for a 15-km radius around the colony (A2–D2; mean ± SE). Asterisks represent individual data points for each colony. A Pup estimated age. B Pup forearm length. C Pup weight. D Pup BMI

We rejected the possibility that urban pups are born at the same time as those in the wild but simply grow faster than rural pups, as there was no significant interaction between urbanization and time from the beginning of the season in explaining pup size. Pup size therefore did not increase differentially in the two environments (GLMM with pup forearm length as the response and colony type and time as fixed effects with an interaction between time and colony, p = 0.7 for the interaction, n = 208, Additional file 1: Table S3). Note that the spring sampling occurred over a span of ~ 4 weeks, enabling us to perform this analysis and assess the urban and rural growth rates. We additionally determined whether the size of the mother (forearm length) had any effect on the size of the pups, by adding the forearm length of the mother as a fixed factor in the pup size models (weight and forearm), but found no significant effect (GLMM, p > 0.5, n = 123 pups).

We also examined several alternative models (using the AIC to perform model selection). To account for the distribution of colonies—all urban colonies were near the coast and at lower elevations—we examined models that included ambient temperature (near the colony) as a fixed factor instead of or in addition to human density. Based on their AIC score, these models did not fit the data nor the urbanization level models (which were based on the human population density Additional file 1: Table S3). Adding bat population size to the model also did not improve model fit.

We only measured the temperature inside the colony in five colonies (four urban and one rural), so we could not compare these models to the models above. Interestingly, however, internal colony temperature in the urban colonies significantly correlated with pup age but not with the other pup measurements (p = 0.03, Additional file 1: Table S3).

Pups in rural colonies had significantly higher BMIs (Fig. 2D and Table 2; GLMM, p < 0.001, n = 123 pups). However, a model of the change in pup BMI over time (Methods) suggested that, for the relevant ages of the sampled pups, the BMI decreases by 0.0001 g/mm2 per day, making this difference likely a result of the urban pups being older rather than being due to some resource deficiency in cities.

There was no correlation between parasite load and pup BMI (GLMM, p = 0.54, n = 95 pups). Spring adult parasite load was significantly positively correlated with adult BMI (GLMM, BMI: p < 0.001, n = 244; weight: p < 0.001, n = 244, Table 2) and also with reproductive status, with lactating bats suffering from higher parasite loads than both pregnant and non-pregnant bats and no difference between the two latter groups (GLMM, p < 0.001 for lactating, n = 244 bats, lactating vs. non-pregnant, p < 0.001 and lactating vs. pregnant, p = 0.003). There was also no difference in the parasite load of urban and rural bats in spring. The continuous model showed a significant difference in parasite load but the effect size was negligible and the categorical model with the same AIC did not show a significant difference (continuous—by density, p = 0.03; categorical—by type, p = 0.12, n = 244 bats).

There was no significant effect of urbanization on the reproductive yield, which was defined as the proportion of adult females in the colony that were actively reproductive (i.e., either pregnant/lactating/with pup; GLMM, null model, intercept p < 0.001, n = 364 adult female bats, Table 2). This measurement sought to validate the hypothesis that urban bats giving birth earlier does not result in lower overall reproduction rates due to more miscarriages or higher pup mortality.