Assessing the biodiversity and the impact of pollinators on carob production

ones in the SE 1 to the landrace Apostolika, but we note that in many cases it is not possible to separate the two landraces based on morphology alone [34]. None of the landraces used in the pollination study were hermaphroditic. To assess the contribution of insects to pollination, groups of inflorescences on trees in each carob grove were assigned to two different treatments: a) open pollination, and b) wind pollination. In the open pollination, carob tree inflorescences were not subjected to any treat- ment, while in the wind pollination inflorescences were enclosed in an insect-proof mesh bag (1 mm size mesh opening) before the start of blooming. Although the insect-proof mesh bags are reducing wind movement and potentially pollen delivery to the flowers, there is no other method for preventing insect access to the flowers. For each of the 20 trees per carob grove, two branches were randomly selected with one group of inflorescences per branch assigned to open pollination and another group assigned to wind pollination (20 trees x 2 branches per tree x 1 group of inflorescences per treatment per branch). Carob inflorescences are found in groups on branches. The number of inflorescences per group varied from five to 30 (median = 15) in the SE Lowlands, four to 16 in the S. Plateau (median = 9), and six to 20 in NW Coast (median = 10). The mesh bags were removed at the end of November, after the end of blooming. All mature carob pods produced were harvested on Julian day 182 of 2018 (July 1st, 2018). Carob

pods were transferred to the lab, where we measured the weight, length, and number of seeds produced per pod. To examine the short-term transfer of pollen by the wind from male to female trees we placed pollen traps on one male (the only male tree present in the carob grove) and five female trees in the SE Lowlands carob grove. Female trees were located at 12, 60, 63 109 and 172 m from the male tree. Pollen traps were constructed using 10 cm diameter and 4.5 cm depth plas- tic bowls, the base of which was covered with 3 cm height gypsum plaster to limit shaking by the wind. A 9 cm Petri dish, with a filter paper covering its base and filled with 30 ml of 99.5% glycerin was placed within the bowl. The pollen traps were hung on the tree branches at the same height, ca. at 2 m from the ground, with a total of four traps for each of the male and five female trees (24 traps in total). The traps remained in the trees for 24 h (18/11/2017-19/11/ 2017). After collection of the traps, the glycerin was transferred into 50 ml falcon tubes and stored in the lab at -20˚C. To retrieve pollen grains present in the glycerin, falcons were centri- fuged at 3500 g at 4˚C for 20 min. If present, a small part of the pollen pellet was removed, dis- solved in 50 μl water, stained with fuchsin and observed under a microscope to confirm the presence or absence of pollen grains. The presence of carob pollen grains was confirmed based on morphology using pictures available in the Pollen Atlas of the Honey Flora of Cyprus [35]. Statistical analyses

All statistical analyses were carried out in the open-source R language and environment for statistical computing (R Core Team, 2021). Data were curated and plotted using the tidyverse package [36]. The bee presence-absence data were analyzed in a generalized linear mixed-effects model framework using the package lme4 (function glmer) [37], with a log link and a binomial distri- bution. The response variable was wild bee presence (present or absent). Two analyses were carried out. First an analysis of the 2016 data for all three regions with fixed factors region (NW Coast, S. Plateau, S. Coast), week (5 sampling weeks), tree sex (Male or Female), tree side (N, S, E, W), A. mellifera presence, and wasp presence (present or absent). Τhe second analysis included the NW Coast–S. Lowlands groves which were sampled both in 2016 and 2017. The fixed factors for the model were region (N. Coast, S. Lowlands), year (2016 or 2017), week (5 PLOS ONE Biodiversity and the impact of pollinators on carobs PLOS ONE | https://doi.org/10.1371/journal.pone.0291431 October 20, 2023 4 / 16