The uploaded dataset contains all the data to run the analyses and figures of the publication "Contrasting impacts of climbing plants on host tree reproduction in a drought-stressed forest". Climbing plants, or climbers, are known to negatively affect the survival and reproduction of tropical and temperate humid forest trees through competition and structural parasitism. These impacts are attributed to their growth strategy, which relies on other plants for mechanical support and allows them to divert resources away from structural investment toward vegetative and reproductive functions. Such negative interactions may ultimately influence the composition and dynamics of plant and animal communities around them. Effects of climbers on hosts may differ in drought-stressed systems because investments in foliage and reproductive structures could favor facilitative interactions, such as abiotic stress amelioration through shading or pollinator attraction through synchronous flowering. Knowledge of climber-host interactions in Mediterranean and drought-stressed forests in general is limited. To test the hypothesis that climbers can facilitate host reproduction via synchronous flowering in a drought-stressed Mediterranean forest, we assessed the effects of climber cover on the probability of flowering, fruiting, and final fruit set of Crataegus aronia, a common insect-pollinated tree. Overall, climber cover was negatively related to flowering probability but positively related to fruiting probability and final fruit set. These effects differed between climbers with distinct flowering phenology relative to the host. Non-co-flowering climber effects on flowering probability were negative. However, co-flowering climber effects on fruiting probability and final fruit set were positive. These positive effects on C. aronia are in stark contrast with pervasive negative effects of lianas in tropical moist forests. The differing effects of climbers with distinct phenology suggest involvement of pollinator attraction. Although limited to the studied tree species in a Mediterranean forest, the documented positive effects of climbers in this drought-stressed forest and their relation with synchronous flowering highlight the need for trait-based analyses of climber –host interactions across systems, particularly to expand our understanding of their ecological roles under varied climate conditions.

We conducted a foot survey of an area of approximately 88,000 square meters in the south-eastern part of the reserve to map potential Crataegus aronia individuals. We visited the research site weekly, starting a couple of weeks before the known flowering time of C. aronia, and recorded the geographical locations of potential individuals using a smartphone GPS. We found a total of 162 individuals throughout the survey and selected 123 individuals for sampling based on the following criteria: they were taller than 1.5 m (shorter individuals did not flower during the study regardless of the presence or absence of climbers, E. Fein pers. obs., and were considered immature); they were free from infestation or overtopped by non-climbing plants; and they had at least one branch 50 cm or longer for quantifying flowering and fruiting. Measurements on selected trees were conducted from the onset of flowering. Individuals identified in the second census following the onset of flowering were included in the sample only if their peak flowering occurred after their initial census. All mapped individuals were assessed for the presence or absence of climbers. When climbers were present, each climber species was identified, and its respective cover on the host tree crown was visually estimated. Cover estimates were recorded separately for each climber species as the percentage of the total tree crown volume occupied, enabling analyses incorporating differences in species flowering phenology (described in the Data analysis section below). Host tree size was measured by the stem circumference at the base, as several individuals branched below the standard diameter at breast height.

During 10 weekly censuses conducted between March and May 2023, the number of flowers and fruits found** on a single branch per tree was recorded. For each tree, the branch with the highest number of flowers in the first census was selected for all subsequent counts. These weekly censuses allowed us to determine, with a week-level resolution, the timing of peak flowering and final ripe fruit production, which were used to estimate fruit set as detailed in the Data analysis section. Censused branches were marked with a plastic or metal tags secured with metal wire. Counts were performed on the distal 50 cm of each branch, including any daughter branches. Flower counts included all flower buds and open flowers with petals. Fruit counts included all swollen ovaries and fully ripe fruits. Additionally, the orientation of the branch relative to the cardinal points and the branch circumference 50 cm from its distal end were recorded.

We assessed the effect of climber load on C. aronia reproduction by modeling the probability of flowering, the probability of fruiting, and the final fruit set as a function of the hosts’ climber load. Climber load was calculated as the sum of the individual cover percentages estimated separately for each climber species present on the host crown, which occasionally resulted in climber load, or total climber cover, exceeding 100% due to overlapping layers of climbers. Crataegus aronia individuals were classified as flowering if they had at least one flower recorded in any census. Individuals were classified as fruiting if they had ripe fruits at the final census. Because we did not track the fate of individual flowers, final fruit set was estimated as the ratio of the final count of ripe fruits to the maximum number of flowers recorded on sampled branches across all censuses (Giménez‐Benavides et al., 2007; Rominger et al., 2021). All censused trees were included in the analysis of effects on flowering. For fruiting effects, only trees that produced flowers (n=68) were included. This excludes trees unable to produce fruits and provide information on the effects of climbers on fruiting success.

The individual probability of flowering and fruiting were modeled as binary responses (success or failure to produce flowers or fruits) using a Bernoulli generalized linear model (GLM) with a binomial family distribution. The final fruit set was modeled with a binomial GLM, with the response variable representing the tree’s fruiting success as the proportion of flowers developing into ripe fruits. Two sets of models were used to analyze the effect of climber load on each response variable. One set examined the effects of total climber cover, representing the sum of the cover of all climber species, while the other set differentiated between climber cover with distinct flowering phenologies, i.e., co-flowering and non-co-flowering climber cover. We defined climbers as co-flowering when they have some temporal overlap in flowering with the host tree, and as non-co-flowering when there is no such overlap. From the total of 123 trees sampled, 70 were hosting vines; 26 of these individuals hosted the co-flowering vine species Rubia tenuifolia, Galium aparine, Lonicera etrusca, Prasium majus and Ephedra foeminea and 66 individuals hosted the non-co-flowering Asparagus aphyllus. From the 68 flowering trees, 30 were hosting vines; 11 trees hosted the co-flowering vine species Rubia tenuifolia , Galium aparine, Prasium majus and Ephedra foeminea and 30 trees hosted the non co-flowering Asparagus aphyllus. Single trees hosted different combinations of climber species.

Each set of models included all possible combinations of the climber cover variable(s) of interest and covariates measured to account for their potential effects on host tree reproductive responses. For the flowering model, control covariates included the host tree’s stem circumference. For the fruiting model, covariates included the host tree’s stem circumference, branch circumference, branch orientation, flowering time, and maximum flower count. The fruit set model used the same covariates as the fruiting model, excluding the maximum flower count.