1. We used exclusion experiments to determine the effect of different pollinator taxa on the yield and quality of pitayas (fruit of Stenocereus queretaroensis (F.A.C. Weber) Buxbaum), a major crop in central Mexico. We studied the three most economically important cultivars and wild individuals in the principal region for pitaya production. For each pollinator taxa we recorded fruit set and measured three key parameters of fruit quality: weight, sucrose concentration and seed set.
2. When bats were excluded from flowers and flowers were pollinated by other taxa (i.e. diurnal birds and insects), pitaya yield decreased by 35%, though pollination dependence varied between cultivars. Fruit quality decreased significantly in the absence of bat pollination across all cultivars, with fruits 46% lighter and 13% less sweet when pollinated by other taxa, thus reducing economic value as size determines market price. Additionally, seed set (an indicator of effective pollination) was significantly lower in the absence of bat pollinators.

We conducted this research in the municipality of Techaluta de Montenegro, Jalisco, Mexico (20.074°, -103.550°), one of the most important areas for pitaya production (Pimienta-Barrios and Nobel 1994), during 2016 and 2017.

We studied wild individuals of S. queretaroensis (cacti of 50+ years grown naturally) as well as three cultivars (Blanco, Mamey and Tenamaxtle) chosen for their economic importance, accounting for the majority of fruit production in the area. The study was carried out in six plantations, each containing all three cultivars; and six ranches with wild cacti.

We carried out exclusion experiments to determine the efficiency of different pollinators, using six pollination treatments to differentiate between both nocturnal and diurnal pollinators, as well as invertebrate and vertebrate pollinators. To exclude certain pollinators, bags of different mesh sizes were placed on flowers either during the day or at night. Bags made from a very fine mesh prevented all pollinators from visiting the flower, and bags made from 2 cm² mesh allowed only insects to pollinate flowers (i.e. vertebrate pollinators excluded).

We randomly selected five cacti of each cultivar in each plantation, and five wild cacti at each ranch. Six different treatments were carried out on each cactus, with each treatment on a separate flower: nocturnal pollinators only (NP: fine mesh bag during the day and unbagged at night), nocturnal insects only (NI: fine mesh bag during the day and large mesh bag at night), diurnal pollinators only (DP: unbagged during the day and fine mesh bag at night), diurnal insects only (DI: large mesh bag during the day and fine mesh bag at night), open pollinated control (OC: unbagged during the day and at night), and closed control (CC: fine mesh bag during the day and at night). Bags were changed at 06:00 and 18:00, with experiments lasting 24 hours. We placed all treatments on flowers opening on the same night where possible and on consecutive nights if not. We used randomised stratification to ensure a range of flower heights for each pollination treatment and recorded flower height.

To assess the impact of treatment on pitaya yield and quality we monitored experimental flowers to record mature fruit set (success or failure). We collected successful fruits to measure six different variables of interest: fruit length, fruit width, pulp weight, fruit weight, sucrose content and seed set. We used the ripening times of the first fruits to mature to establish standardised collection times of 52, 57, 54 and 52 days for Blanco, Mamey, Tenamaxtle and wild fruits respectively. We excluded fruits that were damaged by insects or by local people.

We weighed each fruit without spines, and measured the length and width. We peeled the fruits and weighed the fruit pulp. We chose fruit weight as the final indicator of fruit size, as it showed the strongest correlation with the other size parameters (Table S2). Sucrose content in one quarter of the fruit pulp (by wet weight) was measured using a handheld refractometer. We calculated seed set for each fruit by dividing the total seed number (estimated from counting the seeds in one quarter of the fruit by wet weight, and multiplying by four) by the average number of ovules counted in fifteen extra flowers from each cultivar type and wild individuals (collected from cacti not used in exclusion experiments, but from the same sites).

Treatment key for pollinator exclusion:

CC = closed control. NP = nocturnal pollinators only. NI = nocturnal insects only. DP = diurnal pollinators only. DI = diurnal insects only. OC = open control, pollination under natural conditions.

Cactus type = individual is a cultivar (we studied three cultivars - Blanco, Mamey and Tenamaxtle) or a wild plant.