Neolloydia conoidea. Flowering waves and shared bees

The population and the seasonal rhythm

In the limestone landscape of Tolimán in the Mexican state of Querétaro, Neolloydia conoidea grows with Ariocarpus kotschoubeyanus. The stems of the two species differ greatly, but their open flowers have a similar funnel shape and magenta colour. Both the course of flowering and insects visiting flowers were studied at this site. The results come from specific populations near San Miguel Tolimán, at an elevation of approximately 1400 m. The local climate is semiarid, with rain mainly from May to October. The plants occupy xerophytic scrub on limestone with outcrops of clayey-silty rock. [Arroyo-Pérez et al. 2021, pp. 29–30, Study area, Study species]

Two publications by Arroyo-Pérez and co-authors present different scales of observation. The 2019 paper includes weekly flower censuses of N. conoidea in the 2015 and 2016 seasons. The study published in 2021 compares both species using monthly censuses from March 2015 to April 2016, flower examinations, insect records and controlled pollination. Sample sizes differ between these papers. Their results complement one another. The first reveals flowering waves within the season, and the second shows the distribution of floral resources between species over the year. [Arroyo-Pérez et al. 2019, Arroyo-Pérez 2019, pp. 581–582; Arroyo-Pérez 2021, pp. 30–32][Arroyo-Pérez et al. 2021, Arroyo-Pérez 2019, pp. 581–582; Arroyo-Pérez 2021, pp. 30–32]

Five or six flowering waves

In February 2015, 200 N. conoidea plants were marked in a one-hectare area. Individuals taller than 4 cm produced reproductive structures. The analysis included 146 of these in the first season and 142 in the second. Censuses ran from March to November, and open flowers were recorded from March to September. The seven months included several distinct increases in flower numbers, separated by periods with no recorded flowering. Kernel density estimation and Gaussian distribution components were used to distinguish waves. The boundaries of a wave were defined as the mean of the component plus and minus one standard deviation. [Arroyo-Pérez et al. 2019, pp. 581–583, Fenología de la floración, Fig. 1–2]

Six waves were identified in 2015 and five in 2016. An individual wave lasted from 11 to 23 days. Intervals in which no flowers were found during censuses lasted 4–26 days. The annual flower total was similar, although flower distribution and fruit-setting success changed. The highest number of flowers in both seasons was recorded in the third wave. The statistical definition concerns flowering across the whole population, so the duration of such a wave includes flowers of many plants opening on different days. [Arroyo-Pérez et al. 2019, pp. 582–584, Fig. 2, Tables 1–2]

Most plants returned to flowering after a break. More than one wave included 81% of reproductive individuals in 2015 and 73% in 2016. The population rhythm therefore arose both from asynchronous flowering of individual plants and from repeated flowering by the same individuals. This organisation spreads flower availability over a substantial part of the warm season. For insects, it provides a series of periods offering pollen, and for plants, several separate opportunities to set seeds. [Arroyo-Pérez et al. 2019, pp. 583–585, Fig. 3, Discussion]

Synchrony and reproductive success

Synchrony was assessed in two ways. The Augspurger index describes the overlap of periods in which individuals flower, accounting for flower presence. The Marquis index additionally relates an individual's contribution to population flowering to its flower numbers at successive times. The distinction matters when waves differ greatly in intensity. A plant may flower at the same time as many neighbours but develop most of its own flowers at another time. The two indices therefore describe different aspects of the same phenological pattern. [Arroyo-Pérez et al. 2019, pp. 581–582, definitions of SA and SM]

The mean Augspurger index was 0,19 in 2015 and 0,15 in 2016. The Marquis index was 0,41 and 0,33, respectively. The authors found a difference between years for the first measure. At the same time, the wave analysis did not show a consistent relationship between synchrony and the proportion of flowers forming fruits or ovules forming seeds. In 2016, when complete fruits were available, mean seed set was 0,69. In this reproductive system, events within the flower and subsequent fruit development also matter, and the flowering calendar alone does not describe them. [Arroyo-Pérez et al. 2019, pp. 582–585, Tables 1–3]

Two cacti using different months

For monthly observation, 212 individuals of N. conoidea and 231 of A. kotschoubeyanus were marked. Neolloydia flowered in spring and summer, with the greatest proportion of flowers in May. That month accounted for 42% of the annual flower total. Fruits were observed from June to October. Ariocarpus flowered from October to December, and its fruits appeared in February and March. In the observed cycle, flowering of the two species was separated in time. The overlap of some flower-visiting insects can therefore be analysed together with the seasonal succession of resources. [Arroyo-Pérez et al. 2021, pp. 30, 32–33, Reproductive phenology, Fig. 3]

The number of N. conoidea buds was positively correlated with temperature, relative humidity and precipitation in the same month. No comparable relationships were found for Ariocarpus in this series. These results come from observations of seasonal conditions that varied together. They describe the relationship between phenology and weather in the year studied, while the contribution of individual cues to flower initiation requires an experiment. The separation of flowering months also varies. The authors note an earlier observation from 2014, when Neolloydia was still flowering in October, at the beginning of the Ariocarpus season. [Arroyo-Pérez et al. 2021, pp. 32, 35, Floral phenology, Discussion]

A two-day flower and the sequence of activity in its organs

Flowers of both cacti were active for two days, closed at night and, on the days observed, opened at approximately 9–10 hours and closed at approximately 14–15 hours. Anthers began to dehisce in the first hours of the first day. The stigma lobes became receptive later, constituting partial protandry. The stigma remained above the highest anthers throughout both days of flower opening. This spatial separation, called herkogamy, co-occurred with a partial temporal shift in male and female activity. [Arroyo-Pérez et al. 2021, pp. 31–32, Floral anthesis and morphology]

At the end of the second day, the stigma lobes curved outwards and could touch anthers in which little pollen remained. Such contact shows that flower geometry changes during its short life. Its biological significance was tested separately through pollination. In Neolloydia, the experiment demonstrated the need for pollen from another individual, despite the late convergence of its own organs. In Ariocarpus, a small amount of fruit set was obtained after self-pollination, alongside much greater effectiveness of cross-pollination. [Arroyo-Pérez et al. 2021, pp. 32, 34–35, Table 3]

What 75% shared pollinators means

Insects were observed on 20 flowers of each species for two consecutive days. Recording was conducted on Neolloydia on 23–24 May 2015, and on Ariocarpus on 3–4 November. Bees touched both the anthers and stigma. On this basis, the authors regarded them as pollinators. Three taxa occurred on both cacti, Apis mellifera, Augochlora sp. 1 and Lasioglossum sp. 1. A fourth, Agapostemon sp. 1, was recorded only on Ariocarpus. Three of the four bee taxa were therefore shared, representing 75% of the composition of this group. This percentage applies to the list of observed taxa. [Arroyo-Pérez et al. 2021, pp. 31, 33–34, Floral visitors, Table 2]

Visit frequencies presented a different picture. Neolloydia received 694 visits by all insects, including 237 by honeybees, 102 by Augochlora and 53 by Lasioglossum. Augochlora and Agapostemon predominated on Ariocarpus. The proportional similarity index of the bee assemblages was 0,48, reflecting these differences in their relative activity. Wasps, ants and beetles also visited flowers, but contacted the organs differently. Wasps did not touch the anthers, and ants and beetles remained mainly near the bases of the filaments and inside the floral tube. [Arroyo-Pérez et al. 2021, pp. 31–34, Table 2]

Observing a bee contact the stigma and anthers indicates the possibility of pollen transfer. The amount and origin of pollen deposited by individual taxa would require separate measurement. In this study, confirmation of the importance of pollen from other plants comes from pollination experiments, while the insect comparison describes visitor composition and behaviour. Combining these parts of the study allows the mechanism of pollination in the population and seasonal use of a similar group of bees to be described, while retaining the distinction between a visit and its effectiveness. [Arroyo-Pérez et al. 2021, pp. 31–32, 34–36, observation and controlled pollination methods]

Pollen from other plants and the number of seeds formed

Four treatments were used for each species, with 20 flowers on different plants per treatment. Control flowers remained accessible to insects. Cross-pollinated flowers were isolated and supplied with pollen from five other individuals of the same species, more than 20 m away. The next two treatments were hand self-pollination and isolation without manipulation, allowing autonomous self-pollination to be tested. After flowering ended, fruits were protected from consumption, and mature fruits were collected after three months. [Arroyo-Pérez et al. 2021, p. 32, Mating system]

In N. conoidea, fruits formed only in the control and after cross-pollination. The difference between the proportions of fruiting flowers in these two treatments was not statistically significant, while additional pollen from other plants increased seed set. In the population studied, the number of seeds forming was therefore limited by pollen delivery, although a large proportion of flowers under natural conditions already formed fruits. Fruit set and full realisation of a fruit's seed potential are separate stages of reproduction. [Arroyo-Pérez et al. 2021, pp. 34–36, Table 3, Discussion]

In A. kotschoubeyanus, fruit set was 85% in the control and 100% after hand cross-pollination, but also 15% after hand self-pollination and 5% in isolated flowers without manipulation. The authors described its mating system as mixed, with a predominance of outcrossing. In Neolloydia, the results indicated obligatory use of pollen from other plants under the conditions employed. Similar colour, shape and flower-opening behaviour can therefore coexist with different self-pollination effectiveness. [Arroyo-Pérez et al. 2021, pp. 34–35, Mating system, Table 3]

Spreading reproduction over time

Recurring Neolloydia flowering waves and later Ariocarpus flowering operate at two levels. Within a species, successive waves create several opportunities for reproduction. Between species, the succession of seasons provides similar insects with flowers in different months. The authors consider reduced competition for pollinators and a lower probability of receiving pollen from another species as possible benefits of separating flowering times. These mechanisms are an interpretation of the documented phenology and similarity of flower-visiting insects. [Arroyo-Pérez et al. 2019, Arroyo-Pérez 2019, pp. 584–585; Arroyo-Pérez 2021, pp. 35–36][Arroyo-Pérez et al. 2021, Arroyo-Pérez 2019, pp. 584–585; Arroyo-Pérez 2021, pp. 35–36]

The flowering-wave paper also presents the hypothesis of reproductive risk spreading. If weather, insect activity or resource availability change between waves, repeated flowering may spread the consequences of a single adverse event. The field studies document repeated flowering by individuals and variable success in successive waves. In the observed population, flowers appeared in stages, and seed set depended on pollen from other individuals. [Arroyo-Pérez et al. 2019, Arroyo-Pérez 2019, pp. 584–585, Discussion; Arroyo-Pérez 2021, pp. 34–36][Arroyo-Pérez et al. 2021, Arroyo-Pérez 2019, pp. 584–585, Discussion; Arroyo-Pérez 2021, pp. 34–36]