Data collection and field observations were carried out in the Jenipabu Environmental Protection Area (APAJ), located in the state of Rio Grande do Norte, Brazil. The APAJ consists of 1,881.89 hectares which is composed of a Restinga ecosystem (see Correia et al. 2020). The climate is considered tropical rainy, with rainfall in winter and in the dry period of summer. The average temperature is 26.6º C and the average annual rainfall is 1456.6 mm, with an average temperature of 26.6º C and relative humidity of 70% (Nuc-Idema 2009). The management plan of APAJ (2009), divides area into seven geoenvironmental units: Fixed Dunes, Mobile Dunes, Deflation Plain, River Plain, Flood-Marine Plain, Coastal Deck and Beach Zone (Nuc-Idema 2009). The experiments were carried out in areas of fixed and mobile dunes and coastal board (Marinho et al. 2021), which are areas that suffer great influence from the environment, with strong winds and high temperatures and anthropogenic activities related to tourism and subsistence agriculture (Nuc-Idema 2009). The species Ipomoea asarifolia, whose genus corresponds to the most representative of the family Convolvulaceae, was studied (Muñoz-Rodríguez et al. 2019; Wood et al., 2020) and, in the study area, plays an important ecological role in the maintenance and fixation of dunes and other local ecosystems (Nuc-Idema, 2009). Ipomoea asarifolia is considered self-incompatible (Kill and Ranga 2003) and has hermaphrodite flowers that have reproductive structures with similar heights (Santos et al. in prep.).
To evaluate the reproductive system and the success in the reproduction of I. asarifolia in the study area, the following experiments were performed: I. Apomixis in buds in pre-anthesis that were emasculated and isolated until the end of anthesis (N = 30); II. Natural pollination (control) (N = 49) in freshly opened, marked flowers left free for pollination (Radford et al. 1974); III. Manual self-pollination crosses in previously bagged flowers (N = 32); IV. Manual cross-pollination in flowers previously bagged and pollinated with pollen from other individuals about 100 meters away (N = 55). In all experiments flowers were distributed in three populations, and in approx. 10 to 15 individuals, all about 30 m apart. Subsequently, they were followed for 30 days and the formation of fruits and seeds was recorded (Radford et al. 1974).
To analyze the influence of incompatible pollen on fruit and seed formation, manual crosses were performed in Ipomoea asarifolia using mixtures of pollen grains in approximate amounts of PIntraC (compatible intraspecific pollen) x PIntraI (incompatible intraspecific pollen) (N = 31 flowers) or PIntraC x PInterI (incompatible interspecific pollen) (N = 25 flowers). Manual crosses were also performed with mixtures of pollen grains in 2x higher amounts of PIntraI (N = 32 flowers) and PInterI (N = 36 flowers). The compatible intraspecific pollens came from flowers of other populations (distant ca. 100 meters from the receiving population) and the incompatible intraspecific pollens from the crossed flower itself (i.e., autopollen). For the PInterI crosses, pollen grains of Ipomoea brasiliana. This species occurs in sympathy with I. asarifolia, both have similar floral attributes, have overlapping flowering and share a high frequency of pollinator visits, which are considered main for both species and transfer pollen grains in the same region of the body (Santos et al. in prep.). To determine the appropriate amounts to perform the pollen mixtures, the grains were counted directly under a microscope (N = 10 buds/species, 5–10 individuals). Ipomoea asarifolia presents an average of 2739.1 ± 174.0 (average ± d.p.) grains, while in I. brasiliana the average is approximately twice as high (4476 ± 429,6).
Thus, the following experiments were carried out with the mixtures of pollens:
-
2 anthers of PIntraC + 1 anther of PInterI (approximate quantities of PIntraC x PInterI);
-
1 anther of PIntraC + 1 anther of PIntraI (approximate quantities of PIntraC x PIntraI);
-
1 anther of PIntraC + 1 anther of PInterI (2x greater quantity of PInterI in relation to PIntraC);
-
1 anther of PIntraC + 2 anthers of PIntraI (2x greater quantity of PIntrarI in relation to PIntraC).
The crossings were performed with the aid of eppendorf and brush. The experiments were carried out on flowers of approx. 10 to 15 individuals about 30 m apart and distributed in three populations. The flowers used were previously emasculated and isolated (with “voil” bags) and after crosses were marked and checked for fruit and seed formation after a period of 30 days.
The G2×6 test was used, with Cramer correction, followed by a pair-to-peer post-hoc analysis (adjusted by the Bonferroni method) to verify the association between fruit production and the presence of incompatible pollen grains (general chi-square result), verifying the differences in each level of combination of these two factors (comparison between pairs). Since the seed production was very low from the fruits formed in the experiments of crosses with pollen mixture in approximate quantities (PIntraC x PInterI and PIntraC x PIntraI), the sum of these seeds produced in these two experiments was made to form a single category. As in the experiments 2x PInterI x PIntraC and 2x PIntrarI x PIntraC there was no seed formation (there was no fruit formation), these did not enter into statistical analysis. For the statistical analysis, the results of seeds produced were compared to the number of ovules produced in the species (4.0 ± 0.0; mean ± .sd – information previously collected). Thus, the G 2x2 test was applied, with Cramer correction, followed by a pairwise post-hoc analysis (adjusted by the Bonferroni method) to verify the association between seed production and the presence of incompatible pollen grains, verifying the differences in each level of combination of these two factors (i.e., natural pollination, cross-pollination, PIntraC x PInterI, PIntraC x PIntraI, 2x PInterI x PIntraC, 2x PIntraI x PIntraC). The G test was performed with the DescTools package (Signorell 2023); pairwise post-hoc analysis was performed using the pairwise Nominal Independence function of the r companion package (Mangiafico, 2022).