not-yet-known not-yet-known not-yet-known unknown 2 MATERIALS AND METHODS 2.1 Study area and analyzed plant material The study was carried out at the São João farm, which has about 90 ha and is located in the rural area of the municipality of Quebrangulo - AL. The forest in the area is drier than the humid Atlantic forest, but not as dry as the xerophyte forest of the Caatinga (Nusbaumer et al., 2015). The average annual rainfall rate is 431.8 mm and the rainy season is concentrated between November and April (Mascarenhas et al., 2005). Senna aversiflora is an endemic shrub species of the Brazilian flora found only in the Northeast of Brazil, occurring in Caatinga areas of the states of Paraíba, Pernambuco, Alagoas, and Bahia (Flora e Funga do Brasil, 2024), and is frequently found across the São João farm. Twenty flowers and 10 buds from each of ten individuals within the farm were collected, totaling 200 flowers and 100 buds. Intact and damaged flowers and buds were included in the sample. Flowers and buds were fixed in FAA 70 (formaldehyde, acetic acid and 70% ethanol) for 24 hours and subsequently stored in 70% alcohol for light microscopy and scanning electron microscopy (SEM) analyses. Vegetative samples of the branches containing flowers and buds were also collected as testimony material, then herborized and deposited in the UFPE herbarium under number 88.954. 2.2 Microscopy Fragments of about 1 cm2 from sterile floral whorls (sepals and petals, adaxial and abaxial view) and the whole reproductive whorls (anthers and pistils) from the fixed material were used for SEM analysis. The fragments were adhered onto aluminum stubs (1/2”) using a double-sided adhesive tape. The samples were observed, analyzed, and photomicrographed using the HITACHI TM4000Plus scanning microscope. The identification and terminology of epicuticular waxes follow the classification of Barthlott et al. (1998). For light microscopy, we used samples of the material fixed in FAA 70% (n = 60), preserved in 70% ethanol. We prepared paradermal (adaxial and abaxial surfaces) and transverse sections of the sepals and petals and transverse sections of the ovary, style, feeding anthers, and pollinating anthers from three intact (n = 30) and three damaged (n = 30) flowers per individual. The sections were cut freehand using a steel razor blade and Styrofoam cubes as support. They were then clarified in 50% sodium hypochlorite, washed three times in distilled water, and neutralized in 1% acetic acid solution. The paradermal sections were stained with Safranin and the transverse sections were submitted to double staining with Astra blue/Safranin (Kraus et al., 1998) and later assembled between a slide and a coverslip in hydrated medium containing 50% glycerin (Kraus & Arduin, 1997). Flower buds were embedded in paraffin before sectioning. Samples of the materials fixed in FAA 70% and preserved in 70% ethanol were submitted to ethanol-tertiary butanol series (50% to 100%) and then transferred to pure butanol (overnight). After this step, we performed the paraffin infiltration and inclusion in a butanol-paraffin series (3:1, 1:1, 1:3) followed by 2-3 changes of pure paraffin (Kraus & Arduin, 1997). The included material was sectioned in a rotary microtome and the sections were submitted to double staining with Astra blue/safranin (Kraus et al., 1998) and mounted on permanent slides with Canada balsam (Bukatsch, 1972). 2.3 Histochemistry Histochemical tests were performed to compare the substances produced by intact and damaged flowers. Five sections (embedded in paraffin as described above) from one intact flower and one damaged flower of each of three different individuals were used for these tests, totaling 15 sections from intact flowers and 15 sections from damaged flowers. The following reagents were used: Xylidine Ponceau for proteins (Vidal, 1970); Lugol’s solution for starch (Johansen, 1940); Sudan IV for lipid substances (Pearse, 1972); phenylhydrazine hydrochloride/acetate (Patt, 1989) and Fehling (Kraus & Ardwin, 1997) for glucose and fructose; ferric chloride for phenolic compounds (Johansen, 1940); Wagner’s reagent for alkaloids (Furr & Mahlberg, 1981); NADI for essential oils and oleoresins (David & Carde, 196); hydrochloric vanillin for tannins (Johansen, 1940); and ruthenium red for pectins and mucilage (Johansen, 1940). After treatment with the reagents, the sections were mounted between slides and coverslips in glycerin (50%) and analyzed and micrographed using an optical microscope (Leica DM7500) coupled to a video camera (Leica ICC50). 2.4. Analyses 2.4.1 Anatomy The main characters were observed and recorded in a Zeiss photographic microscope. The following characters, considered to influence the attack of insects, were investigated: emergence and/or suppression of cells and tissues, collapse of cells, emergence and/or suppression of cellular inclusions such as crystals and chemical compounds, deformations in tissues and structures, and any type of alteration in damaged flowers. Anatomical and histochemical images were analyzed using Leica Application Suite software for image processing. Three intact flowers and three damaged flowers were collected from each of ten individuals of S. aversiflora (n = 60) for measurement of anatomical traits. The following anatomical traits of the petals were measured: (i) adaxial (µm) and abaxial (µm) cuticle thickness; (ii ) adaxial (µm) and abaxial (µm) epidermis thickness; (iii ) mesophyll thickness (µm). Ten measurements of each anatomical attribute were made in each intact flower and damaged flower, totaling 600 measurements. Measurements were made using an ocular micrometer on a Zeiss microscope. 2.4.2 Histochemistry To determine the levels of primary and secondary metabolites in intact and damaged flowers, we converted qualitative data into quantitative data. We counted the number of cells that reacted positively in the histochemical tests in each of the five sections analyzed per flower (Table 1) and the average number of the five sections were converted to a numerical scale of 0-4, as detailed in Table 2. Excel graphs are used to illustrate the results. 2.4.3 Resource production A total of 20 young buds from 10 individuals (n = 200) were marked and followed until anthesis. During the follow-up, when florivory was detected, the damaged buds were marked with ribbons of a specific color for each stage of development of the bud (pink = initial, red = intermediate, yellow = pre-anthesis). At the end of the follow-up, before the opening of the flowers, the buds were isolated in voile bags to avoid visiting pollinators. One day after the opening of the bud, the anthers of each flower were removed and stored in eppendorfs (one flower per eppendorf) containing 70% alcohol. To assess whether florivory influences resource production, we counted the number of pollen grains in feeding and pollinating anthers of intact and damaged flowers using a Neubauer chamber, following the usual protocol (adapted from Maêda, 1985). 2. 5 Statistical analyses To check the normality and homoscedasticity of the data, we used the Anderson-Darling and Bartllet test. The thickness of the adaxial and abaxial epidermis and the mesophyll and the amount of pollen were compared between intact and damaged flowers with non-parametric tests due to lack of normality and homoscedasticity (p < 0.05). Thus, the Kruskall-Wallis test and the post-hoc Dunn’s test were used to compare thickness data of petal tissues and the amount of pollen. As for pollen production at different stages of floral development, the data had a normal distribution and thus ANOVA was used to compare the means. All tests were performed with 95% confidence level using R (RcoreTeam). Means were square-root transformed for better graphical representation in the analyses of pollen production. 3 RESULTS