Flower, fruit phenology and flower traits in Cordia boissieri (Boraginaceae) from northeastern Mexico

Study area

The study was performed from October 2009 to September 2011. We worked with a population of 256 individuals of Anacahuita in a fragment of Tamaulipan thornscrub in Northeast Mexico, inside the experimental area of Facultad de Ciencias Forestales (Universidad Autónoma de Nuevo León, 24°46′43″N99°31′39″W) at an elevation of 370 m above sea level. The area has an average temperature of 21 °C, with a maximum extreme temperature in summer >40 °C and <0 °C in winter. The annual rainfall average is 805 mm, and dominant soils in the area are vertisols of alluvial-colluvial origin (SPP–INEGI, 1986; Alanís-Rodríguez et al., 2008).

Species description

Cordia boissieri is a native North American shrub or small tree, 5 to 8 m tall. It has simple, alternate and ovate leaves from 15 to 20 cm in length, with a pilose-velvety surface (Vines, 1986; Gilman & Watson, 1993). The flowers are trumpet shaped, in groups from five to eight, white with a yellow nectar guide, up to 45 mm in length, with five rounded lobes and five stamens joined at the base within the corolla tube (Vines, 1986; Gilman & Watson, 1993; Alvarado et al., 2004). In addition, the anthers are oblong, filiform, glabrous, and yellow–greenish; the pistil usually varies in length and narrows towards the apex, ending with two stigmas. Flowering occurs throughout the year, with peaks in the rainy season from late spring to early summer (Vines, 1986; Gilman & Watson, 1993; Alvarado et al., 2004).

Prospective visits were made in order to identify and mark all individuals >1.5 m in height. Each individual was marked with a metallic label and flagging tape with progressive numbers. Five plants were randomly selected to follow the life of 10 flowers in each one, from bud opening to flower senescence, in order to determine flower life span and avoid duplicating data during subsequent samplings.

Flowering and rainfall

In order to determine variations on flower and fruit production of Anacahuita through time and how it relates to rainfall, we recorded flowering of individuals within the population and the monthly amount of rainfall (mm) in the study area. During the study we quantified monthly: highly flowering plants (as seen below), plants with fruits, and plants without flowers and fruits. Flowering in Anacahuita showed two very distinct patterns, one was a very dense flower production that covers the entire canopy and the other is the production of a few flowers in one or two branches (Fig. 1). Only plants with dense flower cover were considered as highly flowering plants. The same criterion was used for fruit production. To determine the relationship between highly flowering plants and fruiting with rainfall events, we performed a GLM analysis using STATS package for R software (R Core Team, 2014), where the dependent variables were plants with flowers and plants with fruits and the independent variable was rainfall. We considered rainfall of one month prior to flowering and two months prior to fruit set following our observations that flowering and fruiting roughly occur after this time.

Flower measurements

To determine phenotypic variation of flowers within plants and between plants and in different flowering events, we selected five flowers at random in each cardinal point, obtaining data from 20 flowers for each one of five plants during each flowering event. The floral attributes evaluated were as follows: Whether flowers were longistylus or brevistylus (style length) (ST); style number (SN); stamens number (SAN); total flower length (TL; mm with a 0.05 mm accuracy); corolla length (CL; mm); corolla diameter (CD; mm); nectar guide diameter (NGD; mm); number of lobes (NL); nectar guide color (NGC; by direct observation of changes in patterns and color tones in the flower tube) and sepal length (SL; mm), obtained through the difference of the total and corolla lengths. Style length was determined in 1,500 flowers from 75 plants across flowering events (>5% of high flowering plants). To reduce color bias due to perception, this trait was assessed by only one person in full day light against a white sheet of paper.

Recorded data were tested for normal distribution using Kolmogorov–Smirnov test. We performed a nested ANOVAs to test for differences in floral traits among plants and flowering events (P = 0.05) for quantitative variables. Kruskal-Wallis tests were performed for categorical and variables not normally distributed (Zar, 2010). We used a binomial test for style length and goodness of fit using χ² tests (Ríus-Díaz et al., 1998) for nectar guide colors to determine if there were variations in the phenotypic expressions of the studied plants.

Flowering and rainfall events

C. boissieri flowering generally resulted after one month of rainfall and fruiting after two months (Fig. 2); the GLM analyses showed a significant relationship between rainfall events and the number of plants with flowers (F_1,22 = 817.33; P = 0.003) and fruits (F_1,22 = 143.13; P < 0.001). We recorded nine flowering events, one in 2009 (October), five in 2010 (April, June, July, August, September and October) and three in 2011 (April and July), we also observed isolated flowering plants in November 2009 (three plants), March 2010 (six plants), and March 2011 (five plants). Four flowering events included over 80% of plants (215). These had a longer duration (Fig. 2).

We observed an overlapping of flowering and fruit production in three of the four flowering events with >80% of the plants, flower and fruit set were observed as separate processes in the other flowering event with >80 of the plants with many flowers (April 2011) (Fig. 2). The latter was the only event that occurred after no rainfall.

Flower trait variation

We determined that anthesis (opening to senescence of flowers) of C. boissieri was from three to five days. We observed heterostyly on C. boissieri flowers (Supplemental Information 1): there were 760 brevistylus, flowers (from 38 plants) and 740 longistylus flowers (from 37 plants). The style length did not vary within individuals; however, the Kruskal-Wallis test showed that there was a significant variation between plants (H_1,74 = 1, 499; P < 0.001). In addition, the binomial test for style length showed similar numbers longistylus (51%) and brevistylus flowers (49%; P = 0.624). Style length and nectar guide color (NGC) did not vary between flowering events.

The phenotypic expression most commonly observed was two styles, five stamens and five lobes. However, in 2,267 observed C. boissieri flowers we found flowers from one to four styles ($\bar{x}=2.02\pm 0.135$ SD), with two to nine stamens ($\bar{x}=5\pm 0.381$ SD) and two to nine lobes ($\bar{x}=5.01\pm 0.381$ SD, as seen in Supplemental Information 2). We found five nectar guide colors (Supplemental Information 3), in the 1,500 flowers from 75 plants. There were 760 flowers with yellow, 380 with yellow-orange, 260 with yellow-white, 60 with yellow-white-orange and 40 with an orange-yellow nectar guide. These numbers were significantly different to chance variations (χ² = 1149.333; df = 4; P < 0.001), and varied between plants (H_1,74 = 1, 499; P < 0.001).

Cordia boissieri flowers ranged in total length from 15.45 mm to 58.9 mm ($\bar{x}=35.49\pm 5.78$ SD), corolla varied from 6.45 mm to 41.1 mm ($\bar{x}=24.08\pm 5.26$ SD), and sepal from 4.45 mm to 29.5 mm ($\bar{x}=11.4\pm 2.57$ SD). Corolla diameter varied from 16.3 mm to 77.0 mm ($\bar{x}=47.98\pm 9.33$ SD), whereas nectar guide diameter ranged from 4.9 to 30.6 mm ($\bar{x}=11.64\pm 2.22$ SD). Nested ANOVAs for each floral attribute showed that most variations occurred between plants than between flowering events. Total flower length (F_2,81 = 18.217, P < 0.001), corolla length (F_2,81 = 14.532, P < 0.001), corolla diameter (F_2,81 = 47.316, P < 0.001) and nectar guide diameter (F_2,81 = 42.335, P < 0.001) differed between plants but not between flowering events. Sepal length varied both between plants (F_2,81 = 5.697, P < 0.001) and between flowering events (F_81,1596 = 3.156, P = 0.048).

There were significant variations between plants for style length (H_2,27 = 1679.000, P < 0.001), style number (H_2,27 = 112.868, P < 0.001), number of anthers (H_2,27 = 126.521, P < 0.001), number of lobes (H_2,27 = 143.071, P < 0.001), and color of nectar guide (H_2,27 = 1679.000, P < 0.001). Number of lobes also differed between flowering events (H_2,27 = 15.618, P < 0.001). No intra plant variations in floral traits were detected comparing the four sides (cardinal points) of the canopies (P > 0.05).

Flowering and rainfall events

The phenological characteristics of flowering (intensity, duration and overlap) are important aspects of the reproductive effort of plant (Richards, 1986; Guitián & Sánchez, 1992), and their variation (including opening time and number of flowers) commonly correlates with temperature, moisture, and day length (Justiniano & Fredericksen, 2000; Elzinga et al., 2007; Yadav & Yadav, 2008; Hódar, Obeso & Zamora, 2009); edaphic and biotic factors (Borchert, 1983) like pollinators (Trapnell & Hamrick, 2006; Medel & Nattero, 2009) seed dispersers and herbivores (Mahoro, 2002; Lacey et al., 2003). C. boissieri is described in the literature as producing flowers and fruits all year, with two peaks of blooms in late spring and early summer (Vines, 1986; Gilman & Watson, 1993; Alvarado et al., 2004); this may be a result of a bimodal rainfall pattern (Chapman et al., 1999). Our results agree with the literature, because the plants of Cordia boissieri showed four flowering peaks during the two years of observation and were approximately in late spring and early summer. The general coincidences of flower traits across time suggest that the species might be pollinated by the same agents.

We observed that flowering was more abundant one month after heavy rainfall events, but also one abundant flowering event occurred after a dry season. This flowering coincides with reports form the literature of highest blooming events in plants of an arid tropical community during the dry and warm season (León De La Luz, Coria–Benet & Cruz–Estrada, 1996). The common pattern for angiosperms is one annual flower production and only for some plants, flowering may occur more than once a year to the point that some species produce flowers and fruits all year (Wolfe & Burns, 2001). The variation in time of year for flowering indicates that unlike for other studies (i.e., Borchert et al., 2005) the variation of moisture is very important in triggering flowering in this C. boissieri population. The observed flowering events of C. boissieri could be induced by previous rainfall events that may ensure seed production (Elzinga et al., 2007). Our results for Cordia boissieri agree with those observed in Cordia alliodora and C. elaeagnoides that bloom at the end of the rainy season (Bullock & Solis-Magallanes, 1990) and were similar to C. multispicata that is described as producing flowers and fruits during most of the year (Vieira & Silva, 1997). Further long-term studies might consider temperature and rainfall or perhaps evapotranspiration or other water availability indicator as a predictor of flower and fruit production.

Flower trait variation

We observed that Anacahuita flowers had a lifespan of five days, this may be due to the receptivity of flowers, because the time to anthesis of each species varies with pollination systems; for example there are plants that have flowers that are receptive from only a few hours (Armbruster & Muchhala, 2009) to more than two weeks (Steinacher & Wagner, 2010). C. boissieri showed a diurnal and nocturnal display, suggesting diurnal and nocturnal pollinators. C. boissieri flowers showed a heterostyly pattern, belonging to the group of distyly flowers, coinciding with previous findings for Cordia to avoid self-pollinating (Gibbs & Taroda, 1983; McMullen, 2012). Boraginaceae is one of the 24 families that exhibit heterostyly, with the Cordia species displaying distyly and tristyly (Ganders, 1979).

Previous studies indicate that it could be misleading to characterize the phenotype of a single flower, because there could be an intra-plant floral variation, which may even be greater than between plants within populations (Medel & Nattero, 2009). In contrast, floral traits measured in our study showed more variation between plants than within the same plant or between flowering events.

The observed flowers in this study allowed us to find some differences from the flower description for the species. For example, in terms of styles, the general morphotype or most dominant is of two styles per flower (Vines, 1986; Gilman & Watson, 1993; Alvarado et al., 2004); however; we found flowers with one to four styles. Anacahuita flowers are described as having five stamens and five lobes with a length of 45 mm, with flowers with a yellow nectar guide (Vines, 1986; Gilman & Watson, 1993; Alvarado et al., 2004); in contrast, we found a flower length from 15.45 mm to 58.9 mm and five phenotypic expressions of nectar guide color, including the yellow variation cited in the literature. This could be considered for a new description of C. boissieri flowers.

In conclusion, three of the four flowering events with >80% of C. boissieri plants occurred one month after heavy rainfall events. Fruit set occurred after two months of rainfall. The species showed overlapping of flowering and fruiting only after heavy rainfall events but not during the dry season. The highest variation of phenotypic expression of C. boissieri flowers occurred between plants and not between flowering events. We suggest considering new flower phenotypes described here for future flower descriptions of C. boissieri. Further longer term studies may help understand if variations in flower traits are the result of similar insects pollinating Anacahuita flowers throughout the year.