Morphological analysis of pollen grains from heterodynamous stamens of some Aeschynomene L . ( Leguminosae : Papilionoideae-Dalbergieae )

(Morphological analysis of pollen grains from heterodynamous stamens of some Aeschynomene L. (Leguminosae – Papilionoideae – Dalbergieae)). Flowers with heterodynamous stamens can present diff erences in the pollen grains of each stamen size group. Sp ecies of Aeschynomene L. present didynamous stamens (fi ve long and fi ve short) but li ttle is known about their pollen dimorphism. The objective of this study was to increase the knowledge about the pollen characteristics in Aeschynomene and emphasize the possible morphological diff erences between the pollen grains of long and short stamens in order to contribute to ecological and taxonomic studies. Po llen grains from the two groups of stamens size of ten species were analyzed separately, according to the standard methodology for studies of pollen morphology. In addition, analyses of variance, comparison of means and base index were performed. Th e results showed that the amb, shape, endoaperture type and sexine ornamentation did not vary in the pollen grains of the long and short stamens in the same specimen, but they varied among the species. However, in relation to the size of the pollen grains of the two groups of stamens, four species presented signifi cant diff erences regarding the size of the polar and equatorial axes. The pollen morphological data obtained considering the heteromorphism of the stamens can contribute to the knowledge of the reproductive dynamics of the genus, and to the systematic studies.


Introduction
The tribe Dalbergieae sensu latu Klitgaard and Lavin (2005) in Papilionoideae subfamily of Fabaceae comprises 49 genera, wich Aeschynomene L. as the third most representative of the tribe .The Aeschynomene genus presents pantropical distributions, with ap proximately 180 species in the world (Rudd 1981, Lewis et al. 2005).Brazil is a centre of diversity, with 49 accepted species, from which 26 are endemic (Lima et al. 2015).The genus' habit varies from herbaceous to shrubby and is divided into two sections and nine series (Rudd 1955, 1959, Fernandes 1996).However, the combined analysis by Ribeiro et al. (2007) and others (Lavin et al. 2011) demonstrated the paraphyly of genus.Aeschynomene presents papilionoid corolla, which is related to the co-evolution between the hymenopterans and the Papilionoideae (Arroyo 1981, Judd et al. 2007), whose flowers' androecium has heterodynamous stamens (five long and five short) and uniform anthers (Rudd 1955).Tucker (1996) studied many species of Leguminosae and pointed out that stamens with dimorphisms are commonly found in this family, possibly differing in size, filament length, anther shape and moment of dehiscence.The presence of different stamen sizes was diagnosed in Aeschynomene in research carried out by Burkart (1939), Fernandes (1996), Sampaio (2005), Silva & Antunes (2014).Carvalho & Oliveira (2003) found significant differences in the pollen grains-dimensions among heteromorphic anthers of Senna sylvestris (Vell.)H.S. Irwin & Barneby (Leguminosae) and Luo et al. (2008) describe differences in sexine ornamentation among long and short stamens of the same specimen of Melastoma malabathricum L. (Melastomataceae).
It is also known that polymorphism occurs in the size of the pollen grain of longistylous and brevistylous flowers (Furtado et al. 2014), in the number of apertures (Stanski et al. 2016) and sexine ornamentation of Psychotria capitata (Ruiz & Pav.) (Rubiaceae) (Ganders 1976, Stanski et al. unpublished data).These characters that change depending on the floral morphology, such as the number and aperture type and sexine ornamentation are important in the delimitation of pollen types in ecological and palynotaxonomic studies (Salgado-Labouriau 1973, Lorscheitter 2006).
Studies on genus concerning the morphological differences between pollen grains from heterodynamous groups of stamens are inexistent.In this context, the aim of this research was to carry out an unprecedented analysis of the pollen morphology of ten species from the genus Aeschynomene L.; with attention to the possible morphological differences between long and short stamens of the same specimen to contribute to the studies of reproductive biology, pollination and taxonomy to the genus.Groups with long (+) and short (-) stamens were segregated for palynological analysis.To avoid contamination between groups (+) and (-) samples were dissected under stereomicroscope, with sterilized tweezers and just flowers with anthers in pre anthesis were studied.These samples were prepared according to Erdtman (1960) and Melhem et al. (2003).

Materials and methods
For each stamen group (+) and (-) of each specimen the polar and equatorial axes in equatorial view were measured in 25 random pollen grains.The measurements of length and width of colpus and endoapertures, thickness of sexine layer (tectum + columelae) and nexine layer, total exine (sexine + nexine) and lumen length were based on ten pollen grains.The description of pollen morphology to the genus level followed Barth and Melhem (1988), Punt et al. (2007) and Hesse et al. (2009).
For the 25 measurements we performed the arithmetic mean (x), with the standard deviation pattern (sx), the sample deviation standard (s) and the coefficient of variation (V).The comparisons of means were performed through the analysis of confidence interval at 95% of probability (Vieira 1981), with software Minitab 14.For the ten measurements the arithmetic mean was calculated, except for lumen size at which the variation range was used.
The values of axes were subjected to variance analysis (ANOVA) and significant differences between means were identified by the Tukey test at 5 % probability.The base index (BX) was calculated and consists of the subtraction of each of the 25 measurements of an axis (MS) by the smallest value of variation range (M).Classes were established: with values lower than 2.0 considered as a low value (< 2.0 = B-) and values higher than 2.0 as a high value (> 2.0 = B+).This index was described along with the sample standard deviation, coefficient of variability and the statistical difference between the stamens of the groups (+) and (-) per species, to establish a standard among the four variables.
Photomicrographs were obtained using an OLYMPUS BX 50 microscope with camera and CellSens software for Windows.For the SEM images the pollen grains were deposited in a metal stub, coated by cathodic spray (Leica EM ACE 600) using Au with thickness of 120 nm.Samples were observed under the SEM JEOL JSM-IT300LV (Tokyo Japan) operating with 20 kV electron beam and the images were digitized.

Pollen morphology of short and long stamen groups
These characteristics (polarity, amb, apertures, ornamentarion) did not vary between the two sets of stamens, but rather between the species.The pollen grain differences among the species demonstrate that the palynological characteristics can be used in the segregation of the species with palynotaxonomic value.
Although the pollen grains of the long and short stamens are in the same size class in Aeschynomene denticulata, A. elaphroxylon, A. montevidensis and A. rudis, they present significant differences concerning the size of their polar and equatorial axes, when submitted to the analysis of variance and subsequent Tukey's test (table 1).When submitted to the same analyses, the specimens of A. americana, A. brevipes, A. fluminensis, A. martii, A. selloi and A. sensitiva they did not present significant differences.Carvalho & Oliveira (2003) report that the long anther of Senna sylvestris has the largest pollen grains when compared to those of the other short anthers.Furtado et al. (2014) demonstrated that the longistylous morpho of Psychotria capitata presented smaller pollen grains (12.87 ± 1.28 μm) than the brevistylous morpho (18.73 ± 0.60 μm).We did not observe here the pattern long stamen/largest pollen grain axes in all species.According to our study, the pollen grains of the short stamens not necessarily present low values of the axes, as in A. elaphroxylon (figures 16-17, table 1), whose pollen grains of the short stamens presented a higher mean to polar and equatorial axes of pollen grains from the long stamen group.In the case of A. denticulata and A. montevidensis, the long stamens presented the pollen grains with the largest polar axes, whereas in A. elaphroxylon and A. rudis it was the short stamens that presented the highest values for the same axis (table 1).Concerning the equatorial axis, the pollen grains from these four species, the short stamens always had the highest values.
The base index (B) demonstrated the difference of the data values between the lowest sample values compared with the other values measured for each axis of the pollen grains (table 1).The highest B indexes were found in A. elaphroxylon (≥ 3.3), followed by A. martii (≥ 2.7), A. brevipes (≥ 2.2) and A. rudis (≥ 2.2).In these species, the highest values for the standard deviation of the sample and for the coefficient of variation were also observed, however not all species presented significant differences between the pollen grain axis of the (+) and (-) stamens group (A.brevipes, A. martii).Although they have B+, in A. martii, A. brevipes and A. rudis, these values were found only in one of the axes from one of the groups of stamens, whereas the others axes were B-.The lowest B indexes were found in A. montevidensis (≤ 0.7), followed by A. denticulata (≤ 0.9), with significant differen between stamen group (+) and (-).It is worth mentioning that between the two groups of stamens, A. americana, presented the highest index of the basis (B), standard deviation of the sample and coefficient of variation than some species with significant differences.Therefore, it is not possible to establish a constant pattern among the four variables compared.
Concerning their axes size and the B values, the pollen grains of the long stamens of Aeschynomene denticulata, A. elaephroxylon, A. montevidensis and A. rudis presented significant differences in polar and equatorial axes when compared to the short stamens.The pollen grains of the other species studied did not present statistically significant differences in these axes.It is recommended that new statistical treatments and models be used, as well as studing more species and specimens of Aeschynomene aiming to complement the data obtained.As the pollen Table 1.Measurements (μm) in equatorial view of pollen grains of Aeschynomene L. (n = 25).Long stamen (+) and short stamen (-), confidence interval at 95% of probability of the lowest sample values (CI-) and highest sample values (CI+), arithmetic mean (x), average standard deviation (sx), mean of base index (Bx), standard deviation of sample (s), coefficient of variation (V%), ratio between polar and equatorial axes (P/E), arithmetic mean (n=10) of lenght and width of colpus, arithmetic mean (n = 10) of length and width of endoaperture, thickness of nexine layer (Nex), thickness of sexine layer (tectum + columelae) (Sex), total thickness of exine (Ex) and variation range of length of lumen (Lum).
Obs. Species in bold present significant differences concerning the size of their polar and equatorial axes by Tukey's test.The means (x) followed by the same letter (a) in the same species and on the same axis (P or E) were not differentiated by Tukey test at 5% of probability (p < 0.05).grains showed variations in the characteristics of the apertures and exine ornamentation between species the results of which are of the utmost importance for systematic studies.

Figure 18 .
Figure 18.Representation of confidence interval of means at 95% of probability of Polar axis (μm) in equatorial view of Aeschynomene pollen grains.The higher and lower boundaries show the confidence interval; the average circle shows the arithmetic mean.

Figure 19 .
Figure 19.Representation of confidence interval of means at 95% of probability of Equatorial axis (μm) in equatorial view of Aeschynomene pollen grains.The higher and lower boundaries show the confidence interval; the average circle shows the arithmetic mean.