MORPHOMETRIC ANALYSIS OF NECTARIES AND THEIR POTENTIAL USE IN THE TAXONOMY OF THE JOVIBARBA HEUFFELII COMPLEX ( CRASSULACEAE )

The aim of this study was to quantify the morphological variation of nectaries among 14 populations of Jovibarba heuffelii based on multivariate statistics, and to establish whether nectaries possess taxonomic significance for differentiating taxa within the J. heuffelii complex. To this end, we measured the width of the nectary, its height, the angle between the carpels and nectary, the shape of the nectary and the distance between nectaries were measured and analyzed. Descriptive statistics, the Tukey HSD (honest significant difference) of homogenous groups for the unequal N post-hoc test, canonical discriminant analysis (CDA) and multiple correspondence analysis (MCA) were used. Morphometric analysis showed that the quantitative and semiquantitative characteristics of nectaries in the J. heuffelii complex are highly morphologically variable, both within and between populations, and that they are unreliable as taxonomic characters for taxon differentiation within the J. heuffelii complex.


INTRODUCTION
Floral nectaries are relatively simple structures of diverse origin, partaking in the pollination process.They may appear in almost all parts of a flower and produce sweet-tasting exudates with different quantitative and qualitative characteristics (Bernadello, 2007).Although the shape, structure and position of nectaries have been used for taxonomic and phylogenetic studies for more than a century (Behrens, 1879;Bonnier, 1879;Knuth, 1906;Brown, 1938;Norris, 1941;Fahn, 1953;Brown, 1961), the value of nectaries in systematics was recognized only fairly recently (Zandonella, 1977;Fahn, 1979;Vogel, 1981;Cronquist, 1981Cronquist, , 1988;;Smets, 1986Smets, , 1988;;Smets and Cressens, 1988;Smets et al., 2000Smets et al., , 2003)).Fahn (1979) classified floral nectaries into 9 types according to their position within the flower.Representatives of the Crassulaceae (Sempervivum) family have nectaries of the 4 th type, characterized as "nectary as a disc surrounding base of ovary".Bernadello (2007) criticizes the use of the term "disc" as it has been used for various structures in the form of a ring within the flower, and they may or may not belong to the nectaries (e.g., Daumann, 1931;Fahn, 1979;Cronquist, 1981;Smets, 1986;Schmid, 1988;Smets and Cresens, 1988;Vogel, 1998).Schmid (1988) has based his classification of floral nectaries solely on the position of the nectaries, explaining that more general classification systems are better understood and more practical for use.According to his criteria, nectaries in the J. heuffelii complex belong to the type of floral receptacular nectaries.Bernadello (2007) supplements this classification with the terms intrastaminal, extrastaminal and interstaminal, depending on whether the nectaries are situated on the receptacle between the staminal whorl and the ovary (intrastaminal), or between the perianth and the androecium, respectively (extrastaminal).The receptacular nectaries located exactly between the stamens may be referred to as interstaminal.According to this criterion, the nectaries in J. heuffelii are intrastaminal receptacular nectaries.
Molecular analyses have shown that the Crassulaceae family is monophyletic (Morgan and Soltis, 1993;Soltis and Soltis, 1997), while the presence of nectary scales in this family is a synapomorphic characteristic together with succulent leaves and anysocitic stomata.Jovibarba heuffelii (Schott) A. Löve & D. Löve is a complex of insufficiently clarified taxa, distributed in the central part of the Balkan Peninsula and central and southern parts of the Carpathians (Meusel, 1965;Jalas et al., 1999).
Pančić (1874) used the shape and position of nectaries to differentiate the taxon Sempervivum kopaonikense from Mt. Kopaonik in Serbia from   the taxon Sempervivum patens from the southern Carpathians in Romania.For S. kopaonikense, he claimed that its nectaries are "at right angles, rounded or tridented at the top", while in S. patens they are "directed toward the top, cut off straight or shallowly indented".Preager (1932) also believed that nectaries are a significant trait of the population from Kopaonik, which he treated as a form of the species J. heuffelii.The mentioned author stated that S. kopaonikense differs from the typical forms by "hypogynous scales spreading instead of erect with a corresponding bulge at the base of the calyx and corolla, giving the flower a ventricose appearance."The morphological characteristics of nectaries in representatives of the genus Jovibarba Opiz have never been studied before.
The aim of this study was to quantify the morphological variation of nectaries between populations of J. heuffelii based on multivariate statistics, and to find out whether nectaries may have taxonomic significance in the differentiation of taxa within the J. heuffelii complex.

Plant material
289 individuals from 14 populations of Jovibarba heuffelii s.l. were collected from their natural habitats for morphological analysis.The samples were taken from Serbia, Macedonia, Bulgaria and Romania (Table 1., Fig. 1).The plants in the flowering stage were collected in July and August during two vegetation seasons (2010,2011).Voucher specimens were deposited in the Herbarium at the Institute of Botany and Botanical Garden "Jevremovac", at the Faculty of Biology, Univer-sity of Belgrade -BEOU (Thiers, 2014).The inflorescences were stored in glycerol:96% ethanol (50:50 v/v) until measurements were taken.

Morphometric analysis
For morphometric analysis, 4 nectaries from 4 different flowers from the inflorescence were separated per individual (1156 nectaries in total).The nectaries were photographed using a LEICA DM 1000 microscope.For measuring, the angle of the nectary with a carpela was isolated per individual and photographed on a LEICA MZ16 A stereomicroscope.All the characters were measured using the LeicaQWin image analyzing program (Leica image software).

Statistical analysis
Descriptive statistics (mean, maximum, minimum, variance, standard error and standard deviations, coefficient of variation) for the quantitative characters were done.The Tukey HSD of homogenous groups for the unequal N post-hoc test was used to analyze the differences in the nectary morphology of 14 populations.Canonical discriminant analysis (CDA) was used to test the hypothesis for the separation of the analyzed populations.CDA analyses were performed on individuals for 4 characters (width of the nectary, height of nectary, the angle between carpel and nectary, shape of the nectary), and two different data sets.The first data set included only populations of S. kopaonikensis and S. patens from localities Kopaonikand Domogled, while the second included all analyzed populations.Canonical scores for each case were calculated in order to estimate the distances between individuals that were used to visualize the relationship among the 14 populations, which were defined as a priori groups.The qualitative characters of all the investigated populations were statistically analyzed by multiple correspondence analysis (MCA) and presented on a scatterplot diagram.All statistical analyses were performed using the Statistica 5.1 (Statsoft, 1996) package.

RESULTS
Detailed morphological analysis has shown that the nectaries in individuals from J. heuffelii populations are whitish, somewhat lighter in color than the carpels, and situated on the receptaculum in the base of the apocarpous gynoecium.Their upper surface varies from flat to convex to indented, with two recognizable teeth (Figs.2A-C).The two teeth may be symmetrical or one may be larger than the other.In some individuals, the teeth are slightly rounded and close to each other, so the upper side of the nectary is cordate (Fig. 2C), while in other individuals the teeth converge at the top, forming a beak shape (Fig. 2D).Certain individuals have nectaries with widely spaced, slightly rounded teeth (Fig. 2F).Some individuals within the populations have three more or less pronounced teeth on the upper side of the nectaries (Fig. 2E).These populations include SR-Treska with only one individual with such nectaries, SR-Basarski kamik with 2 individuals, MA-Treskavec with 4 individuals, MA-Ma-vrovo and RO-Domogled with one individual each, and BU-Trojanski prolaz with 3 individuals bearing 3-teeth on the nectaries.The distance between nectaries also varies from one population to another.It may be greater than the width of the individual nectary (Fig. 3A) or smaller than the width of the nectary (Fig. 3B).There are records of the upper part of the nectary curving toward the inner side, toward the carpel (Fig. 3C).The results have shown that most individuals have two teeth on the upper part of the nectary, but in certain populations, there are also individuals with level surfaces on the nectaries or with three teeth.

Descriptive statistics
The width of the nectary varies from 0.39 to 1.16 mm.The maximum value of this character was registered in the SR-Basarski kamik population.The smallest width of the nectary was noticed in the MA-Treskavec population.This character showed a moderate coefficient of variability ranging from 11.12 to 17.18% (Table 2).The height of the nectary varies from 0.23 to 0.85 mm.The minimum value of this character was registered in the RO-Domogled and SR-Studenica populations, while the maximum value was noticed in the SR-Nebeske Stolice population.The coefficient of variability of this character was in a range of 11.38% to 22.87%.The angle between the carpel and nectary is the most variable character, ranging from 11.53° to 92.70°.
The maximum angle was registered in the SR-Suvaja population and the minimum value in the SR-Gradac population.The coefficient of variability for this character was in a range of 17.77% to 50.17%.Very high variability of this character was shown by the SR-Suvaja population (50.17%).

Tukey HSD post-hoc test
This test showed statistically significant variability in the mean values of characters (p<0.05) between certain pairs of populations (Table 3).Within the first homologous group, the MA-Treskavec, SR-Pljačkovica and SR-Suvaja populations stand out, as they do not show statistically significant differences in the mean values of the character width of nectary (W).Each of these three populations shows statistically significant differences to the other analyzed populations.For the character height of nectary (H), the unequal N HSD test showed separation of the SR-Studenica, SR-Suvaja and RO-Domogled populations as the most similar, while these populations show statistically significant differences to other analyzed populations within the first homologous groups.The angle between the carpel and nectary (A) is a character showing statistical similarities for the SR-Gradac, SR-Pljačkovica, SR-Radan, SR-Besna Kobila, RO-Domogled and MA-Mavrovo populations.Each of these populations shows statistically significant differences to the other analyzed populations.The unequal N HSD test for the character shape of nectary showed the least differences among populations within the J. heuffelii complex for all the analyzed characters.There are no statistically significant differences in variations of this character in most populations, except those from Macedonia (MA-Treskavec, MA-Mavrovo) and Bulgaria (BU-Trojanski prolaz), which show statistically significant differences from the other analyzed populations in the mean values of this character (Table 3).

Canonical discriminant analysis (CDA)
Analysis of populations from Kopaonik (type locality for taxon S. kopaonikense) and Domogled (type locality for taxon S. patens), taken as a priori groups, showed a complete lack of differentiation between these two groups of populations accord-ing to the quantitative characters of nectaries (Fig. 4B, Table 4).The CDA including all analyzed populations also resulted in no statistically significant differentiation of populations (Fig. 4A, Table 4).

Multiple correspondence analysis (MCA)
This analysis showed that the first axis carried 16.17% of the total variability in the sample, while the second axis carried 9.52% (Fig. 5).Based on the character distance between nectaries, the SR-Suvaja, SR-Nebeske Stolice, SR-Gradac and SR-Basarski kamik populations were arranged into one group in the positive part of first correspondent axis (Fig. 5).The nectaries of these populations are spaced at a distance greater than the width of an individual nectary (Fig. 3A).The additional characters responsible for grouping these populations were convex surface of the nectary (Csurf: 1) and flat surface of the nectary (Fsurf: 1) (the surface of nectaries in individuals from these populations is either flat or convex).
All the other populations were grouped around the characters two teeth of nectary (TwoT: 1) and space between nectaries less than the width of the nectary (Sless: 1) on the negative part of the first MCA axis, indicating that these characters had the greatest impact on population grouping.Individuals from these populations are characterized by the presence of two teeth on the surface of the nectaries and the distance between the nectaries is smaller than the width of a single nectary (Fig. 3B).

DISCUSSION
The more recent literature such as Parnell and Farvager (1993), Jalas et al. (1999) andHart et al. (2003), does not support the separation of several taxa within the J. heuffelii complex.On the other hand,    Marhold (2011) recognizes two infraspecies taxa: J. heuffelii subsp.heuffelii and J. heuffelii subsp.glabra (Beck & Szyszył.)Holub.The main character used to distinguish subsp.heuffelii and subsp.glabra has been the absence or presence of hairs on the adaxial and abaxial sides of the rosette leaves (Gajić, 1972).Lectotypification of all names within the J. heuffelii complex (Nikolić et al., 2014) was performed in order to solve the taxonomic issues within this group.Nectaries, as systematic characters useful for taxonomy in the J. heuffelii group, were first mentioned by Pančić (1874) and Praeger (1932), while in other flora they were neglected.
This morphometric analysis has shown that the quantitative and semiquantitative characteristics of nectaries in the J. heuffelii complex are highly morphologically variable, both within and between populations.In certain cases, as for example in the species Brassica rapa L., there is a particularly high variability in the morphology of the nectaries due to ploidy, so within the same species haploid, diploid and tetraploid individuals vary according to the morphological characteristics of their nectaries (Davis et al., 1996).In the species J. heuffelii only cases of diploidy have been recorded (Uhl, 1961), which indicates that ploidy is not the cause of such high variability.On the other hand, ecological factors may cause high morphological variability in nectaries, as well as variability in the other qualitative characteristics of nectaries such as production and chemical composition of the nectar.A positive correlation between the intensity of nectar secretion and nectary size was recorded in certain genera, e.g.Citrus and Lamium (Fahn, 1949;Gulyas, 1967).High temperatures and favorable soil moisture increase nectar secretion (Bonnier, 1879;Zander, 1921;Ostashenko-Koodryavzeva, 1928;Fahn, 1949;Kartashova, 1965).Environmental factors such as water availability may be very important for nectary shape in certain representatives of the Lamiaceae family in the Mediterranean, since in areas that are more arid the nectar-excreting stomas decrease in size, the volume of nectaries is altered and they are generally smaller than in temperate regions (Petanidou et al., 2000).Petanidou et al. (2000) also remarked that nectary volume is positively correlated with the volume of nectariferous tissue.As J. heuffelii is a succulent plant with a CAM metabolism, in an ecological sense a xerophyte that may survive without water for a long period of time, the water regime of the habitat should not play a crucial role in the size of the nectaries.Our previous studies have shown that the size of the vegetative and reproductive organs in Jovibarba heuffelii increases as altitude decreases, and in serpentinite and silicate substrates as compared to limestone substrate (Dimitrijević et al., 2011;Spasić, 2012).The significance of altitude on the composition of the nectar in bee-pollinated flowers was recorded by Andrejeff (1932), Hocking (1968) and Heinrich and Raven (1972), as flowers in plants growing at higher altitudes produce nectar of greater energy value than flowers at lower altitudes, indicating that altitude may influence the qualitative characteristics of nectaries and thereby also indirectly influence the quantitative morphological characteristics.The amount of nutrients in the soil may also influence nectar production, since nectar secretion is higher under low nitrogen supply (Shuel, 1955).The high level of variability in nectaries and the position of nectaries within the flower are very commonly caused by the type of pollination and behavior of pollinators (Nepii, 2007).
The quantitative characters width of nectary (W) and height of nectary (H) showed moderate variability, which is in accordance with previously obtained results for the morphometric characters of reproductive organs.The coefficient of variability for these characters had low to moderate values (Dimitrijević et al., 2011).
It is important to note that populations with the greatest angles between the nectaries and carpels at the same time also show the greatest variation quotient for that character.This means that within the same populations there are individuals with right angles (Fig. 3D) and very sharp angles between the nectaries and carpels, indicating that this character is not stable and that it lacks the taxonomic significance assigned by Pančić (1874) and Preager (1932).According to the characteristic shape of nectaries and the distance between two nectaries, it can be concluded that there are two distinct morphological types of nectaries that can be used for grouping the analyzed populations and that are commonly observed in individuals of these analyzed populations.The first type of nectaries is those with two teeth, very close to each other at a distance smaller In a taxonomic sense, the characteristics of nectaries were not the cause of differentiation between the taxa Sempervivum kopaonikense and S. patens, but the semiquantitative characters shape of nectary and distance between nectaries were shown as the most important in the separation of populations into two groups, indicating that they may be used as additional characters in the analysis and taxonomy of the J. heuffelii group.Were the study area increased to include the whole range of the species, then knowledge about the variability of nectaries would definitively be better.Further research would need to include the qualitative characteristics of nectaries (production of nectar and its quantitative-qualitative characteristics) as well as the pollination process.
d e 0.50 b c d 42.84 b c 2.89 a b c 7 SR-Basarski kamik 0.84 d e 0.55 d 39.22 b c 2.21 a 8 SR-Radan 0.75 c d e 0.53 b c d 30.86 a b 3.00 a b c 9 SR-Besna Kobila 0.72 c 0.52 c d 32.89 a b 2.60 a b c 10 SR-Stara planina 0.74 c d 0.47 b c d 42.86 b c 2.70 a b

Table 1 .
Essential characteristics of the localities from which populations of J. heuffelii were obtained.Vouchers were deposited in the Herbarium at the Institute of Botany, Faculty of Biology, University of Belgrade (BEOU).

Table 2 .
Results of descriptive statistics for the analyzed characters of nectaries.

Table 3 .
Results of Unequal N HSD test for the variables: width, height, angle and shape of nectaries.Different letters(a, b, c, d)indicate statistically significant differences of variable means at p<0.05.

Table 4 .
Factor loadings of the variables and eigenvalues of the first two axes of canonical discriminant analysis (CDA) for Aall analyzed populations and B -Sempervivum kopaonikensis vs. S. patens.