Variability of beech cupules in Serbia

Beech is the most important tree species in forests of Serbia. However, despite its significance, the taxonomic status of beech in Serbia is unclear. Morphology of the cupules can be successfully used to distinguish species and within species taxa. In this paper we report results on cupules morphology, measuring the same attributes reported by Mišić (1955): 1) Length of the longest valve of cupule, 2) Width of that valve (at widest point), 3) Distance between base of the longest valve and peduncle, 4) Length of cupule without peduncle, and 5) Length of peduncle. The length of cupule and peduncle length from 12 populations in Serbia are in the range reported for Fagus sylvatica in Serbia and western Eurasia. Results of cluster analysis shows a grouping of populations in two groups: 1) the southeast group, and 2) group consist of populations from northwest, east and southeast of Serbia. Populations from this southeast group also consist the group of populations on altitude over 850 m, indicating presence of ecotypes. The exception is population from Stara Planina (1,520 m a.s.l.) which is grouped with populations from altitudes under 850 m.

Although Fagus moesiaca is described as a separate species based on biometrical studies (Czeczott 1933) its morphological description is unclear and authors disagree on morphological attributes which separate this species from Fagus sylvatica.Some of used characters are leaf and cupule size and shape (see Jovanović 2000).However, morphotypes which were described as Fagus sylvatica, are well within the range of Fagus sylvatica subsp.sylvatica (Denk 1999).Even in Serbian forestry practice some populations are described as Fagus moesiaca and others as Fagus sylvatica, without any spatial pattern.
Morphology of the cupule/nut complex can be successfully used to distinguish living species of Fagus (Denk and Meller 2001).However, there are just a few researches on cupules size and shape (Mišić 1955(Mišić , 1957;;Istratii 1980;Denk 1999;Denk et al. 2005).This research is a part of a larger project aiming to determine the taxonomic status of beech in Serbia, using anatomical, morphological, and molecular markets.In this paper we report results on cupules morphology, measuring the same attributes reported by Mišić (1955Mišić ( , 1957)).

Material and method
This research is a part of a larger project where beech samples (leaves, buds, and cupules) were sampled from 20 seed trees distanced no less 50 m from each other, in 12 populations in Serbia (Figure 4).During that occasion, cupules were collected according to their abundance, resulting with unequal sample size.In addition, due to lack of masting in populations Golija, Javor, and Tara, only a minimal bulk samples were collected.For analysis of cupules within and between population variability, following attributes were measured (Mišić 1955, Figure 1): 1) Length of the longest valve of cupule, 2) Width of that valve (at widest point), 3) Distance between base of the longest valve and peduncle, 4) Length of cupule without peduncle, and 5) Length of peduncle.Descriptive statistics (mean value, standard variation, minimal and maximal values, and variance), as well as results of ANOVA and Tukey Unequal N HSD multiple comparison test on population and individual level, are given in Appendix 1.Because of a bulk samples of minimal size, populations Golija, Javor, and Tara were excluded from analysis on individual level.We used a Variance Components and Mixed Model ANOVA with "population" and "tree" as a random effects, in a hierarchically nested random effects design to estimate the components of variance.The populations were clustered based on combined mean value of measured attributes using a Joining (Tree Clustering) following the rule of Single Linkage on a non-standardized Euclidean distances.All statistics was done in Statistica 7 (StatSoft 2004) software.

Results
The mean value of length of the longest valve of cupule for all populations is 19.17 mm (SD=3.5),and range from 7.1 mm to 29.8 mm (Appendix 1).The mean value of width of that valve (at widest point) for all populations is 10.44 mm (SD=1.56),and range from 5.3 mm to 21.1 mm.The mean value of distance between base of the longest valve and peduncle for all populations is 5.69 mm (SD=1.53),and range from 1.9 mm to 12.6 mm.The mean value of length of cupule without peduncle for all populations is 24.66 mm (SD=2.99),and range from 10.1 mm to 36.5 mm.The mean value of peduncle length for all populations is 12.90 mm (SD=5.58),and range from 1.4 mm to 35.9 mm.

Senjski rudnik Žagubica
Fruška Gora Stara Planina Boranja Goč Kopaonik Golija Javor Tara Populations with largest cupules are Fruška Gora and Boranja, and populations with smallest are Golija and Tara (Figure 2).The length of peduncle shows the largest variability, followed by length of cupule.Except for length of the longest valve of cupule, individual component of variance overcomes the population component (Figure 3).In addition, except for length of the longest valve of cupule, error is the largest component of variance for the measured attributes of cupules.Results of cluster analysis shows a grouping of populations in two groups (Figure 4 -left).The southeast group of populations consists of Golija, Javor, Kopaonik, Tara, and Goč, while the second group consist of populations from northwest, east and southeast of Serbia (Figure 4 -middle).Populations from this southeast group also consist the group of populations on altitude over 850 m (Figure 4 -right).The exception is population from Stara Planina (1,520 m a.s.l.) which is grouped with populations from altitudes under 850 m (Boranja, Senjski rudnik, Žagubica, Miroč and Fruška Gora).

Single Linkage Euclidean distances
4 Discussion Mišić (1957) reports a wide polymorphism of beech cupules, especially on shape, length and width of valve, length of peduncle, and shape of the cupule base.Although our research did not involve the shape of cupules, our findings support Mišić's (1957) report that length of peduncle is the most variable attribute.
The length of cupule and peduncle length from 12 populations in Serbia are in the range reported for Fagus sylvatica in Bulgaria (Denk 1999) and other morphotypes in western Eurasia, as described by Denk et al. (2005).The length of the longest valve in this research shows the wider range compared to results reported by Mišić (1957), probably because of the larger sample regarding number of populations and cupules measured in our research.
Grouping of populations in this research support the finding of Mišić (1957) on presence of altitudinal ecotypes of beech in Serbia.The exception is population Stara Planina which is grouped with populations from altitudes under 850 m, is consistent with claim that for proper testing of hypothesis about altitudinal ecotypes requires a detailed research of beech populations from different altitude zones inside the same mountain massif (Ivetić 2009).
Results of this research, combined with results of leaf morphology, anatomical markers, and DNA markers can be used in defining of taxonomic status of beech in Serbia.We did not find evidences on existence of a separate species Fagus moesiaca in Serbia and our results suggest that beech in Serbia, or at least in 12 studied provenances, belong to Fagus sylvatica.

Figure 1 .
Figure 1.Measured attributes of cupules: 1) Length of the longest valve of cupule, 2) Width of the longest valve (at widest point), 3) Distance between base of the longest valve and peduncle, 4) Length of cupule without peduncle, and 5) Length of peduncle.

Figure 2 .
Figure 2. Mean values of five measured attributes of beech cupules from 12 populations in Serbia.Vertical bars shows confidence interval (CI=95%).

Figure 3 .
Figure 3. Relative Variance Components (in Percent) for: LLV -Length of the longest valve of cupule, WV -Width of the longest valve (at widest point), DBP -Distance between base of the longest valve and peduncle, LC -Length of cupule without peduncle, and LP -Length of peduncle.