(Berberis L.) from the Balkan Peninsula and Sicily

(2020) 44 (2): 137-148

DOI: https://doi.org/10.2298/BOTSERB2002137K journal homepage: botanicaserbica.bio.bg.ac.rs

Original Scientific Paper Morphological variability of leaf and shoot traits of four barberry taxa (Berberis L.) from the Balkan Peninsula and Sicily

Dario Kremer1, Renata Jurišić Grubešić1, Faruk Bogunić2, Eleni Elefheriadou3, Dalibor Ballian2, Ivan Kosalec1, Marko Randić4, Jadranka Vuković Rodríguez1 and Ksenija Karlović5✳

1 University of Zagreb, Faculty of Pharmacy and Biochemistry, A. Kovačića 1, HR-10000 Zagreb, Croatia 2 University of Sarajevo, Faculty of Forestry, Zagrebačka 20, BIH-71000, Bosnia and Herzegovina 3 Faculty of Forestry and Natural Environment, St. Illia Chatzakou 59, GR-54124 Salonika, Greece 4 Public institution “Priroda”, Grivica 4, HR-51000 Rijeka, Croatia 5 University of Zagreb, Faculty of Agriculture, Svetošimunska 25, HR-10000 Zagreb, Croatia ✳ correspondence: [email protected]

Keywords: ABSTRACT: Berberidacae, Berberis aetnensis, Leaf and shoot characteristics of the following four European barberry taxa Berberis cretica, Berberis croatica, from the Balkan Peninsula and Sicily were investigated in the present study: Berberis vulgaris, morphology, Berberis croatica, B. vulgaris, B. aetnensis and B. cretica. Analyses were based on morphometry, multivariate analysis, 10 populations of B. croatica, five ofB . vulgaris and two populations of both B. taxonomy aetnensis and B. cretica. Populations were randomly selected within the natural distribution area of these species. Eight leaf traits, three shoot traits and the blade length/width ratio were analysed. Multivariate analysis (principal component analysis, canonical discriminant analysis and cluster analysis) distinguished B. cretica and B. aetnensis populations and, to a lesser extent, the populations of B. croatica and B. vulgaris. ANOVA showed that the analysed populations of both B. aetnensis and B. cretica were homogeneous within the species. All populations of both B. croatica and B. vulgaris showed different degrees of intraspecies variability. Lack of complete separation, the observed grouping of populations and high intraspecies variability in B. vulgaris and B. croatica may reflect the fact that the sampled B vulgaris and B.croatica populations were located at environmentally variable sites (unlike B. aetnensis and B. cretica), resulting in high phenotypic plasticity in those populations. Even though the observed patterns of morphological variation support the idea of four barberry UDC: 582.675.3(292.464+450.82):581.4 taxa on the Balkan Peninsula and in Sicily, because of suspected adaptive phenotypic plasticity of the analysed Berberis taxa, the true taxonomic status of Received: 21 November 2019 these taxa needs to be additionally confirmed by molecular methods. Revision accepted: 06 March 2020

Introduction cretica L. were randomly selected on the basis of published data (Kušan 1969; Phitos & Strid 2002; Kar- The genusBerberis L. includes about 500 species native lović et al. 2009) within the area of natural distribu- to Asia, North Africa, America and Europe (Ahrendt tion of these species (Fig. 1A). Species identification was 1961). Taxonomically, Berberis is considered to be a very based on descriptions and keys provided by Webb (1964), complex genus and has even been called a “taxonomic Kušan (1969), Trinajstić (1973), Pignatti (1982) and black hole” (Landrum 1999), with variable characters Akeroyd & Webb (1993). On the basis of morphologi- in its species (Sodagar et al. 2012). Many authors used cal characters, two of the populations (one from Greece morphological traits to examine inter- and intraspecies and one from Bosnia and Herzegovina) were assigned varibility of Berberis, mostly on species growing in South to different taxa after the identification procedure. The America (Bottini et al. 1998; Landrum 1999; Arena et population from Greece that was reported as B. vulgaris al. 2011; Radice & Arena 2015; Giordani et al. 2017), in the literature was treated as B. croatica (Bc_Gr), and but also in Europe and elsewhere (Rivas-Martínez et the population from Bosnia and Herzegovina that was al. 1985; Karlović et al. 2009; Jannatizadeh & Khad- reported to be B. croatica in the literature was treated as ivi-Khub 2016). As a result of the large size of the genus B. vulgaris (Bv_Os; Table 1). and its ubiquitous distribution, the number of barberry Ten plants were sampled from each population, ex- species in the world and in Europe remains controver- cept for the small populations Bc_Cr and Bc_Sn, which sial. In 1931, Dermen cited Berberis vulgaris L. as the included samples from seven and six plants, respective- only European species, while Rikli (1946) mentioned ly. In total, 183 individuals were surveyed in the study. four subspecies in the European Mediterranean area: B. Voucher specimens of plant material were deposited in vulgaris subsp. hispanica, B. vulgaris subsp. aetnensis, the “Fran Kušan” Herbarium, maintained by the Facul- B. vulgaris subsp. cretica and B. vulgaris subsp. cratae- ty of Pharmacy and Biochemistry, University of Zagreb, gina. Later Webb (1964) recognised four European bar- Croatia. berry species: B. vulgaris L., B. aetnensis C. Praesl, B. The following eight leaf traits were measured on dry hispanica Boiss. & Reut. and B. cretica L. Around the and pressed plant material: number of leaves on short same time, some authors suggested that Berberis croat- shoots, blade length and width, number of teeth on the ica Horvat, was a separate species occurring in Croatia left and right sides of the blade, length of the longest (Kušan 1969), Bosnia and Herzegovina (Šilić 1996), teeth, the greatest distance between two teeth and peti- the Republic of North Macedonia (Trinajstić 1973) ole length. Blade length and width were used to calculate and Montenegro (Grlić 1979). Other authors, in con- the blade length/width ratio. A higher ratio indicates a trast, recognised B. croatica as a subspecies of B. vulgaris narrower blade, while a lower ratio indicates a round- (Anić 1946) or B. aetnensis (Anić 1946; Forenbacher er blade. In addition, three shoot traits were measured: 1990). More recently, Akeroyd & Webb (1993) included number of spines on five consecutive nodes starting only B. vulgaris and B. cretica as European barberry spe- from the twig top, spine length and internode length. cies, while considering B. aetnensis and B. hispanica as subspecies of B. vulgaris. Berberis croatica, on the other Data analysis. Descriptive statistics of each morpholog- hand, was not even listed as a European barberry species ical trait were calculated to obtain basic parameters of by Akeroyd & Webb (1993), while a database like IPNI the studied populations/taxa. Morphological variation (2020) does not list it and the Euro+Med (2006) data- of the sampled taxa was evaluated using principal com- base lists it as a synonym for B. vulgaris. ponent analysis (PCA), cluster analysis (CA) and canon- In this way, the number and classification of Berberis ical discriminant analysis (CDA). Leaf and shoot traits taxa in Europe remains controversial due to their mor- were averaged for each individual to construct two data phological diversity and extensive distribution. Berberis matrices: (1) the “population matrix” and (2) the “indi- vulgaris seems to be the only European barberry species vidual matrix”. The “population matrix” was based on whose taxonomic status is not in doubt. To clarify un- means of all morphological characters at the popula- certainties in the classification of other barberry species, tion level regardless of their taxonomic affiliation. The the present study analysed leaf and shoot morphological populations were used as units in PCA and CA in or- traits of the barberry taxa from the Balkan Peninsula der to pre-specify their affinities for a taxonomic group. and Sicily, i.e., B. croatica, B. vulgaris, B. aetnensis and On the other hand, the “individual matrix” was based B. cretica. on individual means of morphological characters using individual plants as units in PCA and CDA. PCA per- Materials and methods formed on the “individual matrix” aimed at revealing the overall pattern of morphological variation and rela- Plant material. Ten populations of Berberis croatica tionships among individuals originating from the speci- Horvat, five populations ofB . vulgaris L., two popula- fied groups. CDA based on the “individual matrix” with tions of B. aetnensis C. Presl and two populations of B. four groups was performed to determine morphological D. Kremer et al.: Morphological variability of Berberis taxa | 139 traits discriminating the studied taxa and to classify leaf trait was the blade length/width ratio (ranging from each individual into an a priori specified group (taxon). CV = 15.3% in Bae_Li to CV = 28.2% in Bc_Ca). Prior to running analyses, all data were standardised The PCA performed on the “population matrix” due to different scales of character scoring Quinn( & showed separation of populations mostly corresponding Keough 2009). Spearman and Pearson correlation co- to their taxonomic affiliation (Fig. 1 B). The first three efficients were calculated to find very highly correlat- components accounted for 82.24% of the total variance: ed character pairs since they may distort the results of 49.91% for PC1, 19.74% for PC2 and 13.29% for PC3 (Ta- discriminant analysis (Legendre & Legendre 1998). ble 2). Several traits (none of them being highly correlat- Since the characters blade length and number of teeth ed), viz., blade length, number of teeth on the right side on the left side were highly intercorrelated with blade and blade width, contributed the highest value for PC1. width and number of teeth on the right side, respec- Spine length, number of leaves and distance between tively (r>0.8), the latter ones were excluded from CDA. teeth contributed most to the second PC axis, while the The PCA of populations/individuals was computed on maximum score for PC 3 was obtained for length of the correlation matrix of all scored traits, and the axes teeth and the blade length/width ratio (Table 2). corresponding to principal components with Eigen val- The PCA based on the “individual matrix” displayed ues >1 were retained in analysis. Cluster analysis by the a similar pattern as for the “population matrix” (Fig. 1D). unweighted pair-group method with arithmetic means In a biplot of PC1 and PC2, B. aetnensis and B. cretica (UPGMA) was calculated using Euclidean distances showed a separation from B. croatica and B. vulgaris (Fig. (Sokal & Rohlf 2003). CDA computation was based on 1B). Several individuals of B. cretica intermingled with Mahalanobis distances of 10 variables. The relationships individuals of B. croatica and vice-versa. Along the first among individuals of the studied taxa were visualised on axis, B. croatica differentiated fromB. vulgaris, but these the bidimensional plot of discriminant function scores. two groups were not clearly separated (Fig. 1D). Individ- The classification discriminant function was then de- uals mostly from the populations Bv_Pr, Bc_Me and Bv_ rived and used to classify each individual into one of CL were dispersed between the two main clusters. the a priori determined groups by the cross-validation Groupings of populations into taxa were also validat- procedure. Between-population variation was evaluated ed by cluster analysis based on Euclidean distances. Al- by analysis of variance (ANOVA) and Scheffe’s post-hoc most all populations were clearly defined as falling into test, with p<0.05 defined as the threshold of significance. the corresponding taxon. An exception was the Bv_Pr Descriptive statistics, CDA and cluster analysis com- population (Fig. 1C), which was connected with B. cro- putations were performed in Statistica 7 (StatSoft Inc. atica accessions. According to the conducted cluster 2004), while PCA was done in PAST, ver. 3.14 (Hammer analysis, the sampled populations could be divided into 2016). The PCA and CDA graphs, phenogram and distri- four main clusters. The first included B. vulgaris pop- bution map were edited for better performance in Adobe ulations. The second cluster grouped populations ofB. Illustrator CS6. cretica, the third included populations of B. aetnensis. The fourth cluster was represented by B. croatica pop- RESULTS AND DISCUSSION ulations, including the Bv_Pr population. The constant intermixing (both in PCA and CA) of the Bv_Pr popula- From descriptive statistics of the analysed traits (Supple- tion with B. croatica accessions might be an indicator of mentary Material 1), it can be seen that B. cretica differed spontaneous hybridisation and introgression in the past, from the other taxa in having the smallest and narrow- which affected formation of the Prokike population. At est leaves with almost no teeth. Berberis aetnensis, on the the same time, the Prokike population grows in envi- other hand, differed from the other taxa in having the ronmental conditions which are more similar to those longest spines. Populations of B. croatica were generally of B. croatica (although at a low altitude of 593 m a.s.l., characterised by short spines; however, that was not a uni- the population inhabits a slope on shallow and stony soil form trait since the Bosnian Kamešnica population (Bc_ with southeastern exposition in drier conditions on an Ka) was an exception to the rule. Even though possession open, wind-exposed site with more sun available). of the the largest leaves was a distinctive trait for some The CDA model and discrimination were signifi- populations of B. vulgaris, that was not a uniform feature cant (Wilks’ λ = 0.011, df = 27, p<0.001). The first three throughout B. vulgaris populations. Shoot traits were canonical axes extracted 51.80, 28.70 and 19.50% of generally less variable than leaf traits. The most variable total variance among four groups (Table 3). The CDA shoot trait was spine number, for which CV ranged from based on nine morphological characters resulted in 25.2% in Bcre_O to 105.7% in Bv_Os. The least variable four groups separated along three canonical axes. The shoot trait was internode length, with CV ranging from first two axes clearly separated B. cretica from the other 14.4% in Bae_MV to 37.0% in Bc_Bl. The most variable species (Fig 1E). Berberis croatica and B. vulgaris repre- leaf trait was petiole length (ranging from CV = 38.8% in sented two groups with some intermixing individuals, Bc_Cr to CV = 81.2% in Bv_RP), while the least variable while B. aetnensis was weakly separated from B. vulgar- 140 | vol. 44 (2)

Fig. 1. Geographical distribution of the analysed Berberis populations (A); PCA ordination of the investigated Berberis populations (B); UPGMA dendrogram of the investigated populations of Berberis taxa (C); PCA ordination of the investigated Berberis spp. individuals (D); scatterplot of canonical scores on the first and second (E), and first and third (F) canonical axes for the investigated Berberis individuals. D. Kremer et al.: Morphological variability of Berberis taxa | 141

Table 1. Origin and collection data of investigated Berberis L. spp. samples.

Population Number of Voucher no. Latitude; Altitude Abbreviation (Collectors´ names and sampling date) analysed plants Longitude m a.s.l. B. croatica – Vela Pliš, Croatia 10 HFK-HR-221–2015 N 45°23’; E 14°35’ 1141 Bc_VP (Randić, Karlović, Kremer; July 2015) B. croatica – Međuvrhi, Croatia 10 HFK-HR-222–2015 N 45°26’; E 14°34’ 1396 Bc_Me (Randić, Karlović, Kremer; July 2015) B. croatica – Obli Vrh, Croatia 10 HFK-HR-223–2015 N 44°44’; E 15°01’ 1515 Bc_Ov (Kremer, Kosalec; August 2015) B. croatica – Šatorina, Croatia 10 HFK-HR-224–2015 N 44°38’; E 15°02’ 1610 Bc_Sa (Kremer, Kosalec; August 2015) B. croatica – Crnopac, Croatia 7 HFK-HR-225–2015 N 44°15’; E 15°50’ 1350 Bc_Cr (Kremer, Kosalec; August 2015) B. croatica – Sniježnica, Croatia 6 HFK-HR-226–2015 N 42°34’; E 18°21’ 1125 Bc_Sn (Karlović, Kremer; July 2015) B. croatica – Kamešnica, Bosnia and Herzegovina 10 HFK-HR-142–2016 N 46°11’; E 15°54’ 1421 Bc_Ka (Ballian, Bogunić; August 2016) B. croatica – Blidinje, Bosnia and Herzegovina 10 HFK-HR-143–2016 N 45°54’; E 15°55’ 1210 Bc_Bl (Ballian, Bogunić; August 2016) B. croatica – Čabulja, Bosnia and Herzegovina 10 HFK-HR-144–2016 N 45°38’; E 15°56’ 1626 Bc_Ca (Ballian, Bogunić; August 2016) B. croatica – Mt Grammos, Greece 10 HFK-HR-278–2015 N 40º18’; E 20°55’ 1970 Bc_Gr (Elefheriadou; July 2015) B. vulgaris – Rakov Potok, Croatia 10 HFK-HR-263–2015 N 45°44’; E 15°47’ 130 Bv_RP (Karlović, Kremer; July 2015) B. vulgaris – Skrad, Croatia 10 HFK-HR-264–2015 N 45°25’; E 14°54’ 695 Bv_Sk (Karlović, Kremer; July 2015) B. vulgaris – Crni Lug, Croatia 10 HFK-HR-265–2015 N 45°25’; E 14°42’ 710 Bv_CL (Randić, Karlović, Kremer; July 2015) B. vulgaris – Prokike, Croatia 10 HFK-HR-266–2015 N 44°59’; E 15°04’ 593 Bv_Pr (Kremer; July 2015) B. vulgaris – Ostrožac, Bosnia and Herzegovina 10 HFK-HR-141–2016 N 45°40’; E 15°45’ 370 Bv_Os (Ballian, Bogunić; August 2016) B. aetnensis – Mt. Vetore, Etna, Italy 10 HFK-HR-75–2015 N 37°70’; E 14°98’ 1700 Bae_MV (Karlović; July 2015) B. aetnensis – Linguaglossa, Etna, Italy 10 HFK-HR-76–2015 N 37°83’; E 15°13’ 1730 Bae_Li (Karlović; July 2015) B. cretica – Kanto Olympos, Kallipefki, Greece 10 HFK-HR-126–2015 N 39°58’; E 22°28’ 1074 Bcre_K (Elefheriadou; July 2015) B. cretica – Olympos, Kokkinopilos, Greece 10 HFK-HR-127–2015 N 40°05’; E 22°15’ 1071 Bcre_O (Elefheriadou; July 2015) 142 | vol. 44 (2)

Table 2. Component loadings for the first three principal Table 3. Standardised coefficients for canonical variables. components.

Variable PC1 PC2 PC3 Variable DF1 DF2 DF3 Number of leaves 0.089 0.477 0.148 Internode length -0.43 -0.63 -0.28 Blade length 0.393 0.043 -0.138 Spine length -0.39 -0.38 0.79 Blade width 0.394 -0.047 -0.043 Length of teeth -0.24 0.35 0.26 Blade length/width -0.182 0.358 -0.450 Distance between teeth -0.89 0.71 -0.12 Petiole length 0.357 0.037 -0.121 Number of teeth – left -1.28 0.47 0.19 Number of teeth – left 0.387 0.128 0.072 Blade length 0.76 -0.64 0.13 Number of teeth – right 0.390 0.121 0.065 Petiole length 0.23 -0.07 -0.46 Length of teeth 0.129 0.053 0.725 Blade length/width -0.12 -0.26 0.17 Distance between teeth 0.070 -0.462 0.231 Number of spines 0.27 0.04 0.05 Number of spines -0.271 0.281 0.173 Eigenvalue 5.71 3.28 2.04 Spine length -0.067 0.536 0.255 Cumulative proportion (%) 51.80 80.50 100 Internode length 0.343 0.153 -0.232 Eigenvalue 5.90 2.36 1.59 % variance 49.21 19.74 13.29

Table 4. Classification matrix of correctly classified individuals (N).

% of correctly B. croatica B. vulgaris B. aetnensis B. cretica classified cases B. croatica 90 0 2 1 96.7 B. vulgaris 4 45 1 0 90 B. aetnensis 1 0 19 0 95 B. cretica 0 0 0 19 100 Total 95 45 22 20 –

is (Fig. 1E). However, ordination of the first and third cation (≥90%). Only B. cretica resulted in a 100% correct axes displayed clear separation of B. aetnensis and B. classification, while 95% (N = 19) of B. aetnensis indi- cretica from the two other species (Fig. 1F). Although viduals were correctly determined (Table 4). A higher B. croatica and B. vulgaris represented two groups mor- percentage of misclassified individuals was evident for phologically, a portion of individuals from Bv_Pr inter- B. vulgaris (10%, N = 5) and B. croatica (13%, N = 3). The mixed within B. croatica. The first function was mostly greatest number of misclassified individuals originated determined by the following traits: number of teeth on from the Bv_Pr population. the left side, distance between teeth and blade length. It is evident that the Greek population of B. croatica The second was mostly determined by internode length, (Bc_Gr), which according to literature sources belongs to blade length and distance between teeth, and the third B. vulgaris, noticeably clustered with B. croatica popula- by spine length (Table 3). tions. This fact might be interpreted in two ways. Firstly, B. Classificatory discriminant analysis confirmed the croatica has a larger natural distribution than is currently CDA results and yielded a high rate of correct classifi- believed (Kušan 1969; Trinajstić 1973; Grlić 1979; Šilić D. Kremer et al.: Morphological variability of Berberis taxa | 143

Table 3. Standardised coefficients for canonical variables. 1996). Secondly, if we traditionally treat this population differences in the majority of analysed shoot traits. The as B. vulgaris, then it inhabits contrasting environmental shoot trait showing the greatest difference among the in- conditions, which presumably induced phenotypic traits vestigated populations was spine length, while the num- of the plants. Such changes commonly occur in nature un- ber of spines showed the smallest difference. der different selection pressures Gratani( 2014). The Bc_ Our results indicate that the population Bv_Os, first Gr population featured small leaves with few leaves per described as B. croatica by Kušan (1969), on the basis shoot, fewer teeth on the blade edge and short internodes. of leaf and shoot traits belongs to B. vulgaris and not B. These traits are typical of extreme harsh climates such as croatica. Conversely, the population Bc_Gr (tradition- those in mountainous areas. Climate is considered to be a ally considered as B. vulgaris) on the basis of the ana- primary selective force inducing leaf morphology change lysed morphological traits belongs to B. croatica and not (Wilf 1997). Plants growing in drier climates (such as al- B. vulgaris. However, the finding that the population pine environments) tend to have smaller leaves (Bonan collected in Greece (Bc_Gr) grouped with the other B. 2002) to reduce evaporative cooling, while larger leaves croatica populations and not with B. vulgaris may also are common in more humid climates because the result- reflect the fact that Bc_Gr grows at 1970 m a.s.l. under ing water loss is less critical (Givnish 1984). Svriz et al. environmental conditions that are more similar to those (2014) confirmed differential action of light on the growth of B. croatica populations, which grow above 1000 m of B. darwinii Hook. To be specific, the authors found that a.s.l., than to those of the other B. vulgaris populations shoot internode length was smaller in plants growing in in the present research, all of which grow at 130–710 forest gaps than in those growing at the forest edge or m a.s.l. The Prokike population of B. vulgaris (Bv_Pr), under the tree canopy. Regarding the leaf margin, plants which grows at a lower altitude (593 m a.s.l.) but in dri- with toothed leaves have growth advantages, especially in er conditions on open wind-exposed ground with more non-optimal environments. Leaves with more teeth show available sunlight, differed from the other B. vulgaris more active photosynthesis; however, they are disadvan- populations and clustered with B. croatica populations. tageous in xeric, water-stressed environments because of The Bv_Pr population, just like Bc_Gr, featured more higher transpiration (Xu et al. 2008). The size and number compact leaves with fewer teeth and shorter internodes of teeth correlate negatively with mean annual tempera- than the other B. vulgaris populations. The clustering of ture (Royer et al. 2005, 2009; Royer & Wilf 2006). populations and species observed here probably reflects ANOVA was carried out for leaf traits (Table 5) and high phenotypic plasticity of the examined Berberis taxa. shoot traits (Table 6) to gain insight into differences be- Phenotypic plasticity is the capability of a genotype to tween populations for the investigated characters. In produce diverse phenotypic expression under different general, leaf traits differed among populations to a great- environmental conditions (Radice & Arena 2018), act- er extent than shoot traits. The morphological variability ing to increase the performance of plants under stress noted in this research was also confirmed byGoodarzi (Xu et al. 2008). Berberis microphylla grown under envi- et al. (2018), who documented high variations among ronmental conditions of higher temperatures and weaker the studied accessions of B. vulgaris var. asperma Willd. sunlight showed modifications in leaf morphology and for most of the phenotypic, pomological and chemical structure (Radice & Arena 2015). In the more recent re- traits analysed. The differences observed in the present search conducted by Radice et al. (2018), B. microphylla analysis may be due to the small number of analysed showed pronounced phenotypic plasticity in most of the B. cretica and B. aetnensis populations and their distri- variables studied. These findings were further reinforced bution at similar altitudes and under similar environ- by Radice & Arena (2018), who reported adaptive plas- mental conditions, in contrast to the larger number of ticity of B. microphylla in regard to the adjustment of its sampled B. croatica and B. vulgaris populations, which morphological but also physiological characteristics to are found over a wider range of altitudes and growing the environmental conditions of growth. Giordani et al. conditions. As Giordani et al. (2017) noted for B. mi- (2017) reported significant correlation among some mor- crophylla G. Forst., short geographic distance and sim- phological characters of Berberis microphylla with envi- ilar site characteristics can influence plant morphology. ronmental factors. For example, light intensity exerted To support this thesis, there are reports of plants with significant influence on leaf roundness and compactness, characters intermediate between B. aetnensis and B. vul- altitude affected leaf elongation and wind influenced leaf garis found in the Alps and Southern France, while some compactness. Even inclination of a terrain influenced results even distinguish the typical B. aetnensis found morphological traits: leaf elongation was higher in plants in Italy and Sicily from plants growing in Corsica and sited in an inclined area as opposed to flat areas. Follow- Sardinia (Webb 1964). ing this line of thought, we feel that marked elongation of The number of teeth on the blade edge was the trait the spines of B. aetnensis could also be an expression of showing the greatest difference among populations, while phenotypic plasticity. Spine length has often been linked tooth length showed the smallest difference. Most of the with the protection of leaves. According to Kariyat et B. croatica and B. vulgaris populations showed significant al. (2017), spines may play a significant role in defence 144 | vol. 44 (2) Bcre_K ABCD Bae_Li ABCDE A ABCD ABCDE Bae_MV BDE BCD BCDE BCDE Bv_Os ABE ADE Bv_Pr ABCD ABCDE ABCDE ABE BCD ABDE ABCD ABCD Bv_CL ABCDE D ABD Bv_Sk ABDE ABDE ABCD ABCD ABCDE DE ABDE ABDE ABDE ABCD ABCD Bv_RP ABCDE CE CD ABD ABD ABCE Bc_Gr ABCD ABCD ABCDE ABCDE DE CDE ABD BCDE ABDE ACDE Bc_Ca ABCD ABCD ABCDE ABCDE ABE ABD Bc_Bl ABCE ABCE ABDE ABDE ABDE ABDE ABCD ABCD ABCDE BC BCD BCD ABD ABD ABDE ABDE ABDE ABDE ABDE Bc_Ka ABCD ABCDE CE DE DE ABD ABD ABD ABD ACD ACD Bc_Sn ABDE ABDE ABCD ABDCE E E D D AB ABD ABD ABDE ABDE Bc_Cr ACDE ABCD ABCD ABCD ABCDE D CD ABE ACE ABC ABC BCD Bc_Sa ABCE BCDE ACDE ABCD ABCD ABCD ABCD ABCDE C DE AB CD BCE BDE ABD ABD BCD ABD ABCE ABCD ABCD ABCD Bc_Ov ABCDE ABCDE BCE ACE ABC ABC ABD ABCE BCDE ABDE ABCD Bc_Me ABCDE ABCDE ABCDE ABCDE ABCDE ABCDE ABCDE ABCDE D CE DE CD CDE ABD ABD ABD ACD BCDE ACDE ABCD ABCD ABCD Bc_VP ABCDE ABCDE ABCDE ABCDE . Between-population variability leaf for traits: blade length blade width blade length/width (A), (B), number ratio teeth of (C), right on and length side (D) teeth of The (E). Bc_Bl Bc_Sa Bv_Pr Bv_Sk Bc_Sn Bc_Cr Bc_Gr Bc_Ka Bv_Os Bc_Ca Bv_RP Bv_CL Bae_Li Bc_Ov Bc_Me Bcre_K Popula. Bcre_O Bae_MV Table 5 Table presence of a letter indicates significance at p≤0.05.For abbreviations,1. Table see D. Kremer et al.: Morphological variability of Berberis taxa | 145 Bcre_K B B Bae_Li B B Bae_MV BC AC AC ABC Bv_Os C AC AC ABC ABC Bv_Pr C AC ABC ABC ABC ABC Bv_CL C C AC AC ABC ABC Bv_Sk B C C BC BC ABC ABC ABC ABC Bv_RP B BC BC AC AC AC ABC ABC Bc_Gr C C A AB AB AB BC AB BC BC Bc_Ca C AB AB AB BC AB BC BC BC ABC Bc_Bl B B B C A C C A A AB BC AC Bc_Ka B B C C A AB AB BC AB BC BC Bc_Sn B C C AB BC AB BC BC BC AB AC ABC Bc_Cr B A A A A BB BC AB AB AC AC ABC ABC ABC Bc_Sa B B B B A A BC AB AB AC AC ABC ABC Bc_Ov B B B B C C A A A BC AB BC AC AC ABC Bc_Me B C A AB AB AB BC AB BC BC BC AB AB AB Bc_VP Bc_Bl Bc_Sa Bv_Pr Bv_Sk Bc_Sn Bc_Cr Bc_Gr Bc_Ka Bv_Os Bc_Ca Bv_RP Bv_CL Bae_Li Bc_Ov Bc_Me Bcre_K Popula. Bcre_O Bae_MV Table 6. Between-populationTable variability shoot for traits: number spine spines length of (A), and internode (B) lengthpresence The letterof a (C). indicates significance at p≤0.05. abbreviations,For 1. see Table 146 | vol. 44 (2)

against insect herbivores by restricting herbivore move- these taxa in order to improve performance in a stressful ment and increasing the time taken to access feeding environment. Definitive conclusions about taxonomic sites. Moreover, Gowda & Raffaele (2004) reported relationships between these taxa require further genetic that spines are significantly longer in three Berberis spe- analysis using appropriate molecular methods, but also cies re-sprouting after fire compared to before fire. The cultivation of the researched taxa under the same envi- combination of longer spines after fire with no elongation ronmental conditions. of leaves in burned plants results in a significantly higher portion of leaves protected by the spines. Because both Acknowledgements − This work was supported by the investigated B. aetnensis populations are growing on Mt. Ministry of Science, Education and Sports of the Re- Etna, in a region frequently crossed by lava flows, it is public of Croatia (project no. 006–0000000–3178) and possible that one of the traits that distinguished B. aetn- the Ministry of Education, Science and Youth of Canton ensis from the other taxa, i.e., long spines, is merely an Sarajevo (grant no. 11/05-14-277750-1/19). We thank A. expression of adaptive phenotypic plasticity. As reported Chapin Rodríguez for helpful comments on the manu- by Sodagar (2012), spines and margins of leaves are very script and for improving the English style. changeable in Berberis. The morphological analysis of leaves and shoots de- References scribed here differentiated populations of B. cretica and B. aetnensis, but also B. croatica and B. vulgaris popu- Ahrendt L. 1961. Berberis and Mahonia, a taxonomic lations, even though to a lesser extent. The observed revision. Botanical Journal of the Linnean Society 57: partial overlap of B. croatica and B. vulgaris populations 1–410. and the grouping of Bv_Pr and Bc_Gr populations with Akeryod JR & Webb A. 1993. Berberis L. In: Tutin TG, B. croatica likely reflects environmental factors and site Burges NA, Chater AO, Edmondson JR, Heywood adaptation. There are two possible explanations for the VH, Moore DM, Valentine DH, Walters SW case of the Bc_Gr population from Greece, which be- & Webb DA (eds.), Flora Europaea 1, Psilotaceae to longs to B. vulgaris according to published sources and Platanaceae, 2nded, pp. 295–296, Cambridge University to B. croatica according to morphological traits: either Press, Cambridge. the distribution of B. croatica is broader than conven- Anić M. 1946. Dendrologija. In: Šafar J (ed.), Šumarski tionally considered, or else it is possible that B. vulgaris priručnik, pp. 475–582, Poljoprivredni nakladni zavod, shows pronounced adaptability to environmental con- Zagreb. ditions, as reflected in high phenotypic plasticity. Be- Arena ME, Giordani E & Radice S. 2011. Flowering, cause of the suspected adaptive plasticity of the analysed fruiting and leaf and seed variability in Berberis Berberis taxa, morphological characters are probably buxifolia, a native Patagonian fruit species. In: Marin not enough or could even be misleading for recognition L & Kovac D (eds.), Native species, pp. 1-20, Nova of the given taxa without using appropriate molecular Science Publishers Inc., New York. methods. In order to test the hypothesis presented here, Bonan G. 2002. Ecological climatology. Cambridge further investigation should concentrate on cultivation University Press, Cambridge. of the analysed taxa under the same environmental con- Bottini MCJ, Orsi MC, Greizerstein EJ & Poggio L. ditions to assess how much of the observed variability is 1998. Relaciones fenéticas entre las especies de Berberis site-related and how much is species-specific. (Berberidaceae) del Noroeste de la región Patagónica. Darwiniana 35: 115–129. CONCLUSIONS Dermen H. 1931. A study of chromosome number in two genera of Berberidacae: Mahonia and Berberis. Multivariate analyses of leaf and shoot traits successful- Journal of the Arnold Arboretum 12: 281–287. ly distinguished four investigated Berberis taxa from the Euro+Med. 2006–. Euro+MedPlantBase– the Balkan Peninsula and Sicily. It clearly differentiated the information resource for Euro-Mediterranean plant B. cretica and B. aetnensis populations and, to a lesser diversity. Published on the Internet - http://ww2. degree, those of B. croatica and B. vulgaris. A NOVA re- bgbm.org/EuroPlusMed/ http://ww2.bgbm.org/ vealed that two populations of B. cretica and two popu- EuroPlusMed/ [Accessed on January 28th 2020] lations of B. aetnensis were similar within species limits, Forenbacher S. 1990. Velebit i njegov biljni svijet. while all populations of both B. croatica and B. vulgaris Školska knjiga, Zagreb. varied to a greater or lesser extent. Similar site charac- Giordani E, Müller M, Gambineri F, Paffetti teristics could be a factor influencing the observed uni- D, Arena M & Radice S. 2017. Genetic and formity of B. aetnensis and B. cretica. Equally, diverse morphological analysis of Berberis microphylla G. phenotypic expression of the B. vulgaris and B. croatica Forst. accessions in southern Tierra del Fuego. Plant populations under different environmental conditions Biosystems 151: 715–728. could represent a manifestation of adaptive plasticity of D. Kremer et al.: Morphological variability of Berberis taxa | 147

Givnish TJ. 1984. Leaf and canopy adaptations in Radice S & Arena ME. 2018. Reproductive shoots of tropical forests. In: Medina E, Mooney HA & Berberis microphylla G. Forst. in relation with the floral Vasquez-Yanes C (eds.), Physiological ecology of bud development and the fruit set. Heliyon 4: e00927. plants of the wet tropics, pp. 51–84, Dr. W. Junk Rikli M. 1946. Das Pflanzenkleid der MittelmeerLänder. Publishers, Hague. Verlag Hans Huber, Bern. Goodarzi S, Khadivi A, Abbasifar A & Akramian Rivas-Martínez A, Loidi J & Arnáiz C. 1985. Berberis M. 2018. Phenotypic, pomological and chemical L. (Berberidaceae) Hispaniae. Lazaroa 8: 5–9. variations of the seedless barberry (Berberis vulgaris Royer DL, Kooyman RM, Little SA & Wilf P. 2009. L. var. asperma). Scientia horticulturae 238: 38–50. Ecology of leaf teeth: A multi-site analysis from an Gowda J & Raffaele E. 2004. Spine production is Australian subtropical rainforest. American Journal of induced by fire: a natural experiment with three Botany 96: 738–750. Berberis species. Acta Oecologica 26: 239–245. Royer DL & Wilf P. 2006. Why do toothed leaves Gratani L. 2014. Plant phenotypic plasticity in correlate with cold climates? Gas exchange at response to environmental factors. Advances in leaf margins provides new insights into a classic Botany ID 208747: 1–17. paleotemperature proxy. International Journal of Plant Grlić Lj. 1979. Samoniklo jestivo bilje. Prosvjeta, Sciences 167: 11–18. Ljubljana. Royer DL, Wilf P, Janesko DA, Kowalski EA & Hammer Ø. 2016. PAST:3.12. Natural History Museum, Dilcher DL. 2005. Correlations of climate and plant University of Oslo. Available at: http://folk.uio.no/ ecology to leaf size and shape: Potential proxies for ohammer/past/ [Accessed on February 2nd 2020]. the fossil record. American Journal of Botany 92: IPNI 2020. International Plant Names Index. http:// 1141–1151. ipni.org/ [Accessed on January 28th 2020]. Sodagar N, Bahrami AR, Memariani F, Ejtehadi H, Jannatizadeh A & Khadivi-Khub A. 2016. Vaezi J & Khosravi AR. 2012. Biosystematic study Morphological variability of Berberis integerrima of the genus Berberis L. (Berberidaceae) in Khorassan, from Iran. Erwerbs-Obstbau 58: 247–252. NE Iran. Plant Systematics and Evolution 298: 193–203. Kariyat RR, Hardison SB, De Moraes CM & Sokal RR & Rohlf FJ. 2003. Biometry. The principles Mescher MC. 2017. Plant spines deter herbivory by and practice in biological research. 3rd ed., W.H. restricting caterpillar movement. Biology Letters 13: Freeman and Co., New York. 201720176. StatSoft Inc. 2004. Statistica 7. Tulsa, Oklahoma, USA. Karlović K, Kremer D, Liber Z, Šatović Z & Vršek Svriz M, Damascos MA, Lediuk KD, Varela SA & I. 2009. Intra- and interpopulation variability and Barthélémy D. 2014. Effect of light on the growth taxonomic status of Berberis croatica Horvat. Plant and photosynthesis of an invasive shrub in its native Biosystems 143: 40–46. range. AoB Plants 6: plu033. Kušan F. 1969. Nova vrsta žutike (Berberis) u flori Šilić Č. 1996. Spisak biljnih vrsta (Pteridophyta Hrvatske. Acta Botanica Croatica 28: 423–434. i Spermatophyta) za Crvenu knjigu Bosne i Landrum LR. 1999. Revision of Berberis Hercegovine. Glasnik Zemaljskog Muzeja Bosne i (Berberidaceae) in Chile and adjacent southern Hercegovine Prirodne Nauke 31: 323–367. Argentina. Annals of the Missouri Botanical Garden Trinajstić I. 1973. Berberis L. In: Trinajstić I (ed.), 86: 793–834. Analitička flora Jugoslavije 1, pp. 377–381, Šumarski Legendre P & Legendre L. 1998. Numerical ecology. fakultet Sveučilišta u Zagrebu, Sveučilišna naklada 2nd ed., Elsevier, Amsterdam. Liber, Zagreb. Phitos D & Strid A. 2002. Berberis L. In: Strid A & Webb A. 1964. Berberis L. In: Tutin TG, Heywood Tan K (eds.), Flora Hellenica 2, pp. 83–84, A.R.G. VH, Burges NA, Valentine DH, Walters SM & Gantner Verlag, Ruggell. Webb DA (eds.), Flora Europaea 1 Lycopodiaceae to Pignatti S. 1982. Flora d’Italia. Edagricole, Bologna. Platanaceae. 2nd ed., p. 245, Cambridge University Quinn PG & Keough MJ. 2009. Experimental design Press, Cambridge. and data analysis for biologists, 8th ed. Cambridge Wilf R. 1997. When are leaves good thermometers? A University Press, Cambridge. new case for leaf margin analysis. Paleobiology 23: Radice S, Alonso M & Arena ME. 2018. Berberis 373–390. microphylla: A species with phenotypic plasticity in Xu F, Guo W, Xu W & Wang R. 2008. Habitat effects on different climatic conditions. International Journal of leaf morphological plasticity in Quercus acutissima. Agriculture and Biology 20: 2221–2229. Acta Biologica Cracoviensia Series Botanica 50: 19–26. Radice S & Arena M. 2015. Environmental effect on the leaf morphology and anatomy of Berberis microphylla G. Forst. International Journal of Plant Biology 6: 5677. 148 | vol. 44 (2)

Rezime

Morfološka varijabilnost lista i izbojaka četiri taksona žutike (Berberis L.) sa Balkanskog poluostrva i Sicilije

Dario Kremer, Renata Jurišić Grubešić, Faruk Bogunić, Eleni Elefheriadou, Dalibor Ballian, Ivan Kosalec, Marko Randić, Jadranka Vuković Rodríguez i Ksenija Karlović

Istraživane su karakteristike lista i izdanaka četiri taksona roda žutika s područja Balkanskog poluostrva i Sicilije, Berberis croatica, B. vulgaris, B. aetnensis i B. cretica. Analiza je urađena na deset populacija B. croatica, pet populacija B. vulgaris te po dve populacije B. aetnensis i B. cretica. Populacije su nasumično odabrane unutar prirodnog područja rasprostranjenja istraživanih vrsta. Analizirano je osam karakteristika lista, tri svojstva izdanaka i odnos širine/dužine lista. Multivarijantna analiza (analiza glavnih komponenata, kanonička diskriminativna analiza i klaster analiza) ukazuje na razlike populacije vrsta B. aetnensis i B. cretica te, u manjoj meri, populacije vrsta B. croatica i B. vulgaris. ANOVA je pokazala unutarvrsnu homogenost populacija B. aetnensis i B. cretica dok su sve populacije B. vulgaris i B. croatica pokazale različite stepene unutarvrsne varijabilnosti. Nedostatak potpunog razdvajanja, obrazac grupisanja populacija i visoka unutarvrsna varijabilnost kod B. vulgaris i B. croatica mogli bi da odražavaju činjenicu da su uzorkovane populacije ovih dvaju taksona bile locirane na sredinski različitim mestima (za razliku od B. aetnensis i B. cretica) što je rezultovalo visokom fenotipskom plastičnošću populacija. Premda uočeni način morfološke varijacije podupire ideju o postojanju četiri taksona roda Berberis na području Balkanskog poluostrva i Sicilije, adaptivna fenotipska plastičnost je razlog zbog čega bi taksonomski status ovih entiteta trebalo dodatno potvrditi budućim molekularnim istraživanjima.

Ključne reči: Berberidacae, Berberis aetnensis, Berberis cretica, Berberis croatica, Berberis vulgaris, morfologija, morfometrija, multivarijantna analiza, taksonomija