JOURNAL OF FOREST SCIENCE

Transkript

1 Volume 52, No. 8 (2006) JOURNAL OF FOREST SCIENCE KŘEPELA M., PET RÁŠ R.: CONTENS Comparison of Norway spruce (Picea abies [L.] Karst.), Scots pine (Pinus sylvestris L.) and European larch (Larix decidua Mill.) stem shape by means of geometrical methods KULA E., BUCHT A I., STRÁNSKÝ P.: Frost cracks and their effect on the stability of birch stands in the Krušné hory Mts URBAN J.: Occurrence, development and economic importance of Phratora (= Phyllodecta) vitellinae (L.) (Coleoptera, Chrysomelidae)

2 JOURNAL OF FOREST SCIENCE, 52, 2006 (8): Comparison of Norway spruce (Picea abies [L.] Karst.), Scots pine (Pinus sylvestris L.) and European larch (Larix decidua Mill.) stem shape by means of geometrical methods M. Křepela 1, R. Petráš 2 1 Faculty of Forestry and Environment, Czech University of Agriculture in Prague, Prague, Czech Republic 2 National Forest Centre Forest Research Institute Zvolen, Zvolen, Slovak Republic ABSTRACT: In this article the stem shape is compared in three coniferous tree species: Norway spruce, Scots pine and European larch. Stem is investigated by means of geometrical methods. Simplified Bookstein coordinates (stem shape diameters) and Procrustes coordinates were used for variability investigation. The material, originating from the Czech and Slovak territories, involved in total 3,346 spruce stems, 3,082 pine stems and 1,403 larch stems. The accordance of mean stem vectors was assessed by means of Hotelling s T 2 two-sample test. For stem shape diameters and Procrustes tangent coordinates, the variability was examined using the method of principal components analysis. The three most important principal components were diagrammatized and described. The relationship between the stem shape and its size was also investigated, and inflection points of morphological stem curve were described for all three tree species. Keywords: Norway spruce; Scots pine; European larch; stem shape; stem shape diameters; Procrustes coordinates; principal components analysis Stem shape is one of the most important subjects of forest mensuration. Traditional forest mensuration describes the stem shape using form quotients, form series, stem profiles, form factors, geometrical bodies (neiloid, paraboloid, cone). In the last 25 years, not only multivariate morphometrics but also geometrical methods have been developed in biology. These methods use a finite number of points, called landmarks, for description of an object s shape. The landmark is a point of correspondence on each object that matches between and within populations (Dryden, Mardia 1998). The shape is intuitively defined as general geometrical information that remains when the location, scale and rotational effects are filtered out of the object. In this investigation we used simplified Bookstein coordinates (stem shape diameters) and Procrustes coordinates for stem shape illustration. Data processing is given in Fig. 1. Specific location of landmarks on a morphological stem curve enables to simplify the calculation of Bookstein coordinates since the effect of shifting and rotation need not to be removed (Křepela et al. 2005). Two objects have the same shape if they can be transformed, rescaled and rotated to each other so that they match exactly, i.e. if the objects are similar. In morphometry, definition of average shape and structure of shape variability is often necessary in a dataset. For that purpose, we mostly use multivariate analysis of stem shape diameters, generalized Procrustes analysis (GPA) (Gower 1975, Ten Berge 1977 in Dryden, Mardia 1998) and principal components analysis. Stem in traditional forest mensuration is the term related to relative form factors and relative form series. Šmelko (2000) on page 74 presents an example J. FOR. SCI., 52, 2006 (8):

3 remove scale (baseline) Shape (Bookstein coordinates, stem shape diameters) Original Configuration configuration remove translation (new origin centroid of the object) remove scale (centroid size) remove rotation Shape (Procrustes coordinates) Fig. 1. Scheme of shape evaluation of two spruce stems. The first stem is 20 m, the other is 10 m high. These stems have the same relative form series and the same relative form factors at 1/10 of the height. The size of relative form series is removed by means of diameter at 1/10 of the height and in the case of relative form factors by means of the volume of an ideal cylinder. Then according to traditional dendrometry the shape of both stems is the same. But according to geometrical methods the situation is different. In Figs. 2 and 3 both stems, represented by means of stem shape diameters and Procrustes coordinates, are not obviously of the same shape. The aim of this paper is to compare mean stem vectors for three coniferous tree species, to investigate shape variability by means of principal components and to attempt to explain an ecological impact on principal components. The relationship between shape and stem volume will also be investigated as well as inflection points of morphological stem curve. MATERIAL AND METHODS The empirical material involved 3,346 spruce stems, 3,082 pine stems and 1,403 larch stems originating from the Czech and Slovak Republic territories. This material is nearly identical with the material used by Petráš (1989) for the construction of a mathematical model for shape height of stems. In this article stem diameters over bark and stem heights were used for assessments. Diameters were measured in 2m or 1m sections, stump diameter and dbh were also measured. Diameters were measured to the nearest m, and stem length and stump height were found out as well. Stem length was measured to the nearest 0.1 m, stump height to the nearest 0.01 m. Characteristics of the experimental material are given in Table 1. Diameters in improper sections were interpolated. The lower stem part (stump diameter, diameters at Stem No. 1 (h = 20 m) Stem No. 2 (h = 10 m) Stem No. 1 (h = 20 m) Stem No. 2 (h = 10 m) Fig. 2. Shape of both spruce samples (Šmelko 2000) illustrated by means of stem shape diameters 338 J. FOR. SCI., 52, 2006 (8):

4 Fig. 3. Shape of both spruce samples (Šmelko 2000) illustrated by means of full Procrustes coordinates Stem No. 1 No. (h = 20 1 m) (h = 20 m) Stem No. 2 No. (h = 10 2 m) (h = 10 m) m, 1.3 m and 2 m, 3 m) was balanced by means of power function y = ax k, parameters of this function were derived for individual stems and then diameters at 1/100 and 1/40 of height were calculated. The remaining diameters (1/20 h, 1/10 h, 2/10 h,.., 9/10 h) were calculated by linear interpolation. The stem can be described as a multidimensional object by means of stem shape diameters. Thus, the stem shape diameters (b m ) are the diameters at the relative sections (in this case m = 1/100, 1/40, 1/20, 1/10, 2/10,, 9/10 of the stem height) divided by the stem height (h), therefore b m = d m /h. Division by the height is in fact the elimination of the size from the object in the sense of intuitive definition of the shape. Individual stem is therefore taken as a sample from n objects described by m dimensions (stem shape diameters (b i,j )). Hence: b i = (b i,1,..., b i,m ) T, i = 1,..., n For this selection, it is possible to set a sample vector for mean values µˆ given by the following equation: 1 n µˆ = Σ b i (1) n i=1 Estimation of variance-covariance matrix is ruled by the following equation: n 1 s = Σ (b µˆ ) (b i i µˆ ) T (2) n 1 i=1 The test of hypothesis that the data are derived from multidimensional normal distribution In this article, we use a test based on multidimensional skewness (g 1,m ) and kurtosis (g 2,m ), as described in Meloun and Militký (1998). We test the simultaneous validity of a hypothesis about symmetry (H 01 :g 1,m = 0) and about normality of kurtosis (H 02 :g 2,m = m(m + 2)) of the examined variable distribution. The estimation of sample skewness is given by the following equation: ĝ 1,m = n n 1 n Σ Σ 2 d 3 i=1 j=1 i j (3) where: d ij = (b i µˆ ) T S 1 (b j µˆ ) is squared Mahalanobis distance. Considering the H 01 hypothesis valid, then the test statistics U n 1 = ĝ 6 1,m (4) Table 1. Characteristics of experimental material: h total tree height, d 1.3 overbark diameter at breast height, SD standard deviation Norway spruce (n = 3,346) Scots pine (n = 3,082) European larch (n = 1,403) h (m) d 1.3 (cm) h (m) d 1.3 (cm) h (m) d 1.3 (cm) Mean Median SD Min Max J. FOR. SCI., 52, 2006 (8):

5 has asymptotically chi-square distribution χ 2 m (m + 1)(m + 2)/6 The estimation of sample kurtosis is given by the following equation: 1 n ĝ 2,m = Σ d2 i i (5) n i=1 Considering the H 02 hypothesis valid, then the test statistics: U 2 = (ĝ 2,m g 2,m )/(8m((m + 2)/n)) 0.5 (6) has asymptotically normal distribution N(0,1). This approximation can be used providing the following condition is satisfied: ĝ 2,m > m(m + 2)(n 1)/(n + 1) (7) Multivariate test of equality of covariance matrices For k multivariate populations, the hypothesis of equality of covariance matrices is H 0 : Σ 1 = Σ 2 =... = Σ k The test Σ 1 = Σ 2 for two groups is treated as a special case by setting k = 2. We assume independent samples of size n 1, n 2,..., n k from multivariate normal distribution. To make the test, we calculate S 1 v 1 /2 S 2 v 2 /2... S k v k /2 M = (8) S p Σ i v i /2 where: v i = n i 1, S i the covariance matrix of the i-th sample, S p the pooled sample covariance matrix. If S 1 = S 2 = S p, then M = 1. As the disparity between S 1 and S 2 increases, M approaches zero. Box (1949, 1950) in Rencher (2002) gave χ 2 approximation for the distribution of M. This test is referred to as Box s M-test. We calculate k 1 1 2m 2 + 3m 1 c 1 = [ ][ ] (9) i=1 v i k i =1 v i 6(m + 1)(k 1) then u = 2(1 c 1 ) ln M has approximately χ 2 ( k 1)m(m + 1)/2 distribution, where: M defined in (8), and 1 2 i=1 2 i=1 1 k k ln M = v i ln S i ( v i )ln S p (10) We reject H 0, if u > χ 2 ( k 1)m(m + 1)/2 (α). The test of the hypothesis that the mean vectors are equal Consider two independent random samples b x,1,...,b x,n and b y,1,...,b y,n. The vectors b i = (b i,..., b m ) T, i = 1,..., n are stem shape diameters. In this case m = 12 and n = 3,346 for Norway spruce, 3,082 for Scots pine and 1,403 for European larch. The test is provided for three pairs of tree species. We expect that stems from these populations have mean shapes µ x and µ y. The test of the hypothesis on mean vectors equality (H 0 :µ x = µ y versus H 1 :µ x µ y ) can be carried out using Hotelling s T 2 two-sample test. Let us use the following test statistics: n 1 n 2 (n 1 + n 2 m 1) F stat = (µ x µ y ) T S p 1 (µ x µ y )(11) (n 1 + n 2 )(n 1 + n 2 2)m where: (n 1 1) S 1 + (n 2 1) S 2 Sp = (12) n 1 + n 2 2 is the pooled variance-covariance matrix and S 1 and S 2 are variance-covariance matrices for individual samples. We can express the squared Mahalanobis distance of equation (11) as m d xy = (µ x µ y ) T S p 1 (µ x µ y ) = s 2 j /λ j (13) j=1 where: s j = (µ x µ y ) T γ j the scores in the direction of the observed group difference, γ j eigenvectors of matrix S p, λ j corresponding eigenvalues. High values of s 2/λ indicate which directions of j j shape variability are associated with the difference between the groups. Provided the null hypothesis is valid, the test statistics F stat has Fisher s distribution with m and n 1 + n 2 m 1 degrees of freedom. However, this test can be used only in the case of normality of both sets and homogeneity of variance-covariance matrices. The assumption of normality and equal covariances turned out to be questionable. Therefore, a Monte Carlo test was carried out with the null hypothesis that the groups had equal mean shapes. The data were randomly split into two groups of the same size as the groups in the data, and the test statistic F stat was evaluated for B random permutations T 1,..., T B. The ranking r of the observed test statistic F obs was then used to give the p-value of the test: r 1 p-value = 1 B J. FOR. SCI., 52, 2006 (8):

6 Variability The principal components analysis (PCA) was used to analyze the shape variability. In principal components analysis, we seek to maximize the variance of a linear combination of the variables. The first principal component is the linear combination with maximal variance; we are essentially searching for such a dimension that the observation maximally separates or around which the observation data are spread out. The second principal component is the linear combination with maximal variance in a direction orthogonal to the first principal component, and so on. The orthogonal eigenvectors of variance-covariance matrix, denoted by γ i, i = 1, 2,..., j, are the principal components of variance-covariance matrix with corresponding eigenvalues λ 1 λ 2... λ j 0 where: j = min (n 1, m). The principal components are in fact transformed variables and the principal component (PC) score represents transformed objects. PC score for the i-th individual on the j-th principal component is given by s ij = γ jt (b i µˆ ) (14) The standardized PC scores are s ij c ij = (15) λ j Allometry Allometry involves the study of relationships between shape and size, and in particular the manner how shape depends on size. Traditional methods in allometry involve the fitting of linear or non-linear regression equations between size and shape measures. Allometry can also be investigated in our geometrical framework, using regression. Principal component score is chosen as a response variable and total tree height is used as the explanatory variable. Inflection point of morphological stem curve The landmarks are investigated by geometrical methods. When we want to get the morphological stem curve, we must balance these individual points. Balancing was provided by the same function as used by Petráš (1989) for the construction of a mathematical model for the stem shape of coniferous tree species: d i (h i ) = b 1 h i p 1 + b 2 h i p 2 + b 3 (16) where: d i the stem diameter at height h i, b 1, b 2, b 3, p 1, p 2 parameters of the function. This function has the only inflection point. In this point the morphological stem curve transfers from the convex to concave course. If we want to find the height coordinates of inflection point I (d I, h I ), we define the second function derivation equal to zero: d I " (h I ) = b 1 p 1 (p 1 1)h I p b 2 p 2 (p 2 1)h I p 2 2 = 0 (17) then b 1 p 1 (p 1 1) h 1 = p2 p1 b (18) 2 p 2 (p 2 1) RESULTS AND DISCUSSION All the following tests are calculated for stem shape diameters. In the case of Norway spruce, Scots pine, European larch, the sample skewness is ĝ 1,12 = 54.14, and 44.87, resp. The test statistics U 1 thus equals 30,190, 18,560, 10,490, which is more than the critical value of χ 2 (0.05) = Sample 364 kurtosis is ĝ 2,12 = 330.8, 285.4, Test statistics U 2 = 256.9, 177.8, and the critical value of standardized normal distribution on the significance level of 0.05 is The criterion (7) is satisfied because ĝ 2,m > 167.9, 167.9, In both quantities, skewness and kurtosis, we therefore reject the coincidence with normal distribution. We also did Box s M-test for (a) Norway spruce and Scots pine sets, (b) Norway spruce and European larch sets, (c) Scots pine and European larch sets. Test statistics: (a) u = 2(1 c 1 )lnm = 6,692.3 > χ 2 (0.05) = 99.3, 78 p-value < We reject H 0. (b) u = 2(1 c 1 )lnm = 2,147.2 > χ 2 (0.05) = 99.3, 78 p-value < We reject H 0. (c) u = 2(1 c 1 )lnm = 2,913.5 > χ 2 (0.05) = 99.3, 78 p-value < We reject H 0. The Monte Carlo test was used for assessing the coincidence of mean vectors because of problems with normality and doubtfulness of equal variances. The coincidence of mean shapes was tested for (a) Norway spruce and Scots pine sets, (b) Norway spruce and European larch sets, (c) Scots pine and European larch sets. For each pair of samples, 2,000 random permutations were performed. In all three cases p-value < 0.01 and therefore we reject the null hypothesis about coincidence of mean vectors and accept the hypothesis about the difference between mean shape vectors. J. FOR. SCI., 52, 2006 (8):

7 Norway spruce spruce Scots pine pine European larch larch Fig. 4. Average stem shape of all three tree species investigated by means of stem shape diameters and balanced by function No Mean shapes for stem shape diameters are graphically represented in Fig. 4, and full Procrustes mean shapes for all three species are shown in Fig. 5. Only landmarks express the shape. However, these landmarks do not give a very clear illustration, therefore in the case of stem shape diameters they were balanced by function No. 16 and in the case of full Procrustes coordinates they were connected by means of abscissas. Both figures are very similar to each other. The landmarks are placed at 1/100 h, 1/40 h, 1/20 h, 1/10 h, 2/10 h,, 9/10 h. Norway spruce has the largest root swelling. The function of European larch behaviour is closer to Norway spruce than to Scots pine. Scots pine is wider between 1/40 h and 2/10 h, and narrower than Norway spruce and European larch between 2/10 h and 6/10 h (7/10 h). Then Scots pine is markedly wider the widest at 9/10 h. We also found intersection points of all three curves illustrated in Fig. 4 by means of FindRoot function in the programme Mathematica. The intersection point of the curve at about 0.2 h is noteworthy. Norway spruce with Scots pine intersects at h, Norway spruce with European larch at h, Scots pine with European larch at h. The height coordinates of inflection points were calculated for all three tree species according to equation No. 18. They are for Norway spruce, Scots pine, and European larch: 0.29 h, 0.43 h and 0.29 h, resp. Demaerschalk and Kozak (1977) and Perez et al. (1990) found the relative height Z of the inflection point to range between 0.20 and 0.25 and between 0.15 and 0.35, respectively. In another study by Allen (1993), who used an average of 0.30, the inflection point was found to vary between 0.29 and 0.32 for small and medium-sized Caribbean pine Norway spruce Scots pine European larch Norway spruce Scots pine European larch Fig. 5. Stem shapes of all three tree species expressed by full Procrustes mean shapes. The landmarks are connected by abscissas 342 J. FOR. SCI., 52, 2006 (8):

8 Table 2. F stat partition of Equation (11) for 12 principal components, for three pairs of tree species, d ij is the squared Mahalanobis distance No. of principal components Norway spruce and Scots pine sets F stat partition F stat d ij Norway spruce and European larch sets Scots pine and European larch sets (Pinus caribaea Morelet var. hondurensis Barret and Golfari) trees. The graphic effect of the first three principal components is the same in stem shape diameters as in Procrustes tangent coordinates. For better illustration and with regard to complex programmes prepared by Dryden (2000), we carried out an analysis of the first three principal components in Procrustes tangent coordinates as shown in Dryden and Mardia (1998); the definitions were introduced in the same way as in Křepela et al. (2004). Proportional expressions of variability explained by eigenvalues of the variance-covariance matrices of stem shape diameters and Procrustes tangent coordinates are given in Table 3. Fig. 6 illustrates the graphic effect of the first three principal components for all three tree species. The first three principal components in all three sets show the same graphic effect. The first principal component (PC 1 see Fig. 6) has a symmetrical graphical effect the whole stem extension (narrowing) and explains the great deal of variability (on average 95%). Its cause is not explainable from our data, but based on the preceding research (Křepela et al. 2001; Křepela 2002), we take it for a competition effect: above-level the thicker trees, below-level the narrower ones. The second principal component (PC 2 see Fig. 6) has an asymmetrical graphical effect: the lower stem part versus the upper part. For Norway spruce up to 1/40 h to one side and for 1/20 h up to the other side. For European larch the boundary lies between 1/10 h and 2/10 h and so higher than for Norway spruce. For Scots pine the boundary is at the highest level at 2/10 h. The second principal component for Norway spruce expresses 3.9% of variability the most for all the tree species. For spruce its effect is the highest at 1/100 h, and it is the effect of root swellings. For Norway spruce it is the lowest on the stem but is of the greatest importance. The graphical effect of the third principal component (PC 3 see Fig. 6) changes its direction three times. At 1/100 h one direction, at 1/40 h null effect, then the reverse direction follows that is up to 4/10 h for Norway spruce, up to 5/10 h for European larch and up to 5/10 h for pine, then the reverse effect follows up to 9/10 h for Norway spruce and European larch. For Scots pine also with the exception of 6/10 h, Table 3. Proportional expression of variability explained by eigenvalues of the variance-covariance matrices of stem shape diameters and Procrustes tangent coordinates from the investigated sets Norway spruce Scots pine European larch Eigenvalue Stem shape diameters (%) Procrustes tangent coordinates (%) Stem shape diameters (%) Procrustes tangent coordinates (%) Stem shape diameters (%) Procrustes tangent coordinates (%) p λ j / λ j 100 j=1 p λ j / λ j 100 j=1 p λ j / λ j 100 j=1 p λ j / λ j 100 j=1 p λ j / λ j 100 j=1 p λ j / λ j 100 j=1 λ λ λ J. FOR. SCI., 52, 2006 (8):

9 Scots pine PC 1 : 96.1% PC 2 : 2.2% PC 3 : 1% Norway spruce PC 1 : 93.7% PC 2 : 3.9% PC 3 : 1.3% European larch PC 1 : 96.1% PC 2 : 2.3% PC 3 : 1.1% Fig. 6. The first three principal components (PC) with configurations evaluated for 1 (PC 1) and 3 (PC 2, 3) standard deviations along each PC from the full Procrustes mean shape, including proportions of explained variability where we see the null graphical effect of the third principal component. Another problem which principal components differ from each other the most was solved. Table 2 contains the components of test statistics F stat, calculated for individual principal components. Components of Mahalanobis distance s 2 j /λ j indicate which directions of shape variability are associated with the difference between the groups. Spruce and pine differ the most from each other above all in the 4 th principal component, pine and larch in the 4 th principal component, spruce and larch in the 3 rd principal component. We also investigated the relationship between principal component score and total tree height (allometry). Principal component score is chosen as a response variable, and total tree height is used as an explanatory variable. The Spearman coefficients of correlation for both variables are calculated in Table 4. Scots pine has the lowest dependence between shape and height. The Norway spruce correlation coefficient between 4 PC score and height is worth 344 J. FOR. SCI., 52, 2006 (8):

10 Table 4. The Spearman coefficient of correlation (rs) for dependence between principal component score and total tree height. Significance is the significance level PC score Norway spruce Scots pine European larch rs significance rs significance rs significance E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E E-01 of attention having the value 0.49, which is a moderate dependence. For larch the first three coefficients document a moderate dependence. Generally it can be said that the height influences the shape of all three tree species only very weakly and therefore the influences on the stem shape must be searched outside the stem. CONCLUSION This study showed that the stem shape of three investigated coniferous tree species (Norway spruce, Scots pine, European larch) differed. The term stem is understood in the sense of geometrical methods. In practice the simplified Bookstein coordinates (stem shape diameters) were used. Shape variability was investigated by the method of principal components using stem shape diameters and Procrustes tangent coordinates. The first principal component explains the prevailing part of variability, has a symmetrical graphical effect and we relate it to the competitive pressure onto individual trees. The second principal component has an asymmetrical graphical effect; we connect it with the effect of root swelling. The relationship between stem size (height) and shape showed to be weak, the weakest for Scots pine. Inflection points on the morphological stem curve were situated for Norway spruce, Scots pine, and European larch at: 0.29 h, 0.43 h and 0.29 h, resp. R e f e r e n c e s Allen P. J., Average relative stem profile comparisons for three size classes of Caribbean pine. Canadian Journal of Forest Research, 23: Demaerschalk J.P., Kozak A., The whole bole system: A conditioned dual-equation system for precise prediction of tree profiles. Canadian Journal of Forest Research, 7: DRYDEN I.L., MARDIA K.V., Statistical Shape Analysis. Chichester, New York, Weinheim, Brisbane, Singapore, Toronto, John Wiley & Sons: 345. DRYDEN I.L., Statistical shape routines. University of Nottingham. KŘEPELA M., SEQUENS J., ZAHRADNÍK D., Dendrometric evaluation of stand structure and stem forms on Norway spruce (Picea abies [L.] Karst.) sample plots Doubravčice 1, 2, 3. Journal of Forest Science, 47: KŘEPELA M., Point distribution form model for spruce stems (Picea abies [L.] Karst.). Journal of Forest Science, 48: KŘEPELA M., SEQUENS J., ZAHRADNÍK D., A contribution to the knowledge of Scots pine (Pinus sylvestris [L.]) stem shape. Journal of Forest Science, 50: KŘEPELA M., ZAHRADNÍK D., SEQUENS J., Possible methods of Norway spruce (Picea abies [L.] Karst.) stem shape description. Journal of Forest Science, 51: J. FOR. SCI., 52, 2006 (8):

11 MELOUN M., MILITKÝ J., Statistické zpracování experimentálních dat. Praha, East Publishing: 839. Perez D.N., Burkhart H.E., Stiff C.T., Avariableform taper function for Pinus oocarpa Schiede in Central Honduras. Forestry Science, 36: PETRÁŠ R., Matematický model tvaru kmeňa ihličnatých drevín. Lesnictví, 35: Rencher A.C., Methods of Multivariate Analysis. Chichester, New York, Weinheim, Brisbane, Singapore, Toronto, John Wiley & Sons: 708. ŠMELKO Š., Dendrometria. Zvolen, TU: 399. Received for publication March 16, 2006 Accepted after corrections April 14, 2006 Porovnání tvaru kmene smrku ztepilého (Picea abies [L.] Karst.), borovice lesní (Pinus sylvestris L.) a modřínu opadavého (Larix decidua Mill.) pomocí geometrických metod M. Křepela 1, R. Petráš 2 1 Fakulta lesnická a environmentální, Česká zemědělská univerzita v Praze, Praha, Česká republika 2 Národné lesnícke centrum Lesnícky výskumný ústav Zvolen, Zvolen, Slovenská republika ABSTRAKT: Článek se zabývá porovnáním tvaru kmene tří jehličnatých dřevin: smrku ztepilého, borovice lesní a modřínu opadavého. Tvar je zkoumán pomocí geometrických metod. Byly použity zjednodušené Booksteinovy souřadnice (kmenové tvarové průměry) a pro zkoumání variability také Prokrustovy souřadnice. Materiál tvořilo celkem kmenů smrku, kmenů borovice a kmenů modřínu. Tento materiál pochází z území České a Slovenské republiky. Shoda středních tvarových vektorů byla zkoumána pomocí Hotellingova T 2 testu vždy pro dva výběry. Variabilita byla zkoumána pro kmenové tvarové průměry a pro Prokrustovy tangentové souřadnice pomocí metody hlavních komponent. Byly znázorněny a popsány tři nejdůležitější hlavní komponenty. Dále byl zkoumán vztah mezi tvarem a velikostí a byly popsány inflexní body morfologické křivky kmene pro všechny tři dřeviny. Klíčová slova: smrk ztepilý; borovice lesní; modřín evropský; tvar kmene; kmenové tvarové průměry; Prokrustovy souřadnice; analýza hlavních komponent Pro popis tvaru kmene jsou v článku použity geometrické metody (Dryden, Mardia 1998), které vycházejí z přesné definice tvaru. Tvar je definován jako geometrická informace o konfigurační matici po odstranění efektu posunutí, otočení a přeškálování. Tato konfigurační matice je složena ze souřadnic hraničních bodů, umístěných na morfologické křivce kmene. Klasická lesnická dendrometrie používá k vyjádření tvaru kmene rotační geometrická tělesa, pravé tvarové řady nebo pravé výtvarnice. V článku je na příkladu dvou smrkových kmenů, který uvádí Šmelko (2000), demonstrováno, že kmeny, které mají stejnou pravou tvarovou řadu a stejné pravé výtvarnice, nemusejí mít na základě geometrických metod stejný tvar (obr. 2 a 3). Ke zkoumání tvaru kmene byly z geometrických metod použity zjednodušené Booksteinovy souřadnice (kmenové tvarové průměry) a dále Prokrustovy souřadnice. Všechny statistické testy byly provedeny s kmenovými tvarovými průměry. Prokrustovy souřadnice byly použity zároveň s kmenovými tvarovými průměry pro výzkum variability a znázornění efektu jednotlivých hlavních komponent, a to pro jejich lepší grafickou vypovídací schopnost a také pro srovnání obou geometrických metod. Pokusný materiál tvořilo kmenů smrku, kmenů borovice a kmenů modřínu. Tento materiál pochází z území České a Slovenské republiky. Je téměř totožný s materiálem, který použil Petráš (1989) ke konstrukci matematického modelu tvaru kmene jehličnatých dřevin. Hraniční body byly umístěny na morfologickou křivku kmene do 1/100, 1/40, 1/20, 1/10, 2/10,,9/10 výšky kmene. Podle schématu, znázorněného na 346 J. FOR. SCI., 52, 2006 (8):

12 obr. 1, byly vypočteny tvarové kmenové průměry a Prokrustovy souřadnice pro všechny kmeny. Statistické testy směřovaly k porovnání středních tvarových vektorů tvarových kmenových průměrů všech tří zkoumaných dřevin pomocí Hotellingova T 2 testu. Před vlastním provedením tohoto testu jsme zkoumali vícerozměrnou normalitu kmenových tvarových průměrů jednotlivých dřevin a shodu variančně-kovariančních matic pro tři dvojice dřevin. V obou případech byly nulové hypotézy zamítnuty. Nezbývalo tedy než provést neparametrický Monte Carlo test, kterým jsme zamítli nulovou hypotézu o rovnosti středních tvarových vektorů pro tři dvojice dřevin tedy zkoumané dřeviny nemají ve smyslu geometrických metod stejný tvar. Tvarová variabilita v rámci kmenových tvarových průměrů a Prokrustových souřadnic byla zkoumána pomocí metody hlavních komponent. Přehled výsledků podává obr. 6 a tab. 3. Procenta vysvětlené variability pomocí hlavních komponent jsou pro kmenové tvarové průměry i pro Prokrustovy souřadnice téměř shodná. Také geometrické efekty prvních tří hlavních komponent, které vysvětlují přes 99 % variability, jsou pro všechny tři dřeviny stejné. Ekologický vliv vyjádřený první hlavní komponentou způsobuje rozšiřování (zužování) kmene po celé jeho délce. Stejný efekt první hlavní komponenty popsal Křepela et al. (2001) a Křepela (2002). V těchto pracích byl zkoumán tvar kmene smrku podle Konšelových stromových tříd. Usuzujeme, že i v tomto případě se jedná o efekt konkurence. Nadúroveň silnější stromy, podúroveň užší. Přímý důkaz však nemůže být podán, protože u empirického materiálu nebyly zkoumány stromové třídy ani konkurenční indexy. Druhá hlavní komponenta má asymetrický grafický efekt: spodní část kmene kontra zbývající horní část. Tato komponenta u smrku vyjadřuje 3,9 % variability tedy nejvíce ze všech dřevin. U smrku je její efekt v 1/100 h největší. Jde tedy o efekt kořenových náběhů. U smrku zasahuje na kmeni nejníže, má ale největší váhu, což odpovídá prostému okulárnímu pozorování v přírodě. Grafický efekt třetí hlavní komponenty mění směr třikrát. Jaký ekologický vliv jej způsobuje, nebylo možné z našich dat zjistit. Zajímavý problém představuje zkoumání závislosti mezi velikostí kmene a jeho tvarem. V případě kmenových tvarových průměrů představuje velikost výška kmene, neboť její pomocí je z objektu (kmene) odstraňována jeho velikost. V případě geometrických metod se vztah mezi velikostí a tvarem zkoumá pomocí korelační závislosti mezi velikostí a jednotlivými skóre hlavních komponent, kterých je 12, protože máme na kmeni 12 kmenových tvarových průměrů. Spearmanovy korelační koeficienty jsou demonstrovány v tab. 4. Tato závislost se ukázala nejnižší u borovice, kde ji můžeme s určitým nadhledem charakterizovat jako téměř žádnou. U modřínu se slabá závislost projevuje mezi prvními třemi skóre a výškou, u smrku je zajímavější 4. skóre. Celkově můžeme říci, že tvar kmene zkoumaných jehličnanů je závislý na jiných ekologických vlivech daleko více než na velikosti. V dendrometrických pracích se často spodní část kmene vyjadřuje pomocí geometrického tělesa neiloidu a část nad ní pomocí paraboloidu. Právě inflexní bod, tedy bod, kde jedno těleso přechází v druhé, byl také předmětem našeho zkoumání. Geometrické metody pracují s hraničními body. Tyto hraniční body bylo nejdříve nutno vyrovnat pomocí funkce, která by zachytila morfologickou křivku kmene. Byla zvolena funkce č. 16. Ta má jediný inflexní bod, a jí byly vyrovnány střední tvarové průměry. Pro smrk, borovici a modřín vycházejí inflexní body do 0,29, 0,43 a 0,29 výšky kmene. Corresponding author: Mgr. Ing. Michal Křepela, Ph.D., Česká zemědělská univerzita v Praze, Fakulta lesnická a environmentální, Praha-Suchdol, Česká republika tel.: , fax: , Krepela@fle.czu.cz J. FOR. SCI., 52, 2006 (8):

13 JOURNAL OF FOREST SCIENCE, 52, 2006 (8): Frost cracks and their effect on the stability of birch stands in the Krušné hory Mts. E. Kula, I. Buchta, P. Stránský Faculty of Forestry and Wood Technology, Mendel University of Agriculture and Forestry Brno, Brno, Czech Republic Abstract: Frost cracks which originated in birch due to bending during long-term icing in winter 1995/1996 became the place of entrance for the attack of birch stems by Piptoporus betulinus (Bull. ex Fr.) Karst. and subsequently for one of the causes of birch stand disintegration in the Krušné hory Mts. The hypothesis is substantiated on the basis of regularities of the frost crack dispersion in the stem profile, frequency of occurrence related to the stand age, altitude and cardinal points. The number of frost cracks increased with the transect profile altitude m. One crack on the birch stem predominated (73%) while the higher number of frost cracks occurred at altitudes > 800 m. In stands younger than 20 years, frost crack damage was higher (34 47%) than in older stands (14%). Frost cracks occurred in the lower part of stems with the highest bending stress. Keywords: Betula pendula Roth; frost cracks; icing; Piptoporus betulinus (Bull. ex Fr.) Karst.; Krušné hory Mts. Birch as a highly tolerant species created together with blue spruce the basis of stands of substitute species in the eastern Krušné hory Mts. after 1980 (Kula 2006; Moravčík 1994). The birch stand disintegration at upland locations of the Krušné hory Mts. became evident after 1997 when birch did not come into bud. However, this physiological detriment was preceded by mechanical damage to stands by icing in winter 1995/1996 (Kula, Kawulok 1998; Kula et al. 1999). Causes of the birch stand dieback were dealt with by Šrámek (1998), Kula and Rybář (1998), Martinková et al. (2001), Mauer and Palátová (2003). Frost cracks originate by freezing through a stem at a sudden change of a temperature in the form of longitudinal cracks in the lower part of the stem. Species with wide pith rays are affected (oak, poplar, elm, chestnut Pfeffer 1961). Frost cracks are not frequent in nature. As for oak stands, Lamprecht (1950) mentions the increased frequency of frost cracks in 22% of trees on the northern and southern side. The cracks originating in consequence of drought and water stress are noted even in young spruce stands (20 40 years old) as radial cracks in sapwood. On the stem surface, only small grooves fringed by thin walls occur. High suction stress in sapwood leading water exceeds the resistance of a wood tissue formed from large thin-walled cells (Hartmann et al. 2001). Unlike frost cracks which are orientated towards south or south-west cracks induced by drought occur along the whole stem girth not forming a marked frost crack. They are also substantially longer. On skeleton sites on slope tops, up to 100% trees tend to be damaged (Mrkva 2004; Riedl 2005). The cracks occur mainly on oak, hornbeam, lime and ash. In spite of the reasons of frost cracks in birch in the Krušné hory Mts. (Martinková et al. 2001; Mauer, Palátová 2003) we hypothesize that due to long-term icing and the load of trees bending of Supported by the Ministry of Agriculture of the Czech Republic, Project No. 1G46002, and by the Ministry of Education, Youth and Sports of the Czech Republic, Research Project No. MSM It was sponsored by the following companies and authorities: Netex Co. and Alusuisse Co. in Děčín, Municipal Office in Děčín, Setuza Co. in Ústí nad Labem, ČEZ Co. in Prague, Cement Works Co. in Čížkovice, North-Bohemian Mines Co. in Chomutov and Dieter Bussmann Co. in Ústí nad Labem. 348 J. FOR. SCI., 52, 2006 (8):

14 birch occurred particularly in young stands when frozen tissues were mechanically damaged and subsequently after release, trees straightened and frost cracks incurred became the place of entrance for wood-destroying fungi. material and methods The control plots were chosen in the height transects of birch stands of Forest District Litvínov. The plots were differing in age, altitude and slope orientation. At least 100 trees on each plot were controlled. The study was concentrated on living trees being carried out in three levels: Basic study specified the occurrence of frost cracks on stems (number, position in relation to cardinal points, sample tree dbh (diameter at breast height). Detailed study moreover, the position of cracks was recorded in the tree profile, their length and width, creation of frost cracks, crown size. Special study determination of the extent of rot on felled living trees after their cutting to sections. To assess data obtained ANOVA statistics was used. The study was carried out in 28 sample plots including 2,904 trees of birch (Betula pendula Roth) 1,008 of them showing damage by frost cracks. Description of the area under study Forest stands except of J1 and J2 (Forest District Litvínov, Forest Enterprise Janov) belong to Forests of the Town of Jirkov. The stands are situated at an altitude of m; the areas occur partly on slopes of southern and northern orientation. In 1995, the stand age range was 7 24 years, 20 of 28 studied stands were established after 1980 following the large-area dieback of spruce stands, remaining (%) y = x y = x R 2 = R 2 = Altitude Altitude (m) [m] stands come from the second half of the 70s of the 20 th century. The area passing to a plateau of the Krušné hory Mts. is moderately cool. Mean annual temperature is 5 C depending on altitude. On plateaus, effects of inversions are evident and the highest temperature differences are at the beginning of winter (October December). Mean annual total precipitation ranges from mm, in the Fláje game preserve (Forest District Litvínov) up to 984 mm (Hošek et al. 1999). Icing showed a high proportion in damages to forest stands of Forest Enterprise Janov. In the second half of the 70s, icing culminated in 1977 (14,930 m 3 ); in other years, losses exceeded 5,000 m 3. Also in the first half of the 80s, the height of damages remained in the same extent culminating in 1984 (9,333 m 3 ) and then gradually decreased. Increased negative impacts of icing on forest stands were evident as late as 1996 (3,260 m 3 ). Culmination values were not reached just because spruce stands of appropriate extent did not occur in the icing zone (Kula, Richterová 1999). In the Krušné hory Mts., maximum icing damage occur on SE slopes at an altitude of m (Singer 1916). Icing threatens particularly unprotected main and lateral ridges and steep windward slopes where it occurs at much lower locations. Negative impacts of icing are increased by their uneven deposition in tree crowns particularly on branches subject to air flow. Impacts of icing on spruce stands in 1995/1996 were described by Lomský et al. (1996), on birch stands by Kula and Kawulok (1998) and Kula (1998). results The occurrence of frost cracks was assessed in birch stands on living trees at an altitudinal profile of m regardless of aspect and stand age. Dif- Fig. 1. The proportion of living trees with the occurrence of frost cracks according to altitude J. FOR. SCI., 52, 2006 (8):

15 W NW SW N S Fig. 2. The frequency of the occurrence of frost cracks according to their orientation on a stem 29 NE SE E ferences between particular stands can be induced by the various degree of dieback in trees damaged by icing in 1995/1996 and failed budding in Nevertheless, it was possible to derive trends in the increase of the number of frost cracks with increasing altitude (Fig. 1). A transition between altitudes 700 and 750 m was particularly marked. Probability of the occurrence of frost cracks was there 17% while at altitudes m and m 40 and 43%, respectively. In affected trees (774), frost cracks occurred once (73%), twice (21%), three times (4.8%), four times (1.1%) and six times (0.1%). The frequency of two frost cracks per tree was higher at altitudes above 800 m, the frequency of four and more cracks occurred in trees at altitudes above m. The age of stands at the time of their load by icing (1995/1996) was an important attribute of the potential damage to stands in the form of bending, breakage and windfalls. There is a relationship between the proportion of frost cracks in young stands (up to 10 years 34.2%, years 43.8%, years 47% and years only 14.2%) and a frequency of the number of frost cracks per tree changing also with age. Up to 20 years of the stand age, the occurrence of the only crack predominated (70 73%) and 94% proportion occurred in all older stands. At the occurrence of two cracks, stands younger than 20 years are on the level of 23 21% while in older stands we noted only 4.5%. A marked decrease with age was noted in the occurrence of three frost cracks ( %). The area of frost cracks was expressed by an index which was a product of its length and width. If we ignore the position (720 m a.s.l.) markedly smaller cracks occurred in stands up to 800 m a.s.l. (average index values 67.8 or 81.4) than at higher locations (134.1). Even more marked differences were evident if the index referred to the frequency of the occurrence of cracks when in the lower part of the transect, the average area of damage was four times lower than above 800 m a.s.l. The position of frost cracks predominated unambiguously on the southern side (S 46.7%, SE + SW 22.1%), northern positions were less important (N 10.3%, NE + NW 8%), E 3.7%, W 9% (Fig. 2). The frost crack centre occurred most frequently in the birch stem profile between 25 and 50 cm (25.2%). Cracks in a section above and below the position were in the same frequency (18%). A marked decrease in frost cracks occurred in higher parts of the stem profile ( cm 14.3%, cm 11.2% and in next three 25 cm long sections %). The whole position of the frost crack centre showed a decreasing trend with an increasing altitude (Fig. 3). The frost crack length does not exhibit a statistical dependence on altitude and age but short- Average height of rupture centre (cm) Altitude (m) Fig. 3. The position of the frost crack centre on a stem in relation to altitude 350 J. FOR. SCI., 52, 2006 (8):

16 Average height of frost rib centre (cm) y y = = x R R 2 2 = = Altitude (m) ens with the number of occurring cracks on a stem and on northern aspects. In stands aged up to 10 years, frost cracks occurred most frequently at the stem foot 0 25 and cm (21 and 20%); further in the profile, the proportion of frost cracks continually decreased. In stands aged years, an area with the crack centre occurred at a position of cm (27%) and stands aged years can be characterized similarly (28.3%). In trees even with the course of 10 years Fig. 4. The average position of frost cracks on a stem in relation to altitude short cracks predominated up to 10 cm (28%), cm (25%) and cm (24%). The position of the crack centre decreased with altitude (Fig. 4) and the crack length with age (Fig. 5). The frost crack width decreased only with the higher frequency of cracks on a sample tree. Effects of cardinal points on the crack width decreased from southern to northern locations (Fig. 6). The occurrence of frost cracks is connected with rot caused by Piptoporus betulinus (Bull. ex Fr.) Karst. Average length of frost crack (cm) y = x R 2 2 = = Stand age Fig. 5. The frost crack length in relation to the stand age Average width of rupture (cm) N NE E SE S SW W E Rupture aspect Fig. 6. The frost crack width according to the orientation on a stem J. FOR. SCI., 52, 2006 (8):

17 Average stem diameter (cm) trees without frost ribs trees with frost ribs 6 1_5 J01 1_4 J02 2_3 J15 J12 1_3 2_5 J13 J05 J11 J14 1_1 2_2 3_3 J03 J09 J10 J04 2_4 3_2 J07 1_2 J06 J08 2_1 3_1 Fig. 7. The dependence of the living tree dbh on the occurrence of frost cracks Site which passes through from the point of injury up to the crown level spreading rather quickly and affecting growth properties, dbh (breast-height diameter) of living and dead trees, living trees with frost cracks and without them decreased with altitude (Fig. 7) mean dbh of trees with frost cracks being lower (Fig. 8). It means that more than the actual position the occurrence of rot is important in attacked trees which effects not only growth properties but was also an important factor participating in the dieback of damaged stands and their disintegration. In this context, Mauer and Palátová (2003) draw attention to the presence of honey fungus [Armillaria gallica Marxmüller et Romagn, A. ostoyae (H. Romagnesi) Herink.] on the root system of birch with markedly decreased foliage. discussion Frost cracks on birch in connection with the necrosis of tissues and decreased regeneration are mentioned by Martinková et al. (2001). To the primary water stress a secondary water stress was added (drought in crown, excessive water content in the stem base) as well as a tertiary stress as the consequence of reserve exhaustion for budding and formation of annual shoots. However, it refers to the description of the dieback of trees affected by failed budding in 1997 and not the description of causes of the origin of frost cracks. The primary stress was very probably caused by a specific pollutant disturbing dormancy. Stresses mentioned above are a side effect of the response of heavily weakened trees (Kula et al. 2000; Kula 1999). Average stem diameter (cm) 20 y y = = x x trees without frost ribs R 2 = R2 = trees with frost ribs y y = x RR2 2 = = Altitude (m) [m] Fig. 8. Comparisons of the living sample tree dbh with a frost crack and without a frost crack in relation to altitude 352 J. FOR. SCI., 52, 2006 (8):

18 Mauer and Palátová (2003) mention air pollution stress (SO 2 ) as a primary factor of the dieback of birch, however, at the same time, they also refer to the occurrence of frost cracks on stems with the subsequent enter of honey fungus to the root system of birch. Icing caused extensive damage to spruce stands in the eastern Krušné hory Mts. in winter 1995/1996 (Lomský et al. 1996) and also markedly affected birch stands (Kula, Kawulok 1998; Kula 1998). Depending on stand age, altitude and stocking damage to birch by bending, stem and crown breakage was differentiated. The tree crown, shape, size and thus also the position of the centre of gravity (Zach, Kula 1999) have a decisive importance for the irreversible damage to a stem. Extensive part of birch stands at the turn of the first and second age classes was bent by the mass of icing. After the receding of icing, particularly young stands of the higher degree of stocking returned to their original position and breakage damage was not important (Kula, Kawulok 1998). In 1995, there was the high proportion of young birch stands in the eastern Krušné hory Mts. at medium and higher locations where stem and crown breakage was not so frequent as at locations m a.s.l. There, growth conditions were more favourable for birch and the species reached higher height and dbh with the lower plasticity for bending in the same age category (Kula 1998). In assessing the impact of icing on birch stands nobody dealt with frost cracks because they were not sufficiently striking and thus they escaped attention. The age of stands in the studied region is not sufficiently differentiated because the majority of stands was established after 1979 and stands aged over 30 years, if affected by icing, did not exhibit frost cracks but damage due to stem and crown breakages. Extensive cracks at higher locations over 800 m can be related to the lower regeneration potential of birch affected by other forms of stress, e.g. failure of budding (Kula, Stoklasa 2002), higher site requirements (climatic, soil) and genetic lability due to the unsuitable transfer of seed. Trends in the increase of frost cracks with altitude can be related to the fact that stands situated at higher altitudes are characterized by lower height and dbh at the same age as stands at lower locations. Thus, they became more plastic and bent and returned to a great extent to their original position (Kula 1998; Kula, Kawulok 1998). The origin of frost cracks is noted from the southern and northern side (Pfeffer 1961) unlike cracks induced by drought (Mrkva 2004). If the distribution of frost cracks agreed on all aspects in the dominant proportion on the southern side of stems regardless of other site factors (stocking) a phenomenon of the single-sided creation of icing cannot be excluded. The formation of icing was increased by an exceedingly long inversion situation accompanied by southeastern air flow and low temperatures. Such situation was observed in Forest District Děčín (formerly Sněžník) on the plateau of Tisá (600 m a.s.l.) in winter 1995/1996. Trees 3 6 m tall lay bent for a period of even 8 weeks because also a snow layer functioned simultaneously and crowns of the trees froze into the layer. The start of icing in the eastern Krušné hory Mts. is dated with the permanent occurrence from 24, to and from to For the whole period with continuous icing, temperature was below the freezing point. In the first stage, temperature oscillated between 1 and 12 C; in its end, the intense freezing out of tissues occurred and subsequently the second stage of icing followed ( ) when temperatures decreased to 2 up to 14 C (Měděnec). A hypothesis on freezing out the birch tissues, their mechanical damage during long-term bending and return to the original position after the end of icing and thawing out the tissues is based on findings mentioned above. The origin of frost cracks due to insolation and abrupt changes in temperatures could not occur because in 1996, temperatures persisted below the freezing point till mid-april and then they did not occur any more. Frost cracks are recorded in the same extent not only on southern slopes but also on slopes with northern orientation. After 1996, new frost cracks did not occur any more in localities under study. Characteristics of frost cracks from the viewpoint of the length, width and position of the crack centre in the stem profile appear to be little dependent on aspect, altitude and age with respect to the relatively low amplitude and extent of altitudes as well as age classes. Concentration of cracks in the lower part of the tree stem corresponds to places most exposed to bending stress and with the lowest capacity to resist the bending load. The position and frequency of cracks agree with data given by Pfeffer (1961) on the occurrence of frost cracks. Although the length of cracks and cracks can reach even 2.5 m (Hartmann et al. 2001) cracks of shorter lengths predominated. Peeling off bark should be considered to be important as well. Although crack frosts are created in sapwood, the phloem part is also disturbed and the entrance of fungal pathogens cannot be excluded. J. FOR. SCI., 52, 2006 (8):

19 With the exception of Mauer and Palátová (2003) describing the occurrence of honey fungus on a root system nobody dealt with the health condition of birch according to the occurrence of Piptoporus betulinus mentioned in overmature birch stands. Under conditions of impaired and damaged birch stands in young age the fungus became an important factor accelerating the dieback of trees and disintegration of stands. conclusion A hypothesis has been proved that frost cracks originated on birch as a result of bending during long-term icing in winter 1995/1996 became the entrance point for the attack of birch stems by Piptoporus betulinus (Bull. ex Fr.) Karst. which was one of the causes of birch stand disintegration at higher locations of the Krušné hory Mts. Climatic situations (inversions, SE air flow, low temperatures) made possible to form long-term icing and due to mechanical load bending occurred when tissues were damaged and frost cracks originated. The number of frost cracks increased with the transect profile altitude m the only crack on a stem predominating (73%). The higher frequency of frost cracks on a stem occurred at locations above 800 m a.s.l. A different response to the formation (size) of frost cracks can be linked with stress events (failure of budding) at higher locations and less favourable site conditions. In stands aged < 20 years, damage by frost cracks was higher (34 47%) than in older stands (14%). Frost cracks occurred in the lower part of stems where the highest bending stress occurs. Frost cracks appear to be the entrance place for Piptoporus betulinus which rose in living trees up to the crown part of trees thus contributing to the dieback of trees and disintegration of stands. R e f e r e n c e s Hartmann G., Nienhaus F., Butin H., Atlas poškození lesních dřevin. Praha, Nakladatelství Brázda: 289. Hošek J. et al., Chráněná území ČR, Ústecko, svazek I. Praha, Agentura ochrany přírody a krajiny ČR: 350. Kula E., Vliv porostních a stanovištních podmínek na poškození březových porostů východního Krušnohoří námrazou. Lesnícky časopis Forestry Journal, 44: Kula E., Regenerační schopnost nevyrašených bříz. Zprávy lesnického výzkumu, 44: Kula E., Činitelé ovlivňující stabilitu porostů břízy ve východním Krušnohoří. In: Slodičák M., Novák J. (eds.), Lesnický výzkum v Krušných horách. Sborník z celostátní vědecké konference, Teplice Opočno, VÚLHM, VS: Kula E., Kawulok T., Poškození porostů břízy námrazou v oblasti lesní správy Sněžník. Lesnictví-Forestry, 44: Kula E., Richterová D., Ovlivňují skutečně pevné imise tvorbu námrazy? Journal of Forest Science, 45: Kula E., Rybář V., Proč odumírá bříza v Krušných horách. Lesnická práce, 77: Kula E., Rybář V., Kawulok T., Stanovištní podmínky a dynamika vývoje zdravotního stavu porostů břízy postižených nevyrašením. Sborník z konference Problematika zachování náhradních dřevin v imisní oblasti Krušných hor, Most České Budějovice, Firma MVDr. Václav Prokop-INPROF: Kula E., Rybář V., Zabecka J., Dynamika zdravotního stavu porostů břízy ve východním Krušnohoří. Sborník ze semináře Výsledky a postupy výzkumu v imisní oblasti SV Krušnohoří, Phare program, Teplice Opočno, VÚLHM, VS: 3 6. Kula E., Stoklasa M., Vyhodnocení dynamiky změn zdravotního stavu porostů břízy ve východním Krušnohoří z družicových snímků. Sborník z konference Výsledky lesnického výzkumu v Krušných horách v roce 2001, Teplice. Opočno, VÚLHM, VS: Lamprecht H., Über den Einfluss von Umweltsfaktoren auf die Frostrissbildung bei Stiel- und Traubeneiche im nordostschweizeizerischen Mittelland. Mitteilungen der Schweizerischen Anstalt für das Forstliche Versuchswesen, 26: Lomský B., Hynek V., Pasuthová J., Uhlířová H., Šrámek V., Badalík V., Poškození lesních porostů v Krušných horách po zimě 1995/1996. Lesnická práce, 75: Martinková M., Maděra P., ÚRadníček L., Strategy of birch (Betula L.) survival in substitute stands of the Krušné hory Mts., air-polluted region. Journal of Forest Science, 47 (Special Issue): Mauer O., Palátová E., The role of root system in silver birch (Betula pendula Roth) dieback in the air-polluted area of Krušné hory Mts. Journal of Forest Science, 49: Moravčík P., Development of new forest stands after a large scale forest decline in the Krušné hory Mountains. Ecology Engineering, 3: Mrkva R., Praskliny kůry listnatých dřevin jako pravděpodobný důsledek epizod sucha nový, dosud nesledovaný problém. Sborník z konference Nové trendy v ochraně lesa a krajiny. Zvolen, TU: Pfeffer A., Ochrana lesů. Praha, SZN: 838. Šrámek V., Význam meteorologických faktorů při poškození břízy v Krušných horách v r Lesnická práce, 77: J. FOR. SCI., 52, 2006 (8):

20 Riedl V., Praskliny kmenů listnatých dřevin, pravděpodobná příčina jejich vzniku a rozsah výskytu. [Diplomová práce.] Brno, MZLU, LDF: 55. Singer J., Über Rauhreif- oder Duftbruch im Erzgebirge. Centralblatt für das gesamte Forstwesen, 42: , Zach J., Kula E., Mechanické poškození břízy námrazou. Brno, PAIDO: 63. Received for publication April 12, 2006 Accepted after corrections May 2, 2006 Mrazové trhliny a jejich vliv na stabilitu březových porostů v Krušných horách E. Kula, I. Buchta, P. Stránský Lesnická a dřevařská fakulta, Mendelova zemědělská a lesnická univerzita v Brně, Brno, Česká republika Abstrakt: Hypotéza mrazové trhliny vznikly na bříze v důsledku ohnutí stromu při dlouhodobé námraze v zimě 1995/1996. Poškozené břízy byly následně napadeny březovníkem obecným (Piptoporus betulinus (Bull. ex Fr.) Karst., který je jednou z příčin rozpadu porostů břízy ve vyšších polohách Krušných hor. Hypotéza je zdůvodněna na základě stanovené zákonitosti disperze mrazových trhlin v profilu kmene, četnosti výskytu v závislosti na věku porostu, na nadmořské výšce a světových stranách. Počet mrazových trhlin narůstal s nadmořskou výškou profilu transektu m, přičemž převažovala jedna trhlina na kmen (73 %), vyšší frekvence kýl na kmeni byla v poloze nad 800 m n. m. V porostech do 20 let bylo vyšší poškození mrazovými trhlinami (34 47 %) než v porostech starších (14 %). Mrazové trhliny se nacházely ve spodní části kmene, kde bylo nejvyšší namáhání stromu v ohybu. Klíčová slova: bříza Betula pendula Roth; mrazové trhliny; námraza; Piptoporus betulinus (Bull. ex Fr.) Karst.; Krušné hory Rozpad porostů břízy v náhorních polohách Krušných hor se s vysokou intenzitou projevil po roce 1997, kdy bříza nevyrašila. Této rozsáhlé fyziologické újmě ale předcházelo mechanické poškození porostů námrazou v zimě 1995/1996 (Kula, Kawulok 1998; Kula et al. 1999). Přes udávané příčiny vzniku mrazových kýl u břízy v Krušných horách (Martinková et al. 2001; Mauer, Palátová 2003) vyslovujeme hypotézu, že v důsledku dlouhodobé námrazy a zatížení stromů došlo k ohnutí břízy zvláště v mladých porostech, při němž zmrzlá pletiva byla mechanicky poškozena a následně po uvolnění ze zátěže se stromy narovnaly a vzniklé trhliny (mrazové kýly) se staly vstupem pro dřevokazné houby. Ve výškových transektech LS Litvínov v porostech břízy se šetření soustředilo na živé stromy s cílem vymezit přítomnost mrazových trhlin na kmeni (počet, poloha ke světové straně a v profilu stromu, jejich délka a šířka, vytvoření závalu, přítomnost houbových patogenů, výčetní tloušťka). K hodnocení získaných dat bylo užito statistiky ANOVA. Šetření se uskutečnilo na 28 zkusných plochách a zahrnuje jedinců břízy bělokoré (Betula pendula Roth), z nichž bříz bylo poškozeno mrazovou trhlinou. Porosty (LS Litvínov, LHC Janov a Lesy města Jirkova) v nadmořské výšce m, J a S expozice, v r věkové rozpětí 7 24 let. Území přecházející v náhorní plošinu, roční průměrná teplota je 5 C, průměrný roční úhrn srážek mm (Hošek et al. 1999). Vysoký podíl na škodách v lesních porostech má námraza (Kula, Richterová 1999), jejíž dopady v zimě 1995/1996 popsal ve smrkových porostech Lomský et al. (1996), v porostech břízy Kula a Kawulok (1998) a Kula (1998). Byl odvozen trend vzestupu počtu mrazových trhlin s narůstající nadmořskou výškou (obr. 1). Zvláště výrazný byl přechod mezi polohou m n. m., kde byla pravděpodobnost výskytu trhliny 17 %, zatímco v poloze a m n. m. 40 % a 43 %. Na postižených stromech (774) byla mrazová J. FOR. SCI., 52, 2006 (8):

21 trhlina jednou (73 %), dvakrát (21 %), třikrát (4,8 %), čtyřikrát (1,1 %) a šestkrát (0,1 %). Frekvence dvou kýl na strom byla vyšší v poloze nad 800 m n. m., frekvence čtyř a více trhlin se projevila na stromech v poloze m n. m. Věk porostů v době jejich zatížení námrazou (1995/1996) byl významným atributem potenciálního poškození ve formě ohnutí, zlomů a vývratů. Podíl mrazových trhlin se zastoupením profiloval v mladších porostech (do 10 let 34,2 %, let 43,8 %, let 47 % a let pouze 14,2 %) s tím, že frekvence počtu kýl na stromech se s věkem rovněž měnila. Do 20 let věku porostu měl dominantní postavení výskyt jediné trhliny (70 73 %) a 94% podíl byl ve starších porostech, při výskytu dvou trhlin se shodují porosty do 20 let na úrovni %, zatímco ve starších porostech jsme zaznamenali pouze 4,5 %. Výrazný ústup s věkem byl zaznamenán při výskytu tří kýl (6,5 4,6 3,6 1,5 %). Poloha mrazových trhlin se profilovala jednoznačně na jižní stranu (J 46,7 %, JV + JZ 22,1 %), severní polohy byly méně významné (S 10,3 %, SV + SZ 8 %), V 3,7 %, Z 9 % (obr. 2). Střed mrazové trhliny se nejčastěji nacházel v profilu kmene břízy mezi cm (25,2 %). Ve shodné frekvenci (18 %) byly trhliny v sekci pod tímto prostorem a nad ním. Výše v profilu kmene se projevil výrazný pokles mrazových trhlin ( cm 14,3 %, cm 11,2 % a v následujících třech 25 cm dlouhých sekcích 7,5 4,7 0,8 %). Celková poloha středu mrazové trhliny měla sestupný trend s narůstající nadmořskou výškou (obr. 3). Délka mrazové trhliny nevykazuje statistickou závislost na nadmořské výšce, věku, ale zkracuje se s počtem vyskytujících se trhlin na kmeni a na severní expozici. S nadmořskou výškou se snižovala poloha středu trhliny (obr. 4) a délka trhliny s věkem (obr. 5). Šířka mrazové trhliny se snižovala pouze při vyšší četnosti trhlin na vzorníku. Vliv světové strany na šířku trhliny byl klesající od jižních poloh k severním (obr. 6). Přítomnost mrazových trhlin je spojena s hnilobou březovníka obecného [Piptoporus betulinus (Bull. ex Fr.) Karst.], který prostupuje od místa poranění až do úrovně koruny, šíří se relativně rychle a ovlivňuje růstové vlastnosti. Výčetní tloušťka jedinců odumřelých a živých, živých s mrazovými trhlinami a bez nich klesla s nadmořskou výškou (obr. 7), přičemž průměrná výčetní tloušťka stromů s mrazovými trhlinami byla nižší (obr. 8). Znamená to, že více než samotná poloha je významný výskyt hniloby u napadených stromů, který nejen ovlivňuje růstové vlastnosti, ale byl i významným činitelem podílejícím se na odumírání poškozených porostů a jejich rozpadu. Vyslovená hypotéza byla potvrzena. Corresponding author: Prof. Ing. Emanuel Kula, CSc., Mendelova zemědělská a lesnická univerzita v Brně, Lesnická a dřevařská fakulta, Lesnická 37, Brno, Česká republika tel.: , fax: , Kula@mendelu.cz 356 J. FOR. SCI., 52, 2006 (8):

22 JOURNAL OF FOREST SCIENCE, 52, 2006 (8): Occurrence, development and economic importance of Phratora (= Phyllodecta) vitellinae (L.) (Coleoptera, Chrysomelidae) J. Urban Faculty of Forestry and Wood Technology, Mendel University of Agriculture and Forestry Brno, Brno, Czech Republic ABSTRACT: The paper summarizes results of the study of the occurrence, development and harmfulness of Phratora (= Phyllodecta) vitellinae (L.). The majority of studies was carried out in 1998 to 2005 in riparian and accompanying stands of the Svitava and Svratka rivers in the region of Brno and in a laboratory. Imagoes leave hibernation hiding places at the end of April and at the beginning of May. In captivity, they lived on Salix fragilis about 2.5 months damaging on average 28.6 cm 2 leaf blades and laying on average 293 eggs. In the excessively warm growing season of 2005, imagoes lived about 3.5 months after hibernation, however, already after one month of feeding they fell in a month diapause at the beginning of June. Before its start, they damaged on average 12.8 cm 2 (after the diapause 14.4 cm 2 ) leaves and laid on average 389 eggs (of this number, 260 eggs before and 129 after the diapause). Larvae damage about 4 cm 2 leaves during 2 to 3 weeks (in the laboratory during 12 to 13 days). After 2 to 3 weeks (in the laboratory after 10 to 12 days) from the cessation of feeding young beetles appear on trees. Imagoes of the 1 st generation occur from mid-june to the beginning of October. During about 55 days of life, they damaged 19 cm 2 leaves and laid on average 182 eggs. Imagoes of the 2 nd generation occur from mid-august to the end of the growing season. After 10 to 14 days of feeding (without previous copulation), they take shelter in wintering places. In the laboratory, however, these imagoes damaged about 19 cm 2 leaves during 2 months and laid about 190 eggs. Wintering places were looked up by imagoes of the 3 rd generation which damaged on average 4.2 cm 2 leaves before hibernation. In the Czech Republic, P. vitellinae is usually bivoltine the 2 nd generation being always incomplete. Keywords: Chrysomelidae; Phratora (= Phyllodecta) vitellinae; occurrence; host species; development; generation conditions; harmfulness In the region of the CR, 5 species of chrysomelids of the genus Phratora Chevr. (= Phyllodecta Kirby) occur. Members of the genus create morphologically, biologically and ecologically a uniform group which is distributed mainly in the northern temperate zone. Their phylogenesis broadly coincides with the species spectrum and chemical composition of host species. As for nutrition, particular chrysomelids specialized in various degree to species of the family Salicaceae. P. vittellinae (L.) ranks among the most abundant and forestry-most important domestic members of the genus and of the whole family of Chrysomelidae. This chrysomelid occurs on many species of willows (Salix spp.) and poplars (Populus spp.). Under favourable conditions, it reproduces often on a mass scale and then heavily damages particularly young trees. Temperate and dry winters and excessively dry and warm springs participate in its activation. In the CR, population densities of P. vitellinae and a number of other phyllophagous species of chrysomelids generally increased in obvious connection with climatic-meteorological anomalies and the primary physiological weakening of trees during last decades. The occurrence, biology and harmful- Supported by the Faculty of Forestry and Wood Technology, Mendel University of Agriculture and Forestry Brno within the Research Plan MSM No J. FOR. SCI., 52, 2006 (8):

23 ness of some forestry-important chrysomelid species which gradated in Moravia recently were studied by Urban (1997, 1998a,b, 1999, 2000, 2005, 2006, etc.). This paper deals with findings obtained through field and laboratory studies of P. vitellinae. P. vitellinae occurs in the best part of Palaearctic and Nearctic zoogeographic regions. Thus, in principle, it is a Holarctic (according to Warchalowski 1973 Eurosiberian) species its centre of distribution being the Eurosiberian subregion. Already Calwer (1876) mentioned its occurrence in Germany, Hungary, Illyria (i.e. in the part of the present Croatia and Slovenia), Italy, France, England and Sweden. According to Escherich (1923), the area of its occurrence extends from France up to the mouth of the Amur river and from the Caucasus up to Lapland and in North America. Hellén et al. (1939) rank it among broadly distributed Fennoscandinavian species (including Karelia and Lapland). According to the author, it is even more distributed there than a sympatric species P. vulgatissima (L.). It is also considered to be a very abundant Holarctic species with similar economic importance as P. vitellinae. Considering rather similar ecological requirements both species occur often in the same localities and even on the same species. Arnoldi et al. (1955) complete the natural range of P. vitellinae mentioning it in Algeria, whole Europe and the Caucasus (with the exception of the steppe zone of the European part of the former USSR), Siberia, the seabord of Far East and North America. Similarly, the geographical occurrence of P. vitellinae is mentioned by Vasiljev et al. (1974). According to the authors, the chrysomelid was also found in Kamchatka and the Kurile Islands. Its vertical distribution reaches from lowlands up to high locations above the forest limit (Maisner 1974). P. vitellinae together with P. vulgatissima and Plagiodera versicolora (Laich.) belong to the most abundant and most harmful mainly blue or green coloured chrysomelids on tree species from the family Salicaceae. Various aggregate publications and specialized entomological and entomological-forest protection papers deal with its occurrence, development, harmfulness and possibilities of protection. During last three decades, about 150 scientific papers on P. vitellinae were published in Europe (particularly in Belgium, Great Britain, Germany, Finland and Sweden). The greatest attention was paid to its feeding selection, i.e. the effect of physical properties and chemical composition of leaves of host species on the occurrence, fecundity, rate of development and mortality. According to Ratzeburg (1839), P. vitellinae often damages leaves of willows. According to Kaltenbach (1874), it occurs abundantly on smooth-leaf willows and Populus nigra L. For example, Henschel (1876) and Kuhnt (1913) rate the beetle among generally leaf-eating species on willows while Calwer (1876) rate it among leaf-eating species on P. tremula L. A number of authors (Reitter 1912; Fleischer ; Rubner et al. 1942; Javorek 1947; Medvedev, Šapiro 1965; Mohr 1966; Warchalowski 1973; Maisner 1974; Vasiljev et al. 1975) limits itself to the general statement of the beetle occurrence on willows and poplars. However, e.g., Nüsslin and Rhumbler (1922) mention that P. vitellinae attacks in preference S. purpurea L. and allegedly also S. caprea L. and poplars. S. triandra L. is considered to be resistant (however, in case of lack it damages reputedly its hybrids). Also Escherich (1923) mentions that not all species of willows are attacked. He regards S. purpurea, S. viminalis L. and S. caprea as very favourite and S. triandra as quite neglected. In Austria, Pernersdorfer (1941) found P. vitellinae on S. fragilis L., S. purpurea and P. tremula. However, she observed intense feeding only on S. fragilis. Except S. purpurea also S. alba L. var. vitellina (= S. vitellina L.) (Blunck et al. 1954) are main host plants. According to Arnoldi et al. (1955) it occurs not only on S. purpurea, S. viminalis and poplars but also on Alnus hirsuta (Spach.) Rupr. Vasiljev et al. (1974) rank S. purpurea, S. viminalis, P. nigra L., P. nigra L. var. italica and sometimes also Alnus sp. among its most searched hosts. The chrysomelid is sporadically reported from Betula sp. (Schaufuss 1916; Roubal ). It was obviously related to accidental finds of imagoes of a wintering generation during their posthibernation migrations. P. vitellinae is known particularly as the pest of willows in osier plantations, i.e. in willow stands with a 1(2)-year rotation intended for the production of wicker for basketry (Escherich 1923; Rubner et al. 1942; Schwerdtfeger 1944; Živojinovič 1948; Nejedlý 1950; Blunck et al. 1954; Gäbler 1955; Schnaider 1957; Wagner, Ortmann 1959; Kadłubowski, Czalej 1962; Urban 1982, etc.). Recently, it becomes to occur as the pest of willows and poplars in energy plantations. Young intensively growing trees create also optimum conditions for the gradation of many other pest herbivores. P. vitellinae frequently heavily damages plantations of S. cv. Americana imported to Europe (Wagner, Ortmann 1959; Kadłubowski, Czalej 1962, etc.). Also Czerniakowski (2002) found its dominant proportion (within the genus Phratora) in a plantation of S. cv. Americana. In agreement with Köpf et al. (1998) the author assumes that it refers to a species with rather broad trophic plasticity. 358 J. FOR. SCI., 52, 2006 (8):

24 Goidanich (1983) mentions damage to willows particularly at medium and higher altitudes of Italy. According to Rowell-Rahier and Pasteels (1982), P. vitellinae occurs mainly on S. nigricans Sm. (= S. myrsinifolia Sal.) the leaves of which contain phenolic glucosides (mainly salicin) and usually do not occur on the abaxial face densely covered with trichomes. The author ranks S. cinerea L. and S. caprea among neglected species not containing these secondary metabolites, however, their leaves are densely covered with trichomes on the abaxial face. Larvae of P. vitellinae excrete in danger a defensive secretion from 2 pairs of thoracic and 7 pairs of abdominal dorsal glands containing salicylaldehyde which is derived from salicin in leaves of host species (Soetens 1997; Pasteels et al. 1983, 1988, 1990; Brückmann et al. 2002, etc.). The larval secretion is considered to be a defensive substance against nonspecific predators and parasitoids. Hilker (1989) demonstrated that the secretion of a chrysomelid operated as a repelent even on conspecific imagoes and imagoes of other competitive phyllophagous species of chrysomelids (e.g., as against Plagiodera versicolora). Active social behaviour of larvae of P. vitellinae is an ecological adaptation to physical and chemical properties of host species and to the protection from some natural enemies and unfavourable weather (Grégoire 1988). The gregariousness increases the defence of particular larvae both through the cumulative effect of toxins and due to conspicuity and colour polymorphism of groups of larvae created by the fancy mixture of white and black individuals. The gregariousness also increases rather limited potential of egg larvae to bite their way out through the intact leaf epidermis. A mutual contact and tigmotaxis stimulate the feeding of larvae inhibiting their movement (Grégoire 1988). The secretion increases the immunity system of larvae showing antibacterial and fungicidal effects (Gross et al. 2004). Unlike larvae imagoes of P. vitellinae synthesize their own defensive substances incorporating them into eggs. However, the absence of salicin as the precursor of salicylaldehyde is not decisive for food intake. It is, for example, demonstrated by laboratory experiments of Rowell-Rahier and Pasteels (1982) where larvae willingly consumed leaves of S. caprea deprived of trichomes, however, did not excrete secretions containing salicylaldehyde. The secretion of salicylaldehyde was restored after the addition of salicin to leaves. Rowell-Rahier (1984a) mentions P. vitellinae mainly on S. nigricans and S. hegetschweileri Heer, less frequently on S. purpurea, P. tremula and P. trichocarpa Torr et Gray P. deltoides Marsh. Defoliation obviously does not induce the resistance of S. nigricans to P. vitellinae. Rowell-Rahier (1984b) found (in the laboratory) the following order of host plants according to the decreasing trophic preference: S. nigricans, S. purpurea, P. nigra, P. tremula, S. alba, S. caprea and S. cinerea. Leaves of S. nigricans are richest in salicin and leaves of three least consumed willows are densely pilose on the abaxial face. In populations of P. vitellinae obtained from geographically different localities in Central Europe, there are certain dietary specializations which can be induced by laboratory conditions. The resistance of some poplar clones to P. vitellinae was studied by Finet and Gregoire (1981, 1982), Finet et al. (1982, 1983) in a glasshouse, nature and laboratory. The authors proved the very high resistance of a clone P. nigra Ghoy 1 and the considerable resistance of a clone P. trichocarpa Fritzy Pauly. On the other hand, clones P. trichocarpa deltoides Unal and Beaupré were very disposed. According to Finet and Gregoire (1981) imagoes are equally active at night as during the day most active being at dusk. Trophic relationships between P. vitellinae and the spectrum and concentration of phenolic glucosides of 4 autochthonous and 4 introduced species of willows were studied by Tahvanainen et al. (1985) in Finland. The chrysomelid preferred a native smooth-leaved S. nigricans the leaves of which contained extraordinary high concentrations of phenolic glucosides (mainly salicortine and salicin). It consumed intensively medium-densely pilose leaves of imported S. cv. Aquatica and S. dasyclados Wimm. with the medium concentration of glucosides. P. vitellinae refused to consume smooth leaves of S. triandra which contain the medium concentration of glucosides but a different dominant glucoside salidrosid. Also S. pentandra L. the glabrous leaves of which contain considerable amounts of mostly little known or unknown glucosides was rather resistant. Sericeous pilose leaves of S. viminalis and glabrous leaves of S. phylicifolia L. show the low general concentration of phenolic glucosides. The trophic affinity of imagoes and larvae to S. phylicifolia is evidentially lower than that to S. pentandra (Rank et al. 1998). S. fragilis is characterized by the high concentration of usual glucosides and glabrous leaves. Therefore, S. fragilis is far more damaged than S. alba the leaves of which are densely pilose being characterized by the low concentration of phenolic glucosides (Soetens, Pasteels 1988; Soetens et al. 1991). During oviposition, females of P. vitellinae highly preferred S. fragilis to S. viminalis and S. dasyclados (Denno et al. 1990). Nevertheless, larvae developed well on leaves of J. FOR. SCI., 52, 2006 (8):

25 S. viminalis which were poor in salicylates. Considering the absence of a defensive secretion these larvae are, however, vulnerable towards general predators. Effects of phenolic glucosides on the trophic behaviour of P. vitellinae were also demonstrated by Kolehmainen et al. (1995), etc. Hallgren (2003) and Hallgren et al. (2003) dealt with the effect of the hybridization of S. caprea and S. repens L. on the concentration of secondary compounds in leaves and on insect herbivores (including P. vitellinae). Interactions between P. vitellinae and a primary host plant S. nigricans (or secondary host plant S. phylicifolia) and P. vitellinae in the environment with elevated UV radiation were studied by Veteli et al. (2002). An increase in UV-B radiation did not affect any measurable characters of the food quality. However, the abundance of imagoes on treated trees was higher than on control trees. While the growth of larvae was not affected by UV radiation on treated trees of S. nigricans, on S. phylicifolia was retarded. Thus, the authors assume that specialized herbivores (i.e. also P. vitellinae) can be more sensitive to secondary changes in secondary hosts than in primary hosts. Veteli et al. (2002) dealt with the study of the effect of increased temperature and CO 2 concentration on the metabolism of S. nigricans and feeding of imagoes (including the growth of larvae) of P. vitellinae. With the increase of both climatic elements the above-ground biomass increased and the total concentration of phenolic substances in leaves decreased. With the increased CO 2 concentration the content of water and N in leaves decreased. With the increased CO 2 concentration the consumption of food and the period of life of imagoes can increase and the relative rate of the growth of larvae decrease (unlike temperature accelerating the growth of larvae). Some other problems of the biology of P. vitellinae including its life cycle, natural enemies, harmfulness and control were also studied. For example, egg laying and their embryonal development were studied by Zachvatkin (1967). Temperature requirements of some chrysomelids (including P. vitellinae) were determined by Kadłubowski and Dudik (1968). Kendall and Wiltshire (1998) mentioned the life cycle of P. vitellinae in a plantation of S. viminalis. Sage et al. (1999) described the posthibernation migration from wintering habitats to poplar plantations and the secondary colonization of plantations. According to Peacock et al. (2003) ecological niches of P. vitellinae and P. vulgatissima are considerably different due to differences in the food preference. However, both species occur sometimes together and on the same tree species (in Great Britain, e.g. on S. burjatica Nas. Germany and in early spring on P. trichocarpa Torr et Gray Trichobel) and sometimes even mate with one another. However, the reproduction isolation of both species is perfect because larvae hatch from laid eggs sporadically and always die during a week (Peacock et al. 2004). An unexpected ecological role of the secretion of P. vitellinae larvae in the trophic behaviour of a hover fly Parasyrphus nigritarsis (Zett.) (Syrphidae) was demonstrated by Köpf et al. (1997). Larvae of this specialized predator are surprisingly attracted both by the secretion of larvae of the chrysomelid and by pure salicylaldehyde (i.e. the main component of the secretion). Thus, the hover fly uses the chrysomelid secretion to localize its prey. For general predators, such as predaceous ants salicylaldehyde in the chrysomelid secretion is an effective repellent (Pasteels et al. 1983). Thus, larvae of P. vitellinae on S. nigricans are killed by predaceous ants less than general leaf-eating herbivores on S. phylicifolia (Sipura 2002). According to Palokangas and Neuvonen (1992) P. vitellinae is attacked by spiders much less on P. tremula and S. nigricans than P. polaris Schn. on Betula pubescens Ehrh. (free of glucosides). Receptivity of 5 species of imagoes (including P. vitellinae) to Phylloscopus trochilus L. (Aves) was tested by Lundvall et al. (1998) in an aviary. In species with the different type of chemical defence feeding on Salix, birds destroyed somewhat more Galerucella lineola (F.) than P. vitellinae and Gonioctena viminalis (L.). In another experiment with chrysomelids of the same type of defence birds preferred Phratora polaris Schn. (from Betula sp.) to P. vitellinae (from Salix sp.). They looked for least Chrysomela lapponica L. (with a marked aposematic colour). P. vitellinae is an economically important pest of young willows and poplars (Tuinzing 1946; Györfi 1952; Charvát, Čapek 1954; Timčenko, Treml 1963; Kasap 1988; Sage, Tucker 1997; Glavas et al. 1997; Gruppe et al. 1999; Aslan, Ozbek 1999; Jaskiewicz et al. 2004, etc.). Recently, problems of the effective control of the pest were, e.g. dealt with by Attard (1979), Jodal (1985), Allegro (1989), Moraal (1989) and Karp (2003). MateriAl and methods The majority of field and laboratory studies was carried out in 1998 to In nature, the chrysomelid was studied mainly in riparian and accompanying stands of the Svitava and Svratka rivers in the region of Brno. The investigated stands are situated at an altitude of 190 to 230 m. Mean annual tem- 360 J. FOR. SCI., 52, 2006 (8):

26 peratures are 8.4 C and mean annual precipitation 547 mm. In the mixture of species of various age the family of Salicaceae was represented, e.g. by S. fragilis, S. rubens Schr., S. alba, S. triandra, S. purpurea, S. rubra Huds., S. viminalis, P. nigra, P. nigra var. italica, P. alba, P. tremula, P. canescens (Ait.) Sm., etc. Field inspections were carried out throughout the growing season, namely usually in 2-week intervals. Occasionally, trees were also checked in forest stands of the Křtiny Training Forest Enterprise and of the Židlochovice Forest Enterprise. In 1969 to 1975, integrated research was carried out into insect pests in 6 osier plantations (with 12 cultivated clones of willows) in Moravia. Urban (1982) briefly reported on the occurrence, bionomy and harmfulness of P. vitellinae. Findings from nature were compared with results of extensive laboratory rearings on S. fragilis and tentatively also on some other species of Salicaceae. For the rearings, glass vessels of a diameter of 10 (20) cm and height 5 (10) cm were used. Into the vessels, fresh leaves (more sporadically 5 to 15 cm long filiaged sections of 1-year-old shoots) were put in regular 2-day intervals (in rearings of larvae in 1-day intervals). These were always taken from the same tree and from the same part of a crown. Leaf petioles (or lower ends of shoot sections) were wrapped by a bit of moistened cotton wool. Less frequently, they were put into small vessels with water necks of the vessels being sealed by cotton wool. In individual and mass laboratory rearings of imagoes carried out throughout the growing season the damaged area of leaves was regularly determined every second day using planimetry. The number of laid eggs (including the number of eggs destroyed by imagoes) and the number and localization of egg groups on a leaf blade were recorded. The length was measured of male and female bodies and in dead females, the number was determined of unlaid eggs using the microscopic dissection of ovaries. In 2005, the number of frass pellets and their dimensions were recorded. During long-term rearings, findings were obtained on the life of imagoes of particular generations and the course of feeding and egg laying. Attention was paid to the preimaginal (particularly larval) development of P. vitellinae. Larvae were reared on leaves of S. fragilis (or other species of Salicaceae), viz 1 to 30 pieces from May to October. Every day, the size was recorded of the instar of larvae as well as mortality and damaged leaf area. Instars were determined by means of micrometry according to the width of cranium. In 2005, the number and dimensions of frass pellets of larvae of particular instars were recorded daily. Earth for pupation was put in the bottom of vessels for growing up larvae. In case of the small volume of earth it was possible to observe pupation and maturation of newly hatched imagoes through glass walls of vessels in pupal chambers. results and discussion Host species Tahvanainen et al. (1985), Soetens et al. (1991), Kolehmainen et al. (1995), etc. note that chemical and physical characteristics of leaves of host species are a good indicator of the abundance of insect herbivores. Unlike P. vulgatissima and Galerucella lineola a chrysomelid P. vitellinae occurs most frequently on species with smooth leaves on the abaxial face and with leaves medium-rich and rich in phenolic glucosides. The feeding selection is in correlation with the content and species spectrum of these compounds. The absence or lack of secondary substances in the food of P. vitellinae (e.g. in leaves of S. viminalis and S. caprea) is not critical for food intake being, however, decisive for the defensive secretion of larvae. Relation to glucosides manifests in the trophic specialization which evidently did not reach such a degree as in other members of the genus Phratora. In osier plantations in Moravia, P. vitellinae was found most frequently on S. purpurea, S. rubens and frequently on S. viminalis and S. rubra and S. cv. Americana. In riparian and accompanying stands of the Svitava and Svratka rivers in the Brno region, the chrysomelid occurred most frequently on S. fragilis, S. purpurea and S. rubens, less frequently on S. viminalis and P. tremula. It was not found on S. triandra, S. caprea, S. alba, P. nigra and P. canescens. Neglected species are characterized either by the considerable concentration of unusual glucosides in leaves perhaps even by their small amount or absence or densely pilose abaxial face of the leaf blade. Findings from nature generally agree with results of laboratory tests. In addition to the above mentioned primary host species imagoes consumed, in case of the lack of more suitable food, also S. caprea, S. alba and S. acutifolia but always refused S. triandra. If imagoes had the possibility of selection they always willingly consumed S. fragilis, S. purpurea and S. rubens and refused S. viminalis, S. caprea, S. alba, P. nigra, P. nigra var. italica and P. alba. In the laboratory (i.e. under conditions of the absence of natural enemies), larvae developed quite normally on S. viminalis and S. caprea. For example, larvae of the 1 st generation fed on S. viminalis on average J. FOR. SCI., 52, 2006 (8):

27 12 days. Within the time, they damaged on average 2.7 cm 2 leaves and produced on average 400 frass pellets. On S. caprea, larvae of the same generation fed on average 14 days damaging on average 3.0 cm 2 leaves and defecated on average 300 frass pellets. On a trophically optimum host species S. fragilis, larvae developed on average 12 days, damaged on average 2.9 cm 2 leaves and defecated on average 600 frass pellets. There are many factual data concerning the disposal of trees to P. vitellinae damage some of them differing diametrically from each other. Therefore, determination of the objective order of species according to the degree of resistance is not yet possible. In a simplified fashion it is possible to state that S. fragilis, S. nigricans and S. rubens rank among the most favourite domestic species. In our country, S. purpurea and obviously also P. tremula are damaged very frequently. Nevertheless, Rowell- Rahier (1984a) assumes that S. purpurea is not suitable for egg laying and the development of larvae. S. viminalis, S. rubra and S. cv. Americana are often damaged. Evidently only rarely or even sporadically (e.g. in insect outbreak on primary hosts), the pest occurs on S. caprea, S. alba, S. alba var. vitellina and S. cinerea. For example, S. triandra, P. alba, some clones of P. nigra etc. are quite resistant. Hibernation of beetles and leaving winter habitats According to Nüsslin and Rhumbler (1922) imagoes of P. vitellinae winter in the ground, in bark fissures, between buds in whorls of pines and in hollow plant stems. Localization of hibernation places was specified by Escherich (1923). According to the author imagoes winter mostly on protected elevated places, viz e.g. between non-fallen twisted leaves, between terminal buds of pines, in hollow stems of plants, under loosened bark of trees and in corridors of bark beetles. For example Wagner and Ortmann (1959) mention the most frequent wintering under bark, in dry leaves on trees, in fissures of fences and stubs of sticks or in soil between fallen leaves. According to Schnaider (1972), in autumn imagoes fly from osier plantations to neighbouring forests and shrubs where they winter in fissures of bark, in litter etc. Kendall and Wiltshire (1998) found imagoes in cancerous tumours on willows in plantations and in fissures of wooden posts and in cardboard hibernation traps in the surroundings up to 100 m from a plantation. Also a related P. vulgatissima was found by the authors mostly in the close vicinity of plantations (within 100 m) while part of Table 1. Average week area of leaves of S. fragilis damaged by imagoes of P. vitellinae after wintering. Imagoes were caught at the beginning of the period of feeding in Bílovice nad Svitavou (in 1999* in Adamov and in 2001 in Žďár nad Sázavou). Laboratory rearings on S. fragilis Week Period * 2001 (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) 1 st 4 10 May nd May ?? 3 rd May th May th 1 7 June th 8 14 June th June th June th 29 5 July th 6 12 July th July th July th 27 2 August th 3 9 August th August 0 0 Total (36.1) Number of / 8/8 10/6 4/6 3/4 362 J. FOR. SCI., 52, 2006 (8):

28 Fig. 2. A detail of damage to the abaxial face of leaves of S. fragilis by imagoes of P. vitellinae. Bílovice nad Svitavou, 15 May 1998 Fig. 1. Imagoes of Phratora vitellinae during feeding on the abaxial face of leaves of Salix fragilis. Laboratory rearing, 18 May 1998 imagoes hibernated in plantations (e.g. in fissures of willow stems, in dead vegetation and in litter). It is possible to suppose that with the increasing distance of suitable hibernation shelters imagoes will show higher tendency for hibernation in plantations (Kendall, Wiltshire 1998). Sage et al. (1999) mention the considerable flying activity of imagoes of a wintering generation. The authors found that majority of imagoes wintered under the bark of fullgrown coniferous and broadleaved trees at a distance up to 250 m from plantations. According to the authors imagoes winter in shelters either individually or in variously numerous close aggregations (up to 500 beetles). With the increasing length of a light day and growth of spring temperatures beetles of the wintering generation become active. Usually at the end of April (Achremowicz 1960) (under conditions of colder weather already at the beginning of May Kadłubowski, Czalej 1962), they occur on host species. According to Escherich (1923), Živojinovič (1948), Nejedlý (1950), Schnaider (1972), etc. beetles occur after hibernation usually in April, according to Maisner (1974) sporadically already in March. According to Sage et al. (1999) they fly mainly on the border 8-meter-wide zone of plantations where about 80% beetles are accumulated. During next about 3 weeks, the beetles gradually spread in a stand (secondary dispersion). In the region of southern Moravia, beetles leave wintering shelters usually at the end of April; in the region of central Moravia usually at the end of April and at the beginning of May. Beetle feeding after wintering After the very long (about 8 months) period of hibernation, beetles are extremely starved and weakened. Therefore, as soon as the chrysomelids reach host species they begin to damage unfolding (or newly flushed) leaves. In case of the retarded start of budbreak they damage buds and fine bark at the end of young shoots. They eat petty irregular holes into the blade of leaves from the abaxial face (%) st 3rd 5th 7th J. FOR. SCI., 52, 2006 (8): Les 08_06 3 Urban.indd 363 9th 11th 13th 15th 17th 19th Week Fig. 3. The course of damage to leaves of S. fragilis by imagoes of P. vitellinae after their hibernation in nature (in % of the total damaged area). Laboratory rearings, 1999 (dash line), 2005 (solid line) :34:31

29 Table 2. Average week area of leaves of S. fragilis damaged by imagoes of P. vitellinae after wintering. Imagoes were caught at the beginning of the period of feeding on S. fragilis in Bílovice nad Svitavou. Laboratory rearings on S. fragilis, 2005 Week Period Rearing 1 Rearing 2 Rearing 3 Rearing 4 Rearing 5 (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) 1 st 4 10 May nd May rd May th May th 1 7 June th 8 14 June th June th June th 2 5 July th 6 12 July th July th July th 27 2 August th 3 9 August th August th August th August th 31 6 September th 7 13 September th September 0 0 Total Number of / 1/1 1/1 1/1 0/2 2/0 (in the laboratory, some 20% of the blade area also from the adaxial face). Budding leaves are always perforated by the beetles; newly unfolded leaves are partly perforated and partly skeletonized. Older leaves are always skeletonized, all venation of leaves and upper epidermis being left intact (Figs. 1 and 2). Beetles damage mostly semi-developed or newly unfolded leaves. Little flushed or old leaves are mostly neglected. Particularly old leaves are nutritionally the very poor source of food for beetles and their reproduction (Raupp, Sadof 1991). Holes in little developed leaf blades are on average about 4 mm long and 3 mm wide. Holes in full-grown blades are on average 0.9 mm long and 0.6 mm wide. Imagoes of the wintering generation caught in 1998 to 2001 at the beginning of the period of feeding lived in laboratory rearings on S. fragilis 1.5 to 3.5 (on average 2.5) months (Table 1). It follows (based on the average week leaf area damaged by male and female imagoes sex ratio 1:1) that imagoes took (%) Males Females (mm) Fig. 4. The body length of imagoes of P. vitellinae. Mean length of males is 3.95 mm and of females 4.2 mm 364 J. FOR. SCI., 52, 2006 (8):

30 Table 3. Average number (%) of frass pellets produced by imagoes of P. vitellinae during particular week intervals. Imagoes were caught at the beginning of the period of feeding on S. fragilis in Bílovice nad Svitavou. Laboratory rearings on S. fragilis, 2005 Week Period Rearing 1 Rearing 2 Rearing 3 Rearing 4 Rearing 5 No. (%) No. (%) No. (%) No. (%) No. (%) 1 st 4 10 May nd May rd May th May th 1 7 June th 8 14 June th June th June th 29 5 July th 6 12 July th July th July th 27 2 August th 3 9 August th August th August th August th 31 6 September th 7 13 September th September Total 2, , , , , Number of / 1/1 1/1 1/1 0/2 2/0 Mean length/width of a frass pellet (mm) Mean volume of one frass pellet (mm 3 ) Mean volume of frass pellets (mm 3 ) 0.8/ / / / / food more or less continually in these years (i.e. without long-term interruption). Imagoes damaged on average 19.2 to 36.1 (on average 28.6) cm 2 leaves. Within 1 hour, they damaged 0.1 to 2.7 mm 2 leaves. According to a preliminary investigation carried out by Finet and Gregoire (1981) imagoes Table 4. Mean number and volume of frass pellets (mm 3 ) produced by male and female imagoes of P. vitellinae (1:1) during feeding in particular months from damaged area of 1 cm 2. Mean volume of one frass pellet = mm 3. Laboratory rearings on S. fragilis, 2005 Imagoes Frass May June July August September October Mean After wintering 1 st generation 2 nd generation 3 rd generation number volume number volume number? volume? number? volume? J. FOR. SCI., 52, 2006 (8):

31 Number Fig. 5. Mean number of frass pellets produced by imagoes of P. vitellinae (sex ratio 1:1) in particular week intervals (after their hibernation in nature). Laboratory rearings on leaves of S. fragilis, st 3 rd 5 th 7 th 9 th 11 th 13 th 15 th 17 th 19 th 21 th Week feeding on P. trichocarpa P. deltoides damaged 2 to 2.5 mm 2 per 1 hour. In an extraordinary warm year 2005, imagoes fell (after 4- to 5-week feeding) into a spring/summer diapause taking about 3 to 5 weeks (Table 2) at the beginning of June. Diapausing imagoes lived 3 to 4 (on average 3.5) months (i.e. by one month longer) and damaged 13.0 to 36.8 (on average 27.2) cm 2 leaves (i.e. roughly the same area as non-diapausing imagoes). Before the diapause, they damaged on average 12.8 cm 2 (47.1%) and after the diapause 14.4 cm 2 (52.9%) (Fig. 3). With respect to on average larger size of females (Fig. 4) and much higher energy expenditures for the creation of eggs the average consumption of food in females is significantly (about 2.4 ) higher than in males. Imagoes spend nearly 90% time during food intake (Finet, Gregoire 1981). According to Peacock et al. (2003), there are certain geographical differences in the amount of consumed food in various host plants dimensions and weight of the body (including the percentage of fat in dry matter) being higher in northern localities. During feeding, imagoes produce 1,323 to 3,662 (on average 2,464) frass pellets of a total volume of 38.4 to mm 3 (Table 3). The number and the total volume of male frass pellets ranged in the lower half of the range (in males frass pellets in the upper half). Frass pellets are oblong, cylindrical to fusiform. At first, they are dark green, later dark brown to black. Their mean length is 0.8 mm and width 0.25 mm. Mean dimensions of male frass pellets are mm, those of female frass pellets mm (Table 3). The mean number and volume of frass pellets defecated by male and female imagoes during particular months (from damage of a size of 1 cm 2 ) significantly increases in August and September (Table 4). This increase is undoubtedly related to the trophic value of food gradually worsening during the growing season. The course of the defecation of imagoes of a wintering generation (sex ratio 1:1) in particular week intervals in 2005 (i.e. in the year with a month spring/summer diapause) is illustrated in Fig. 5. Last year s beetles occur on their host plants usually from the end of April to the beginning of September (in the laboratory till mid-august Fig. 6). It is probable that also in nature beetles can enter the diapause under extremely warm springs and after the change of climatic conditions to continue again in feeding. In such a case, the occurrence of last year s Beetles of 3 rd generation Pupae Larvae Eggs Beetles of 2 nd generation Pupae Larvae Eggs Beetles of 1 st generation Pupae Larvae Eggs Last year s beetles on willows May June July August September October Month Fig. 6. Basic scheme of the development of P. vitellinae on S. fragilis. Laboratory rearings, J. FOR. SCI., 52, 2006 (8):

32 Table 5. Mean period of feeding of P. vitellinae females of particular generations from the beginning of feeding on S. fragilis to the beginning of egg laying. Mean period of egg laying and the mean period of life of females (males) of particular generations after the end of oviposition (all in days). Laboratory rearings, Generation After wintering (without a spring/ summer diapause) of feeding to the beginning of oviposition Mean period of oviposition of life of ( ) after the end of oviposition (17) After wintering (with a diapause) (6) 1 st generation (12) 2 nd generation (?) 3 rd generation (9 + 14) (57) (after hibernation) (8/17/) Table 6. Mean number (%) of eggs laid by females of P. vitellinae during particular week intervals. Imagoes were obtained at the beginning of the period of feeding on S. fragilis in Bílovice nad Svitavou (in 1999* in Adamov and in 2001 in Žďár nad Sázavou). In dead females, the number of non-laid eggs was found by means of the dissection of ovaries. Laboratory rearings on S. fragilis Week Period * 2001 No. (%) No. (%) No. (%) No. (%) 1 st 4 10 May nd May ?? 3 rd May th May th 1 7 June th 8 14 June th June th June th 29 5 July th 6 12 July th July th July th 27 2 August th 3 9 August th August 0 0 Laid Non-laid Total Number of / 8/8 10/6 4/6 3/4 beetles on trees could extend until the beginning of October. Egg laying of the 1 st generation Eggs are created in meroistic ovarioles of a telotrophic type, viz only during feeding. There are 20 ovarioles in ovaries, i.e. 10 in each of the ovaries (Zachvatkin 1967). After about one-week feeding, imagoes mate on leaves for the first time and after next week females lay the first eggs (Table 5). Females which did not mate in spring lay always unfertilized eggs. Kendall and Wiltshire (1998) came to the same conclusion in a related P. vulgatissima. In localities under study, the first eggs of the first generation were found about 6 May. In the period of the most intensive egg laying (in the 2 nd half of May and in June), females lay eggs nearly every day. In the laboratory, females (without the spring/summer diapause) mated repeatedly and laid eggs for a period of almost 2 months. Females with the diapause reproduced for a period of 2.5 months (of this J. FOR. SCI., 52, 2006 (8):

33 (%) Fig. 7. The course of egg laying by females of P. vitellinae in particular 24 week intervals (after their hibernation in nature). Laboratory rearings on 18 leaves of S. fragilis, 1999 (dash line), (solid line) st 3 rd 5 th 7 th 9 th 11 th 13 th 15 th 17 th 19 th Week 1 month before the diapause and 1.5 months after the diapause) (Tables 6 and 7, Fig. 7). Male imagoes (without a diapause) lived in captivity on average 69 (female 57) days. Male imagoes (with a diapause) lived on average 93 (female 100) days (Table 8). Several days after the termination of egg laying imagoes died (Table 5). Males in rearings quaintly tried to copulate even with dead females. Females place eggs into 8- to 20-member groups on the abaxial face of leaves (Figs. 8 and 9). The majority of eggs (about 52%) is laid on the basal third of a leaf blade (Fig. 10). A group of eggs of a medium size is laid within 30 min. In laboratory rearings, females of a wintering generation laid only about 43% eggs on the abaxial face of leaves. Almost 45% eggs were laid on glass walls (mainly on caps) of rearing Table 7. Mean number (%) of eggs laid by females of P. vitellinae during particular week intervals. Imagoes were obtained at the beginning of the period of feeding on S. fragilis in Bílovice nad Svitavou. In dead females, the number of non-laid eggs was found by means of the dissection of ovaries. Laboratory rearings on S. fragilis, 2005 Week Period Rearing 1 Rearing 2 Rearing 3 Rearing 4 No. (%) No. (%) No. (%) No. (%) 1 st 4 10 May nd May rd May th May th 1 7 June th 8 14 June th June th June th 29 5 July th 6 12 July th July th July th 27 2 August th 3 9 August th August th August th August th 31 6 September th 7 13 September th September 0 0 Laid Non-laid (mean) Total Number of / 1/1 1/1 1/1 0/2 368 J. FOR. SCI., 52, 2006 (8):

34 Fig. 8. Egg-laying of P. vitellinae on the abaxial face of leaves of S. fragilis. Bílovice nad Svitavou, 15 May 1999 Fig. 9. Egg-laying of P. vitellinae on the abaxial face of leaves of S. fragilis. Eggs are covered on their surface by the secretion of accessory glands. Bílovice nad Svitavou, 11 May 1998 plates and 12% on the adaxial face of leaves (Table 9). In particular rearings, females laid eggs either individually or in small groups (on average 3.4 eggs each) (Table 10). Eggs of P. vitellinae are oval, grey-white to yellowish and glossy. Newly hatched eggs are 0.86 mm long and 0.43 mm wide, sticky on their surface. By means of an air-stiffening thickening secretion the eggs are stuck to a leaf, namely always flat, to serried double series with poles close to each other. The whole group of eggs is often covered with a thin transparent cuticle created by the stiffened secretion of accessory glands (Fig. 9). The cuticle cleaves closely on a chorion and sometimes partly passes to a leaf. During a one-week embryonal development in the laboratory, the mean length of eggs increases to 0.97 mm and width to 0.47 mm (Table 11, Fig. 11). Larvae usually do not hatch from eggs laid on walls of plates. It means that the close contact of eggs with the abaxial surface of the living leaf blade significantly stimulates the embryonal development. On the other hand, Zachvatkin (1967) claims that embryos allegedly develop even without additional moistening in eggs stuck on glass or miller silk. In rearings of imagoes, cannibalism which showed itself in the consumption of less than 3% eggs was observed only rarely (Table 12). Table 8. Mean period of life of male and female imagoes of wintering generations of P. vitellinae without a spring/summer diapause and with the diapause (days). Imagoes were caught on 4 May in Bílovice nad Svitavou (in 2001 on 18 May in Žďár nad Sázavou). The table also gives the mean number of laid eggs and the mean area of damaged leaves. Laboratory rearings on S. fragilis Sex/Number Imagoes without a spring/summer diapause Imagoes with a diapause Mean Males Females Mean Number of eggs Damaged area (cm2) Table 9. Localization of eggs of the 1st and 2nd generations of P. vitellinae on leaves of S. fragilis and on walls of rearing vessels. Laboratory study, Localization Eggs of the 1st generation Eggs of the 2nd generation No. (%) No. Abaxial leaf face 3, ,420 (%) 63.0 Adaxial leaf face Walls of rearing vessels 3, Total 7, , J. FOR. SCI., 52, 2006 (8): Les 08_06 3 Urban.indd :34:39

35 Table 10. Numerical proportion (%) of groups with the various number of eggs laid by females of P. vitellinae after wintering and by females of the 1 st generation. Laboratory rearings on S. fragilis, Number of eggs in a group Imagoes after wintering Imagoes of the 1 st generation number of groups (%) number of groups (%) Total 1, Based on laboratory studies the fecundity of females of the wintering generation of P. vitellinae on S. fragilis amounts to 191 to 546 (on average 341) eggs. Females which did not enter the spring/summer diapause laid 191 to 546 (on average 293) eggs (Table 6). Females which underwent a diapause laid 258 to 546 (on average 389) eggs. Before the diapause, they laid about 67% eggs and after the diapause about 33% eggs (Table 7). Ovaries of 52% naturally died females did not contain any developed eggs. In total 48% died females showed ovaries with 1 to 18 (on average 3) unlaid eggs. There are very few more accurate literature data on egg laying and fecundity of P. vitellinae. For example, according to Kaltenbach (1874), there are 13 to 18 eggs in groups, according to Blunck et al. (1954) and Schnaider (1957) 10 to 30 eggs, according to Escherich (1923), Živojinovič (1948), etc. about 20 eggs and according to Zachvatkin (1967) 8 to 36 eggs. The same number of eggs in groups (8 to 20) as this paper author gives Gäbler (1955). Kendall and Wiltshire (1988) found that in clutches of a related species P. vulgatissima there were on average only 10.2 eggs. According to available literature, females lay in total 200 to 300 eggs. Thus, based on our study, the average fecundity of females on S. fragilis (about 341 eggs) is significantly (on average 1.4 times) higher. It is known that imagoes of P. vitellinae prefer newly unfolded leaves near shoot tops while eggs of the chrysomelid are mostly laid on somewhat older leaves in proximal parts of shoots. According to Rowell-Rahier (1984a) females also sometimes lay eggs on old leaves at bases of shoots. Females are very movable and after laying one clutch (which takes only a short time) they quickly leave the lower space of shoots (Finet, Gregoire 1981). The marginalization of young leaves during egg laying limits 370 J. FOR. SCI., 52, 2006 (8):

36 Table 11. Length and width of eggs of P. vitellinae during embryonal development (mm). Laboratory rearing, Eggs Length Width from to mean from to mean Immediately after egg-laying In the half of an embryonal development Immediately before hatching Total (%) Table 12. Mean mortality of eggs (in % of the total number of eggs) caused by imagoes of P. vitellinae in common rearings of individuals of both sexes on leaves of S. fragilis. Laboratory rearings, Number of imagoes Eggs of the 1 st generation Eggs of the 2 nd and 3 rd generations Total Basal Middle Apical Leaf third third third blade Fig. 10. Localization of eggs of P. vitellinae on the abaxial face of leaves of S. fragilis. Laboratory rearings, 1998, 1999 feeding competition between imagoes and larvae and evidently also significantly decreases mortality caused by various (particularly insect) predators and parasitoids. Development of the 1 st generation After 8 to 11 days (according to Schnaider 1957 after 1 to 2 weeks) following the oviposition, egg envelopes crack at the cranium pole by a lengthwise fissure (reaching up to the half of the egg length) and egg larvae gradually hatch. In the laboratory, larvae hatched as early as after 6 to 7 days (Fig. 11). Newly hatched larvae are about 1.1 mm long, yellowish, with black head and legs. Their cranium is 0.39 to 0.46 mm wide (Fig. 12). During 2 to 3 hours, they show dye and then concentrate in the immediate vicinity of abandoned egg envelopes (Fig. 13). Newly hatched larvae never consume empty egg envelopes (as against larvae of Plagiodera versicolora, Chrysomela populi, etc.). They line closely side by side and soon begin jointly to skeletonize leaves. During this frontal feeding, they bite out minute irregular holes of an average length 0.3 mm and width 0.2 mm into leaf blades (Fig. 14). The holes usually reach up Day May June July August September October Month 7.8 Fig. 11. The mean period of embryonal development (dark columns) and the mean period of development of P. vitellinae from the end of feeding of larvae till the occurrence of imagoes on trees (light columns). Laboratory rearings on leaves of S. fragilis, 1999, 1999, 2001 and 2005 J. FOR. SCI., 52, 2006 (8):

37 Number Divisions (mm) Fig. 12. The width of cranium of larvae of the 1 st to the 3 rd instar of P. vitellinae (1 division = mm) (mm) st 2 nd 3 rd Imago Instar stadium Fig. 13. Larvae of the 1 st instar of P. vitellinae soon after hatching from eggs on the abaxial face of leaves of S. fragilis. Bílovice nad Svitavou, 22 May 1998 Fig. 14. Mean length of feeding marks (dark columns) and the width of feeding marks (light columns) of larvae of particular instars and imagoes of P. vitellinae on leaves of S. fragilis (mm). Laboratory rearings, 1998 to 2005 Table 13. Mean period of the development of larvae of the 1 st to the 3 rd instar of P. vitellinae on S. fragilis (in dyas and %). A period is given from hatching of larvae from eggs to the end of feeding. Laboratory rearings, Table 14. Mean leaf area of S. fragilis damaged by larvae of the 1 st to the 3 rd instar of P. vitellinae during May to September. Laboratory rearings, 2005 Instar Instar Period of development Mean (%) 1 st nd rd Mean May June July August September (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) (cm 2 ) (%) 1 st nd rd Total J. FOR. SCI., 52, 2006 (8):

38 Fig. 15. Larvae of the 1st instar of P. vitellinae during feeding on the abaxial face of leaves of S. fragilis. Bílovice nad Svitavou, 25 May 1998 Fig. 16. Larvae of the 2nd instar of P. vitellinae during feeding on the abaxial face of leaves of S. fragilis. Laboratory rearings, 4 June 1998 to the upper epidermis not disturbing the epidermis, leaf veins and vein anastomoses. Part of the leaf blade (about 20%) between particular feeding marks remains undamaged (Fig. 15). During 4.5 days, larvae reach a length of 2 mm (Table 13). Within the period, they damage on average 0.2 cm2 leaf blade of S. fragilis (Table 14). They produce on average 109 fusiform frass pellets of a mean length of 0.18 mm and width 0.07 mm (Tables 15 and 16). Grown up larvae of the 1st instar gradually leave the front of feeding larvae roughly 3 hours before ecdysis. Through the terminal part of their abdomen they attach themselves to the undisturbed part of the leaf blade. The ecdysis of larvae takes about 15 to 20 minutes. The cranium of larvae of the 2nd instar is 0.57 to 0.69 mm wide and the length of the larvae is 1.5 to 3.8 (on average 2.6) mm (Fig. 12). After about 2 hours of rest, newly hatched larvae begin to feed on the abaxial face of leaves. For a period of about 4 days, they jointly frontally bite out holes mm (Table 13, Figs. 14 and 16). They damage on average 0.5 cm2 leaf blades of S. fragilis and exclude on Fig. 17. Abaxial face of leaves of S. fragilis damaged by larvae of the 2nd instar of P. vitellinae. Laboratory rearings, 31 May 1998 average 168 fusiform frass pellets of a dimension about mm (Tables 14 to 16). They do not damage leaf venation and upper epidermis, however, they consume all leaf parenchyma (Fig. 17). Grown up larvae of the 2nd instar hatch on the abaxial face of leaves (near the place of feeding). Larvae of the 3rd (i.e. last) instar show cranium 0.86 to 1 mm wide and the length of the larvae is 3.4 to 6.2 (on average 5) mm (Fig. 12). For a period of on average 5 days, they bite out holes about mm (Table 13, Figs. 14 and 18) either jointly or in small close groups. They damage on average 2.1 cm2 leaf blades of S. fragilis and exclude on average 335 fusiform frass pellets of an average dimension of mm (Tables 14 to 16, Fig. 19). Also larvae of the 3rd instar skelotonize leaves from the abaxial face (in the laboratory, however, also from the adaxial face) and leave both the upper epidermis and venation intact. If we displace young larvae of the 3rd instar to the adaxial face of leaves then 80% larvae begin to skelotonize leaves from this face and about 20% larvae move to the abaxial face. Table 15. Mean number and volume of frass pellets produced by larvae of the 1st to the 3rd instar of P. vitellinae during May to September. Mean volume of frass pellets from 1 cm2 damaged leaf area (mm3/cm2). Laboratory rearings, 2005 Instar May June July No. (mm3) No. (mm3) 1st nd rd Total (mm3/cm2) 1.90 J. FOR. SCI., 52, 2006 (8): Les 08_06 3 Urban.indd 373 August September (mm3) No. (mm3) No. (mm3) , , No :34:44

39 Fig. 19. Abaxial face of leaves of S. fragilis damaged by larvae of the 3rd instar of Lochmaea capreae (L.) (left) and P. vitellinae (right). Laboratory rearings, 8 June 1996 Fig. 18. Larvae of the 3rd instar of P. vitellinae during feeding on the abaxial face of leaves of S. fragilis. Laboratory rearings, 11 June 1998 The first generation o larvae occurs in nature usually from the 2nd decade of May to mid-august (at a supposed spring/summer diapause probably till midseptember). The development of larvae (from hatching from eggs till a departure to the earth) takes 2 to 3 weeks, in the laboratory 12 to 13 days (Table 13, Fig. 20). Also Schnaider (1957) mentions the 2 3-week development of larvae. Within the period, larvae damage in total 2.8 to 5.6 (on average 4) cm2 leaves of S. fragilis, i.e. on average 7.4 times less than last year s imagoes after wintering (Table 14). Of the total damaged area larvae of the 1st instar damaged Table 16. Mean dimensions and volume of frass pellets defecated by larvae of the 1st to the 3rd instar and imagoes of P. vitellinae. Laboratory rearings on S. fragilis, Instar Length (mm) Width (mm) Volume (mm3) 1st nd rd Imago about 6%, larvae of the 2nd instar about 16% and larvae of the 3rd instar about 78% (Fig. 21). During feeding in May and June, larvae produce about 600 frass pellets, in July about 1,000 frass pellets (%) 90 Day May June July August September October Month Fig. 20. The mean period of development of larvae of P. vitellinae from hatching to the end of feeding the larvae (in days). Laboratory rearings on leaves of S. fragilis, 1998 to 2005 Les 08_06 3 Urban.indd st 2nd 3rd Instar Fig. 21. Leaf area of S. fragilis (in % of the total damaged area) damaged by larvae of the 1st to the 3rd instar of P. vitellinae (light columns). Dark columns depict the volume of frass pellets (in % of the total frass pellets volume). Laboratory rearings, 2005 J. FOR. SCI., 52, 2006 (8): :34:46

40 (%) st 2nd 3rd Instar Number of larvae Fig. 22. Mortality of particular instars of larvae (dark columns) and the total mortality of larvae of P. vitellinae in relation to the number of larvae in a group (light columns) (in %). The mean mortality of larvae is depicted by a dash line. Laboratory rearings on leaves of S. fragilis, 1998, 1999, 2001 and 2005 (Table 15). In rearings, on average 12.2% larvae died and the highest mortality showed larvae of the 1st instar (Fig. 22). With the increasing number of larvae in groups the mortality of larvae significantly decreased. Larvae of P. vitellinae live gregariously in stable close groups. Gregariousness facilitates the use of food sources (particularly in the initial period of feeding) and significantly strengthens the active and passive protection of larvae from natural enemies. As already mentioned the abundance of gregariously living larvae is in positive correlation with the high content of some phenolic glucosides and the low (%) June July August September Month Fig. 24. Mortality of prepupae, pupae and imagoes of P. vitellinae during their development in soil. Mean mortality is depicted by a dash line. Laboratory rearings, 2005 J. FOR. SCI., 52, 2006 (8): Les 08_06 3 Urban.indd 375 Fig. 23. Pupae of P. vitellinae in earth pupal chambers. Laboratory rearings, 8 June 1998 density of trichomes on the abaxial face of leaves. Glucosides (particularly salicin and salicortin) are used by larvae for the formation of salicylaldehyde which is the main compound of the defensive secretion of larvae secreted by eversible dorsal glands (Soetens, Pasteels 1988; Soetens et al. 1991, etc.). However, high concentrations of usual phenolic glucosides in leaves function on some other insect herbivores (e.g. P. vulgatissima and Galerucella lineola) as repellents (Kendall et al. 1996). Grown up larvae of the 3rd instar hide in the earth in the vicinity of host species. There, they create an oval pupal chamber with smoothed walls in the surface layer of soil (at a depth of 1 to 4 cm). In the chamber, they prepare about 4 days (in the laboratory 2.5 days) for pupation. The stage of a pupa (Fig. 23) takes about 8 days (in the laboratory 5.5 to 6 days) after which imagoes hatch. Imagoes remain in chambers about 4 days (in the laboratory 2.5 days) and then climb from soil and appear on host trees. In nature, imagoes of the 1st generation appear in the open after 2 to 3 weeks (in the laboratory after 10 to 12 days) after the termination of feeding (Fig. 11). Prepupae, pupae and young off-colour imagoes in pupal chambers are very sensitive to adequate soil moisture, structure and acidity. In excessively moist soil, they pupate at the very surface of soil (in the laboratory even on the surface) and in the acid humous soil mostly die. Also young imagoes can die in too dry and hard pupal chambers. Imagoes fail to bite out from them. In the laboratory, on aver :34:48