Genetic analysis of inherited reduced susceptibility of Fraxinus excelsior L. seedlings in Austria to ash dieback

Genetic analysis of inherited reduced susceptibility of Fraxinus excelsior L. seedlings in... Abstract Hymenoscyphus fraxineus causes massive dieback of common ash (Fraxinus excelsior L.) across populations. Previous common garden trials have revealed differences in susceptibility among individuals, suggesting a genetic basis for reduced susceptibility to the pathogen. The aim of the study was to identify any correlation between damage intensity of mature trees and their offspring in natural ash stands. Crown and shoot damage of naturally infected trees and saplings were assessed in two geographically isolated stands in Austria, and parentage analysis was carried out with molecular markers. No significant correlation could be detected using Spearman’s rank correlation analysis, suggesting that this approach would need higher numbers of adult–offspring pairs present to compensate for environmental and genetic variability at the sites. Likewise, an in situ estimate of heritability was nearly zero. The results thus support the results of other studies, i.e. that highly resistant individuals occur only at low frequency within European ash populations. While most of the previous studies were conducted in progeny trails or seed orchards and suggested a fairly strong genetic component, results from our investigation support a more complex mechanism of susceptibility differences under natural, heterogeneous conditions. Further analyses are needed to obtain a better understanding of gene–environment interactions and individual infection pressure of ash dieback in natural environments; such studies would need to be based on much higher sample numbers. Identification and propagating of non-susceptible ash trees is an important challenge to halt large-scale dieback of common ash. Introduction Common ash (Fraxinus excelsior L.) is an ecologically and economically important hardwood tree species widely distributed throughout temperate Europe. The species tolerates a wide range of environmental conditions from riparian to mountain habitats (Dobrowolska et al., 2011). Compared with beech (Fagus sylvatica L.), common ash is adapted to sites that are either moister or drier and it prefers more nutrient-rich soils (Marigo et al., 2000; Thomas, 2016). Although the seedlings are relatively shade tolerant, good light conditions are needed to compete with other broadleaved tree species (Marigo et al., 2000). Common ash has a complex reproductive system with male, female and hermaphrodite individuals and its pollen and seeds (samaras) are wind-dispersed (Morand-Prieur et al., 2003; Wallander, 2008). Recently, common ash has become highly threatened by a fungal disease (ash dieback) caused by the ascomycete Hymenoscyphus fraxineus (Gross et al., 2014), previously known as H. pseudoalbidus, with its anamorph stage Chalara fraxinea (Kowalski, 2006; Queloz et al., 2011). The pathogen was most likely introduced to Europe from East Asia (Zhao et al., 2012; Gross et al., 2014) and was first observed in Poland in the early 1990s (Przybyl, 2002; Pautasso et al., 2013). Since then, the pathogen has spread quickly towards Western and Southern Europe. The first ash dieback symptoms were observed in Austria in 2005 at a few sites in Lower Austria, Upper Austria and Styria, and subsequently the disease spread to all federal provinces of Austria by 2009 (Cech, 2006a, b; Cech et al., 2012). The symptoms of ash dieback range from necrotic lesions and wilting on ash leaves and petioles to necrotic lesions on branches, shoots and stem, leading to wood discoloration and crown dieback, and in most severe cases to death of the tree (Cech, 2006b; Bakys et al., 2009). Through wind dispersal, the fungus infects ash leaves during summer, growing into petioles and shoots and overwintering on infected petioles in the ground litter (Kräutler and Kirisits, 2012; Landolt et al., 2016). Ash dieback attacks trees of all age classes, although symptoms progress more severely and more rapidly in younger individuals, causing problems for natural regeneration (Keßler et al., 2012; Heinze et al., 2017). As the pathogen causes high mortality of ash trees, ash dieback has become a serious problem raising concerns about the future of the species. In several European countries, recent studies in seed orchards, progeny trials and in natural stands have aimed to increase understanding of the pathogen and its impact on common ash. Although there is an overall progression of the disease over the years in individual trees, several studies have revealed differences in symptom intensity among individuals in clonal or progeny trials, providing evidence for genetically based variation in susceptibility, and have suggested the existence of resistance or tolerance (McKinney et al., 2011; Pliūra et al., 2011; Kjær et al. 2012; Stener, 2013). Results from inoculation studies showed that the pathogen was less penetrative in tolerant individuals because of a better response mechanism of the host (McKinney et al., 2012; Cleary et al., 2014). The extent of possible resistance against ash dieback was previously estimated based on covariance and estimated heritability among half-sibs and full-sibs in progeny trails, suggesting that parents that are less susceptible transmit the resistance to their offspring (Kjær et al. 2012; Lobo et al., 2014). We planned and executed most of the study in the year 2015, when this was the state of knowledge. We thus expected that some insight may be gained from directly observing and comparing trees and their natural offspring in woodlands. It is as yet not entirely clear whether expression of tolerance in nursery seedlings is sufficiently related to that under natural site conditions. This would possibly allow us to see whether genetic selection is already at work there, i.e. if the seedlings that are present and healthy are so because they derive from tolerant adults. A positive result would significantly speed up the identification of tolerant genotypes, because the field testing of offspring, possibly over several years, requires significant resources and time. Clearly, being able to make inferences without having to grow seedlings and assessing them for several consecutive years would therefore be an advantage. Phenotypic selection of adults is likely not sufficient for identification of tolerant genotypes; but combined assessment of adults and their offspring at the same sites may increase precision (and so decrease the amount of field testing of seedlings), we thought. We also wanted to see whether the presence of healthy seedlings at a site would correlate with the presence of related, tolerant adults, i.e. whether genetic relationships can be established among young and adult plants in similar disease classes (like in a sort of inverse Janzen–Connell effect, see e.g. Comita et al., 2014), and thus check for any juvenile–adult correlations. Alternatively, site conditions may contribute strongly to disease prevalence; genotype by environment interaction (G×E) could interfere with this analysis; or size effects could also play a role (bigger trees might be more tolerant than smaller or non-dominant ones). We explored whether sites in Austria would be small and isolated enough so that sufficient numbers of adult-sapling pairs can be detected. Because of the strong interest in this pathogen system, new insights have been gained on heritability and on the genetic structure of tolerance since we conducted our study. Lobo et al. (2015) and Muñoz et al. (2016) have furthered our understanding and found out that while there is a family component of tolerance in open-pollinated offspring, there is also great variability within families, thus requiring large sample numbers for sizing the effects. Several possible mechanisms of the genetic basis of tolerance have very recently also been described with the help of whole-genome sequencing (Sollars et al., 2017). We thus looked at our data again, with a view on how to improve our approach, if possible at all. In this study, we investigated if transmission of resistance to ash dieback from mature trees to saplings is detectable in two naturally regenerating common ash populations in Austria. We used parentage analysis and analysed if and how closely symptom intensity of saplings and their parents are correlated by direct parent–offspring comparisons, and by analysing correlation between kinship and damage intensity of trees and saplings. Positive correlations should imply a likely genetic resistance to ash dieback and reveal its relative contribution to observed susceptibility under field conditions. As this is likely a strategy with several crucial issues in comparison to offspring testing under common garden conditions (e.g. environmental and age heterogeneity, and numbers of related seedlings present), our study is a first attempt to identify if the necessary conditions for this approach are given in forest situations in Austria. Methods Study sites and sampling Requirements were defined a priori for selecting suitable sampling sites. We searched for optimal sites representing fairly isolated ash stands (so that seedlings would likely be descendents of the local adults) with substantial natural regeneration and a range of healthy to severely damaged mature trees. Among several candidates (suggested by colleagues involved in monitoring ash dieback and in research in natural forest reserves in Austria), two study sites were finally chosen (Figure 1). Figure 1 View largeDownload slide Location of the study sites in eastern Austria; 1: Johannser Kogel (Lainzer Tiergarten, southwest Vienna); 2: Siegenfeld (Heiligenkreuz, Wienerwald) Figure 1 View largeDownload slide Location of the study sites in eastern Austria; 1: Johannser Kogel (Lainzer Tiergarten, southwest Vienna); 2: Siegenfeld (Heiligenkreuz, Wienerwald) The first sampling site is a small ash stand within Johannser Kogel, which is a strict natural forest reserve of ~45 ha located in the northwestern part of Lainzer Tiergarten (Türk and Pfleger, 2008, Table 1). The Lainzer Tiergarten is a protected Natura 2000 site of 2460 ha with near-natural old-growth forests and interspersed, extensively used grasslands; it is located in the western outskirts of Vienna and belongs to the biosphere reserve Wienerwald (Forst-und Landwirtschaftsbetrieb der Stadt Wien, 2017). The natural forest reserve Johannser Kogel is dominated by an oak (Quercus sp.)-hornbeam (Carpinus betulus L.) forest with iconic old oak trees which are up to 400 years old (Türk and Pfleger, 2008). The ash stand covers a core area of ~2 ha and is found on the top of the hill in the centre of the reserve, it is further described as a so called ‘hilltop ash forest’ (‘Gipfeleschenwald’). Hilltop ash forests are often pure stands on hilltops and on northern slopes (Willner, 1996). The hilltops may represent atypical sites where ash finds better (locally and temporarily moister) conditions and out-competes other forest trees (Willner, 1996). The core ash forest at Johannser Kogel is mixed with field maple (Acer campestre L.) and some hornbeam, and surrounded by old-age oak-hornbeam forest. Ash is rare in the immediately surrounding oak-hornbeam woods (at the scale of hundreds of metres), and parts of Johannser Kogel are bordered by meadows. However, ash is a common component of the further surrounding broadleaf forest landscapes on a kilometre scale. Table 1 Study sites, stand information and sampling Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 Table 1 Study sites, stand information and sampling Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 The second sampling site Siegenfeld is a small, nearly pure ash stand of ~1 ha area located in a forest between Siegenfeld and Heiligenkreuz at the eastern rim of the Alps, c. 30 km south of Vienna (Figure 1 and Table 1). It is surrounded by a spruce (Picea abies Karst.) forest in the north, west and south and adjacent to a forest road and meadow in the east. Other tree species in the ash stand are rare and include single beech, maple (Acer sp.) and larch (Larix decidua Mill.) trees. The altitude is ~400 m and mean annual precipitation is ~688 mm (data from Climate-data.org, 2017). In 2007, the Federal Research Centre for Forests (BFW) conducted a monitoring program on 50 plots in Lower Austria where crown dieback intensity of common ash was estimated on 20 trees per plot. Monitoring was continued on a sub-set of 16 plots in later years. In 2009, the Austrian Forest Inventory included 1200 plots in an effort to assess the status of ash dieback all over Austria. Siegenfeld in Lower Austria was included in all these assessments (Cech et al., 2012). Most of the 20 trees assessed in detail at Siegenfeld showed decreasing symptom intensity from 2007 to 2009 (Cech et al., 2012), which singled out Siegenfeld as one of the relatively healthier monitoring plots by now (trees at many other sites were not in a good health state at all; Katharina Schwanda pers. comm.). Saplings occurred in small patches at both sites that seemed to depend on penetrance of the canopy by sunlight. Biological material for DNA analysis was collected in June, July and August 2015 at both sites as follows. One leaf was sampled from randomly chosen saplings across the site (resulting in 80 saplings per site). Either a leaf, brought down by a slingshot if this was possible, or otherwise a bark plug per mature tree (from all mature ash trees at the site – resulted in a total of 134 trees) was collected (Table 1). The bark plug, from which cambium tissue can be sliced, was cut out of the trunk with a 1 cm diameter leather punch. Cambium tissue as well as leaf tissue contains the same genomic DNA. The material was immediately dried in silica gel and kept at room temperature prior to DNA extraction. Ash dieback was recorded as the degree of crown dieback (loss of crown foliage; see Table 2). Mature trees were classified into one of six damage classes by visual inspection (Figure 2A). There were also six damage classes for sapling assessment, but these were based on shoot damage instead (percentage of the shoots of a sapling affected by the disease; Figure 2B and Table 2). Additionally the diameter at breast height (DBH) of each tree and the height of each sapling were measured. Saplings were only sampled when they exceeded the height of ~60 cm, so that symptom identification and damage class categorization were possible with higher confidence. Table 2 Definition of damage classes Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Table 2 Definition of damage classes Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Figure 2 View largeDownload slide Illustration of the visually assessed damage classes (d.c.) (A) for mature trees, no tree was assessed for damage class 6 (photos from Siegenfeld by Alexandra Wohlmuth); (B) for saplings (d.c. 5 is not shown) (Photos by Thomas Kirisits, University of Natural Resources and Life Sciences, Vienna, Austria). Figure 2 View largeDownload slide Illustration of the visually assessed damage classes (d.c.) (A) for mature trees, no tree was assessed for damage class 6 (photos from Siegenfeld by Alexandra Wohlmuth); (B) for saplings (d.c. 5 is not shown) (Photos by Thomas Kirisits, University of Natural Resources and Life Sciences, Vienna, Austria). DNA extraction On average, 25–45 mg dried leaf or cambium tissue per individual was put into 2 mL tubes together with two glass balls of 3 mm diameter and one of 4 mm (for cambium tissue, steel balls were used), one spatula tip each of glass powder, activated charcoal, polyvinyl pyrrolidon (PVP 40 000) and sodium metabisulfite (pro analysi grades were used for all chemicals, most of which were purchased from Sigma-Aldrich, St. Louis, MO, USA). For homogenization, the material was frozen in liquid nitrogen for two minutes and ground with a TissueLyser shaking mill (QIAGEN, Hilden, Germany) at 25 Hz for 2 min. The process was repeated a second time. The DNA was extracted using the Invisorb® Spin Plant Mini Kit (STRATEC Molecular, Birkenfeld, Germany), applying the protocol recommended for the kit, but replacing Lysis Buffer P (provided by the kit) with a mixture of 800 μL 2× CTAB Buffer (20 g/L cetyl trimethyl ammonium bromide [CTAB], 100 mM Tris–HCL pH 8.0, 1.4 M NaCl, 25 mM EDTA pH 8.0) plus 1.6 μL β-mercaptoethanol and 1 μL proteinase K (QIAGEN) for lysis. Additionally 40 μL RNAse A (10 mg/mL; QIAGEN) were added to each sample before the binding process. To determine DNA concentrations, a NanoDrop 1000 Spectrophotometer (Thermo Fischer Scientific, Ulm, Germany) was used. DNA amplification and fragment analysis DNA concentrations for all samples were below 20 ng μL−1. Therefore, DNA extracts were employed undiluted for polymerase chain reaction (PCR). Nine nuclear microsatellite loci were used for amplification (with annealing temperatures: Femsatl-4 – 60°C, Femsatl-10 – 50°C, Femsatl-11 – 55°C, Femsatl-12 – 55°C, Femsatl-16 – 62°C, Femsatl-19 – 58°C, M230 – 57°C, FR639485 – 55°C and FR646655 – 60°C; Brachet et al., 1999; Lefort et al., 1999; Beatty et al., 2015). Each forward primer was labelled with a fluorescent dye. PCRs were carried out using the QIAGEN Type It Microsatellite Kit. The amplification reactions were performed on a PTC-100 Thermal Cycler (BIO-RAD, Vienna, Austria) under the following conditions: An initial denaturation step of 5 min at 95°C, 28 cycles of denaturation at 95°C for 30 s, annealing with corresponding temperatures (see above) for 90 s and extension at 72°C for 30 s, and a final extension step at 60°C for 30 min. A CEQ 8000 Beckman-Coulter (Vienna, Austria) Sequencer was used to visualize the PCR products based on fragment length polymorphism, as compared with CEQ DNA Size Standard Kit-400 (Beckman-Coulter). Allele assessment, calling and binning were carried out using the fragment analysis tool of GenomeLab GeXP Beckman-Coulter software (version 10.2.3), with additional visual inspection and binning of peaks. Parentage analysis CERVUS 3.0.7 software (Marshall et al., 1998; Kalinowski et al., 2007; Field Genetics Ltd, London, UK) was used for parentage analysis and for calculating other parameters, such as expected (He) and observed (Ho) heterozygosity, polymorphic information content (PIC), average non-exclusion probability for one candidate parent (NE-P1), average non-exclusion probability for a candidate parent pair (NE-PP), Hardy Weinberg equilibrium (HW) and estimated null allele frequency (NULL). All of these parameters give information about the loci and their suitability for parentage analysis. For parentage assignment, the CERVUS program uses likelihood ratio, a well-established statistical method (Marshall et al., 1998). Parentage is assigned to a candidate parent if the likelihood is large relative to the likelihood of alternative candidate parents. The likelihood ratio is expressed as LOD scores (logarithm of the likelihood ratio; Marshall et al., 1998). Candidate parents with positive LOD scores are more likely to be the true parents, and candidate parents with negative LOD scores are less likely to be the true parents. If two or more parents have positive LOD scores, Marshall et al. (1998) defined delta (difference in LOD scores) as an assessment criterion. Delta is the difference in LOD scores between the most likely candidate parent and the second most likely candidate parent. This parameter is useful when two candidate parents have a positive LOD score. If delta is high enough, parentage can be assigned to the candidate parent with the higher LOD score. The advantage in the use of CERVUS lies in the allowance for mistyping and missing data for individuals at a specified number of loci. Parentage analysis for one parent (implemented by the ‘maternity analysis’ function of CERVUS) and parent-pair analysis were carried out for both sites separately. Prior to parentage analysis, the simulation of parentage was performed. This is important, because it is used to check the feasibility of the parentage analysis and it calculates values of likelihood ratios, so that the confidence of parentage assignment can be determined. In short, in a simulation appropriate LOD and delta scores for valid parentage assignment are generated for the parentage analysis with the real data from genotyping. The simulation in this study was performed with 10 000 offspring, an error rate of 0.01 at strict (95 per cent) and relaxed (80 per cent) confidence levels. As additional parameters of the simulation, the number of candidate parents was set to 100 for Johannser Kogel with a 0.75 proportion of candidate parents sampled, and 60 candidate parents with a 0.90 proportion of candidate parents sampled for Siegenfeld. The proportion of sampled candidate parents were estimated by field observation, as the occurrence of unsampled trees in the proximity of the stands cannot be determined with certainty. Unsampled candidate parents were assumed (for the purpose of the CERVUS simulation) to be present in moderate frequency at Johannser Kogel and at low frequency at Siegenfeld. Data analysis We performed a correlation analysis on the relationship between the DBH of mature ash trees and the number of their offspring in order to see whether tree size explains seedling numbers and thus possibly confounds health state correlations. The damage classes of trees in different DBH classes at both sites were plotted as a ‘box and whisker plot’, with a similar intention (to check for possible covariance). We tabulated the number of offspring per damage class for parents in each damage class. We also calculated the frequency distribution of the damage classes of local saplings (one or both parents assigned locally) and immigrant saplings (no local parents assigned, thus descents exclusively of trees outside of the stand) to see whether there was a difference in the health performance between these cohorts (local and immigrant), using a Kolmogorov–Smirnov test to check for significance. The boxplot (DBH for damage classes) was done in SPSS, whereas the tables and the bar diagram were done in Microsoft Excel 2013 (Redmond, WA, USA). To estimate the significance of correlation between damage class of parent and offspring (categorial data), a one-tailed Spearman’s rank correlation analysis was calculated using SPSS Statistics 23 software (IBM, Vienna, Austria). The calculation was applied separately for those offspring where both parents were assigned (Spearman’s rank correlation coefficient between the damage class of the offspring and the averaged mean damage class of both parents) and for those where only one parent was assigned (Spearman’s rank correlation coefficient between the damage classes of the offspring and the one parent). Alternatively, an attempt was made to estimate heritability in situ in the sense of Ritland (2000): a pairwise matrix of differences in damage class for each possible pair of individuals in our study was regressed onto a pairwise kinship coefficient matrix, using the program SpaGeDi 1.05 (Hardy and Vekemans, 2002). The slope (b) and intercept of the regression, as well as the coefficient of determination (r2) were calculated. This coefficient estimates the degree to which similarity in damage classes is determined by kinship, thus resembles a heritability value. The data set was permutated 1000 times in order to estimate P-values. Results Tree and sapling dimensions, and damage assessment The mean diameter of the mature trees at breast height (DBH) was 39.7 cm at Johannser Kogel and 35 cm at Siegenfeld. Johannser Kogel hosts some particularly old trees with DBH diameters of around 60 cm (Table 1). At Johannser Kogel mean height of saplings was 96.5 cm with a range from 65 to 190 cm, and at Siegenfeld, 99.3 cm ranging from 70 to 170 cm. Disease symptoms were present at both sites in adults and saplings. Shoot dieback and crown defoliation were detectable, but there were few, if any, stem collar necroses. Both sites show a similar range of slightly to severely damaged individuals, with most mature trees having a crown damage intensity between 10 and 50 per cent and assigned to damage classes 2 and 3 (67 per cent of the trees at Johannser Kogel and 54 per cent at Siegenfeld). Percentages of assessment to damage class 1 were higher in Siegenfeld than in Johannser Kogel (Figure 3). However, the distribution of damage classes of saplings and mature trees showed no significant differences across both sampling sites according to Kolmogorov–Smirnov tests (saplings: P = 0.692; trees: P = 0.897). No tree was observed to have lost all its crown foliage (no damage class 6). Trees in the healthiest damage class at Johannser Kogel had higher DBH, but the ranges of DBH in the other damage classes at both sites were overlapping (Figure 4). Figure 3 View largeDownload slide The percentage of ash individuals that belong to damage classes 1–6 shown separately for mature trees and saplings from each study site; SJ: saplings from Siegenfeld; SA: mature trees from Siegenfeld; JJ: saplings from Johannser Kogel; JA: mature trees from Johannser Kogel. Figure 3 View largeDownload slide The percentage of ash individuals that belong to damage classes 1–6 shown separately for mature trees and saplings from each study site; SJ: saplings from Siegenfeld; SA: mature trees from Siegenfeld; JJ: saplings from Johannser Kogel; JA: mature trees from Johannser Kogel. Figure 4 View largeDownload slide Boxplot of the assignment of mature ash trees of different size (measured by their diameter at breast height, DBH) to different damage classes at the two study sites. Figure 4 View largeDownload slide Boxplot of the assignment of mature ash trees of different size (measured by their diameter at breast height, DBH) to different damage classes at the two study sites. Allele frequencies A total of 285 individuals of 294 sampled were successfully genotyped for the nine microsatellite loci. DNA from one ash tree and eight saplings (including all of the six dead saplings from Siegenfeld) could not be amplified, likely due to insufficient DNA quality. These individuals had to be excluded from further analysis. The loci showed high allele numbers (N) and most of them were highly polymorphic (Table 3). The highest allele variation was found at locus Femsatl-10 with 41 alleles, and the lowest variation with six alleles at locus Femsatl-16. In total, 183 alleles were detected in 161 individuals from Johannser Kogel with an average number of 20.3 alleles per locus, and 166 alleles were detected for 124 individuals from Siegenfeld (average 18.4 alleles per locus; Table 3). The mean proportion of loci typed exceeded 0.99 for both sites. Mean expected heterozygosity (He) was 0.75 for Johannser Kogel and 0.80 for Siegenfeld. Observed heterozygosity (Ho) ranged from 0.38 to 0.85 for Johannser Kogel, and from 0.29 to 0.87 for Siegenfeld. Expected heterozygosity was higher than observed in most of the loci and in both sites. Mean polymorphic information content was 0.71 for Johannser Kogel and 0.77 for Siegenfeld. Deviations from Hardy–Weinberg equilibrium were detected at five loci in Johannser Kogel and at one locus in Siegenfeld. The estimated frequency of null alleles ranged from −0.003 (FR639485, Johannser Kogel) to 0.5 (Femsatl-12, Siegenfeld) with a mean value of 0.072. Table 3 Information about the microsatellite loci used and the analysed parameters Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Number of alleles (N); Ho, observed heterozygosity; He, expected heterozygosity; PIC, polymorphic information content; NE-1 P, average non-exclusion for one candidate parent; NE-PP, average non-exclusion probability for a candidate parent-pair; HW, significance of deviation from Hardy–Weinberg equilibrium; NS = not significant, ND = not determined. *Significant at 5 per cent level, **significant at 1 per cent level, ***significant at 0.1 per cent level. Null: Estimated null allele frequency. Table 3 Information about the microsatellite loci used and the analysed parameters Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Number of alleles (N); Ho, observed heterozygosity; He, expected heterozygosity; PIC, polymorphic information content; NE-1 P, average non-exclusion for one candidate parent; NE-PP, average non-exclusion probability for a candidate parent-pair; HW, significance of deviation from Hardy–Weinberg equilibrium; NS = not significant, ND = not determined. *Significant at 5 per cent level, **significant at 1 per cent level, ***significant at 0.1 per cent level. Null: Estimated null allele frequency. Parentage assignment The assignments suggested by CERVUS (including LOD scores and delta values) are given in Supplementary Data File S1. Combined non-exclusion probabilities for the first parent and for the parent pair were 5.70E-03 and 6.00E-07 for Johannser Kogel and 8.92E-04 and 8.73E-09 for Siegenfeld if Femsatl-12 was included. Without Femsatl-12, the values were less favourable (by approximately one order of magnitude in most cases): 8.82E-03 and 2.24E-06 (Johannser Kogel first parent/parent pair) and 2.00E-03 and 8.00E-08 (Siegenfeld first parent/parent pair). Maternity analysis, i.e. the assignment of only one parent, for Johannser Kogel was successful in 42 cases or 52.5 per cent (20 of the cases also at high stringency); for Siegenfeld, it resulted in a 58.3 per cent assignment rate (42 saplings; all at strict confidence level). The assignment rate for the parent-pair analysis in Johannser Kogel was 7.5 per cent, implying that five saplings from Johannser Kogel were assigned to both parents (among the local adult trees) at relaxed confidence level (80 per cent) and one additional sapling at strict confidence level (95 per cent). In Siegenfeld, the assignment rate for the parent-pair analysis was higher with 27.7 per cent – both parents were determined for 20 saplings from Siegenfeld at a relaxed confidence level (10 of which also at strict confidence level). Parentage analysis (‘maternity’ or one parent only) without marker Femsatl-12 (at less stringent non-exclusion probabilities) resulted in 39 saplings assigned to candidate parents at Johannser Kogel at 80 per cent confidence level (19 of which at strict level). For Siegenfeld, the numbers increased as well, two additional saplings were assigned to a single parent (at 80 per cent), and eight more at 95 per cent, but one of the original ones lost significance of assignment (total, 50 saplings assigned), while for two saplings, alternative parents were assigned. Parent-pair assignment without Femsatl-12 resulted in the following changes: Johannser Kogel, six more saplings assigned at 80 per cent, one more at 95 per cent, one improved from 80 to 95 per cent, but three lost at 80 per cent significance. At Siegenfeld, there were 11 more saplings assigned at 80 per cent, one more at 95 per cent, four improved, one worsened in significance, and three seedlings were assigned to different parent pairs (numbers and details in Table 4 and Supplementary Data File S1). Table 4 Spearman’s rank-correlation analyses for offspring for which one parent was assigned, and for offspring for which both parents were assigned, calculated including and excluding marker Femsatl-12 Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* NS: not significant, *significant at the 0.05 level. Table 4 Spearman’s rank-correlation analyses for offspring for which one parent was assigned, and for offspring for which both parents were assigned, calculated including and excluding marker Femsatl-12 Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* NS: not significant, *significant at the 0.05 level. Data analysis The data show a very weak positive correlation of parents with higher DBH with the number of their offspring (R2 = 0.0101; Supplementary Figure S1; single-parent and parent-pair results combined; FEMSATL 12 included). The numbers of offspring per disease class of adult trees are rather equally distributed across damage classes (Supplementary Table S1). The frequency distribution of the damage classes of local and immigrant saplings showed no apparent differences in the health status between these cohorts at both sites (Supplementary Table S2; not significantly different in Kolmogorov–Smirnov tests). Spearman’s rank correlation for parentage vs damage class was generally low (−0.293 to 0.039, Table 4) and thus showed no apparent association between damage classes of offspring and parents, neither for parent-pair, nor for maternity analysis, and only one of the coefficients – with a negative sign! – was significantly different from zero (−0.293 without Femsatl-12 for single parent/maternity, P = 0.039; Table 4). The value for in situ heritability (r2) obtained by the regression of damage class differences on kinship (calculated including Femsatl-12; distribution shown in Supplementary Figure S2) was very close to zero (4.69E-006) and not statistically significant after permutations (P = 0.34). Discussion Genetic parameters and parentage assignment The majority of the loci used in this study were highly polymorphic, and expected heterozygosity was high. Similar observations were made for the same microsatellite loci in previous population genetic studies in common ash, suggesting high genetic diversity within populations across Europe (Heuertz et al., 2001, Hebel et al., 2006; Beatty et al., 2015). The values for genetic variation are typical for studies with these microsatellite markers in European ash (see Heinze and Fussi, 2017 for a comparison). While the deviations from Hardy–Weinberg equilibrium, and the high frequency of null alleles in Femsatl-12 are not ideal, the combined non-exclusion probabilities are very much appropriate (given the numbers of candidate parents present). Femsatl-12 seems to show an increase of null alleles from Western to Eastern Europe (Heinze and Fussi, 2017, their Table 4). Null alleles do not amplify consistently during PCR and thus cannot be reliably detected, leading to false assessments of heterozygotes as homozygotes (Kalinowski and Taper, 2006). They are problematic in parentage analysis and appear sometimes simultaneously with deviations from the Hardy–Weinberg equilibrium (Pemberton et al., 1995), which was also the case for five loci in Johannser Kogel and at one locus in Siegenfeld. However, estimation of null allele frequencies does not necessarily imply that a null allele is present. If no parent–offspring relationship is definitely known and there is no Hardy–Weinberg equilibrium, it is difficult (or impossible) to decide about the presence of null alleles with certainty (Dakin and Avise, 2004). We do not expect strict Hardy–Weinberg equilibria at our sites, as we observed incoming seed and pollen in the young generation. But even Femsatl-12 adds power to this analysis (better non-exclusion probabilities), because undetected null alleles do not lead to (wrong) exclusions of candidate parents; so the number of parent–offspring pairs is not underestimated because of the null alleles. It could be overestimated, but the simulations in CERVUS take the null frequencies into account for suggesting thresholds of LOD and delta values. The deviations from Hardy–Weinberg equilibria are already a hint for absence of panmixia (which would require equal parental contributions to the offspring generation) and/or the presence of immigrant saplings (see below). Both our sites had nearly the same set of alleles with few private alleles (alleles that only occur in one population) at each site, and little differentiation (like in Heuertz et al., 2001; Hebel et al., 2006; Ballian et al., 2008). Surprisingly, positive assignments of at least one parent were made only for slightly more than half of the saplings under various conditions of the analysis, and only 17 per cent of the saplings were assigned to both parents within their stands, despite their relative isolation. A possible technical explanation for the low assignment rate could be genotyping or binning errors which would probably lead to undetected parent–offspring trios or duos, although CERVUS uses likelihood equations that take account of such errors. High combined exclusion probability (99 per cent) across all loci for unrelated parents suggests low error rate in parentage assignment and shows that the microsatellite set used was adequate. Nevertheless, we tested our data analysis procedure omitting the critical Femsatl-12 marker. The parent non-exclusion probabilities then changed to less stringent values (Supplementary Data File S1) and slightly higher rates for parentage assignment resulted, so that more saplings could be assigned to parents. However, the correlation (Spearman’s rank) did not change in a meaningful way. Thus our conditions are conservative in the sense that more relaxed assignment conditions, if they lead to higher errors in parent–offspring assignments, possibly underestimate correlations (but provide a wider data basis for them). Although we made an effort to sample all putative parents in both fairly isolated stands, it is possible that these missing parents are actually scattered trees outside the stands. Therefore, a more likely explanation for the low assignment rates is that the rest of the saplings had been sired by these unsampled trees. High seed and pollen inflow from the surroundings is a phenomenon often detected in forest tree stands, even if we did not expect it here due to the scarcity of ash trees in the surroundings, and because of the generally low physical disposition of ash seed and pollen for passive transport beyond a few hundred metres (Richards, 1997). Nevertheless, in this respect our findings resemble those of Lobo et al. (2015), where paternity analysis in two Danish seed orchards resulted in an assignment rate of less than 40 per cent. They suggested the small size of the seed orchard as one possible explanation (12 putative parents in the orchard). Analysis of seed dispersal in two natural ash woodlands in Ireland carried out by Beatty et al. (2015) indicates the potential for seed dispersal over several hundreds of metres. Bacles et al. (2005) and Bacles and Ennos (2008) estimated pollen dispersal in fragmented ash populations in Scotland, suggesting effective pollination within 300 m, but detected rare pollination events over distances of up to 3 km. While these studies on pollination were carried out in isolated tree stands in open landscapes, where pollen can be transported easier with less barriers being present, a study by Heuertz et al. (2003) in a Romanian continuous ash woodland estimated average distances of 14 m between seeds and their female parents, and average pollen flow below 140 m distance (between seed mothers and pollinating trees). However, their common ash stand was much denser (~200 mature stems per ha), and their gene flow estimates were indirectly derived from genetic data of adults only (decrease of kinship with increasing distance), while we analysed saplings (and thus the ‘realized gene flow’ of the transition of one generation to the next). We infer from our low assignment rate for parent-pair analysis and the moderate assignment rate for maternity analysis that seed and pollen dispersal within mixed landscapes is probably higher than the estimates of Heuertz et al. (2003) for continuous woodlands and more in the range of the estimates of Beatty et al. (2015) in Irish woodlands (but lower than those for very open landscapes). It is also the experience of foresters in Austria in similar forest types that (i) either a single or a few ash trees are able to produce dense seedling patches, if light and other conditions are good (‘seed shadows’), and (ii) these patches can be found up to a few hundred metres away from the source (especially following the main wind direction; Herfried Steiner, Werner Ruhm, pers. comm.). Further analysis is needed in order to infer the precise influence of landscape features on these parameters. Variation in susceptibility The presence of seeds that originate entirely from outside the stands investigated allowed us to calculate if there are ‘population differences’ in susceptibility between offspring of the local trees vs immigrant saplings. The frequency distribution of the damage classes between these cohorts showed no apparent differences (Supplementary Table S2). This is a further indication that there is no selection (yet) for highly tolerant saplings at our sites (or for more tolerant adults with a consequent higher success in contributing to seed production in the wider surroundings). It also follows that there is no reason to believe that different susceptibilities of trees outside our stands (compared with the adults within our stands) have led to any systematic error in the correlations. No significant relationship (with the exception of one negative) was found based on Spearman’s rank correlation between damage intensity of parents and their offspring (Table 4), neither for parent–offspring duos (offspring with one parent), nor for parent–offspring trios (offspring with both parents). Previous studies suggested genetically based variation as an explanation for observed differences in susceptibility to dieback symptoms based on the observation of clones and single-tree offspring in progeny tests, and they concluded that these differences are transferred across ash generations (Bakys et al., 2009; McKinney et al., 2011; Lobo et al., 2014; Muñoz et al., 2016). Most of these studies were performed in clonal seed orchards or established progeny trials. The seedlings were raised in a nursery prior to planting (Kjær et al., 2012; Muñoz et al., 2016) and the clones for seed orchards were grafted onto rootstocks and also further cultivated in a nursery, and rather intensively managed after planting (with wider spacing; McKinney et al., 2011; Stener, 2013). Family size in the study of Muñoz et al. (2016) was between 8 and 68 half-sibs per mother tree. Lobo et al. (2015) used 2–48 offspring per half-sib family in inoculation tests (8–48 in maternal families). We expected to arrive at numbers similar to the lower part of this range in our settings, but that was not the case. Progeny trails and seed orchards (and especially sapling inoculations) represent more controlled environments and they often do not fully resemble natural woodland conditions. In contrast, this study was conducted in naturally regenerating ash stands, and the more heterogeneous conditions there likely led to higher genotype × (micro-)environment interactions., which in turn lowered the genetic correlations observed (see also below) Lobo et al. (2014) estimated the heritability of damage intensity in two progeny trails in Denmark. Both trials included progenies from the same mother trees, but one trial was left unfenced and progenies were exposed to strong competition from vegetation and to browsing animals, whereas the other was protected by fencing. Lower estimated heritability values of the unfenced (0.2) compared with the fenced site (0.42–0.53) indicated that estimates of heritability under heterogeneous conditions may be lower (Lobo et al., 2014). Johannser Kogel is also fenced, but the fence is obviously penetrable at some points, as wild boars were observed inside the fence during sampling. Pliūra et al. (2014) tested clones from seven different sites from Lithuania exposed to different ash-dieback infection pressure. They calculated the genotype–environment interaction on phenotypic variation and found significant contributions of genetic variation in plasticity and reaction norms of clones across a range of infection pressure environments (Pliūra et al., 2014). This indicates the presence of variation of individual response to the disease across sites depending on site conditions (Pliūra et al., 2011, 2014), or different levels of disease pressure. Results from previous studies, together with our current ones, suggest that precise estimates of heritability are only possible in more controlled environments, or with much higher sample numbers (sapling half-sib families detected and assigned to parents). Therefore, environmental conditions and sample sizes are likely more relevant and need to be strongly considered for inferences on susceptibility in field observation studies. It seems from all cited studies that up to now, only a small fraction of ash individuals maintain potential resistance to the disease, as practically all trees are infected at some point. Kjær et al. (2012) found high susceptibility and mortality among trees in two progeny trails in Denmark. They further estimated that only 1 per cent of the trees have the potential to pass lower susceptibility to their offspring. In three Lithuanian progeny trails almost 90 per cent of the trees died during the observation period of eight years (Pliūra et al., 2011). In Austria, such high mortality rates have not yet been reported from observation sites (Heinze et al., 2017). In 50 ash dieback monitoring plots in Lower Austria (mainly adult stands), mean crown dieback intensity reached 18.1 per cent in 2009 and 17.6 per cent in 2010 (Kirisits and Freinschlag, 2012). Three seed orchards in Austria showed moderate mean crown dieback intensity in 2011 (14.2 per cent, 13.5 and 31 per cent), but no clone was totally unaffected (Heinze et al., 2017). Crown and shoot dieback intensity was also relatively moderate at Johannser Kogel and Siegenfeld (66 and 72 per cent of the individuals, respectively, showed lower than 50 per cent damage intensity). It is likely that the mature trees that were identified as parents did not express assessable resistance due to lower infection pressure (especially for the higher crowns of thicker trees, Figure 4), and consequently produced more offspring because of their greater size and the better health condition associated with it (Supplementary Figure S1), while infection pressure was higher for saplings close to the forest floor, where the presence of petioles and more humid conditions would favour high spore concentrations. This could be one reason for the low correlation between damage classes of parents and offspring in our study. However, this effect may impact the severity, but not so much the direction of susceptibility; healthier mature trees should then still produce healthier offspring, though the average damage class rating would be shifted between old trees and young saplings. Probably the small sample size of parent–offspring trios (N = 26–50) and of half-sib families (only nine mature trees produced more than two offspring) may have led to underestimation of actual correlations of damage intensity between parent and offspring for possible reduced susceptibility. Muñoz et al. (2016) found that there are family effects in open-pollinated offspring, but they also calculated individual breeding values for the offspring, and most of the genetic variation was found within families. Thus, our small family sizes may not allow estimating heritabilities with great precision. The in situ heritability estimation based on all kinship relationships among trees and saplings also resulted in a coefficient of determination of very close to zero. It may not necessarily be the case that related trees have very similar disease tolerance levels. Breeding values, which represent the average effect of the parent genotype, estimated from the performance of its offspring (McKinney et al., 2014), were calculated by Kjær et al. (2012) in a Danish progeny trail, and by Muñoz et al. (2016) in France. Only one of 101 tested mother trees in Denmark had breeding values for ‘susceptibility’ below 10 per cent and was estimated to produce healthy offspring (and four trees with breeding values below 20 per cent were estimated to produce fairly healthy offspring). Another study carried out by McKinney et al. (2011) supported these findings, with only one of 39 tested clones exhibiting breeding values below 10 per cent. Breeding values for susceptibility in Kjær et al. (2012) were normally distributed, suggesting that possible resistance is based on expression of several genes rather than on one alone (McKinney et al., 2014). The data from France (Muñoz et al., 2016) also suggest that many genes contribute to resistance, as the largest part of genetic variation was found within families. This also means that much larger numbers of saplings are necessary to estimate heritability in our approach. Harper et al. (2016), in contrast, discovered a number of single nucleotide polymorphism (SNP) and gene expression markers that are associated with crown damage in infected trees using Associated Transcriptomics. With three markers (concerning transcription factor genes) they were able to predict individuals with a low level of susceptibility to the disease (combined R2 value of 0.28). Sollars et al. (2017) very recently published the sequence of a low-heterozygosity ash tree, and further improved these markers (which, in their data, explained Danish disease scores with r2 = 0.25) and suggested that according to these markers, native trees in Great Britain would have a lower chance of becoming susceptible. They also reported on biochemical substances that have a higher prevalence in ‘resistant’ vs susceptible tree. Despite these advances, the exact genetic background(s) of low susceptibility still remains unclear and it may well be that more than one gene network or metabolic pathway is involved, or that different such mechanisms are at work in different regions of Europe. The robustness of the observed potential resistance in previous field studies needs to be re-evaluated. Precise genetic tests of their offspring are necessary to determine which of the field-resistant trees really transmit this trait to seeds. Large offspring numbers are necessary for this purpose, which make such tests costly. The alternative approach we tested would, however, require sites with high numbers of seedlings with established parentage, much higher than the ones we identified. Thus, such sites should have many seedlings, but few immigrants among them. This may be difficult to find. Although the distribution data of ash seedlings in the wider area of Johannser Kogel suggest the presence of ‘seed shadows’ of single adults (Herfried Steiner, pers. comm.), it remains to be seen whether numbers are high enough at a particular site. The higher workload of laboratory testing would require more time (and it would increase costs), but the approach may still be faster in total than field planting of seedlings to be tested for disease tolerance. It is uncertain whether low susceptibility in single, but infrequent individuals could hinder massive decline of ash in Europe, and if that resistance would sustain massive infection pressure at natural forest sites. However, selecting and breeding non-susceptible trees in a timely manner throughout Europe seems the only way to overcome the disease; the approach we discuss here may help to accelerate selection. Alternatively, selecting ash stands that combine many environmental and demographic factors that disadvantage high infection pressure and the progress of the disease (e.g. hot and dry summers, dry soils during the sporulation period of the fungus, big trees with large, dominating crowns or stands were leaf litter does not persist until the next summer) may emerge as a conservation strategy in situations where genetic tolerance is generally low (Heinze et al., 2017). Conclusion In this study, no significant correlation could be found between ash dieback damage intensity of parent and offspring in two natural stands in Austria. This can reflect low power of estimating heritability under natural in situ conditions, where the micro-environment may have a stronger influence on the trait than in planted tests. Given this, and the recent insights into the genetics of disease tolerance, much higher numbers of offspring with identified parenthood (than the ones we could find) are necessary for our approach. Other possible causes could be that there were no very tolerant parent trees at our sites, or that juvenile–adult correlations are low for this trait. We further suggest that the genetic basis for variation in susceptibility might be more complex under natural heterogeneous conditions than in more controlled environments like planted progeny trials or seed orchards, where most of the previous studies were implemented. The robustness of the observed potential low susceptibility in previous studies should therefore be critically re-evaluated after planting offspring from tolerant trees in production forests. The influence of environmental conditions should be considered when inferring on susceptibility (i.e. when assessing field resistance). A scenario of large-scale decline of common ash trees in Europe is becoming more and more likely. It will be important to identify and propagate healthy ash individuals throughout Europe, to find out about the genetic mechanism of any resistance in more detail and about the environmental conditions that favour low susceptibility, in order to implement a management program that accounts for these aspects. Supplementary data Supplementary data are available at Forestry online. Acknowledgements Special thanks to Renate Slunsky, Daniela Jahn, MSc, and the Genome Research unit at the Federal Research Centre for Forests (BFW) for their kind support and also to the Natural Forest Reserve and the Phytopathology units at BFW, especially to Dr Katharina Schwanda, Christian Neureiter and Mag. Herfried Steiner. ‘European Cooperation in Science and Technology (COST)’ Action FP1103 ‘FRAXBACK’ is acknowledged for providing a stimulating environment of meetings and discussions. Furthermore we would like to thank Dr Thomas Kirisits, University of Natural Resources and Life Sciences in Vienna, who provided his photos of the damage classes of saplings, and the Editor in Chief and three reviewers for their helpful comments and suggestions. Conflict of interest statement None declared. References Bacles , C.F.E. , Burczyk , J. , Lowe , A.J. and Ennos , R.A. 2005 Historical and contemporary mating patterns in remnant populations of the forest tree Fraxinus excelsior L . Evolution 59 ( 5 ), 979 – 990 . Google Scholar PubMed Bacles , C.F.E. and Ennos , R.A. 2008 Paternity analysis of pollen-mediated gene flow for Fraxinus excelsior L. in a chronically fragmented landscape . Heredity 101 ( 4 ), 368 – 380 . Google Scholar CrossRef Search ADS PubMed Bakys , R. , Vasaitis , R. , Barklund , P. , Ihrmark , K. and Stenlid , J. 2009 Investigations concerning the role of Chalara fraxinea in declining Fraxinus excelsior . Plant Pathol. 58 ( 2 ), 284 – 292 . Google Scholar CrossRef Search ADS Ballian , D. , Monteleone , I. , Ferrazzini , D. , Kajba , D. and Belletti , P. 2008 Genetic characterization of common ash (Fraxinus excelsior L.) populations in Bosnia and Herzegovina . Period. Biol. 10 , 323 – 328 . Beatty , G.E. , Brown , J.A. , Cassidy , E.M. , Finlay , C.M.V. , McKendrick , L. , Montgomery , W.I. et al. . 2015 Lack of genetic structure and evidence for long-distance dispersal in ash (Fraxinus excelsior) populations under threat from an emergent fungal pathogen: implications for restorative planting . Tree Genet. Genomes 11 ( 3 ), 53 . Google Scholar CrossRef Search ADS Brachet , S. , Jubier , M.F. , Richard , M. , Jung-Muller , B. and Frascaria-Lacoste , N. 1999 Rapid identification of microsatellite loci using 5′ anchored PCR in the common ash Fraxinus excelsior . Mol. Ecol. 8 ( 1 ), 160 – 163 . Cech , T.L. 2006 a Auffallende Schadfaktoren an Waldbäumen im Jahr 2005 [Striking damaging agents on forest trees in 2005] . Forstschutz Aktuell (BFW) 35 , 6 – 7 . Cech , T.L. 2006 b Eschenschaden in Österreich [Ash dieback and premature leaf shedding in Austria] . Forstschutz Aktuell (BFW) 37 , 18 – 20 . Cech , T.L. , Kessler , M. and Brandstetter , M. 2012 Monitoring des Zurücksterbens der Esche in Österreich [Monitoring ash dieback in Austria] . Forstschutz Aktuell (BFW) 55 , 56 – 58 . Cleary , M.R. , Andersson , P.F. , Broberg , A. , Elfstrand , M. , Daniel , G. and Stenlid , J. 2014 Genotypes of Fraxinus excelsior with different susceptibility to the ash dieback pathogen Hymenoscyphus pseudoalbidus and their response to the phytotoxin viridiol—a metabolomic and microscopic study . Phytochemistry 102 , 115 – 125 . Google Scholar CrossRef Search ADS PubMed Climate-data.org. 2017 Climate Data for Siegenfeld. https://de.climate-data.org/search/?q=Siegenfeld (accessed on 22 August, 2017). Comita , L.S. , Queenborough , S.A. , Murphy , S.J. , Eck , J.L. , Xu , K. , Krishnadas , M. , et al. . 2014 Testing predictions of the Janzen-Connell hypothesis: a meta-analysis of experimental evidence for distance- and density-dependent seed and seedling survival . J. Ecol. 102 , 845 – 856 . Google Scholar CrossRef Search ADS PubMed Dakin , E.E. and Avise , J.C. 2004 Microsatellite null alleles in parentage analysis . Heredity 93 , 504 – 509 . Google Scholar CrossRef Search ADS PubMed Dobrowolska , D. , Hein , S. , Oosterbaan , A. , Wagner , S. , Clark , J. and Skovsgaard , J.P. 2011 A review of European ash (Fraxinus excelsior L.): implications for silviculture . Forestry 84 , 133 – 148 . Google Scholar CrossRef Search ADS Forst-und Landwirtschaftsbetrieb der Stadt Wien . 2017 Lage, Größe, Geologie und Klima—Lebensraum Lainzer Tiergarten [Situation, size, geology and climate—habitat Lainzer Tiergarten; online; in German]. https://www.wien.gv.at/umwelt/wald/erholung/lainzertiergarten/lebensraum/lage.html (accessed on 22 August, 2017). Gross , A. , Holdenrieder , O. , Pautasso , M. , Queloz , V. and Sieber , T.N. 2014 Hymenoscyphus pseudoalbidus, the causal agent of European ash dieback . Mol. Plant Pathol. 15 ( 1 ), 5 – 21 . Google Scholar CrossRef Search ADS PubMed Hardy , O.J. and Vekemans , X. 2002 SPAGeDi: a versatile computer program to analyse spatial genetic structure at the individual or population levels . Mol. Ecol. Notes 2 , 618 – 620 . Google Scholar CrossRef Search ADS Harper , A.L. , McKinney , L.V. , Nielsen , L.R. , Havlickova , L. , Li , Y. , Trick , M. , et al. . 2016 Molecular markers for tolerance of European ash (Fraxinus excelsior) to dieback disease identified using Associative Transcriptomics . Sci. Rep. 6 , 19335 . Google Scholar CrossRef Search ADS PubMed Hebel , I. , Haas , R. and Dounavi , A. 2006 Genetic variation of common ash (Fraxinus excelsior L.) populations from provenance regions in southern Germany by using nuclear and chloroplast microsatellites . Silvae Genet. 55 ( 1 ), 38 – 44 . Google Scholar CrossRef Search ADS Heinze , B. and Fussi , B. 2017 Pre-disease levels of genetic diversity and differentiation among ash (Fraxinus excelsior L.) seedlots in Austria . Balt. For. 23 ( 1 ), 198 – 208 . Heinze , B. , Tiefenbacher , H. , Litschauer , R. and Kirisits , T. 2017 Ash dieback in Austria—history, current situation and outlook. In Dieback of European Ash (Fraxinus spp.)—Consequences and Guidelines for Sustainable Management . Vasaitis R. and Enderle R. (eds). , 2017. Swedish University of Agricultural Sciences , pp. 33 – 52 ISBN (print version) 978-91-576-8696-1. Heuertz , M. , Hausman , J.-F. , Tsvetkov , I. , Frascaria-Lacoste , N. and Vekemans , X. 2001 Assessment of genetic structure within and among Bulgarian populations of the common ash (Fraxinus excelsior L.) . Mol. Ecol. 10 ( 7 ), 1615 – 1623 . Google Scholar CrossRef Search ADS PubMed Heuertz , M. , Vekemans , X. , Hausman , J.-F. , Palada , M. and Hardy , O.J. 2003 Estimating seed vs. pollen dispersal from spatial genetic structure in the common ash . Mol. Ecol. 12 ( 9 ), 2483 – 2495 . Google Scholar CrossRef Search ADS PubMed Kalinowski , S.T. and Taper , M.L. 2006 Maximum likelihood estimation of the frequency of null alleles at microsatellite loci . Conserv. Genet. 7 ( 6 ), 991 – 995 . Google Scholar CrossRef Search ADS Kalinowski , S.T. , Taper , M.L. and Marshall , T.C. 2007 Revising how the computer program CERVUS accommodates genotyping error increases success in paternity assignment . Mol. Ecol. 16 ( 5 ), 1099 – 1106 . Google Scholar CrossRef Search ADS PubMed Keßler , M. , Cech , T.L. , Brandstetter , M. and Kirisits , T. 2012 Dieback of ash (Fraxinus excelsior and Fraxinus angustifolia) in Eastern Austria: disease development on monitoring plots from 2007 to 2010 . J. Agric. Ext. Rural Dev. 4 ( 9 ), 223 – 226 . Kirisits , T. and Freinschlag , C. 2012 Ash dieback caused by Hymenoscyphus pseudoalbidus in a seed plantation of Fraxinus excelsior in Austria . J. Agric. Ext. Rural Dev. 4 , 184 – 191 . Kjær , E.D. , McKinney , L.V. , Nielsen , L.R. , Hansen , L.N. and Hansen , J.K. 2012 Adaptive potential of ash (Fraxinus excelsior) populations against the novel emerging pathogen Hymenoscyphus pseudoalbidus . Evol. Appl. 5 ( 3 ), 219 – 228 . Google Scholar CrossRef Search ADS PubMed Kowalski , T. 2006 Chalara fraxinea sp. nov. associated with dieback of ash (Fraxinus excelsior) in Poland . For. Pathol. 36 ( 4 ), 264 – 270 . Google Scholar CrossRef Search ADS Kräutler , K. and Kirisits , T. 2012 The ash dieback pathogen Hymenoscyphus pseudoalbidus is associated with leaf symptoms on ash species (Fraxinus spp.) . J. Agric. Ext. Rural Dev. 4 , 261 – 265 . Landolt , J. , Gross , A. , Holdenrieder , O. and Pautasso , M. 2016 Ash dieback due to Hymenoscyphus fraxineus: what can be learnt from evolutionary ecology? Plant Pathol. 65 ( 7 ), 1056 – 1070 . Google Scholar CrossRef Search ADS Lefort , F. , Brachet , S. , Frascaria-Lacoste , N. , Edwards , K.J. and Douglas , G.C. 1999 Identification and characterization of microsatellite loci in ash (Fraxinus excelsior L.) and their conservation in the olive family (Oleaceae) . Mol. Ecol. 8 ( 6 ), 1088 – 1089 . Google Scholar CrossRef Search ADS Lobo , A. , Hansen , J.K. , McKinney , L.V. , Nielsen , L.R. and Kjær , E.D. 2014 Genetic variation in dieback resistance: growth and survival of Fraxinus excelsior under the influence of Hymenoscyphus pseudoalbidus . Scand. J. For. Res. 29 ( 6 ), 519 – 526 . Google Scholar CrossRef Search ADS Lobo , A. , McKinney , L.V. , Hansen , J.K. , Kjær , E.D. and Nielsen , L.R. 2015 Genetic variation in dieback resistance in Fraxinus excelsior confirmed by progeny inoculation assay . For. Pathol. 45 ( 5 ), 379 – 387 . Google Scholar CrossRef Search ADS Marigo , G. , Peltier , J.-P. , Girel , J. and Pautou , G. 2000 Success in the demographic expansion of Fraxinus excelsior L . Trees 15 ( 1 ), 1 – 13 . Google Scholar CrossRef Search ADS Marshall , T.C. , Slate , J. , Kruuk , L.E.B. and Pemberton , J.M. 1998 Statistical confidence for likelihood-based paternity inference in natural populations . Mol. Ecol. 7 ( 5 ), 639 – 655 . Google Scholar CrossRef Search ADS PubMed McKinney , L.V. , Nielsen , L.R. , Collinge , D.B. , Thomsen , I.M. , Hansen , J.K. and Kjær , E.D. 2014 The ash dieback crisis: genetic variation in resistance can prove a long-term solution . Plant Pathol. 63 ( 3 ), 485 – 499 . Google Scholar CrossRef Search ADS McKinney , L.V. , Nielsen , L.R. , Hansen , J.K. and Kjær , E.D. 2011 Presence of natural genetic resistance in Fraxinus excelsior (Oleraceae) to Chalara fraxinea (Ascomycota): an emerging infectious disease . Heredity 106 ( 5 ), 788 – 797 . Google Scholar CrossRef Search ADS PubMed McKinney , L.V. , Thomsen , I.M. , Kjær , E.D. and Nielsen , L.R. 2012 Genetic resistance to Hymenoscyphus pseudoalbidus limits fungal growth and symptom occurrence in Fraxinus excelsior . For. Pathol. 42 ( 1 ), 69 – 74 . Google Scholar CrossRef Search ADS Morand-Prieur , M.-E. , Raquin , C. , Shykoff , J.A. and Frascaria-Lacoste , N. 2003 Males outcompete hermaphrodites for seed siring success in controlled crosses in the polygamous Fraxinus excelsior (Oleaceae) . Am. J. Bot. 90 ( 6 ), 949 – 953 . Google Scholar CrossRef Search ADS PubMed Muñoz , F. , Marcais , B. , Dufour , J. and Dowkiw , A. 2016 Rising out of the ashes: additive genetic variation for crown and collar resistance to Hymenoscyphus fraxineus in Fraxinus excelsior . Phytopathology 106 , 1535 – 1543 . Google Scholar CrossRef Search ADS PubMed Pautasso , M. , Aas , G. , Queloz , V. and Holdenrieder , O. 2013 European ash (Fraxinus excelsior) dieback—a conservation biology challenge . Biol. Conserv. 158 , 37 – 49 . Google Scholar CrossRef Search ADS Pemberton , J.M. , Slate , J. , Bancroft , D.R. and Barrett , J.A. 1995 Non-amplifying alleles at microsatellite loci: a caution for parentage and population studies . Mol. Ecol. 4 , 249 – 252 . Google Scholar CrossRef Search ADS PubMed Pliūra , A. , Lygis , V. , Suchockas , V. and Bartkevicius , E. 2011 Performance of twenty-four European Fraxinus excelsior populations in three Lithuanian progeny trials with a special emphasis on resistance to Chalara fraxinea . Balt. For. 17 , 17 – 34 . Pliūra , A. , Marčiulynienė , D. , Bakys , R. and Suchockas , V. 2014 Dynamics of genetic resistance to Hymenoscyphus pseudoalbidus in juvenile Fraxinus excelsior clones . Balt. For. 20 ( 1 ), 10 – 27 . Przybyl , K. 2002 Fungi associated with necrotic apical parts of Fraxinus excelsior shoots . For. Pathol. 32 ( 6 ), 387 – 394 . Google Scholar CrossRef Search ADS Queloz , V. , Grünig , C.R. , Berndt , R. , Kowalski , T. , Sieber , T.N. and Holdenrieder , O. 2011 Cryptic speciation in Hymenoscyphus albidus . For. Pathol. 41 ( 2 ), 133 – 142 . Google Scholar CrossRef Search ADS Richards , A.J. 1997 Plant breeding systems. 2nd edn . Chapman & Hall , p. 529 . Sollars , E.S.A. , Harper , A.L. , Kelly , L.J. , Sambles , C.M. , Ramirez-Gonzalez , R.H. , Swarbreck , D. , et al. 2017 Genome sequence and genetic diversity of European ash trees . Nature 541 ( 7636 ), 212 – 216 . Google Scholar CrossRef Search ADS PubMed Stener , L.-G. 2013 Clonal differences in susceptibility to the dieback of Fraxinus excelsior in southern Sweden . Scand. J. For. Res. 28 ( 3 ), 205 – 216 . Google Scholar CrossRef Search ADS Thomas , P.A. 2016 Biological flora of the British isles: Fraxinus excelsior . J. Ecol. 104 ( 4 ), 1158 – 1209 . Google Scholar CrossRef Search ADS Türk , R. and Pfleger , H.S. 2008 Die Flechtenflora am Johannser Kogel im Lainzer Tiergarten und in den Steinhofgründen (Wien, Österreich) [The lichen flora at Johannser Kogel in Lainzer Tiergarten and at Steinhofgründe (Vienna, Austria)] . Verh. Zool. Bot. Ges. Österr. Österr. 145 , 83 – 95 . Wallander , E. 2008 Systematics of Fraxinus (Oleaceae) and evolution of dioecy . Plant Syst. Evol. 273 ( 1–2 ), 25 – 49 . Google Scholar CrossRef Search ADS Willner , W. 1996 Die Gipfeleschenwälder des Wienerwaldes [The hilltop ash forests of Wienerwald] . Verh. Zool. Bot. Ges. Österr. 133 , 133 – 184 . Zhao , Y.-J. , Hosoya , T. , Baral , H.-O. , Hosaka , K. and Kakishima , M. 2012 Hymenoscyphus pseudoalbidus, the correct name for Lambertella albida reported from Japan . Mycotaxon 122 , 25 – 41 . Google Scholar CrossRef Search ADS © Institute of Chartered Foresters, 2018. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Forestry: An International Journal Of Forest Research Oxford University Press

Genetic analysis of inherited reduced susceptibility of Fraxinus excelsior L. seedlings in Austria to ash dieback

Loading next page...
 
/lp/ou_press/genetic-analysis-of-inherited-reduced-susceptibility-of-fraxinus-OgEdhCOALz
Publisher
Institute of Chartered Foresters
Copyright
© Institute of Chartered Foresters, 2018. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com.
ISSN
0015-752X
eISSN
1464-3626
D.O.I.
10.1093/forestry/cpy012
Publisher site
See Article on Publisher Site

Abstract

Abstract Hymenoscyphus fraxineus causes massive dieback of common ash (Fraxinus excelsior L.) across populations. Previous common garden trials have revealed differences in susceptibility among individuals, suggesting a genetic basis for reduced susceptibility to the pathogen. The aim of the study was to identify any correlation between damage intensity of mature trees and their offspring in natural ash stands. Crown and shoot damage of naturally infected trees and saplings were assessed in two geographically isolated stands in Austria, and parentage analysis was carried out with molecular markers. No significant correlation could be detected using Spearman’s rank correlation analysis, suggesting that this approach would need higher numbers of adult–offspring pairs present to compensate for environmental and genetic variability at the sites. Likewise, an in situ estimate of heritability was nearly zero. The results thus support the results of other studies, i.e. that highly resistant individuals occur only at low frequency within European ash populations. While most of the previous studies were conducted in progeny trails or seed orchards and suggested a fairly strong genetic component, results from our investigation support a more complex mechanism of susceptibility differences under natural, heterogeneous conditions. Further analyses are needed to obtain a better understanding of gene–environment interactions and individual infection pressure of ash dieback in natural environments; such studies would need to be based on much higher sample numbers. Identification and propagating of non-susceptible ash trees is an important challenge to halt large-scale dieback of common ash. Introduction Common ash (Fraxinus excelsior L.) is an ecologically and economically important hardwood tree species widely distributed throughout temperate Europe. The species tolerates a wide range of environmental conditions from riparian to mountain habitats (Dobrowolska et al., 2011). Compared with beech (Fagus sylvatica L.), common ash is adapted to sites that are either moister or drier and it prefers more nutrient-rich soils (Marigo et al., 2000; Thomas, 2016). Although the seedlings are relatively shade tolerant, good light conditions are needed to compete with other broadleaved tree species (Marigo et al., 2000). Common ash has a complex reproductive system with male, female and hermaphrodite individuals and its pollen and seeds (samaras) are wind-dispersed (Morand-Prieur et al., 2003; Wallander, 2008). Recently, common ash has become highly threatened by a fungal disease (ash dieback) caused by the ascomycete Hymenoscyphus fraxineus (Gross et al., 2014), previously known as H. pseudoalbidus, with its anamorph stage Chalara fraxinea (Kowalski, 2006; Queloz et al., 2011). The pathogen was most likely introduced to Europe from East Asia (Zhao et al., 2012; Gross et al., 2014) and was first observed in Poland in the early 1990s (Przybyl, 2002; Pautasso et al., 2013). Since then, the pathogen has spread quickly towards Western and Southern Europe. The first ash dieback symptoms were observed in Austria in 2005 at a few sites in Lower Austria, Upper Austria and Styria, and subsequently the disease spread to all federal provinces of Austria by 2009 (Cech, 2006a, b; Cech et al., 2012). The symptoms of ash dieback range from necrotic lesions and wilting on ash leaves and petioles to necrotic lesions on branches, shoots and stem, leading to wood discoloration and crown dieback, and in most severe cases to death of the tree (Cech, 2006b; Bakys et al., 2009). Through wind dispersal, the fungus infects ash leaves during summer, growing into petioles and shoots and overwintering on infected petioles in the ground litter (Kräutler and Kirisits, 2012; Landolt et al., 2016). Ash dieback attacks trees of all age classes, although symptoms progress more severely and more rapidly in younger individuals, causing problems for natural regeneration (Keßler et al., 2012; Heinze et al., 2017). As the pathogen causes high mortality of ash trees, ash dieback has become a serious problem raising concerns about the future of the species. In several European countries, recent studies in seed orchards, progeny trials and in natural stands have aimed to increase understanding of the pathogen and its impact on common ash. Although there is an overall progression of the disease over the years in individual trees, several studies have revealed differences in symptom intensity among individuals in clonal or progeny trials, providing evidence for genetically based variation in susceptibility, and have suggested the existence of resistance or tolerance (McKinney et al., 2011; Pliūra et al., 2011; Kjær et al. 2012; Stener, 2013). Results from inoculation studies showed that the pathogen was less penetrative in tolerant individuals because of a better response mechanism of the host (McKinney et al., 2012; Cleary et al., 2014). The extent of possible resistance against ash dieback was previously estimated based on covariance and estimated heritability among half-sibs and full-sibs in progeny trails, suggesting that parents that are less susceptible transmit the resistance to their offspring (Kjær et al. 2012; Lobo et al., 2014). We planned and executed most of the study in the year 2015, when this was the state of knowledge. We thus expected that some insight may be gained from directly observing and comparing trees and their natural offspring in woodlands. It is as yet not entirely clear whether expression of tolerance in nursery seedlings is sufficiently related to that under natural site conditions. This would possibly allow us to see whether genetic selection is already at work there, i.e. if the seedlings that are present and healthy are so because they derive from tolerant adults. A positive result would significantly speed up the identification of tolerant genotypes, because the field testing of offspring, possibly over several years, requires significant resources and time. Clearly, being able to make inferences without having to grow seedlings and assessing them for several consecutive years would therefore be an advantage. Phenotypic selection of adults is likely not sufficient for identification of tolerant genotypes; but combined assessment of adults and their offspring at the same sites may increase precision (and so decrease the amount of field testing of seedlings), we thought. We also wanted to see whether the presence of healthy seedlings at a site would correlate with the presence of related, tolerant adults, i.e. whether genetic relationships can be established among young and adult plants in similar disease classes (like in a sort of inverse Janzen–Connell effect, see e.g. Comita et al., 2014), and thus check for any juvenile–adult correlations. Alternatively, site conditions may contribute strongly to disease prevalence; genotype by environment interaction (G×E) could interfere with this analysis; or size effects could also play a role (bigger trees might be more tolerant than smaller or non-dominant ones). We explored whether sites in Austria would be small and isolated enough so that sufficient numbers of adult-sapling pairs can be detected. Because of the strong interest in this pathogen system, new insights have been gained on heritability and on the genetic structure of tolerance since we conducted our study. Lobo et al. (2015) and Muñoz et al. (2016) have furthered our understanding and found out that while there is a family component of tolerance in open-pollinated offspring, there is also great variability within families, thus requiring large sample numbers for sizing the effects. Several possible mechanisms of the genetic basis of tolerance have very recently also been described with the help of whole-genome sequencing (Sollars et al., 2017). We thus looked at our data again, with a view on how to improve our approach, if possible at all. In this study, we investigated if transmission of resistance to ash dieback from mature trees to saplings is detectable in two naturally regenerating common ash populations in Austria. We used parentage analysis and analysed if and how closely symptom intensity of saplings and their parents are correlated by direct parent–offspring comparisons, and by analysing correlation between kinship and damage intensity of trees and saplings. Positive correlations should imply a likely genetic resistance to ash dieback and reveal its relative contribution to observed susceptibility under field conditions. As this is likely a strategy with several crucial issues in comparison to offspring testing under common garden conditions (e.g. environmental and age heterogeneity, and numbers of related seedlings present), our study is a first attempt to identify if the necessary conditions for this approach are given in forest situations in Austria. Methods Study sites and sampling Requirements were defined a priori for selecting suitable sampling sites. We searched for optimal sites representing fairly isolated ash stands (so that seedlings would likely be descendents of the local adults) with substantial natural regeneration and a range of healthy to severely damaged mature trees. Among several candidates (suggested by colleagues involved in monitoring ash dieback and in research in natural forest reserves in Austria), two study sites were finally chosen (Figure 1). Figure 1 View largeDownload slide Location of the study sites in eastern Austria; 1: Johannser Kogel (Lainzer Tiergarten, southwest Vienna); 2: Siegenfeld (Heiligenkreuz, Wienerwald) Figure 1 View largeDownload slide Location of the study sites in eastern Austria; 1: Johannser Kogel (Lainzer Tiergarten, southwest Vienna); 2: Siegenfeld (Heiligenkreuz, Wienerwald) The first sampling site is a small ash stand within Johannser Kogel, which is a strict natural forest reserve of ~45 ha located in the northwestern part of Lainzer Tiergarten (Türk and Pfleger, 2008, Table 1). The Lainzer Tiergarten is a protected Natura 2000 site of 2460 ha with near-natural old-growth forests and interspersed, extensively used grasslands; it is located in the western outskirts of Vienna and belongs to the biosphere reserve Wienerwald (Forst-und Landwirtschaftsbetrieb der Stadt Wien, 2017). The natural forest reserve Johannser Kogel is dominated by an oak (Quercus sp.)-hornbeam (Carpinus betulus L.) forest with iconic old oak trees which are up to 400 years old (Türk and Pfleger, 2008). The ash stand covers a core area of ~2 ha and is found on the top of the hill in the centre of the reserve, it is further described as a so called ‘hilltop ash forest’ (‘Gipfeleschenwald’). Hilltop ash forests are often pure stands on hilltops and on northern slopes (Willner, 1996). The hilltops may represent atypical sites where ash finds better (locally and temporarily moister) conditions and out-competes other forest trees (Willner, 1996). The core ash forest at Johannser Kogel is mixed with field maple (Acer campestre L.) and some hornbeam, and surrounded by old-age oak-hornbeam forest. Ash is rare in the immediately surrounding oak-hornbeam woods (at the scale of hundreds of metres), and parts of Johannser Kogel are bordered by meadows. However, ash is a common component of the further surrounding broadleaf forest landscapes on a kilometre scale. Table 1 Study sites, stand information and sampling Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 Table 1 Study sites, stand information and sampling Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 Site Johannser Kogel Siegenfeld Geographical coordinates Lat. N 48°11′; Long. E 16°12′ Lat. N 48°11′; Long. E 16°12′ Elevation (above sea level) 290–377 m ~400 m Precipitation ~650 mm 688 mm Exposition South-west and hilltop East, nearly flat Approx. size of plot 2 ha 1 ha Density of target species (adult common ash, Fraxinus excelsior L.) 41 trees/ha 53 trees/ha Diameter (dbh) range of adult ash trees 25–65 cm 20–55 cm Stand top height 25–30 m 20–25 m Other species present Field maple (Acer campestre L.) and hornbeam (Carpinus betulus L.), approximate collective share – 0.3; single, big decaying oak (Quercus sp.) trees Single European beech (Fagus sylvatica L.), maple (Acer sp.), European larch (Larix decidua Mill.) in the western border of stand Adult trees sampled and assessed for damage 82 53 DNA successfully genotyped 81 52 Saplings sampled and assessed for damage 80 80 DNA successfully genotyped 80 72 The second sampling site Siegenfeld is a small, nearly pure ash stand of ~1 ha area located in a forest between Siegenfeld and Heiligenkreuz at the eastern rim of the Alps, c. 30 km south of Vienna (Figure 1 and Table 1). It is surrounded by a spruce (Picea abies Karst.) forest in the north, west and south and adjacent to a forest road and meadow in the east. Other tree species in the ash stand are rare and include single beech, maple (Acer sp.) and larch (Larix decidua Mill.) trees. The altitude is ~400 m and mean annual precipitation is ~688 mm (data from Climate-data.org, 2017). In 2007, the Federal Research Centre for Forests (BFW) conducted a monitoring program on 50 plots in Lower Austria where crown dieback intensity of common ash was estimated on 20 trees per plot. Monitoring was continued on a sub-set of 16 plots in later years. In 2009, the Austrian Forest Inventory included 1200 plots in an effort to assess the status of ash dieback all over Austria. Siegenfeld in Lower Austria was included in all these assessments (Cech et al., 2012). Most of the 20 trees assessed in detail at Siegenfeld showed decreasing symptom intensity from 2007 to 2009 (Cech et al., 2012), which singled out Siegenfeld as one of the relatively healthier monitoring plots by now (trees at many other sites were not in a good health state at all; Katharina Schwanda pers. comm.). Saplings occurred in small patches at both sites that seemed to depend on penetrance of the canopy by sunlight. Biological material for DNA analysis was collected in June, July and August 2015 at both sites as follows. One leaf was sampled from randomly chosen saplings across the site (resulting in 80 saplings per site). Either a leaf, brought down by a slingshot if this was possible, or otherwise a bark plug per mature tree (from all mature ash trees at the site – resulted in a total of 134 trees) was collected (Table 1). The bark plug, from which cambium tissue can be sliced, was cut out of the trunk with a 1 cm diameter leather punch. Cambium tissue as well as leaf tissue contains the same genomic DNA. The material was immediately dried in silica gel and kept at room temperature prior to DNA extraction. Ash dieback was recorded as the degree of crown dieback (loss of crown foliage; see Table 2). Mature trees were classified into one of six damage classes by visual inspection (Figure 2A). There were also six damage classes for sapling assessment, but these were based on shoot damage instead (percentage of the shoots of a sapling affected by the disease; Figure 2B and Table 2). Additionally the diameter at breast height (DBH) of each tree and the height of each sapling were measured. Saplings were only sampled when they exceeded the height of ~60 cm, so that symptom identification and damage class categorization were possible with higher confidence. Table 2 Definition of damage classes Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Table 2 Definition of damage classes Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Damage class Range of crown foliage loss (trees) or percentage of damaged shoots (saplings) 1 No or few symptoms; < 10% loss/damage 2 Between 10 and 25% loss/damage 3 Between 25 and 50% loss/damage 4 Between 50 and 75% loss/damage 5 Between 75 and < 100% loss/damage 6 Trees/saplings died from infection Figure 2 View largeDownload slide Illustration of the visually assessed damage classes (d.c.) (A) for mature trees, no tree was assessed for damage class 6 (photos from Siegenfeld by Alexandra Wohlmuth); (B) for saplings (d.c. 5 is not shown) (Photos by Thomas Kirisits, University of Natural Resources and Life Sciences, Vienna, Austria). Figure 2 View largeDownload slide Illustration of the visually assessed damage classes (d.c.) (A) for mature trees, no tree was assessed for damage class 6 (photos from Siegenfeld by Alexandra Wohlmuth); (B) for saplings (d.c. 5 is not shown) (Photos by Thomas Kirisits, University of Natural Resources and Life Sciences, Vienna, Austria). DNA extraction On average, 25–45 mg dried leaf or cambium tissue per individual was put into 2 mL tubes together with two glass balls of 3 mm diameter and one of 4 mm (for cambium tissue, steel balls were used), one spatula tip each of glass powder, activated charcoal, polyvinyl pyrrolidon (PVP 40 000) and sodium metabisulfite (pro analysi grades were used for all chemicals, most of which were purchased from Sigma-Aldrich, St. Louis, MO, USA). For homogenization, the material was frozen in liquid nitrogen for two minutes and ground with a TissueLyser shaking mill (QIAGEN, Hilden, Germany) at 25 Hz for 2 min. The process was repeated a second time. The DNA was extracted using the Invisorb® Spin Plant Mini Kit (STRATEC Molecular, Birkenfeld, Germany), applying the protocol recommended for the kit, but replacing Lysis Buffer P (provided by the kit) with a mixture of 800 μL 2× CTAB Buffer (20 g/L cetyl trimethyl ammonium bromide [CTAB], 100 mM Tris–HCL pH 8.0, 1.4 M NaCl, 25 mM EDTA pH 8.0) plus 1.6 μL β-mercaptoethanol and 1 μL proteinase K (QIAGEN) for lysis. Additionally 40 μL RNAse A (10 mg/mL; QIAGEN) were added to each sample before the binding process. To determine DNA concentrations, a NanoDrop 1000 Spectrophotometer (Thermo Fischer Scientific, Ulm, Germany) was used. DNA amplification and fragment analysis DNA concentrations for all samples were below 20 ng μL−1. Therefore, DNA extracts were employed undiluted for polymerase chain reaction (PCR). Nine nuclear microsatellite loci were used for amplification (with annealing temperatures: Femsatl-4 – 60°C, Femsatl-10 – 50°C, Femsatl-11 – 55°C, Femsatl-12 – 55°C, Femsatl-16 – 62°C, Femsatl-19 – 58°C, M230 – 57°C, FR639485 – 55°C and FR646655 – 60°C; Brachet et al., 1999; Lefort et al., 1999; Beatty et al., 2015). Each forward primer was labelled with a fluorescent dye. PCRs were carried out using the QIAGEN Type It Microsatellite Kit. The amplification reactions were performed on a PTC-100 Thermal Cycler (BIO-RAD, Vienna, Austria) under the following conditions: An initial denaturation step of 5 min at 95°C, 28 cycles of denaturation at 95°C for 30 s, annealing with corresponding temperatures (see above) for 90 s and extension at 72°C for 30 s, and a final extension step at 60°C for 30 min. A CEQ 8000 Beckman-Coulter (Vienna, Austria) Sequencer was used to visualize the PCR products based on fragment length polymorphism, as compared with CEQ DNA Size Standard Kit-400 (Beckman-Coulter). Allele assessment, calling and binning were carried out using the fragment analysis tool of GenomeLab GeXP Beckman-Coulter software (version 10.2.3), with additional visual inspection and binning of peaks. Parentage analysis CERVUS 3.0.7 software (Marshall et al., 1998; Kalinowski et al., 2007; Field Genetics Ltd, London, UK) was used for parentage analysis and for calculating other parameters, such as expected (He) and observed (Ho) heterozygosity, polymorphic information content (PIC), average non-exclusion probability for one candidate parent (NE-P1), average non-exclusion probability for a candidate parent pair (NE-PP), Hardy Weinberg equilibrium (HW) and estimated null allele frequency (NULL). All of these parameters give information about the loci and their suitability for parentage analysis. For parentage assignment, the CERVUS program uses likelihood ratio, a well-established statistical method (Marshall et al., 1998). Parentage is assigned to a candidate parent if the likelihood is large relative to the likelihood of alternative candidate parents. The likelihood ratio is expressed as LOD scores (logarithm of the likelihood ratio; Marshall et al., 1998). Candidate parents with positive LOD scores are more likely to be the true parents, and candidate parents with negative LOD scores are less likely to be the true parents. If two or more parents have positive LOD scores, Marshall et al. (1998) defined delta (difference in LOD scores) as an assessment criterion. Delta is the difference in LOD scores between the most likely candidate parent and the second most likely candidate parent. This parameter is useful when two candidate parents have a positive LOD score. If delta is high enough, parentage can be assigned to the candidate parent with the higher LOD score. The advantage in the use of CERVUS lies in the allowance for mistyping and missing data for individuals at a specified number of loci. Parentage analysis for one parent (implemented by the ‘maternity analysis’ function of CERVUS) and parent-pair analysis were carried out for both sites separately. Prior to parentage analysis, the simulation of parentage was performed. This is important, because it is used to check the feasibility of the parentage analysis and it calculates values of likelihood ratios, so that the confidence of parentage assignment can be determined. In short, in a simulation appropriate LOD and delta scores for valid parentage assignment are generated for the parentage analysis with the real data from genotyping. The simulation in this study was performed with 10 000 offspring, an error rate of 0.01 at strict (95 per cent) and relaxed (80 per cent) confidence levels. As additional parameters of the simulation, the number of candidate parents was set to 100 for Johannser Kogel with a 0.75 proportion of candidate parents sampled, and 60 candidate parents with a 0.90 proportion of candidate parents sampled for Siegenfeld. The proportion of sampled candidate parents were estimated by field observation, as the occurrence of unsampled trees in the proximity of the stands cannot be determined with certainty. Unsampled candidate parents were assumed (for the purpose of the CERVUS simulation) to be present in moderate frequency at Johannser Kogel and at low frequency at Siegenfeld. Data analysis We performed a correlation analysis on the relationship between the DBH of mature ash trees and the number of their offspring in order to see whether tree size explains seedling numbers and thus possibly confounds health state correlations. The damage classes of trees in different DBH classes at both sites were plotted as a ‘box and whisker plot’, with a similar intention (to check for possible covariance). We tabulated the number of offspring per damage class for parents in each damage class. We also calculated the frequency distribution of the damage classes of local saplings (one or both parents assigned locally) and immigrant saplings (no local parents assigned, thus descents exclusively of trees outside of the stand) to see whether there was a difference in the health performance between these cohorts (local and immigrant), using a Kolmogorov–Smirnov test to check for significance. The boxplot (DBH for damage classes) was done in SPSS, whereas the tables and the bar diagram were done in Microsoft Excel 2013 (Redmond, WA, USA). To estimate the significance of correlation between damage class of parent and offspring (categorial data), a one-tailed Spearman’s rank correlation analysis was calculated using SPSS Statistics 23 software (IBM, Vienna, Austria). The calculation was applied separately for those offspring where both parents were assigned (Spearman’s rank correlation coefficient between the damage class of the offspring and the averaged mean damage class of both parents) and for those where only one parent was assigned (Spearman’s rank correlation coefficient between the damage classes of the offspring and the one parent). Alternatively, an attempt was made to estimate heritability in situ in the sense of Ritland (2000): a pairwise matrix of differences in damage class for each possible pair of individuals in our study was regressed onto a pairwise kinship coefficient matrix, using the program SpaGeDi 1.05 (Hardy and Vekemans, 2002). The slope (b) and intercept of the regression, as well as the coefficient of determination (r2) were calculated. This coefficient estimates the degree to which similarity in damage classes is determined by kinship, thus resembles a heritability value. The data set was permutated 1000 times in order to estimate P-values. Results Tree and sapling dimensions, and damage assessment The mean diameter of the mature trees at breast height (DBH) was 39.7 cm at Johannser Kogel and 35 cm at Siegenfeld. Johannser Kogel hosts some particularly old trees with DBH diameters of around 60 cm (Table 1). At Johannser Kogel mean height of saplings was 96.5 cm with a range from 65 to 190 cm, and at Siegenfeld, 99.3 cm ranging from 70 to 170 cm. Disease symptoms were present at both sites in adults and saplings. Shoot dieback and crown defoliation were detectable, but there were few, if any, stem collar necroses. Both sites show a similar range of slightly to severely damaged individuals, with most mature trees having a crown damage intensity between 10 and 50 per cent and assigned to damage classes 2 and 3 (67 per cent of the trees at Johannser Kogel and 54 per cent at Siegenfeld). Percentages of assessment to damage class 1 were higher in Siegenfeld than in Johannser Kogel (Figure 3). However, the distribution of damage classes of saplings and mature trees showed no significant differences across both sampling sites according to Kolmogorov–Smirnov tests (saplings: P = 0.692; trees: P = 0.897). No tree was observed to have lost all its crown foliage (no damage class 6). Trees in the healthiest damage class at Johannser Kogel had higher DBH, but the ranges of DBH in the other damage classes at both sites were overlapping (Figure 4). Figure 3 View largeDownload slide The percentage of ash individuals that belong to damage classes 1–6 shown separately for mature trees and saplings from each study site; SJ: saplings from Siegenfeld; SA: mature trees from Siegenfeld; JJ: saplings from Johannser Kogel; JA: mature trees from Johannser Kogel. Figure 3 View largeDownload slide The percentage of ash individuals that belong to damage classes 1–6 shown separately for mature trees and saplings from each study site; SJ: saplings from Siegenfeld; SA: mature trees from Siegenfeld; JJ: saplings from Johannser Kogel; JA: mature trees from Johannser Kogel. Figure 4 View largeDownload slide Boxplot of the assignment of mature ash trees of different size (measured by their diameter at breast height, DBH) to different damage classes at the two study sites. Figure 4 View largeDownload slide Boxplot of the assignment of mature ash trees of different size (measured by their diameter at breast height, DBH) to different damage classes at the two study sites. Allele frequencies A total of 285 individuals of 294 sampled were successfully genotyped for the nine microsatellite loci. DNA from one ash tree and eight saplings (including all of the six dead saplings from Siegenfeld) could not be amplified, likely due to insufficient DNA quality. These individuals had to be excluded from further analysis. The loci showed high allele numbers (N) and most of them were highly polymorphic (Table 3). The highest allele variation was found at locus Femsatl-10 with 41 alleles, and the lowest variation with six alleles at locus Femsatl-16. In total, 183 alleles were detected in 161 individuals from Johannser Kogel with an average number of 20.3 alleles per locus, and 166 alleles were detected for 124 individuals from Siegenfeld (average 18.4 alleles per locus; Table 3). The mean proportion of loci typed exceeded 0.99 for both sites. Mean expected heterozygosity (He) was 0.75 for Johannser Kogel and 0.80 for Siegenfeld. Observed heterozygosity (Ho) ranged from 0.38 to 0.85 for Johannser Kogel, and from 0.29 to 0.87 for Siegenfeld. Expected heterozygosity was higher than observed in most of the loci and in both sites. Mean polymorphic information content was 0.71 for Johannser Kogel and 0.77 for Siegenfeld. Deviations from Hardy–Weinberg equilibrium were detected at five loci in Johannser Kogel and at one locus in Siegenfeld. The estimated frequency of null alleles ranged from −0.003 (FR639485, Johannser Kogel) to 0.5 (Femsatl-12, Siegenfeld) with a mean value of 0.072. Table 3 Information about the microsatellite loci used and the analysed parameters Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Number of alleles (N); Ho, observed heterozygosity; He, expected heterozygosity; PIC, polymorphic information content; NE-1 P, average non-exclusion for one candidate parent; NE-PP, average non-exclusion probability for a candidate parent-pair; HW, significance of deviation from Hardy–Weinberg equilibrium; NS = not significant, ND = not determined. *Significant at 5 per cent level, **significant at 1 per cent level, ***significant at 0.1 per cent level. Null: Estimated null allele frequency. Table 3 Information about the microsatellite loci used and the analysed parameters Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Locus Analysed parameters Johannser Kogel Analysed parameters Siegenfeld N Ho He PIC NE-1 P NE-PP HW Null N Ho He PIC NE-1 P NE-PP HW Null Femsatl-4 26 0.665 0.734 0.695 0.65 0.275 NS 0.052 21 0.734 0.808 0.787 0.53 0.162 NS 0.0437 Femsatl-10 41 0.745 0.895 0.884 0.347 0.067 * 0.0924 37 0.772 0.935 0.927 0.24 0.031 ND 0.0913 Femsatl-11 22 0.776 0.873 0.858 0.406 0.096 *** 0.0589 19 0.789 0.883 0.869 0.384 0.085 NS 0.0561 Femsatl-12 17 0.376 0.737 0.701 0.646 0.268 *** 0.3333 15 0.286 0.855 0.837 0.446 0.114 *** 0.5006 Femsatl-16 7 0.547 0.526 0.454 0.858 0.588 NS −0.0227 6 0.459 0.524 0.489 0.85 0.507 NS 0.0853 Femsatl-19 16 0.85 0.802 0.776 0.549 0.186 ** −0.0328 17 0.79 0.877 0.861 0.403 0.096 NS 0.0515 M230 37 0.844 0.891 0.881 0.349 0.066 * 0.0234 36 0.871 0.94 0.933 0.225 0.027 ND 0.0366 FR639485 9 0.638 0.583 0.524 0.816 0.496 NS −0.051 8 0.677 0.654 0.601 0.757 0.405 NS −0.0258 FR646655 8 0.696 0.695 0.647 0.719 0.36 NS −0.0029 7 0.697 0.71 0.66 0.701 0.342 NS 0.008 total 183 166 mean 20.3 0.682 0.748 0.713 0.593 0.267 0.0501 18.4 0.675 0.798 0.774 0.504 0.197 0.0941 Number of alleles (N); Ho, observed heterozygosity; He, expected heterozygosity; PIC, polymorphic information content; NE-1 P, average non-exclusion for one candidate parent; NE-PP, average non-exclusion probability for a candidate parent-pair; HW, significance of deviation from Hardy–Weinberg equilibrium; NS = not significant, ND = not determined. *Significant at 5 per cent level, **significant at 1 per cent level, ***significant at 0.1 per cent level. Null: Estimated null allele frequency. Parentage assignment The assignments suggested by CERVUS (including LOD scores and delta values) are given in Supplementary Data File S1. Combined non-exclusion probabilities for the first parent and for the parent pair were 5.70E-03 and 6.00E-07 for Johannser Kogel and 8.92E-04 and 8.73E-09 for Siegenfeld if Femsatl-12 was included. Without Femsatl-12, the values were less favourable (by approximately one order of magnitude in most cases): 8.82E-03 and 2.24E-06 (Johannser Kogel first parent/parent pair) and 2.00E-03 and 8.00E-08 (Siegenfeld first parent/parent pair). Maternity analysis, i.e. the assignment of only one parent, for Johannser Kogel was successful in 42 cases or 52.5 per cent (20 of the cases also at high stringency); for Siegenfeld, it resulted in a 58.3 per cent assignment rate (42 saplings; all at strict confidence level). The assignment rate for the parent-pair analysis in Johannser Kogel was 7.5 per cent, implying that five saplings from Johannser Kogel were assigned to both parents (among the local adult trees) at relaxed confidence level (80 per cent) and one additional sapling at strict confidence level (95 per cent). In Siegenfeld, the assignment rate for the parent-pair analysis was higher with 27.7 per cent – both parents were determined for 20 saplings from Siegenfeld at a relaxed confidence level (10 of which also at strict confidence level). Parentage analysis (‘maternity’ or one parent only) without marker Femsatl-12 (at less stringent non-exclusion probabilities) resulted in 39 saplings assigned to candidate parents at Johannser Kogel at 80 per cent confidence level (19 of which at strict level). For Siegenfeld, the numbers increased as well, two additional saplings were assigned to a single parent (at 80 per cent), and eight more at 95 per cent, but one of the original ones lost significance of assignment (total, 50 saplings assigned), while for two saplings, alternative parents were assigned. Parent-pair assignment without Femsatl-12 resulted in the following changes: Johannser Kogel, six more saplings assigned at 80 per cent, one more at 95 per cent, one improved from 80 to 95 per cent, but three lost at 80 per cent significance. At Siegenfeld, there were 11 more saplings assigned at 80 per cent, one more at 95 per cent, four improved, one worsened in significance, and three seedlings were assigned to different parent pairs (numbers and details in Table 4 and Supplementary Data File S1). Table 4 Spearman’s rank-correlation analyses for offspring for which one parent was assigned, and for offspring for which both parents were assigned, calculated including and excluding marker Femsatl-12 Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* NS: not significant, *significant at the 0.05 level. Table 4 Spearman’s rank-correlation analyses for offspring for which one parent was assigned, and for offspring for which both parents were assigned, calculated including and excluding marker Femsatl-12 Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* Comparison Number of cases Spearman’s rank correlation coefficient P-value and significance Offspring assigned to parent pairs at 80% confidence including Femsatl-12 – damage class offspring to average of parent pairs 26 0.098 0.318 NS Additional offspring assigned to one parent at 80% confidence including Femsatl-12 – damage classes compared 58 −0.082 0.272 NS Offspring assigned to parent pairs at 80% confidence excluding Femsatl-12 – damage class offspring to average of parent pairs 42 0.059 0.710 NS Additional offspring assigned to one parent at 80% confidence excluding Femsatl-12 – damage classes compared 50 −0.293 0.039* NS: not significant, *significant at the 0.05 level. Data analysis The data show a very weak positive correlation of parents with higher DBH with the number of their offspring (R2 = 0.0101; Supplementary Figure S1; single-parent and parent-pair results combined; FEMSATL 12 included). The numbers of offspring per disease class of adult trees are rather equally distributed across damage classes (Supplementary Table S1). The frequency distribution of the damage classes of local and immigrant saplings showed no apparent differences in the health status between these cohorts at both sites (Supplementary Table S2; not significantly different in Kolmogorov–Smirnov tests). Spearman’s rank correlation for parentage vs damage class was generally low (−0.293 to 0.039, Table 4) and thus showed no apparent association between damage classes of offspring and parents, neither for parent-pair, nor for maternity analysis, and only one of the coefficients – with a negative sign! – was significantly different from zero (−0.293 without Femsatl-12 for single parent/maternity, P = 0.039; Table 4). The value for in situ heritability (r2) obtained by the regression of damage class differences on kinship (calculated including Femsatl-12; distribution shown in Supplementary Figure S2) was very close to zero (4.69E-006) and not statistically significant after permutations (P = 0.34). Discussion Genetic parameters and parentage assignment The majority of the loci used in this study were highly polymorphic, and expected heterozygosity was high. Similar observations were made for the same microsatellite loci in previous population genetic studies in common ash, suggesting high genetic diversity within populations across Europe (Heuertz et al., 2001, Hebel et al., 2006; Beatty et al., 2015). The values for genetic variation are typical for studies with these microsatellite markers in European ash (see Heinze and Fussi, 2017 for a comparison). While the deviations from Hardy–Weinberg equilibrium, and the high frequency of null alleles in Femsatl-12 are not ideal, the combined non-exclusion probabilities are very much appropriate (given the numbers of candidate parents present). Femsatl-12 seems to show an increase of null alleles from Western to Eastern Europe (Heinze and Fussi, 2017, their Table 4). Null alleles do not amplify consistently during PCR and thus cannot be reliably detected, leading to false assessments of heterozygotes as homozygotes (Kalinowski and Taper, 2006). They are problematic in parentage analysis and appear sometimes simultaneously with deviations from the Hardy–Weinberg equilibrium (Pemberton et al., 1995), which was also the case for five loci in Johannser Kogel and at one locus in Siegenfeld. However, estimation of null allele frequencies does not necessarily imply that a null allele is present. If no parent–offspring relationship is definitely known and there is no Hardy–Weinberg equilibrium, it is difficult (or impossible) to decide about the presence of null alleles with certainty (Dakin and Avise, 2004). We do not expect strict Hardy–Weinberg equilibria at our sites, as we observed incoming seed and pollen in the young generation. But even Femsatl-12 adds power to this analysis (better non-exclusion probabilities), because undetected null alleles do not lead to (wrong) exclusions of candidate parents; so the number of parent–offspring pairs is not underestimated because of the null alleles. It could be overestimated, but the simulations in CERVUS take the null frequencies into account for suggesting thresholds of LOD and delta values. The deviations from Hardy–Weinberg equilibria are already a hint for absence of panmixia (which would require equal parental contributions to the offspring generation) and/or the presence of immigrant saplings (see below). Both our sites had nearly the same set of alleles with few private alleles (alleles that only occur in one population) at each site, and little differentiation (like in Heuertz et al., 2001; Hebel et al., 2006; Ballian et al., 2008). Surprisingly, positive assignments of at least one parent were made only for slightly more than half of the saplings under various conditions of the analysis, and only 17 per cent of the saplings were assigned to both parents within their stands, despite their relative isolation. A possible technical explanation for the low assignment rate could be genotyping or binning errors which would probably lead to undetected parent–offspring trios or duos, although CERVUS uses likelihood equations that take account of such errors. High combined exclusion probability (99 per cent) across all loci for unrelated parents suggests low error rate in parentage assignment and shows that the microsatellite set used was adequate. Nevertheless, we tested our data analysis procedure omitting the critical Femsatl-12 marker. The parent non-exclusion probabilities then changed to less stringent values (Supplementary Data File S1) and slightly higher rates for parentage assignment resulted, so that more saplings could be assigned to parents. However, the correlation (Spearman’s rank) did not change in a meaningful way. Thus our conditions are conservative in the sense that more relaxed assignment conditions, if they lead to higher errors in parent–offspring assignments, possibly underestimate correlations (but provide a wider data basis for them). Although we made an effort to sample all putative parents in both fairly isolated stands, it is possible that these missing parents are actually scattered trees outside the stands. Therefore, a more likely explanation for the low assignment rates is that the rest of the saplings had been sired by these unsampled trees. High seed and pollen inflow from the surroundings is a phenomenon often detected in forest tree stands, even if we did not expect it here due to the scarcity of ash trees in the surroundings, and because of the generally low physical disposition of ash seed and pollen for passive transport beyond a few hundred metres (Richards, 1997). Nevertheless, in this respect our findings resemble those of Lobo et al. (2015), where paternity analysis in two Danish seed orchards resulted in an assignment rate of less than 40 per cent. They suggested the small size of the seed orchard as one possible explanation (12 putative parents in the orchard). Analysis of seed dispersal in two natural ash woodlands in Ireland carried out by Beatty et al. (2015) indicates the potential for seed dispersal over several hundreds of metres. Bacles et al. (2005) and Bacles and Ennos (2008) estimated pollen dispersal in fragmented ash populations in Scotland, suggesting effective pollination within 300 m, but detected rare pollination events over distances of up to 3 km. While these studies on pollination were carried out in isolated tree stands in open landscapes, where pollen can be transported easier with less barriers being present, a study by Heuertz et al. (2003) in a Romanian continuous ash woodland estimated average distances of 14 m between seeds and their female parents, and average pollen flow below 140 m distance (between seed mothers and pollinating trees). However, their common ash stand was much denser (~200 mature stems per ha), and their gene flow estimates were indirectly derived from genetic data of adults only (decrease of kinship with increasing distance), while we analysed saplings (and thus the ‘realized gene flow’ of the transition of one generation to the next). We infer from our low assignment rate for parent-pair analysis and the moderate assignment rate for maternity analysis that seed and pollen dispersal within mixed landscapes is probably higher than the estimates of Heuertz et al. (2003) for continuous woodlands and more in the range of the estimates of Beatty et al. (2015) in Irish woodlands (but lower than those for very open landscapes). It is also the experience of foresters in Austria in similar forest types that (i) either a single or a few ash trees are able to produce dense seedling patches, if light and other conditions are good (‘seed shadows’), and (ii) these patches can be found up to a few hundred metres away from the source (especially following the main wind direction; Herfried Steiner, Werner Ruhm, pers. comm.). Further analysis is needed in order to infer the precise influence of landscape features on these parameters. Variation in susceptibility The presence of seeds that originate entirely from outside the stands investigated allowed us to calculate if there are ‘population differences’ in susceptibility between offspring of the local trees vs immigrant saplings. The frequency distribution of the damage classes between these cohorts showed no apparent differences (Supplementary Table S2). This is a further indication that there is no selection (yet) for highly tolerant saplings at our sites (or for more tolerant adults with a consequent higher success in contributing to seed production in the wider surroundings). It also follows that there is no reason to believe that different susceptibilities of trees outside our stands (compared with the adults within our stands) have led to any systematic error in the correlations. No significant relationship (with the exception of one negative) was found based on Spearman’s rank correlation between damage intensity of parents and their offspring (Table 4), neither for parent–offspring duos (offspring with one parent), nor for parent–offspring trios (offspring with both parents). Previous studies suggested genetically based variation as an explanation for observed differences in susceptibility to dieback symptoms based on the observation of clones and single-tree offspring in progeny tests, and they concluded that these differences are transferred across ash generations (Bakys et al., 2009; McKinney et al., 2011; Lobo et al., 2014; Muñoz et al., 2016). Most of these studies were performed in clonal seed orchards or established progeny trials. The seedlings were raised in a nursery prior to planting (Kjær et al., 2012; Muñoz et al., 2016) and the clones for seed orchards were grafted onto rootstocks and also further cultivated in a nursery, and rather intensively managed after planting (with wider spacing; McKinney et al., 2011; Stener, 2013). Family size in the study of Muñoz et al. (2016) was between 8 and 68 half-sibs per mother tree. Lobo et al. (2015) used 2–48 offspring per half-sib family in inoculation tests (8–48 in maternal families). We expected to arrive at numbers similar to the lower part of this range in our settings, but that was not the case. Progeny trails and seed orchards (and especially sapling inoculations) represent more controlled environments and they often do not fully resemble natural woodland conditions. In contrast, this study was conducted in naturally regenerating ash stands, and the more heterogeneous conditions there likely led to higher genotype × (micro-)environment interactions., which in turn lowered the genetic correlations observed (see also below) Lobo et al. (2014) estimated the heritability of damage intensity in two progeny trails in Denmark. Both trials included progenies from the same mother trees, but one trial was left unfenced and progenies were exposed to strong competition from vegetation and to browsing animals, whereas the other was protected by fencing. Lower estimated heritability values of the unfenced (0.2) compared with the fenced site (0.42–0.53) indicated that estimates of heritability under heterogeneous conditions may be lower (Lobo et al., 2014). Johannser Kogel is also fenced, but the fence is obviously penetrable at some points, as wild boars were observed inside the fence during sampling. Pliūra et al. (2014) tested clones from seven different sites from Lithuania exposed to different ash-dieback infection pressure. They calculated the genotype–environment interaction on phenotypic variation and found significant contributions of genetic variation in plasticity and reaction norms of clones across a range of infection pressure environments (Pliūra et al., 2014). This indicates the presence of variation of individual response to the disease across sites depending on site conditions (Pliūra et al., 2011, 2014), or different levels of disease pressure. Results from previous studies, together with our current ones, suggest that precise estimates of heritability are only possible in more controlled environments, or with much higher sample numbers (sapling half-sib families detected and assigned to parents). Therefore, environmental conditions and sample sizes are likely more relevant and need to be strongly considered for inferences on susceptibility in field observation studies. It seems from all cited studies that up to now, only a small fraction of ash individuals maintain potential resistance to the disease, as practically all trees are infected at some point. Kjær et al. (2012) found high susceptibility and mortality among trees in two progeny trails in Denmark. They further estimated that only 1 per cent of the trees have the potential to pass lower susceptibility to their offspring. In three Lithuanian progeny trails almost 90 per cent of the trees died during the observation period of eight years (Pliūra et al., 2011). In Austria, such high mortality rates have not yet been reported from observation sites (Heinze et al., 2017). In 50 ash dieback monitoring plots in Lower Austria (mainly adult stands), mean crown dieback intensity reached 18.1 per cent in 2009 and 17.6 per cent in 2010 (Kirisits and Freinschlag, 2012). Three seed orchards in Austria showed moderate mean crown dieback intensity in 2011 (14.2 per cent, 13.5 and 31 per cent), but no clone was totally unaffected (Heinze et al., 2017). Crown and shoot dieback intensity was also relatively moderate at Johannser Kogel and Siegenfeld (66 and 72 per cent of the individuals, respectively, showed lower than 50 per cent damage intensity). It is likely that the mature trees that were identified as parents did not express assessable resistance due to lower infection pressure (especially for the higher crowns of thicker trees, Figure 4), and consequently produced more offspring because of their greater size and the better health condition associated with it (Supplementary Figure S1), while infection pressure was higher for saplings close to the forest floor, where the presence of petioles and more humid conditions would favour high spore concentrations. This could be one reason for the low correlation between damage classes of parents and offspring in our study. However, this effect may impact the severity, but not so much the direction of susceptibility; healthier mature trees should then still produce healthier offspring, though the average damage class rating would be shifted between old trees and young saplings. Probably the small sample size of parent–offspring trios (N = 26–50) and of half-sib families (only nine mature trees produced more than two offspring) may have led to underestimation of actual correlations of damage intensity between parent and offspring for possible reduced susceptibility. Muñoz et al. (2016) found that there are family effects in open-pollinated offspring, but they also calculated individual breeding values for the offspring, and most of the genetic variation was found within families. Thus, our small family sizes may not allow estimating heritabilities with great precision. The in situ heritability estimation based on all kinship relationships among trees and saplings also resulted in a coefficient of determination of very close to zero. It may not necessarily be the case that related trees have very similar disease tolerance levels. Breeding values, which represent the average effect of the parent genotype, estimated from the performance of its offspring (McKinney et al., 2014), were calculated by Kjær et al. (2012) in a Danish progeny trail, and by Muñoz et al. (2016) in France. Only one of 101 tested mother trees in Denmark had breeding values for ‘susceptibility’ below 10 per cent and was estimated to produce healthy offspring (and four trees with breeding values below 20 per cent were estimated to produce fairly healthy offspring). Another study carried out by McKinney et al. (2011) supported these findings, with only one of 39 tested clones exhibiting breeding values below 10 per cent. Breeding values for susceptibility in Kjær et al. (2012) were normally distributed, suggesting that possible resistance is based on expression of several genes rather than on one alone (McKinney et al., 2014). The data from France (Muñoz et al., 2016) also suggest that many genes contribute to resistance, as the largest part of genetic variation was found within families. This also means that much larger numbers of saplings are necessary to estimate heritability in our approach. Harper et al. (2016), in contrast, discovered a number of single nucleotide polymorphism (SNP) and gene expression markers that are associated with crown damage in infected trees using Associated Transcriptomics. With three markers (concerning transcription factor genes) they were able to predict individuals with a low level of susceptibility to the disease (combined R2 value of 0.28). Sollars et al. (2017) very recently published the sequence of a low-heterozygosity ash tree, and further improved these markers (which, in their data, explained Danish disease scores with r2 = 0.25) and suggested that according to these markers, native trees in Great Britain would have a lower chance of becoming susceptible. They also reported on biochemical substances that have a higher prevalence in ‘resistant’ vs susceptible tree. Despite these advances, the exact genetic background(s) of low susceptibility still remains unclear and it may well be that more than one gene network or metabolic pathway is involved, or that different such mechanisms are at work in different regions of Europe. The robustness of the observed potential resistance in previous field studies needs to be re-evaluated. Precise genetic tests of their offspring are necessary to determine which of the field-resistant trees really transmit this trait to seeds. Large offspring numbers are necessary for this purpose, which make such tests costly. The alternative approach we tested would, however, require sites with high numbers of seedlings with established parentage, much higher than the ones we identified. Thus, such sites should have many seedlings, but few immigrants among them. This may be difficult to find. Although the distribution data of ash seedlings in the wider area of Johannser Kogel suggest the presence of ‘seed shadows’ of single adults (Herfried Steiner, pers. comm.), it remains to be seen whether numbers are high enough at a particular site. The higher workload of laboratory testing would require more time (and it would increase costs), but the approach may still be faster in total than field planting of seedlings to be tested for disease tolerance. It is uncertain whether low susceptibility in single, but infrequent individuals could hinder massive decline of ash in Europe, and if that resistance would sustain massive infection pressure at natural forest sites. However, selecting and breeding non-susceptible trees in a timely manner throughout Europe seems the only way to overcome the disease; the approach we discuss here may help to accelerate selection. Alternatively, selecting ash stands that combine many environmental and demographic factors that disadvantage high infection pressure and the progress of the disease (e.g. hot and dry summers, dry soils during the sporulation period of the fungus, big trees with large, dominating crowns or stands were leaf litter does not persist until the next summer) may emerge as a conservation strategy in situations where genetic tolerance is generally low (Heinze et al., 2017). Conclusion In this study, no significant correlation could be found between ash dieback damage intensity of parent and offspring in two natural stands in Austria. This can reflect low power of estimating heritability under natural in situ conditions, where the micro-environment may have a stronger influence on the trait than in planted tests. Given this, and the recent insights into the genetics of disease tolerance, much higher numbers of offspring with identified parenthood (than the ones we could find) are necessary for our approach. Other possible causes could be that there were no very tolerant parent trees at our sites, or that juvenile–adult correlations are low for this trait. We further suggest that the genetic basis for variation in susceptibility might be more complex under natural heterogeneous conditions than in more controlled environments like planted progeny trials or seed orchards, where most of the previous studies were implemented. The robustness of the observed potential low susceptibility in previous studies should therefore be critically re-evaluated after planting offspring from tolerant trees in production forests. The influence of environmental conditions should be considered when inferring on susceptibility (i.e. when assessing field resistance). A scenario of large-scale decline of common ash trees in Europe is becoming more and more likely. It will be important to identify and propagate healthy ash individuals throughout Europe, to find out about the genetic mechanism of any resistance in more detail and about the environmental conditions that favour low susceptibility, in order to implement a management program that accounts for these aspects. Supplementary data Supplementary data are available at Forestry online. Acknowledgements Special thanks to Renate Slunsky, Daniela Jahn, MSc, and the Genome Research unit at the Federal Research Centre for Forests (BFW) for their kind support and also to the Natural Forest Reserve and the Phytopathology units at BFW, especially to Dr Katharina Schwanda, Christian Neureiter and Mag. Herfried Steiner. ‘European Cooperation in Science and Technology (COST)’ Action FP1103 ‘FRAXBACK’ is acknowledged for providing a stimulating environment of meetings and discussions. Furthermore we would like to thank Dr Thomas Kirisits, University of Natural Resources and Life Sciences in Vienna, who provided his photos of the damage classes of saplings, and the Editor in Chief and three reviewers for their helpful comments and suggestions. Conflict of interest statement None declared. References Bacles , C.F.E. , Burczyk , J. , Lowe , A.J. and Ennos , R.A. 2005 Historical and contemporary mating patterns in remnant populations of the forest tree Fraxinus excelsior L . Evolution 59 ( 5 ), 979 – 990 . Google Scholar PubMed Bacles , C.F.E. and Ennos , R.A. 2008 Paternity analysis of pollen-mediated gene flow for Fraxinus excelsior L. in a chronically fragmented landscape . Heredity 101 ( 4 ), 368 – 380 . Google Scholar CrossRef Search ADS PubMed Bakys , R. , Vasaitis , R. , Barklund , P. , Ihrmark , K. and Stenlid , J. 2009 Investigations concerning the role of Chalara fraxinea in declining Fraxinus excelsior . Plant Pathol. 58 ( 2 ), 284 – 292 . Google Scholar CrossRef Search ADS Ballian , D. , Monteleone , I. , Ferrazzini , D. , Kajba , D. and Belletti , P. 2008 Genetic characterization of common ash (Fraxinus excelsior L.) populations in Bosnia and Herzegovina . Period. Biol. 10 , 323 – 328 . Beatty , G.E. , Brown , J.A. , Cassidy , E.M. , Finlay , C.M.V. , McKendrick , L. , Montgomery , W.I. et al. . 2015 Lack of genetic structure and evidence for long-distance dispersal in ash (Fraxinus excelsior) populations under threat from an emergent fungal pathogen: implications for restorative planting . Tree Genet. Genomes 11 ( 3 ), 53 . Google Scholar CrossRef Search ADS Brachet , S. , Jubier , M.F. , Richard , M. , Jung-Muller , B. and Frascaria-Lacoste , N. 1999 Rapid identification of microsatellite loci using 5′ anchored PCR in the common ash Fraxinus excelsior . Mol. Ecol. 8 ( 1 ), 160 – 163 . Cech , T.L. 2006 a Auffallende Schadfaktoren an Waldbäumen im Jahr 2005 [Striking damaging agents on forest trees in 2005] . Forstschutz Aktuell (BFW) 35 , 6 – 7 . Cech , T.L. 2006 b Eschenschaden in Österreich [Ash dieback and premature leaf shedding in Austria] . Forstschutz Aktuell (BFW) 37 , 18 – 20 . Cech , T.L. , Kessler , M. and Brandstetter , M. 2012 Monitoring des Zurücksterbens der Esche in Österreich [Monitoring ash dieback in Austria] . Forstschutz Aktuell (BFW) 55 , 56 – 58 . Cleary , M.R. , Andersson , P.F. , Broberg , A. , Elfstrand , M. , Daniel , G. and Stenlid , J. 2014 Genotypes of Fraxinus excelsior with different susceptibility to the ash dieback pathogen Hymenoscyphus pseudoalbidus and their response to the phytotoxin viridiol—a metabolomic and microscopic study . Phytochemistry 102 , 115 – 125 . Google Scholar CrossRef Search ADS PubMed Climate-data.org. 2017 Climate Data for Siegenfeld. https://de.climate-data.org/search/?q=Siegenfeld (accessed on 22 August, 2017). Comita , L.S. , Queenborough , S.A. , Murphy , S.J. , Eck , J.L. , Xu , K. , Krishnadas , M. , et al. . 2014 Testing predictions of the Janzen-Connell hypothesis: a meta-analysis of experimental evidence for distance- and density-dependent seed and seedling survival . J. Ecol. 102 , 845 – 856 . Google Scholar CrossRef Search ADS PubMed Dakin , E.E. and Avise , J.C. 2004 Microsatellite null alleles in parentage analysis . Heredity 93 , 504 – 509 . Google Scholar CrossRef Search ADS PubMed Dobrowolska , D. , Hein , S. , Oosterbaan , A. , Wagner , S. , Clark , J. and Skovsgaard , J.P. 2011 A review of European ash (Fraxinus excelsior L.): implications for silviculture . Forestry 84 , 133 – 148 . Google Scholar CrossRef Search ADS Forst-und Landwirtschaftsbetrieb der Stadt Wien . 2017 Lage, Größe, Geologie und Klima—Lebensraum Lainzer Tiergarten [Situation, size, geology and climate—habitat Lainzer Tiergarten; online; in German]. https://www.wien.gv.at/umwelt/wald/erholung/lainzertiergarten/lebensraum/lage.html (accessed on 22 August, 2017). Gross , A. , Holdenrieder , O. , Pautasso , M. , Queloz , V. and Sieber , T.N. 2014 Hymenoscyphus pseudoalbidus, the causal agent of European ash dieback . Mol. Plant Pathol. 15 ( 1 ), 5 – 21 . Google Scholar CrossRef Search ADS PubMed Hardy , O.J. and Vekemans , X. 2002 SPAGeDi: a versatile computer program to analyse spatial genetic structure at the individual or population levels . Mol. Ecol. Notes 2 , 618 – 620 . Google Scholar CrossRef Search ADS Harper , A.L. , McKinney , L.V. , Nielsen , L.R. , Havlickova , L. , Li , Y. , Trick , M. , et al. . 2016 Molecular markers for tolerance of European ash (Fraxinus excelsior) to dieback disease identified using Associative Transcriptomics . Sci. Rep. 6 , 19335 . Google Scholar CrossRef Search ADS PubMed Hebel , I. , Haas , R. and Dounavi , A. 2006 Genetic variation of common ash (Fraxinus excelsior L.) populations from provenance regions in southern Germany by using nuclear and chloroplast microsatellites . Silvae Genet. 55 ( 1 ), 38 – 44 . Google Scholar CrossRef Search ADS Heinze , B. and Fussi , B. 2017 Pre-disease levels of genetic diversity and differentiation among ash (Fraxinus excelsior L.) seedlots in Austria . Balt. For. 23 ( 1 ), 198 – 208 . Heinze , B. , Tiefenbacher , H. , Litschauer , R. and Kirisits , T. 2017 Ash dieback in Austria—history, current situation and outlook. In Dieback of European Ash (Fraxinus spp.)—Consequences and Guidelines for Sustainable Management . Vasaitis R. and Enderle R. (eds). , 2017. Swedish University of Agricultural Sciences , pp. 33 – 52 ISBN (print version) 978-91-576-8696-1. Heuertz , M. , Hausman , J.-F. , Tsvetkov , I. , Frascaria-Lacoste , N. and Vekemans , X. 2001 Assessment of genetic structure within and among Bulgarian populations of the common ash (Fraxinus excelsior L.) . Mol. Ecol. 10 ( 7 ), 1615 – 1623 . Google Scholar CrossRef Search ADS PubMed Heuertz , M. , Vekemans , X. , Hausman , J.-F. , Palada , M. and Hardy , O.J. 2003 Estimating seed vs. pollen dispersal from spatial genetic structure in the common ash . Mol. Ecol. 12 ( 9 ), 2483 – 2495 . Google Scholar CrossRef Search ADS PubMed Kalinowski , S.T. and Taper , M.L. 2006 Maximum likelihood estimation of the frequency of null alleles at microsatellite loci . Conserv. Genet. 7 ( 6 ), 991 – 995 . Google Scholar CrossRef Search ADS Kalinowski , S.T. , Taper , M.L. and Marshall , T.C. 2007 Revising how the computer program CERVUS accommodates genotyping error increases success in paternity assignment . Mol. Ecol. 16 ( 5 ), 1099 – 1106 . Google Scholar CrossRef Search ADS PubMed Keßler , M. , Cech , T.L. , Brandstetter , M. and Kirisits , T. 2012 Dieback of ash (Fraxinus excelsior and Fraxinus angustifolia) in Eastern Austria: disease development on monitoring plots from 2007 to 2010 . J. Agric. Ext. Rural Dev. 4 ( 9 ), 223 – 226 . Kirisits , T. and Freinschlag , C. 2012 Ash dieback caused by Hymenoscyphus pseudoalbidus in a seed plantation of Fraxinus excelsior in Austria . J. Agric. Ext. Rural Dev. 4 , 184 – 191 . Kjær , E.D. , McKinney , L.V. , Nielsen , L.R. , Hansen , L.N. and Hansen , J.K. 2012 Adaptive potential of ash (Fraxinus excelsior) populations against the novel emerging pathogen Hymenoscyphus pseudoalbidus . Evol. Appl. 5 ( 3 ), 219 – 228 . Google Scholar CrossRef Search ADS PubMed Kowalski , T. 2006 Chalara fraxinea sp. nov. associated with dieback of ash (Fraxinus excelsior) in Poland . For. Pathol. 36 ( 4 ), 264 – 270 . Google Scholar CrossRef Search ADS Kräutler , K. and Kirisits , T. 2012 The ash dieback pathogen Hymenoscyphus pseudoalbidus is associated with leaf symptoms on ash species (Fraxinus spp.) . J. Agric. Ext. Rural Dev. 4 , 261 – 265 . Landolt , J. , Gross , A. , Holdenrieder , O. and Pautasso , M. 2016 Ash dieback due to Hymenoscyphus fraxineus: what can be learnt from evolutionary ecology? Plant Pathol. 65 ( 7 ), 1056 – 1070 . Google Scholar CrossRef Search ADS Lefort , F. , Brachet , S. , Frascaria-Lacoste , N. , Edwards , K.J. and Douglas , G.C. 1999 Identification and characterization of microsatellite loci in ash (Fraxinus excelsior L.) and their conservation in the olive family (Oleaceae) . Mol. Ecol. 8 ( 6 ), 1088 – 1089 . Google Scholar CrossRef Search ADS Lobo , A. , Hansen , J.K. , McKinney , L.V. , Nielsen , L.R. and Kjær , E.D. 2014 Genetic variation in dieback resistance: growth and survival of Fraxinus excelsior under the influence of Hymenoscyphus pseudoalbidus . Scand. J. For. Res. 29 ( 6 ), 519 – 526 . Google Scholar CrossRef Search ADS Lobo , A. , McKinney , L.V. , Hansen , J.K. , Kjær , E.D. and Nielsen , L.R. 2015 Genetic variation in dieback resistance in Fraxinus excelsior confirmed by progeny inoculation assay . For. Pathol. 45 ( 5 ), 379 – 387 . Google Scholar CrossRef Search ADS Marigo , G. , Peltier , J.-P. , Girel , J. and Pautou , G. 2000 Success in the demographic expansion of Fraxinus excelsior L . Trees 15 ( 1 ), 1 – 13 . Google Scholar CrossRef Search ADS Marshall , T.C. , Slate , J. , Kruuk , L.E.B. and Pemberton , J.M. 1998 Statistical confidence for likelihood-based paternity inference in natural populations . Mol. Ecol. 7 ( 5 ), 639 – 655 . Google Scholar CrossRef Search ADS PubMed McKinney , L.V. , Nielsen , L.R. , Collinge , D.B. , Thomsen , I.M. , Hansen , J.K. and Kjær , E.D. 2014 The ash dieback crisis: genetic variation in resistance can prove a long-term solution . Plant Pathol. 63 ( 3 ), 485 – 499 . Google Scholar CrossRef Search ADS McKinney , L.V. , Nielsen , L.R. , Hansen , J.K. and Kjær , E.D. 2011 Presence of natural genetic resistance in Fraxinus excelsior (Oleraceae) to Chalara fraxinea (Ascomycota): an emerging infectious disease . Heredity 106 ( 5 ), 788 – 797 . Google Scholar CrossRef Search ADS PubMed McKinney , L.V. , Thomsen , I.M. , Kjær , E.D. and Nielsen , L.R. 2012 Genetic resistance to Hymenoscyphus pseudoalbidus limits fungal growth and symptom occurrence in Fraxinus excelsior . For. Pathol. 42 ( 1 ), 69 – 74 . Google Scholar CrossRef Search ADS Morand-Prieur , M.-E. , Raquin , C. , Shykoff , J.A. and Frascaria-Lacoste , N. 2003 Males outcompete hermaphrodites for seed siring success in controlled crosses in the polygamous Fraxinus excelsior (Oleaceae) . Am. J. Bot. 90 ( 6 ), 949 – 953 . Google Scholar CrossRef Search ADS PubMed Muñoz , F. , Marcais , B. , Dufour , J. and Dowkiw , A. 2016 Rising out of the ashes: additive genetic variation for crown and collar resistance to Hymenoscyphus fraxineus in Fraxinus excelsior . Phytopathology 106 , 1535 – 1543 . Google Scholar CrossRef Search ADS PubMed Pautasso , M. , Aas , G. , Queloz , V. and Holdenrieder , O. 2013 European ash (Fraxinus excelsior) dieback—a conservation biology challenge . Biol. Conserv. 158 , 37 – 49 . Google Scholar CrossRef Search ADS Pemberton , J.M. , Slate , J. , Bancroft , D.R. and Barrett , J.A. 1995 Non-amplifying alleles at microsatellite loci: a caution for parentage and population studies . Mol. Ecol. 4 , 249 – 252 . Google Scholar CrossRef Search ADS PubMed Pliūra , A. , Lygis , V. , Suchockas , V. and Bartkevicius , E. 2011 Performance of twenty-four European Fraxinus excelsior populations in three Lithuanian progeny trials with a special emphasis on resistance to Chalara fraxinea . Balt. For. 17 , 17 – 34 . Pliūra , A. , Marčiulynienė , D. , Bakys , R. and Suchockas , V. 2014 Dynamics of genetic resistance to Hymenoscyphus pseudoalbidus in juvenile Fraxinus excelsior clones . Balt. For. 20 ( 1 ), 10 – 27 . Przybyl , K. 2002 Fungi associated with necrotic apical parts of Fraxinus excelsior shoots . For. Pathol. 32 ( 6 ), 387 – 394 . Google Scholar CrossRef Search ADS Queloz , V. , Grünig , C.R. , Berndt , R. , Kowalski , T. , Sieber , T.N. and Holdenrieder , O. 2011 Cryptic speciation in Hymenoscyphus albidus . For. Pathol. 41 ( 2 ), 133 – 142 . Google Scholar CrossRef Search ADS Richards , A.J. 1997 Plant breeding systems. 2nd edn . Chapman & Hall , p. 529 . Sollars , E.S.A. , Harper , A.L. , Kelly , L.J. , Sambles , C.M. , Ramirez-Gonzalez , R.H. , Swarbreck , D. , et al. 2017 Genome sequence and genetic diversity of European ash trees . Nature 541 ( 7636 ), 212 – 216 . Google Scholar CrossRef Search ADS PubMed Stener , L.-G. 2013 Clonal differences in susceptibility to the dieback of Fraxinus excelsior in southern Sweden . Scand. J. For. Res. 28 ( 3 ), 205 – 216 . Google Scholar CrossRef Search ADS Thomas , P.A. 2016 Biological flora of the British isles: Fraxinus excelsior . J. Ecol. 104 ( 4 ), 1158 – 1209 . Google Scholar CrossRef Search ADS Türk , R. and Pfleger , H.S. 2008 Die Flechtenflora am Johannser Kogel im Lainzer Tiergarten und in den Steinhofgründen (Wien, Österreich) [The lichen flora at Johannser Kogel in Lainzer Tiergarten and at Steinhofgründe (Vienna, Austria)] . Verh. Zool. Bot. Ges. Österr. Österr. 145 , 83 – 95 . Wallander , E. 2008 Systematics of Fraxinus (Oleaceae) and evolution of dioecy . Plant Syst. Evol. 273 ( 1–2 ), 25 – 49 . Google Scholar CrossRef Search ADS Willner , W. 1996 Die Gipfeleschenwälder des Wienerwaldes [The hilltop ash forests of Wienerwald] . Verh. Zool. Bot. Ges. Österr. 133 , 133 – 184 . Zhao , Y.-J. , Hosoya , T. , Baral , H.-O. , Hosaka , K. and Kakishima , M. 2012 Hymenoscyphus pseudoalbidus, the correct name for Lambertella albida reported from Japan . Mycotaxon 122 , 25 – 41 . Google Scholar CrossRef Search ADS © Institute of Chartered Foresters, 2018. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Journal

Forestry: An International Journal Of Forest ResearchOxford University Press

Published: Oct 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off