Breeding Increases the Efficacy of Chondrostereum purpureum in the Sprout Control of Birch

We tested whether the pairing of selected isolates could be used to increase the efficiency of a decay fungus Chondrostereum purpureum (Pers. Ex Fr.) Pouzar to control hardwood sprouting in Finland. We paired C. purpureum strains efficient in sprout control or highly active in laccase production, and tested the efficacy of their progeny in spout control experiments. This procedure resulted in a strain with an efficacy superior to that of the parental strains. The mortality of birch (Betula pendula Roth. and B. pubescens Ehrh.) 1 cm in stump diameter was 78%, 56% and 9% for the best progeny, the best parental strain and the control, respectively. Mortality was only slightly higher for B. pendula than for B. pubescens but no significant differences were found between the number or maximum height of stump sprouts. Our results showed that cross breeding of this decay fungus is a good alternative in attempts to produce efficient biocontrol agents against hardwood sprouting.


Introduction
In Finland, sprouting of broad-leaved trees is a hindrance in spruce (Picea abies [L.] H. Karst.) and pine (Pinus sylvestris L.) regeneration areas, alongside roads and railways, under electric power lines and above gas pipe lines. In regeneration areas, broad-leaved species such as silver and downy birch (Betula pendula Roth. and B. pubescens Ehrh.), decrease the growth of more commercially valuable conifers, and therefore non-crop species are typically cleaned, preferably at an early stage when trees are about 1 m in height [1,2,3]. Next to roads and railways, broadleaved trees form a threat to the safety of traffic as they restrict visibility, cover traffic signs and tempt moose, and are therefore regularly removed. Electric power lines are kept open to ensure continuous electric transmission and gas pipe lines marked with visible signs are frequently cleared in order to avoid unintended excavations. Sprout control of broad-leaved trees costs more than 60 million euros annually in Finnish forest regeneration areas, along roads and railways, and at electric power and gas pipe lines (information gathered from UPM Forest Ltd., The Finnish Transport Agency, Fingrid Ltd. and Gasum Ltd.).
Herbicides were previously successfully used in sprout control [4,5], but the use of chemicals is no longer recommended in Finnish groundwater areas due to their harmful effects [6,7]. from Phases 2 and 3. All data sets and figures are available from the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.f5s6r

Ethics Statement
Permission for the field experiments of Phases 2-4 was granted by land owners, Fingrid Ltd., UPM-Kymmene Ltd. and the Finnish Forest Research Institute. Endangered or protected species were not used in this study. Fungal strains of C. purpureum have been deposited in the culture collection of the Finnish Forest Research Institute.
All C. purpureum isolates were grown on potato dextrose agar Petri plates (PDA: 24 g potato dextrose broth and 15 g agar with 1000 ml deionized water; Becton, Dickinson and Company), and a 6-mm diameter agar plug at the periphery of the grown mycelium was transferred to the middle of an ABTS plate using a sterilized Pasteur pipette and a scalpel. Two replicate plates    Tables 1 and 2, Phase 1). Laccase activities of the 24 progeny isolates were also investigated after 4 d on the ABTS plates as described above, and the best producer was named E+ 1 .

Pairings
First, the best two C. purpureum strains in terms of laccase activity (see Table 1) were paired ( Table 2, Phase 1). The isolates to be paired were cultured on PDA Petri plates, one isolate per plate, for ca. 3-4 weeks until they formed fruiting bodies. A fruiting body was cut from a plate with a sterilized scalpel and transferred onto a new PDA plate lid. One drop of sterilized water was pipetted just next to the fruiting body to activate spore release. One day later, when spores had started to germinate on a PDA Petri plate, 24 spores per isolate JÄ2 and 20 spores per isolate PI7 were successfully isolated from the plates with a modified and sterilized Pasteur pipette and transferred to new PDA plates, one spore per plate. After the occurrence of hyphae their morphology was investigated with a microscope to verify that single spore isolations were successful and that the hypha was homokaryotic (no visible clump connections). A single hyphal tip was further isolated from each plate to verify a homokaryotic state. These isolates were Breeding phase, C. purpureum strain and its spore number in each breeding phase are presented. The best progenies in terms of laccase activity (the strongest color reaction, Phase 1) or in terms of efficacy in sprout control in the field (Phases 2 and 3) three months after C. purpureum application are in bold (see Fig. 2). During Phase 2 the best parent strains based on field experiments were bred and in Phase 3 the best progenies from Phases 1 and 2 were bred. Note: the subscript of a progeny indicates the breeding phase. paired according to Table 2 (Phase 1), i.e., two spores from different isolates were placed on a single PDA Petri plate to allow them to form heterokaryotic mycelium. After one week, an interaction zone between the isolates developed between the inocula, from which hyphae were transferred to a new PDA Petri plate. Successful pairing was verified microscopically by confirming clump connections. A single heterokaryotic hyphal tip was further transferred to a new PDA plate for further use. The three most efficient natural strains tested by Vartiamäki et al. [13], isolates HY4, P4 and 3.11, were originally collected from birches (Betula pendula and B. pubescens) in Juupajoki in 2003, Vantaa in 2004 and Helsinki in 2001, respectively, and used for the second set of pairings. Altogether 10 spores per isolate HY4 and P4, and 9 spores per isolate 3.11 were successfully isolated from the plates and paired as described above (see Table 2, Phase 2).
Third, the best progenies from Phases 1 and 2 were paired according to Table 2 (Phase 3), as described above.

Inoculum for field experiments
The inoculum medium for the field experiments was prepared as follows: 24 g potato dextrose broth (Becton, Dickinson and Company) and 20 g Sipernat 22S (Evonik Degussa) per 1000 ml deionized water was autoclaved in an Agarmatik machine at 121°C for 15 min. Erlenmayer flasks, 250 ml in volume were also autoclaved at 121°C for 15 min. Altogether, 150 ml cooled autoclaved inoculum medium was added to the flasks and C. purpureum hyphae from one PDA cellophane plate (per isolate, Phase 2) was transferred to the flask with a sterilized scalpel. This inoculated medium was incubated in the dark at 18°C for 7-10 d on a rotator shaker (100 rpm). The inoculum was homogenized by Ultra Turrax apparatus for 1.5 min and diluted 1:10 with tap water before treatments in the field.
For Phases 3 and 4, the fungal inoculum was prepared as in Phase 2 except that the weight of hyphae added to the Erlenmayer bottles was equal (0.120 ± 0.024 g and 0.167 ± 0.018 g, mean ± SD, respectively) in each treatment (i.e., for each fungal isolate used in the experiment). All strains were cultivated for 10 d in a shaker at 20°C.

Experiments in the field
The efficacy of all progenies of Phases 2, 3 and 4 was investigated in field experiments. The first experiment (see Table 2, Phase 2) was established on 9 and 10 June 2009 under electric power lines in Porvoo, Hinthaara, southern Finland. Both days were cloudy with a temperature of ca. 15°C. However, 9 June was partly drizzly. The site included plenty of naturally growing birches (Betula pendula and B. pubescens) with a basal diameter (at ca. 15 cm above soil surface) of 2.2 ± 1.4 cm (mean ± SD). The efficacy of 15 different progenies and their parental strains HY4, 3.11 and P4 were tested in this field experiment (see Table 2). Furthermore, sample plots for controls (cutting only) and liquid controls (procedural controls: inoculum medium without C. purpureum) were also established. Altogether, 80 circular sample plots, four per isolate or control treatment, were established randomly on the site. Each sample plot included ca. 20 birch stumps. Thus, altogether 1583 birch stumps were investigated in this experiment (Betula bendula 49%, B. pubescens 49% and unrecognized birches 2%-the proportion of tree species was investigated three months after the treatments and therefore some were already dead). In the C. purpureum sample plots, fungal inoculum was spread on stumps immediately after cutting. Treatments in the liquid control sample plots were similar, but only inoculum medium without any mycelium was sprayed on the stumps. Viability of the fungal inoculum was confirmed before and after field applications by squirting inoculum to PDA Petri plates. All of the C. purpureum isolates were viable before and after the experiment.
Three and a half months later, after one growing season, in September 2009, the number of living sprouts per birch stump and basal diameters of stumps (mm) were measured. Based on the number of sprouts, the two best C. purpureum progenies were chosen for further investigation. The two best ones had the lowest number of stump sprouts per stump and the statistical difference was the highest (although usually not significant) when compared to the control stumps. These progenies were named V1 2 and V2 2 (see Table 2).
In the second field experiment, the efficacy of the progenies and their parental strains in Phase 3 (Table 2) were tested in the field. The efficacy of different C. purpureum strains was tested in three different regeneration areas of spruce in Turenki, southern Finland, including lots of naturally grown birches (Betula pendula and B. pubescens) 1.0 ± 0.4 cm (mean ± SD) in basal diameter (at ca. 15 cm above soil surface). At each site, the efficacy of 12 progenies and the parental strains from different phases, i.e., HY4, P4, 3.11 and V1 2 , V2 2 , E+ 1 , was tested. Furthermore, sample plots for controls (cutting only) and liquid controls (inoculum medium spread without C. purpureum) were established. One sample plot per treatment was randomly placed within each site. Each sample plot included ca. 30 birch stumps. Thus, 60 sample plots including 1796 birch stumps were included in the study (Betula pendula 46%, B. pubescens 53%, and unrecognized birches 1%). Sample plots were treated on 15 and 16 June 2010. 15 June was partly cloudy, sunny and rainy with 15°C whereas 16 June was sunny but windy with 17°C. All C. purpureum strains were viable before and after the experiment.
Three months later, in September 2010, the number of stump sprouts per birch stump and basal diameters of stumps (mm) were measured, and the three best C. purpureum progenies were chosen for further investigation. The three best strains had the lowest numbers of stump sprouts and statistical differences were the highest (although usually not significant) when compared to the control stumps.
In the third field experiment we tested the efficacy of the best C. purpureum strains from Phases 2 and 3 for three growing seasons. The best strains from Phase 3, R3 3 (progeny strain from the pairing between V1 2 and E+ 1 ), R5 3 (V1 2 × E+ 1 ) and R9 3 (V2 2 × E+ 1 ), together with strains V1 2 and V2 2 from Phase 2 were included in the experiment. Furthermore, HY4, the best parental strain originally collected from the field (based on the study by Vartiamäki et al. [13]), and a liquid control (inoculum medium without C. purpureum) were included. Eight regeneration areas of spruce with a high frequency of birches (Betula pendula and B. pubescens) were chosen for the study ( Table 3). Four of the sites were located in Lapinjärvi, and four in Turenki, both in southern Finland. At each site, circular sample plots including ten birch saplings per treatment were established. The order of the sample plots was randomized within each site. Altogether 560 birch saplings, 80 saplings per treatment, were included in the study (Betula pendula 12%, B. pubescens 75% and unrecognized birches 13%). The mean basal diameter of birches was 1.1 ± 0.4 cm (mean ± SD, ca. 15 cm above soil surface). The experiment was established on 10 to 13 May 2011. The weather was sunny with a temperature of 14-21°C.
The mortality of stumps, the number and maximum height of stump sprouts in living stumps and the occurrence of fruiting bodies were investigated one, two and three growing seasons (2011, 2012 and 2013) after the treatments. Furthermore, the basal area of stumps was measured (mm), and the number of other saplings (cut or uncut), and retention trees (i.e., mature trees left on sites, diameter at breast height 5 cm, m 3 ha -1 ) around an investigated stump were measured as these have an effect on sprouting [9]. The number of other stumps and saplings around an investigated sapling was measured within a circular subsample plot 0.5 m in radius whereas the diameters of retention trees (cm) at breast height were measured within a circular sample plot 10 m in radius. Tree volume was calculated using the models by Laasasenaho [29]. The occurrence of moose browsing was recorded per investigated stump to take this into account as it has an effect on the height of stump sprouts.

Statistical analyses
Generalized linear mixed models (GLMMs) were used to investigate differences between the treatments. In Phases 2 and 3 the effects of different treatments (control, liquid control and different C. purpureum strains) on the number of stump sprouts were investigated using a Poisson model with log link function in package lme4 in the statistical program R [30,31]. Thus, the number of stump sprouts per stump three months after the treatments was the response variable and treatments (as a factor) and the basal diameter of a stump were explanatory variables in the models. In Phase 2, the sample plot was used as a random factor to take into account pseudoreplication, i.e., the fact that 20 stumps within each sample plot may be-due to environmental conditions-more similar than randomly chosen saplings. In Phase 3, site (logging unit) and sample plot were treated as nested random factors to take into account the fact that several stumps within the same site and sample plot were investigated. Both in Phase 2 and 3, the control treatment (cutting only) and the liquid control (inoculum medium without a fungus) were compared to the other treatments. The best fungal strains were chosen for further investigation based on the biggest differences in the number of stump sprouts per stump between the control treatments and the fungal treatments (lower in the fungal treatment) and the lowest p-values.
The GLMMs were also used to analyze data relating to the final field experiment, which lasted for three growing seasons from 2011 to 2013 (Phase 4). We estimated the models separately for each year. The package lme4 in R was used to investigate the effects of controls (inoculum without C. purpureum spread on cut stumps) and different C. purpureum strains on the investigated stumps. Effects on the mortality of investigated stumps and the probability of occurrence of fruiting bodies on stumps were analyzed using a binomial model with logit link function [30,31]. Effects on the number of stump sprouts per stump were investigated using a Poisson model with log link function. The effect of different treatments on the maximum height of stump sprouts per stump was investigated using function lme in the nlme package in R [31,32] assuming a normal distribution. All stumps were included in the mortality models whereas only living stumps were included when differences between the number and maximum height of stump sprouts were investigated. Explanatory variables in the models were 1) treatment (as a factor with seven levels; however, the liquid control was excluded from the fruiting body model), 2) the basal diameter of a stump (mm), 3) the number of other stumps and saplings around an investigated stump, and 4) the total volume of retention trees around an investigated sapling (m 3 ha -1 ). For the stump sprout height model 5) browsing (as a factor: 0 = no browsing, 1 = browsing) was also taken into account as it affects the maximum height of stump sprouts. First, differences between the liquid control and the C. purpureum treatments were investigated. Second, differences between the best fungus treatment (the greatest difference from the control) and other C. purpureum treatments were investigated. Correlations between the continuous explanatory variables were low (the strongest Pearson correlation was between the basal diameter of a stump and the total volume of retention trees, r = 0.14), and therefore all of the explanatory variables were included in the models. Forest site and sample plot were used as nested random factors in the models. Differences in mortality, sprout number and maximum height between birch species (Betula pendula and B. pubescens) were investigated as above. However, tree species (as a factor with two levels: B. pendula and B. pubescens) was included as an explanatory variable to the models instead of treatment. The data collected in 2013 including sample plots treated with C. purpureum strains were used in these investigations. Unrecognized dead stumps were removed from the mortality model.

Phase 1: Laccase activity tests and pairings
Altogether 69 C. purpureum strains were collected from birch stumps, and investigated on ABTS Petri plates revealing laccase activity. Based on the color reaction on these plates, ca. 32% of the investigated strains belonged to class 3 (strong color reaction, i.e., high laccase activity, Table 1, see [13]). The best C. purpureum strains in this enzymatic test were JÄ2 and PI7 (the strongest color reaction on plates), which were further paired with each other ( Table 2, Phase 1). The laccase activity of the progenies of JÄ2 × PI7 was further tested on the ABTS plates. Altogether 67% of the strains belonged to class 3 (strong color reaction, i.e., high laccase activity, Table 2). The best of the progenies in this enzymatic test (showing the strongest color reaction on plates) was R7 1 , which was renamed as E+ 1 .

Phase 2: Pairing of the most efficient parental strains available and first generation progeny testing in the field
The most efficient C. purpureum parental strains were paired in Phase 2 ( Table 2). Based on the subsequent field experiment, progenies R12 2 and R15 2 (later on designated as V1 2 and V2 2 , respectively) had the lowest number of stump sprouts and p-values compared to controls. These strains were chosen for further pairings and investigation (see Table 2, Fig. 2a) although the final effect on sprout control can be expected only after several growing seasons [33]. Differences in the number of stump sprouts between the parental strains HY4, P4 and 3.11, and the control treatments were smaller (0.127 p 0.892). Furthermore, the models revealed that the number of stump sprouts increased with an increase in basal diameter of a stump (p < 0.001).

Phase 3: Pairing of the best progenies from Phases 1 and 2 and second generation progeny testing in the field
In Phase 3, strain E+ 1 from Phase 1 was further paired with V1 2 and V2 2 from Phase 2 ( Table 2). The efficacy of the progenies in sprout control was further tested in the field. The best three progenies in this field experiment were R3 3 , R5 3 and R9 3 based on the lowest numbers of stump sprouts on investigated stumps and p-values compared to the controls (see Table 2, Fig. 2b). Differences in the number of stump sprouts between parental strains HY4 and 3.11 and the controls were smaller (p 0.590), but parental strain P4 was quite efficient in this experiment (p 0.110). Furthermore, the number of stump sprouts increased with an increase in basal diameter of a stump (p < 0.001).
Phase 4: Long-lasting field experiment-final testing of the efficacy of the best progenies from Phases 2 and 3

Mortality
In the final field experiment (Phase 4), lasting three growing seasons (2011-2013), we found that one of the tested C. purpureum progenies, R5 3 , was considerably more efficient than other strains in the sprout control of birch. Mortality was clearly higher in stumps treated with R5 3 especially two and three growing seasons after the treatments in 2012 and 2013 (Table 4). This was the only strain that differed statistically significantly from the parental strain HY4 three growing seasons (in 2013) after the treatments. After three growing seasons, the mortality in R5 3 treated stumps was 78% whereas that for the best parental strain, HY4, was 56% and for the control 9% (cutting with inoculum medium without C. purpureum, see Table 4, Figs. 3a and 4a). The efficacy in sprout control was 28% lower for the parental strain HY4 than for R5 3 . Our results also revealed that another progeny isolate (R3 3 ) was highly efficient in sprout control as 60% of the treated birch stumps were dead after three growing seasons. However, the efficacy for R3 3 was 23% lower than for R5 3 . Furthermore, the mortality of stumps was higher in all the C. purpureum treatments than in the liquid control (0.001 p 0.099).

Sprout numbers
Our results showed that the number of stump sprouts (in the living birches) was lower for treatment R5 3 than for treatments V1 2 , V2 2 , and the liquid control in 2011 (see Table 5). After three growing seasons (in 2013), the number of sprouts in living stumps was still the lowest in stumps treated with R5 3 but the differences were not significant any more (see Table 5, Fig. 3b). The number of stump sprouts in living stumps was lower in all the C. purpureum treatments compared to the liquid control (0.001 p 0.085). e., cutting only; LC = liquid control, i.e., cut stumps were spread with the inoculum medium without C. purpureum; HY4, P4 and 3.11 = the best parental strains of C. purpureum; E+ 1 , R1 2 -R11 2 , R13 2 -R14 2 , V1 2 -V2 2 , and R1 3 -R12 3 = progenies of C. purpureum strains, see

Sprout height
Treatment R5 3 was not better than the other C. purpureum treatments in terms of the maximum height of stump sprouts (see Table 6, Fig. 3c). In fact, after the third growing season (in 2013), sprout height was indicatively lower in stumps treated with the progeny R3 3 than with R5 3 . However, the maximum height of stump sprouts in living stumps was lower in all C. purpureum treatments than in the liquid control (0.005 p 0.096).

Fruiting schedule
In 2011, stumps treated with treatment R5 3 had higher occurrence of fruiting bodies (56%) than stumps treated with HY4 (38%), V1 1 (39%), V2 1 (41%), R3 3 (46%) and R9 3 (48%). However, only some statistically indicative differences between the strains were found (see Table 7). In 2012, positive coefficients for the other C. purpureum treatments indicated that the occurrence of fruiting bodies was lower in the R5 3 treatment. In 2013, no differences in the occurrence of fruiting bodies were found. The occurrence of fruiting bodies on stumps treated with C. purpureum decreased with time, ca. 45% of stumps had fruiting bodies in 2011, whereas later on in 2012 and 2013, 39% and 4% of stumps had fruiting bodies, respectively. Only two stumps in the control treatment had fruiting bodies in 2012 (see Fig. 4d).   Table 2). a Liquid control i.e., cut stumps were spread with inoculum medium without C. purpureum. b The best parent stain based on an earlier study [13].

Effects of stump sizes and environmental factors
One and two growing seasons after the treatments, an increase in stump basal diameter increased stump mortality (see Table 4). Furthermore, an increase in the volume of retention trees increased mortality during the study. Three growing seasons after the treatments, an increase in the basal diameter of stumps increased the number and maximum height of stump sprouts in living stumps, whereas an increase in the number of surrounding saplings and the volume of retention trees decreased the maximum height (see Tables 5 and 6). Browsing decreased the maximum height in 2011. The probability of occurrence of fruiting bodies increased with an increase in stump diameter (see Table 7).

Differences between Betula pendula and B. pubescens
Our results revealed that only slight differences were found between the two birch species. Mortality of B. pendula (58% of investigated stumps) was indicatively higher than for B. pubescens (48%, p = 0.097). However, no differences were found when the number and maximum height of stump sprouts were compared (p = 0.574 and p = 0.707, respectively).

Discussion
Our results showed that breeding can be used to increase the efficacy of C. purpureum as a biocontrol agent. After the whole breeding process, C. purpureum strain R5 3 was statistically significantly better than the best original parental strain HY4 and subsequent parental strain V1 2 created during the process. After three growing seasons, the mortality of birch stumps ca. 1 cm in diameter treated with R5 3 was 78% whereas those with parental strains HY4 and V1 2 were only 56% and 55%, respectively. Thus, it seems that our breeding process was successful, supporting our initial hypothesis. The efficacy of HY4 has been investigated in an earlier study with birch stumps ca. 3-4 cm in diameter resulting in more than 90% mortality after two growing seasons [13]. The lower mortalities observed in the present study was expected as it is known that mortality is usually lower for smaller stumps [9]. Therefore, the mortality caused by the best strain from our breeding program (78%) in small stumps with a diameter of about 1 cm can be considered a promising result because in regeneration areas of spruce (Picea abies) and pine (Pinus sylvestris), non-crop species of this size are usually removed. Thus, high biocontrol efficacy with C. purpureum can provide one option to lower costs for sprout control via decreasing the number of repeated cuttings (see [34]). Sprout control also allows better growth conditions for more valuable conifers because competition with broad-leaved trees can reduce their growth and even induce mortality [3,34,35].
It seems that C. purpureum is especially efficient in the sprout control of birch, as for yellow birch (Betula alleghaniensis Britt.) at least 96% of inoculated stumps died within one year [8], similarly as with silver and downy birch after three growing seasons [33], whereas in the study of Roy et al. [14], the mortality of paper birch (Betula papyrifera Marsh.) was ca. 75% after four growing seasons. Also, fruiting bodies have been especially abundant on inoculated yellow (87% of the treated stumps) and paper birch stumps (100%) [8,36].
In our study, after the first growing season in 2011, the occurrence of fruiting bodies was highest (56%) with strain R5 3 indicating that this fungus was able to penetrate wood faster than the parental strain HY4 and the other C. purpureum strains. Investigations with red alder (Alnus rubra Bong.) have revealed a close relationship between the time of mortality and the occurrence of fruiting bodies: peaks in fruiting body formation occur one year before, the same year or one year after mortality, and those trees that died slowly supported fruiting bodies for a longer time [36]. Although significant differences were not found in the number of stump sprouts per living stump between the progeny strain R5 3 and the parental strain HY4, the treatment with R5 3 resulted in a lower number of sprouts in living stumps (see Figs. 3b and 4b). Furthermore, results relating to the progeny strain R5 3 indicate that the number of sprouts per stump was lower than   Table 2). a Liquid control i.e., cut stumps were spread with inoculum medium without C. purpureum. b The best parent stain based on an earlier study [13].
c The best progenies from Phase 2.
d The best progenies from Phase 3.
in the control in every investigated year. However, in terms of the maximum height of sprouts (in living stumps), hardly any differences between the treatments were found (see also [13]). Our results revealed that an increasing volume of trees around an investigated stump increased stump mortality. This is in accordance with earlier findings as shading of neighboring trees has been shown to have a profound effect on shoot growth due to a greater proportion of shoot buds dying under heavy shading [37,38]. Furthermore, our results showed that an increase in the number of surrounding saplings decreased the number and maximum height of stump sprouts (see also [39]). On the other hand, high mortality with C. purpureum inoculated stumps (especially with R5 3 ) may provide more growing space for those stumps that are still living. This may be the reason why significant differences between the treatments were not observed in stump numbers and height.
All C. purpureum strains investigated were originally collected from birch stumps in southern and middle Finland (see Methods section and Table 1). However, investigations from Finland, Canada and New Zealand showed that C. purpureum strains are genetically diverse, and are not associated with a specific host species or ecological region [11,17,18]. Thus, it is possible that the most efficient strain, R5 3 , is efficient also against the sprouting of other tree species and in other geographical areas.   Table 2). a Liquid control i.e., cut stumps were spread with inoculum medium without C. purpureum.
b The best parent stain based on an earlier study [13]. c The best progenies from Phase 2. Furthermore, earlier studies have indicated that a single genotype can be used as inoculum without the risk of introducing genes that differ significantly from local populations [11,17]. The best strain R5 3 was developed by traditional breeding (via consequent pairings of selected mycelia), i.e., similarly like any other C. purpureum strain in the nature. In that sense this strain does not differ from its natural counterparts. Moreover, as the fungus has no asexual stage, the genetic combination of a single exceptionally efficient biocontrol fungus breaks up in meiosis before the spores are released. Thus, the same C. purpureum genotype applied on cut stumps cannot spread further in forest ecosystems, and thus there is no risk for explosion of the biocontrol strain.
Classical breeding experiments with biocontrolling fungi are scarce. However, in addition to ours, Wall et al. [10] compared the efficacy of C. purpureum monokaryons and their dikaryon progenies in causing wood tissue mortality, but could not find any differences. However, their experiment was not tested under field conditions. Another biocontrol fungus, Phlebiopsis gigantea, was bred to improve the efficacy against Heterobasidion root rot in forests [27]. Results of that study indicated that progeny strains had better properties against H. parviporum than the parental strains. Furthermore, classical pairings were successfully used to improve temperature tolerance of fermenting yeasts using a backcross approach [25]. Thus, these studies (including the present one) show that improving by breeding may help in developing more efficient fungal strains for different purposes.  was not estimated because the occurrence of fruiting bodies was too low (4%). All stumps except those in the liquid control have been included in the models. Statistically significant p-values (p < 0.05) for the model coefficients are in bold and indicative results have been underlined (0.05 p 0.10).
The sign of a coefficient indicates whether an explanatory variable has an increasing (+) or decreasing (-) effect on the occurrence of fruiting bodies. See Fig. 4d. Note: the subscript of a C. purpureum progeny relates to the breeding phase (see Table 2). a The best parent stain based on an earlier study [13]. b The best progenies from Phase 2.
c The best progenies from Phase 3. However, breeding studies with field experiments may be time consuming as seen in our study. Time for follow-up is usually limited, which forces us to choose the best strains for subsequent steps after a short time period. For example, in our study, we had to choose progeny isolates for the next pairings after the first growing season (both in 2009 and 2010) based only on the number of stump sprouts in different C. purpureum treatments (a low mortality did not allow for mortality analyses). Moreover, as seen also in the present study, the efficacy of C. purpureum is not tightly associated with the parents, because each parent can produce relatively virulent as well as avirulent strains [10]. This is not surprising as Wall et al. [10] have stated that the inheritance of efficacy is probably multifactorial and complex and may be subject to different modifying factors under field and laboratory conditions. Thus, final efficacy should always be investigated in field experiments. This view was supported by our results as the best strain in terms of laccase activity in the laboratory, E+ 1 , was not efficient in sprout control (see Fig. 2b) although some of its progenies (R5 3 and R3 3 ) in the end showed the best ability to prevent sprouting of broad-leaved trees in the field.

Conclusions
We showed that traditional breeding can increase the efficacy of C. purpureum in the sprout control of birch. Mortality of the treated birch stumps was highest when the progeny strain R5 3 was used to control sprouting. This resulted in 78% mortality in birch stumps 1 cm in diameter. The efficacy of strain R5 3 was significantly higher than that of the investigated parental strain HY4 (56%). The effect of sprout control was especially pronounced when the volume of trees and the number of surrounding saplings was high. However, breeding experiments that aim at increasing the efficacy of sprout control are time consuming because efficacy has to be investigated in long-lasting field experiments.