Effects of seasonality and climate on the sporulation of single Heterobasidion annosum sensu stricto and H. parviporum fruiting bodies in Norway spruce stands of Latvia

Abstract

The root and butt rot fungi Heterobasidion annosum sensu stricto (s.s.) and H. parviporum are reported as some of the most widespread and destructive pathogens of conifer forests in Europe. By using passive spore traps, in this study conducted over 3 years in Latvian Norway spruce mixed forest stands, we explored the temporal dynamics of sporulation of H. annosum s.s. and H. parviporum overall and at the individual fruiting body scale and we assessed the effects of both season and climatic conditions on sporulation rate. Data suggested an overwhelming importance of the Heterobasidion genotype in determining spore loads. Indeed, we detected a single highly sporulating fruiting body overwhelmingly contributing to the overall airspora regardless of the time period. Peaks in spore release were observed in late summer and early fall for most fruiting bodies. Results clearly proved a prevalent role played by some climatic variables, including average temperatures and the relative humidity 2–4 weeks before spore trapping. In contrast, seasonality explained less variability in sporulation. We fitted, tested, and validated models able to predict the sporulation of Heterobasidion spp. based either on climatic variables, or seasonality. However, for risk prediction, we recommend to use models based on short-term climatic variables rather than seasonality.

Introduction

Aerobiology has proved instrumental for the management of fungal plant pathogens infecting their hosts through airborne propagules. Aerobiological methods combining spore trapping and detection of target plant pathogens allowed clarifying seasonal patterns of spore availability, the role of climatic and meteorological conditions on spore release and deposition, and the potential dispersal range of propagules (Gonthier et al., 2001; Kadowaki et al., 2010; Grosdidier et al., 2018; Brūna et al., 2021; Lione et al., 2021; Lione et al., 2022). This enhances our ability to predict sporal infections and design control strategies. Aerobiological studies also unraveled key aspects of biological invasions, such as the association between alien pathogens and forest habitats and the dynamics of species replacement occurring between invasive and native fungal species infecting trees (Gonthier et al., 2012a; Garbelotto et al., 2022).

A relevant model system for which aerobiology was largely explored because of its practical significance are the fungal forest pathogens Heterobasidion annosum sensu lato (s.l.). This species complex includes five allopatrically and sympatrically differentiated species reported as some of the most destructive pathogens of conifer forests worldwide (Garbelotto and Gonthier, 2013). Direct losses caused by the three Heterobasidion species native to Europe, namely Heterobasidion abietinum Niemelä & Korhonen, Heterobasidion annosum sensu stricto (s.s.) (Fr.) Bref. and Heterobasidion parviporum Niemelä & Korhonen, were estimated at about 800 Million Euros per year (Woodward et al., 1998). A study conducted in the Alps reports that the percentage of direct financial losses caused by Heterobasidion spp. is comparable to the average level of disease incidence (Gonthier et al., 2012b). The ongoing biological invasion of the North American species Heterobasidion irregulare Garbel. & Otrosina, currently reported in Italy, is predicted to further increase such economic losses (Garbelotto and Gonthier, 2013; Gonthier et al., 2014; EPPO, 2015). Based on this, H. irregulare is qualified as a pathogen recommended for regulation by the European and Mediterranean Plant Protection Organisation (EPPO). The different Heterobasidion species display distinct host preference. The two most widespread species in Europe, namely H. annosum s.s. and H. parviporum, commonly attack pines (Pinus spp.) and Norway spruce [Picea abies (L.) H. Karst.], respectively, although Norway spruce may be infected by both (Garbelotto and Gonthier, 2013). While there are Norway spruce forests where the rate of trees infected by H. parviporum vs. H. annosum s.s. achieved 56.3% vs. 43.7%, respectively (Vasiliauskas and Stenlid, 1998), the proportion between the two species in most cases is approximately 9 vs. 1 (Korhonen and Piri, 2003; Müller et al., 2007; Gaitnieks et al., 2021, 2022), indicating a clear prevalence of H. parviporum in these stands.

Heterobasidion annosum sensu lato (s.l.) infects trees by means of airborne meiospores either through fresh wounds on standing trees or via freshly cut stump surfaces (Rishbeth, 1951; Garbelotto and Gonthier, 2013). As soon as the mycelium colonizes the root systems, it can spread to neighboring trees through root contacts or grafts, thereby causing root rot or heart rot depending on the tree species (Garbelotto and Gonthier, 2013). Fruiting bodies of H. annosum s.l. are typically perennial and they are commonly found at the base of infected stumps or trees (Garbelotto and Gonthier, 2013). While the presence of sporulating fruiting bodies is a prerequisite for the availability of infectious airborne inoculum, airspora availability and abundance vary depending on the season and climate. By using passive spore traps without an a priori knowledge on presence and distribution of fruiting bodies, peaks in spore deposition were reported during the summer in Fennoscandia (Kallio, 1970) and the autumn in the Alps (Gonthier et al., 2005), while minima were reported in both cases during winter. In the same areas, the highest frequency of stump infection was recorded in summer in Fennoscandia (Brandtberg et al., 1996) and in autumn in the Alps (Gonthier, 2019), thus confirming that aerobiological data are a good proxy of the Heterobasidion infection risk. In the United Kingdom, infections occur during most of the year (Rishbeth, 1957; Meredith, 1959). Other than seasonality, some studies also investigated the underlying climatic variables possibly explaining the temporal spore deposition pattern of Heterobasidion spp. For instance, positive correlations between spore deposition and temperatures were recorded in both Northern Europe and the Alps. Instead, the relative air humidity (RH) was less explanatory (Kallio, 1970; Gonthier et al., 2005), although some correlations were found between spore deposition and RH in the week preceding spore trapping in Finland (Kallio, 1970). All lines of evidence indicate that the temporal dynamics of spore release and deposition vary depending on the geographic area, possibly because of the influence of associated climatic factors. Nonetheless, it should be noted that all studies focusing on Heterobasidion spp. spore deposition conducted so far were unable to disentangle the role of seasonality from that of climatic factors. Indeed, seasonality and climate are correlated, yet this correlation not only varies among geographic areas, but it can vary in the same location over time, due not only to an intrinsic interannual climatic variability, but also to climate change (Kutta and Hubbart, 2016). Hence, experimental approaches able to test and quantify separately the effect size of seasonality and climate (see for instance Lione et al., 2021, 2022) would likely improve the modeling of Heterobasidion spp. sporulation. Modeling could also be enhanced through the collection of data from geographic areas for which information about temporal patterns of sporulation by Heterobasidion is lacking, as for central Europe and the Baltic countries.

In addition to the season and climatic factors, Heterobasidion spp. sporulation might also depend on the fungal genotype, and on the shape and size of the individual fruiting bodies, in particular of their hymenophore (i.e. the actively sporulating surface). As a new pore layer develops each year, the age of the fruiting body may also account for its ability to sporulate. Unfortunately, previous studies aimed at exploring and modeling Heterobasidion spp. temporal dynamics of spore deposition were not designed to appraise the contribution of individual fruiting bodies to the overall airspora. Determining the sporulation from individual fruiting bodies may allow to assess: (i) whether the contribution to the overall airspora is better explained by a few highly sporulating genotypes, or by all fruiting bodies sporulating in a similar manner; and (ii) whether spore release occurs from all fruiting bodies at the same time, or not.

By working in mixed Norway spruce forest stands of Latvia, here we explored the temporal dynamics of sporulation of Heterobasidion spp. overall and at the individual fruiting body scale and we modeled the role of both season and climatic conditions on sporulation. In addition, we examined whether sporulation rates (SRs) might be affected by the fungal genotype.

Methods

Study sites, Heterobasidion fruiting bodies and spore trapping

Heterobasidion spp. sporulation rate (SR) was assessed from January of 2012 to November of 2014 in three mixed stands of Norway spruce along a 4.5 km transect in the central part of Latvia (Table 1). The dominant tree species in all stands was Norway spruce (with an abundance of at least 60%), but scots pine (Pinus sylvestris L.), birch (Betula sp.), and black alder [Alnus glutinosa (L.) Gaertn.] were also present. All stands were on drained soils. The age of stands was 88–108 years (Table 1). Eight Heterobasidion fruiting bodies were included in the experiment, 2–3 per each forest stand (Table 1 and Table 2). Five fruiting bodies were monitored from the beginning of the experiment (04.01.12), while other three newly developed fruiting bodies were added during the course of the experiment (Table 2).

Table 1

Characteristics of the investigated forest stands.

Stand	Coordinates (latitude, longitude, °)	Stand composition (%)	Stand age (years)	Forest type*	Heterobasidion fruiting bodies surveyed (acronym)
A	56.82241; 24.08076	Spruce 60%, pine 20%, birch 20%	108	Mercurialiosa mel.	A1, A2, A3
BC	56.82116; 24.08502	Spruce 60%, black alder 20%, birch 20%	88	Oxalidosa turf. Mel.	B, C1, C2
D	56.81250; 24.15482	Spruce 80%, pine 20%	97	Myrtillosa mel.	D1, D2

^*According to Latvian forest classification system (Bušs, 1997).

Table 2

Characteristics of the investigated Heterobasidion spp. fruiting bodies.

Fruiting body	Localization	Diameter of log/stump (cm)	Length/height of log or stump (cm)	Hymenophore surface (cm2)	Heterobasidion species	Period of spore trapping, (dd.mm.yy)
A1	Stem part of uplifted stump	56	64	805	H. parviporum	04.01.12.-25.11.14.
A2	Log	42	159	250	H. parviporum	04.01.12.-25.11.14.
A3	Cut surface of the basal edge of log	32	249	196	H. parviporum	12.04.13.-17.04.14.
B	Root collar on the uprooted P. abies cut 2.4 m from stem base	39	2435	594	H. parviporum	04.01.12.-25.11.14.
C1	Stem part of uplifted stump	49	110	113	H. parviporum	31.03.13.-25.11.14.
C2	Root collar of uplifted stump	44	52	188	H. parviporum	31.03.13.-20.06.13.
D1	Lateral surface of the standing stump	65	36	31	H. annosum s.s.	04.01.12.-28.03.13.
D2	Under roots of partly uplifted stump	59	39	206	H. parviporum	04.01.12.-25.11.14.

The hymenophore, i.e. sporulating surface, of each fruiting body was measured by drawing its surface on a transparent plastic sheet at the beginning of the survey. A small portion of fruiting body (approximately 0.5 × 0.5 × 0.5 cm) including the context was excised by using a sterile scalpel, placed in a 1.5 ml Eppendorf tube (Sial) and transferred to the laboratory for isolation. In the laboratory, a fragment was taken from the context and transferred to a 9 cm-diameter Petri dish (VWR) filled with malt-extract agar (15 g malt extract and 12 g agar per liter of sterile water). Isolates were paired at least twice in all possible combinations according to the methods previously described (Stenlid and Karlsson, 1991) in the framework of somatic incompatibility tests to determine whether they belonged to the same or different genotypes. Petri dishes were incubated at room temperature for 4–5 weeks and examined periodically.

To investigate sporulation, spore traps in form of 9-cm Petri dishes filled with malt-extract agar were exposed under each fruiting body, as close as possible to the hymenophore (1–4 cm depending on the shape of the fruiting body). Spore traps were exposed between 8.00 and 9.00 a.m. for 7 minutes. In a few instances, the time of exposure was approximately reduced to 6 minutes or increased to 8 minutes, to cope with operational constraints. In 2012 and 2013, spore trapping was performed at least once a week when the air temperature was above 0°C. Based on the outcomes of these 2 years of surveys (see results), in 2014 spore trapping was concentrated at the beginning of sporulation (March/April), at the peak of sporulation (August/September) and at the end of the sporulation period (October/November).

Germinated spores in the Petri dishes were counted using a Leica DM4000B microscope in 30 fields at 50 × magnification. The number of spores counted was converted to the number of spores released and deposited per 1 m² in an hour (i.e. spores∙m⁻²∙h⁻¹), hereafter called sporulation rate (SR).

Identification of fruiting bodies at the species level was accomplished as follows. One single spore isolate per each fruiting body was obtained and paired with the homokaryotic testers 05017/4 (H. annosum s.s.) and 91203/4 (H. parviporum) according to the methods described by Korhonen (1978).

At the end of the experiments, a sample consisting of three fruiting bodies (i.e. A1, A2, and B) was collected and examined in the laboratory for age determination. Each fruiting body was longitudinally sectioned by using a sharp knife and the number of pore layers was counted.

Collection and preprocessing of temporal and climatic data

Temporal and climatic data were collected and pre-processed following the approach described in Lione et al. (2022) and modified as follows. In brief, the date of each spore sampling was recorded along with the corresponding month (m), coded from 1 (January) to 12 (December) and used as temporal variable (i.e. seasonality). Weather data provided from the State limited liability company ‘Latvian Environment, Geology and Meteorology Centre’ were gathered from the meteorological station of the University of Latvia (latitude—56.950784°, longitude—24.116177°, distance—15 km), the closest to the study sites. Weather data included the average air temperature (t in °C), the relative humidity of the air (rh, in %) and the amount of cumulated rainfall precipitation (p in mm) recorded on a daily basis. The 30 days preceding the sampling were labeled with an index i, and for each climatic variable v (i.e. t, rh and p) and index j = {1,7,10,14,30} the corresponding climatic variables |${\mathrm{V}}_{\mathrm{j}}=\frac{\sum_{\mathrm{i}=1}^{\mathrm{i}=\mathrm{j}}{\mathrm{v}}_{\mathrm{i}}}{\mathrm{j}}$| (i.e. T_j, RH_j, and P_j) were calculated. For the sampling day, V was set at |${\mathrm{V}}_0={\mathrm{v}}_0$| with i = j = 0. For each fruiting body, the SR was transformed by calculating the SRN, namely the SR normalized within the range [0,1] through equation 1 (Lantz, 2019):

$$ \begin{equation} \mathrm{SRN}=\frac{\mathrm{SR}-{\mathrm{N}}_{\mathrm{SR}}}{{\mathrm{M}}_{\mathrm{SR}}-{\mathrm{N}}_{\mathrm{SR}}} \end{equation} $$

(1)

with

SRN: sporulation rate normalized (adimensional).

SR: sporulation rate (spores∙m⁻²∙h⁻¹).

N_SR: minimum sporulation rate (spores∙m⁻²∙h⁻¹).

M_SR: maximum sporulation rate (spores∙m⁻²∙h⁻¹).

Statistical analysis and modeling

Overall and per each fruiting body, the whole and monthly averages of SR were calculated as the averages of SR from spore trapping performed during the whole experiment and during every month, respectively. The corresponding minima and maxima were calculated as well.

Data of fruiting bodies A1, A3, C1, C2, D1, and D2 were randomly selected as training set for modeling, while data of fruiting body A2 were used to conduct the external validation (Lantz, 2019; Lione et al., 2021). The effects of season and climatic variables on Heterobasidion spp. sporulation were assessed on the training set by fitting random forest models. Random forest models were built from unbiased binary recursive partitioning tree models based on conditional inference, bootstrap, and bagging (Hothorn et al., 2006; Strobl et al., 2009). The random forest algorithm (Hothorn and Zeileis, 2015) was set as described in Kļaviņa et al. (2023). We included SRN as response variable of random forests, while season (m) and climatic variables (T_j, RH_j, and P_j) were used as input variables. For each of the latter, the variable importance (VI) was calculated as reported in Strobl et al. (2008) to compare input variables with respect to their impact in predicting the response, and to select the best predictors to model the SR of Heterobasidion spp. Three separate runs of the random forest models were conducted to assess the consistency of VI values, each one based on a different seed value (Carsey and Harden, 2014; Lantz, 2019).

Tree models (Hothorn et al., 2006; Strobl et al., 2009; Hothorn and Zeileis, 2015) were fitted on the training data to predict SRN based on: (i) the season variable m, and (ii) the climatic variables T_j, RH_j, and P_j, whose j was selected based on the highest VI value resulting from the previous random forests runs (Strobl et al., 2008). Sporulation rate scores (SRS) were obtained from the clustering of SRN resulting from the tree models. The score SRS is expressed by an integer number ranging from 1 to the number of terminal nodes (N) of the associated tree model, ranking increasing levels of Heterobasidion spp. sporulation (i.e. SRS = 1 for the lowest and SRS=N for the highest sporulation levels). Each SRS was associated to the corresponding m levels and T_j, RH_j, and P_j ranges, whose thresholds were obtained from the significant splits displayed by the tree model graph. Hence, tree models were used to assess the months and climatic conditions associated with significantly different levels of Heterobasidion spp. sporulation (Crawley, 2013). Tree models’ adequacy was assessed by calculating the root mean squared error of prediction (RMSEP) and by testing the significance of the Pearson’s R linear correlation coefficient between the SRN observed and predicted for both training data (i.e. internal validation) and validation data (i.e. external validation; Lantz, 2019). The model was considered as successfully validated if R was positive and significant for both internal and external validation (Lantz, 2019).

A Kendall’s rank correlation test (Crawley, 2013) was conducted between the average SR of each fruiting body and its corresponding hymenophore surface.

Statistical analyses and modeling were conducted with R version 3.4.0 (R Core Team, 2020). The significance threshold was set to 0.05 for all tests (Crawley, 2013). Average values were calculated along with the associated 95% bias-corrected and accelerated bootstrap confidence intervals (CI_95%) based on 10⁴ re-samplings (DiCiccio and Efron, 1996). The following R packages were used for computation: bootstrap (Efron and Tibshirani, 1994), DescTools (Signorell et al., 2016), FSA (Ogle et al., 2023), Metrics (Hamner and Frasco, 2022), partykit (Hothorn and Zeileis, 2015), and strucchange (Zeileis et al., 2002). The R code used to conduct the statistical analyses and modeling is reported in Supplementary material 1.

Results

Heterobasidion spp. monthly sporulation rate overall and per each fruiting body

Based on the outcomes of the somatic incompatibility tests, all fruiting bodies represented different genotypes. The average, maximum, and minimum values attained by each fruiting body and overall during the whole length of the spore trapping experiment is reported in Table 3.

Table 3

Average, maximum, and minimum values of sporulation rate (SR, spores∙m⁻²∙h⁻¹) attained by each fruiting body of Heterobasidion spp. and overall during the whole length of the spore trapping experiment. The absolute minimum is reported within brackets, while the other minimum value refers to the lowest sporulation level detected when sporulation actually occurred. Additional information about the fruiting bodies characteristics are reported in Table 2.

Fruiting body	Average SR (spores∙m−2∙h−1)	Minimum SR (spores∙m−2∙h−1)	Maximum SR (spores∙m−2∙h−1)
A1	4.873∙107	(0) 6.580∙103	3.946∙108
A2	3.753∙107	(0) 6.580∙103	5.038∙108
A3	4.697∙105	(0) 6.580∙103	9.633∙106
B	2.034∙108	(0) 5.982∙103	2.750∙109
C1	3.067∙107	(0) 6.580∙103	4.427∙108
C2	6.723∙106	(0) 1.645∙105	2.145∙107
D1	7.502∙106	(0) 6.580∙103	1.221∙108
D2	4.602∙107	(0) 5.982∙103	3.564∙108
Overall	6.417∙107	(0) 5.982∙103	2.750∙109

Considering the overall temporal pattern of sporulation, average SR peaked in August (2.285∙10⁸ spores∙m⁻²∙h⁻¹), although showing relatively high values also in September (2.042∙10⁸ spores∙m⁻²∙h⁻¹) and, though to a lesser extent, October (1.152∙10⁸ spores∙m⁻²∙h⁻¹) and July (1.099∙10⁸ spores∙m⁻²∙h⁻¹; Fig. 1a). The values of SR were lower in the other months, attaining averages ranging from 2.413∙10³ spores∙m⁻²∙h⁻¹ (in February) to 5.781∙10⁷ spores∙m⁻²∙h⁻¹ (in May).

Figure 1

Monthly sporulation rate (SR, spores∙m⁻²∙h⁻¹) of Heterobasidion spp. The overall SR (a) and the SR of each fruiting body (b) are reported for each month, from January (1) to December (12). Error bars refer to the 95% confidence interval of the mean. Monthly values along with the corresponding minima and maxima are reported in Supplementary material 2.

By looking at the temporal patterns of sporulation displayed by individual fruiting bodies, four of them (A1, A2, B, and D2) sporulated throughout the experiment, while others for a shorter timeframe (see Table 2, period of spore trapping) and their ceased sporulation was associated with a change in color of the hymenophores. A single fruiting body (B) showed higher spore loads than all other fruiting bodies, and this was true for all the periods under investigation (Fig. 1b and Supplementary material 2). There were slight temporal shifts in the SR maxima depending on the fungal genotype. For instance, while the peaks of the average SR were observed in August for fruiting body B (8.881∙10⁸ spores∙m⁻²∙h⁻¹), they were observed in September for the majority of fruiting bodies (A1, A2, A3, C1, and D2 with average SR in the range: 7.261∙10⁵–1.711∙10⁸ spores∙m⁻²∙h⁻¹), and in October for fruiting body D1 (4.728∙10⁷ spores∙m⁻²∙h⁻¹; Fig. 1b).

Assessing the effects of seasonality and climatic variables on Heterobasidion spp. sporulation

The random forest models aimed at assessing the effects of seasonality and climatic variables on Heterobasidion spp. sporulation clearly showed climate to be more important than seasonality. The average temperature of the 30 days before spore trapping (T₃₀) ranked first in terms of variable importance (VI) among all other temperatures. Similarly the best relative humidity predictor was that of the 14 days before spore trapping (RH₁₄), while VI for precipitation scored highest for the rainfalls occurring during the 30 days preceding each spore sampling (P₃₀). The VI attained by T₃₀ was 1.8-fold higher than RH₁₄, the latter achieving a VI value 2.6-fold higher than P₃₀ following the first random forests run. The other two runs provided similar results, with the sampling month (m) displaying a VI value 2- to 10-fold lower than the corresponding variable importance of T₃₀, RH_14, and P₃₀ (Fig. 2).

Figure 2

Variable importance assessing the effects of seasonality and climatic variables on Heterobasidion spp. sporulation. Panels (a), (b), and (c) display the outcomes of the three different runs of the random forest models. Variable acronyms refer to the month (m), average air temperature (T in °C), relative humidity of the air (RH, in %), and the amount of cumulated rainfall precipitations (P in mm). Subscripts indicate the spore sampling day (0), and the 1, 7, 10, 14, and 30 preceding days.

Predicting Heterobasidion spp. sporulation based on seasonality and climatic variables

Tree models fitted on the training data to predict the SRN of Heterobasidion spp. based on the season variable m resulted in significant splits (P < .05) leading to 5 terminal nodes, corresponding to SRS from 1 to 5. The predicted spore loads were highest in August and especially September (SRS = 5), lowest during December, January, February, and March (SRS = 1; Fig. 3). The tree model was successfully validated, displaying positive and significant Pearson’s R linear correlation coefficients between the SRN observed and predicted for both training (0.476, P < .001) and validation data (0.541, P < .001). Comparable RMSEP values were obtained from the tree model runs on training (0.151) and validation data (0.117).

Figure 3

Predicting Heterobasidion spp. sporulation based on seasonality. For each month, the predicted average sporulation rate normalized (SRN) is reported along with its 95% confidence interval. SRS from 1 to 5 are marked with different colors (SRS = 1 for the lowest and SRS = 5 for the highest sporulation levels). Different letters upon the bars indicate significant differences based on the outcomes of the tree model (P < .05).

Tree models fitted on the training data to predict the SRN of Heterobasidion spp. based on the average temperature (T₃₀), relative humidity (RH₁₄), and cumulated precipitations (P₃₀) produced significant splits (P < .05) leading to 5 terminal nodes, thus resulting in SRS from 1 to 5. However, only temperatures and relative humidity were associated with significant splits (P < .05). The highest spore loads with SRS = 5 were predicted when the average temperatures of the 30 days before samplings exceeded 8.9°C and relative humidity of the 14 days before samplings exceeded 75%. Lowest spore loads with SRS = 1 were predicted when the average temperatures of the 30 days before samplings were below 3.3°C (Fig. 4). The tree model was successfully validated, displaying positive and significant Pearson’s R linear correlation coefficients between the SRN observed and predicted for both training (0.575, P < .001) and validation data (0.527, P < .001). Comparable RMSEP values were obtained from the tree model runs on training (0.140) and validation data (0.120).

Figure 4

Predicting Heterobasidion spp. sporulation based on climatic variables. For each climatic condition defined by values of the average temperatures of the last 30 days (T₃₀) and relative humidity of the last 14 days (RH₁₄), the associated predicted average sporulation rate normalized (SRN) is reported along with its 95% confidence interval. SRS from 1 to 5 are marked with different colors (SRS = 1 for the lowest and SRS = 5 for the highest sporulation levels). Different letters upon the bars indicate significant differences based on the outcomes of the tree model (P < .05).

Sporulation observed for fruiting bodies of different size, age, and Heterobasidion species

All fruiting bodies belonged to Heterobasidion parviporum, with the exception of fruiting body D1, belonging to H. annosum s.s. (Table 2). H. parviporum released on average 6.902∙10⁷ spores∙m⁻²∙h⁻¹, while the H. annosum s.s. fruiting body attained a lower value of 7.502∙10⁶ spores∙m⁻²∙h⁻¹.

The surface of the hymenophore of fruiting bodies ranged from 31 to 805 cm². The Kendall’s rank test conducted between the average SR of each fruiting body and its corresponding hymenophore surface resulted in a positive, yet not significant correlation coefficient (τ = 0.500, P > .05).

The age of fruiting bodies A1 and A2 was 9 years, while the age of fruiting body B was 7 years. The average SR was 3.5-fold lower for 9 years old fruiting bodies (A1 and A2) than for the 7 years old fruiting body (B) (3.560∙10⁷  vs. 1.237∙10⁸ spores∙m⁻²∙h⁻¹).

Discussion

Our study elucidates the temporal dynamics of sporulation by Heterobasidion annosum s.l. in a geographic area (Baltic countries) which has so far not been examined. Based on the close proximity of passive spore traps to the hymenophore of each fruiting body there was little opportunity for spores to escape capture. With this method, we ensured that the sporulation rate (SR) is representative of the temporal dynamics of sporulation of each single fruiting body, and not only of the overall SR of Heterobasidion spp. at the forest stand level. In addition, we attempted to minimize the potential effects of diurnal fluctuations of sporulation by applying the spore traps in the same hour of the day throughout the experiment. The exposure time of spore traps lasted 7 minutes, but sometimes field and/or operational constraints imposed some slight shifts, not exceeding 1 minute, hence resulting in a range of 7 ± 1 minutes. However, such constraints rarely occurred, thereby allowing to consider the time of exposure constant. Since spore traps were exposed at the same hour of the day, variability in spore loads might have resulted from changing physical environmental conditions occurring at the same hour of the day in the different seasons. For instance, shifting sunrise time can potentially affect air humidity, temperature or other variables that might influence the sporulation of Heterobasidion spp. Nonetheless, climatic data were recorded, thus accounting for this extra-source of variability.

The results from our study conducted over a 3-year period show that: (i) overall airspora may be driven by individual highly-sporulating fruiting bodies rather than by fruiting bodies contributing in a similar manner to the overall spore loads; (ii) there may be slight temporal shifts in spore release depending on the fungal genotype and possibly on the Heterobasidion species, but peaks in spore release are observed in late summer and early fall for most fruiting bodies; and (iii) spore loads are better explained by specific climatic variables than by seasonality per se.

Overall, peaks in spore loads were observed in the second part of summer (August, followed by September). However, caution should be taken to avoid overinterpreting these results. Indeed, one of the added value of this study compared to most of the others focusing on Heterobasidion spp. aerobiology conducted so far, is that spore trapping targeted individual fruiting bodies. Thereby, our experimental design allowed for an appraisal of the contribution of individual fruiting bodies to the overall airspora. Data clearly showed that a single fruiting body (i.e. B) accounted for most of the airspora, and this was true for all periods under investigation. This is noteworthy because it shows that highly sporulating Heterobasidion genotypes/fruiting bodies may be present in the forest and they may substantially increase the infection risk regardless of the time period. However, the overall peak in spore loads observed in August is overwhelmingly explained by the highly sporulating Heterobasidion fruiting body B, which incidentally showed its spore release peak right in August. This does not seem to be the rule. Indeed, data analysis showed that there were slight temporal shifts in sporulation peaks depending on the fungal genotype. Additionally, the majority of fruiting bodies had their peaks in spore release in September or even October. These findings suggest that overall sporulation peaks may be shifted to the end of summer or early fall by minimizing the effects of the random factor represented by the highly sporulating genotype showing spore load peaks earlier. These observations are fully supported by the models based on the sporulation rate normalized (SRN) and seasonality as the independent variable predicting spore loads to be highest in September and followed by August, then by October and from May to July, November, and April. Spore loads are lowest from December to March. These temporal patterns of sporulation observed in Latvia and mirroring the infection risk by Heterobasidion spp. are closer to those described in the Alpine area (Gonthier et al., 2005; Gonthier, 2010) than to those reported in Fennoscandia (Kallio, 1970).

When studying temporal patterns of sporulation, a major challenge is disentangling the role of climatic variables, which are intrinsically dependent on the season (Trenberth, 1983), with that of the season itself. To overcome this issue, we used the variable importance metrics based on a random forest algorithm (Strobl et al., 2008) able to quantify the unbiased effect of each predictor on the response variable. Noteworthy, this algorithm is based on a non-parametric statistical approach able to successfully cope with collinear variables, such as seasonality and climate (Strobl et al., 2008). Nonetheless, when analysing temporal patterns of sporulation, data transformation may be required to standardize the response variable. For instance, in some aerobiological studies targeting the spore deposition patterns of fungal pathogens (Lione et al., 2021, 2022) the deposition rate was transformed. The transformation allowed to assess the role of seasonality and climate by excluding the extra-source of variability due to the fact that samplings were conducted in different sites, with different overall levels of inoculum pressure (Lione et al., 2021, 2022). Similarly, in our study we controlled the effect of the genotype on the SR when analysing the role of seasonality and climate by transforming the SR into the SRN (Lantz, 2019; Lione et al., 2021, 2022). Hence, for the first time for H. annosum s.l., results of the analyses clearly proved a prevalent role played by specific climatic variables compared to seasonality in explaining sporulation. Consistently with what has been observed in the Alps (Gonthier et al., 2005), the average temperature of the 30 days before spore trapping was the variable most significantly correlated with Heterobasidion spp. spore loads. The average relative humidity of the 14 days before spore trapping was also a statistically significant variable when average temperatures exceeded approximately 9°C.

We acknowledge that the present study was based on a limited number of fruiting bodies, monitored along a well-defined timeframe of 34 months and that sporulation data refer to three forest sites located within the same country (i.e. Latvia). Nonetheless, the statistical approach we proposed may be deemed robust since: (i) it is based on non-parametric tests and models; (ii) it hinges on algorithms specifically designed for the analysis of small samples and a large amount of predictors (i.e. the so-called ‘small n large p’ problem); and (iii) it is unbiased by construction (Hothorn et al., 2006; Strobl et al., 2008, 2009; Hothorn and Zeileis, 2015). In addition, models performance was assessed through internal and external validation. Internal validation was conducted with sporulation data from the 7 fruiting bodies used to build the models, while external validation was performed by using data from one fruiting body, which was randomly selected and excluded from models fitting. Such statistical design has proven its reliability for modeling and validating the sporulation of plant pathogens through aerobiological assays (Lione et al., 2021, 2022), even with a smaller sampling size than that available for this study. Nonetheless, we recognize that additional data collected from fruiting bodies of Heterobasidion spp. in other regions and in different years might provide further external validation metrics other than those we already provided in this study. Thereby, whether and the extent to which predictive models here developed would be suitable for other areas of central Europe and for forests other than Norway spruce mixed stands deserve further investigations.

The observation that a prevalent role on Heterobasidion spp. sporulation is played by some specific climatic variables rather than by the season itself supports the notion that risk predictions based on climatic variables should be preferred over those based on seasonality. From a practical perspective, one of the pre-conditions for using predictive models based on climatic variables is the availability of climatic data recorded continuously in the area. If these data are lacking, models based on seasonality may be used as well. However, seasonality is a predictor whose causal relation with sporulation is weaker than climate, both under a statistical perspective (see variable importance values VI) and under a biological perspective. Indeed, the sporulation of fungi is expected to be driven more by environmental factors (Kendrick, 2017) than by the calendar time, although calendar time is correlated to environmental factors such as climate (Trenberth, 1983). In our study we showed that the seasonality-based model may achieve satisfactory performance, comparable to those displayed by the climate-based model. Indeed, the seasonality-based model scored worse than the climate-based model when considering the internal validation metrics, while the external validation indices were slightly better, despite the VI of seasonality was far lower than that achieved by climatic variables. This apparent contradiction is probably related to the collinearity between climatic and temporal variables, which by construction does not affect the unbiased VI computation (Strobl et al., 2008), while it could affect the validation, especially if the validation set is limited in size, as in this case. As a general rule, climate-based models seem more appropriate for risk assessment, also in light of the current climate change processes often leading to local asynchronies between climatic conditions and seasons, which is likely to increase in the future (Kutta and Hubbart, 2016; Twardosz et al., 2021).

Generalizations about the role played by the species, hymenophore dimension, and age of the fruiting body on the sporulation of Heterobasidion spp. would require a larger sample size than that available for this study. However, our empirical observations suggest some interpretative hypotheses.

Heterobasidion species seems to have a negligible importance compared to the genotype in determining spore loads, as shown by the highly differentiated individual contributions of the single fruiting bodies to the overall airspora. However, interestingly, the H. annosum s.s. fruiting body showed peaks of SR in October, while all H. parviporum fruiting bodies showed peaks earlier (August or September). In agreement with a previous study targeting the overall Heterobasidion airspora (Gonthier et al., 2005), this observation supports the notion that H. annosum s.s. and H. parviporum may slightly differ in their temporal patterns of sporulation, with maximum spore loads of the former more shifted to the end of the host growing season. Previously, based on aerobiological data derived from spore trappings not targeting individual fruiting bodies, it was suggested that H. parviporum could have a higher spore release (Möykkynen et al., 1997) occurring with a different timing (Gonthier et al., 2005) in comparison to H. annosum s.s. Although we observed that on average H. parviporum sporulated 10-fold more abundantly than H. annosum s.s., our data do not allow to infer about differences of sporulation magnitude between the two species, since only one individual was analysed for H. annosum s.s.

We observed a mild positive, yet not significant correlation between the spore loads and the size of hymenophores. This finding could be explained by the vicinity of the hymenophores to the Petri dishes during spore trapping. However, it is worth noting that also fruiting bodies with a narrow hymenophore sporulated abundantly, releasing in some periods spore loads comparable to, or higher than those discharged by fruiting bodies with a wider hymenophore. For instance, this was the case of fruiting body A1, whose hymenophore size was 1.4-fold larger than that of fruiting body B, but whose monthly SR was on average only 30% of that observed for fruiting body B. Also fruiting bodies A2 and A3 showed a similar pattern.

The same is true for the age of fruiting bodies. Although we observed a trend toward a reduction of sporulation with age, we acknowledge that this could be an artifact due to the very limited number of fruiting bodies examined. In addition, incidentally, the youngest fruiting body was the highly sporulating one. Therefore, we deem these results about the role of the species, hymenophore dimension, and age of the fruiting body on the sporulation of Heterobasidion to be at best preliminary. Nevertheless, these findings suggest that this might be a field of research worth to be further explored.

Conclusion

While shedding light on some neglected aspects of Heterobasidion spp. aerobiology at the individual fruiting body scale, this study elucidated the temporal patterns of sporulation of Heterobasidion annosum s.s. and H. parviporum in an underrepresented area, i.e. Baltic countries. From a practical perspective, spore loads and hence the risk of infection are better explained by specific short-term climatic variables than by the season. Therefore, the use of predictive models developed and validated in the present study based on climatic variables is recommended whenever possible in the framework of integrated disease management programs against the Eurasian Heterobasidion spp. Alternatively, seasonal based models can be used as well, in case climatic data are minimal or not available.

Acknowledgements

Authors are grateful to Astra Zaļuma and Natālija Burņeviča for the assistance in data curation.

Conflict of interest

None declared.

Funding

This work was supported by the European Regional Development Fund (ERDF) within the framework of the project Forest Sector Competence Center (project No. 5.1.1.2.i.0/1/22/A/CFLA/007/P12 ‘The use of cord forming basidiomycetes in limiting the spread of root and butt rot in forests on peat soils’) and JSC Latvian State Forests (project No. 5–5.9.1_007q_101_21_79 ‘Investigation of the impact of root rot and reducing risks caused by root rot’).

This work was also supported by the Agritech National Research Center and received funding from the European Union Next-Generation EU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR)—MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4—D.D. 1032 17/06/2022, CN00000022). In particular, our study represents an original paper related to the Spoke 4—‘Multifunctional and resilient agriculture and forestry systems for the mitigation of climate change risks’—Task 4.1.3. This manuscript reflects only the authors’ views and opinions, neither the European Union nor the European Commission can be considered responsible for them.

This research was also partially funded by a grant from the University of Torino (ex 60%—Approfondimenti di modellistica fitopatologica con applicazioni a casi studio di interesse agrario, forestale ed agro-forestale—MOFIT).

Data availability

Data relevant to this study are provided in Supplementary material 3.

Author notes

Guglielmo Lione and Lauma Silbauma contributed equally to this work.