Corresponding author: Mireia GomezGallego (
Academic editor: Christelle Robinet
The sooty bark disease (
We found that an accumulated water deficit in spring and summer lower than 132 mm correlates with
Muller E, Dvořák M, Marçais B, Caeiro E, Clot B, DesprezLoustau ML, Gedda B, Lundén K, Migliorini D, Oliver G, Ramos AP, Rigling D, Rybníček O, Santini A, Schneider S, Stenlid J, Tedeschini E, Aguayo J, GomezGallego M (2023) Conditions of emergence of the Sooty Bark Disease and aerobiology of
Emerging infectious diseases threaten human health, agriculture and biodiversity (
Examples of forest diseases linked to climate extremes that are increasing in Europe are Diplodia tip blight in pine species (
The
The objectives of the present study are therefore: (1) to develop a realtime PCR assay for the detection of
To analyse the emergence of
where
The climatic data were obtained from MétéoFrance (SAFRAN database) computed on a daily basis on an 8km resolution grid throughout France and Switzerland (except for the Tessin region, where these data were not available) (Suppl. material
The samples used as starting material in our aerobiological study consisted of microscope slides with a ca. 48mm portion of Melinex tape (corresponding to 24 h ± 2 h, depending on the sampling time) from Hirsttype volumetric air samplers used to monitor airborne pollen grains and fungal spores by the aerobiology networks of the involved European countries. The Hirsttype air samplers (
7day volumetric air sampler (Burkard Manufacturing Co Ltd, Hertfordshire, UK) in Brno (Czechia) installed on the roof of the University hospital, 15 m above ground to ensure landscapescale monitoring. Photo credit for Aneta Lukačevičová.
We undertook two studies, at a regional and a continental scale, to evaluate the use of permanent aerobiological networks to assess
We selected samplers to cover the
Selected air samplers,
Selected French air samplers for the regional study with different
City  Code  GPS Coordinates  Year of the first record at < 50 km  Year of the first record at < 100 km  Year of the first record at < 180 km 


MUL 

2010  2010  1992 

BAR 

no records  2010  1992 

BES 

no records  2006  1992 

STR 

no records  2010  2010 

ANG 

no records  2016  1991 

AUR 

no records  no records  2014 

AVI 

no records  no records  2002 

GAP 

no records  no records  2002 
In order to align the aerobiological data with the presence of the disease, we used the disease records from the
We selected a total of 12 air samplers across six European countries, spanning a large longitudinal and latitudinal range, in the axis northsouth from Sweden to Portugal, and in the axis westeast from Portugal to Czechia (Table
Locations of European air samplers for aerobiological samples analysed during the period from the 3^{rd} of June to the 25^{th} of September 2018, every 12 days (N = 10).
City  Code  GPS Coordinates  Country  Year first 
Laboratory for DNA extraction 


BRN 

Czechia  2005^{1}  Mendel University (Czechia) 

GAP 

France  1950^{2}  INRAE Bordeaux (France) 

PON 

France  INRAE Bordeaux (France)  

BES 

France  INRAE Bordeaux (France)  

BOR 

France  INRAE Bordeaux (France)  

BOL 

Italy  1952^{3}  IPSPCNR (Italy) 

PER 

Italy  IPSPCNR (Italy)  

GÄV 

Sweden  Not reported  SLU (Sweden) 

VIS 

Sweden  SLU (Sweden)  

LIS 

Portugal  Not reported  SLU (Sweden) 

MÜN 

Switzerland  1991^{4}  

PAY 

Switzerland 
^{1}
Slides for the regional study were extracted in the laboratory of Forest Pathology at INRAE Nancy (France). For the continental study, the slides were extracted in different laboratories (Table
The ITS region sequences with accurate identification were retrieved from GenBank for
Samples were run in triplicate in the regional study and in duplicate in the continental study, and both a negative (no template DNA) and a positive control (
To quantify the spores on each aerobiological sample, we prepared 5fold serial dilutions of a spore solution obtained by adding purified water on the surface of a sporulating culture of a French
We have fitted Bayesian models to test the different hypotheses as follows. To analyse the effect of the climatic conditions on the emergence of the
number of cases of
where
d_{k} ~ Bernoulli (p) Eq. 3
where
We modelled the number of spores detected per week as a function of the distance to the closer disease report (model distance) and as a function of the total sycamore maple basal area in a radius of 50 km from the sampler (model host). We did not include the distance to the disease report and the total sycamore maple basal area as predictors in the same model because their high collinearity prevented model convergence. The two models followed a Poisson distribution (Eq. 5), with lambda varying for each observation following a Gamma distribution to deal with overdispersion (Eq. 6–8). We included a binomial process (Eq. 9) to account for zeros that arise in addition to those modelled by the Poisson process (i.e. sampler’s failure to capture spores even if they are present in the air). Therefore, the model distinguished two potentially different processes that determine the number of
number of spores ~ Poisson (λ_{k} * e_{k}) Eq. 5
where
λ_{k} ~ Gamma (a_{k}, b_{k}) Eq. 6
where a_{k} and b_{i} are the shape and rate of the Gamma distribution, which relate to the mean number of spores and to the standard deviation (sigma) as follows (Eq. 4–5):
a_{k} =
b_{k} =
e_{k} ~ Bernoulli (p) Eq. 9
log(
where
We modelled the probability of disease occurrence in a certain area of influence of the sampler (in a circumference of different radii, from 40 to 130 km of radius, by 10km intervals) as a function of the number of detected spores. The two models followed a Bernoulli distribution (Eq. 10). The deterministic part of the model is shown in Eq. 11.
Probability of disease occurrence in an area of 40 to 130 km radius ~ Bernoulli (
where
log(
where
To validate our models, we simulated data based on the likelihood of each model. We then compared the means, the coefficients of variation and the sums of squares of the residuals of the original dataset with each simulated dataset. The histogram of the differences for each statistic should be zerocentred, with the proportion of negative (or positive) differences being lower than 0.85 for the model to be accepted.
All Bayesian models were implemented using a Markov chain Monte Carlo (
The selected primers and probe used in this study were ccITS2F (AGGTTGTGCTGTCCGGTG), reported in the study by
The climatic variable best explaining the standardised
Coefficient estimates for each climatic variable and their 95% credible intervals in brackets for models predicting the standardised
Variable  Years  Coefficient estimate [95% CI]  Rhat  Deviance [95% CI] 

n1  1.19 [0.74, 1.68]  1.0012  209.1 [193.2, 228.3]  
1.21 [0.77, 1.70]  1.0009  199.1 [183.7, 217.9]  

1.15 [0.78, 1.53]  1.0009  201.0 [185.9, 218.8]  

0.66 [0.43, 0.89]  1.0009  199.0 [183.9, 217.1]  

1.08 [1.38, 0.78]  1.0009  188.3 [172.7, 207.5]  

 



1.35 [1.85, 0.89]  1.0009  192.2 [178.1, 209.5]  
n25  1.16 [0.78, 1.58]  1.0009  208.0 [193.2, 227.1]  
n1 to n2  1.37 [0.86, 1.98]  1.0009  197.9 [182.0, 217.9]  
1.23 [0.77, 1.71]  1.0013  203.0 [187.8, 221.5]  

1.25 [0.87, 1.66]  1.0009  196.1 [181.0, 214.3]  

0.64 [0.41, 0.87]  1.0009  202.8 [187.9, 221.0]  

1.08 [1.50, 0.70]  1.0009  200.0 [185.8, 218.1]  

1.07 [1.50, 0.68]  1.0001  208.6 [193.7, 227.1]  

0.74 [1.26, 0.27]  1.0009  227.0 [212.3, 244.8]  
n25  1.40 [0.94, 1.90]  1.0010  191.0 [175.5, 210.7]  
n1 to n3  0.83 [0.32, 1.43]  1.0009  222.2 [206.5, 241.2]  
1.21 [0.78, 1.65]  1.0009  201.0 [186.6, 218.9]  

1.09 [0.71, 1.49]  1.0009  205.8 [191.6, 222.8]  

0.72 [0.49, 0.95]  1.0009  201.5 [187.1, 218.6]  

0.61 [1.10, 0.16]  1.0009  227.7 [212.5, 246.5]  

0.50 [0.85, 0.16]  1.0009  225.9 [211.7, 244.6]  

0.42 [0.80, 0.06]  1.0009  227.4 [214.2, 244.9]  
n25  1.16 [0.57, 1.88]  1.0009  211.3 [194.8, 231.4] 
The number of
Model prediction of standardised
The number of spores detected per week was more abundant in samplers closer to disease reports (Fig.
Number of spores per day as a function of the distance to the closest disease report (
Parameter estimates and their 95% credible intervals in brackets for models describing spore detection as a function of distance to
Response variable  Detected spores per week  

Parameter  Intercept estimate  Coefficient estimate  Probability of spore capture 
Distance to disease report  2.64 [2.07, 3.07]  0.41 [0.97, 0.06]  0.73 [0.61, 0.85] 
Total maple basal area  2.04 [1.41, 2.55]  0.04 [0.01, 0.10]  0.72 [0.60, 0.85] 
The two models (distance to the disease, and maple basal area) estimated a similar probability of spore capture (0.73 and 0.72, respectively, Bernoulli process in Eq. 8, Table
The probability of disease occurrence increased with the number of detected spores at a given distance (Fig.
Probability of disease report in an area of 60km and 120km radius from the sampler as a function of the number of detected spores per day.
The proportion of positive aerobiological samples based on the quantitative speciesspecific
Proportion of aerobiological samples that tested positive from May to September of 2018 (n = 10, except for Gap where n = 5) in the different European samplers following the natural distribution of the host
Phenology of spore emission of
The present study aimed at analysing the emergence of
Disease peaks increased exponentially in magnitude with time.
The presence of
Distance tended to decrease the number of detected spores, but the magnitude of the effect was low and it was not significant (the confidence interval of parameter estimates contained 0). We detected
The relatively abundant number of spores of
This project has received funding from the European Union’s Horizon 2020 Programme for Research & Innovation under grant agreement No 771271 (HOMED project, “HOlistic Management of Emergent forest pests and Diseases”), and from the research project SIAMOIS (“Smart and Innovative Monitoring Of airborne fungal Invaders by molecular methods”) by the French laboratory of excellence LabEx ARBRE. We greatly appreciate the valuable assistance provided by Etienne BrejonLamartiniere, Laure Dubois, Julie Faivre d’Arcier, Manuela Branco Ferreira, Anaïs Gillet, Quirin Kupper, Aneta Lukačevičová, Miloslava Majerová, Sophie Strohecker, Fabrizio Cioldi. We thank the ARPAE Area Prevenzione Ambientale Metropolitana di Bologna, Italy, for providing part of the Italian samples analysed in this study. The mycology research unit of the ANSES Plant Health Laboratory is supported by a grant managed by the French National Research Agency as part of the French government’s ‘‘Investing for the Future’’ (PIA) programme (ANR11LABX000201, Laboratory of ExcellenceARBRE).
Records of
figure (word document)
Isolates which DNA was extracted and used to confirm the specificity of the primers ccITS2F and SBD3R and probe SBD5P
table (word document)
Standard curve and its correlation coefficient to determine the limit of detection for the realtime PCR assay in tenfolded DNA solutions of
figures (word document)
Zerocentred histogram of the residuals between simulated data and predictions of the model with the water balance (P
figure (word document)
Zerocentred histogram of the residuals between simulated data and predictions of the model with the distance to the closest disease report as a predictor of the number of
figure (word document)
Zerocentred histogram of the residuals between simulated data and predictions of the model with the total sycamore maple basal area in a radius of 50 km from the sampler as a predictor of the number of
figure (word document)
Coefficient estimate for each variable of maple basal area computed for different radius and their 95% credible intervals in brackets for models predicting the number of spores detected per week
table (word document)
Probability of disease report in an area of 40km to 130km radius from the sampler as a function of the number of detected spores per day
figure (word document)