Dispersal patterns of Alternaria conidia in Spain

摘要

Alternaria is a common airborne phytopathogenic fungus that may affect crops in the field or can cause decay of plant products. It can also cause diseases in animals and humans. The study of airborne Alternaria conidia is a necessary step for the control and prevention of the agricultural damage they can provoke. The aim of this paper is to contribute to model the presence and levels of Alternaria conidia in the air using a logistic regression model. Our study is conducted in 12 monitoring stations in Spain corresponding to three geographic regions with different bio-climatic characteristics, which show three different patterns of Alternaria conidia dynamics: a unique main sporulation season from mid spring to autumn in NE Spain, two defined periods (spring and autumn) in SE Spain and a uniform and constant presence in the Canary Islands. Regarding the abundance, NE Spain shows the highest values and the Canary Islands the lowest. Daily Alternaria conidia concentration is positively correlated to daily minimum temperature and daily temperature variation and negatively correlated to daily precipitation. Also, the occurrence of rain in the 3 previous days has a positive effect on Alternaria levels. These effects are modelled in this paper by means of logit regression equations. The three equations used apply to the presence of Alternaria conidia, and to the exceedance of thresholds of 10 and 30conidia/m3. The model is calibrated in the 12 stations using data from years 1995 to 2008 and validated with data from 2009 in 7 stations, showing a reasonable percentage of right prediction (average 78.6%, ranging from 61.3% to 92.5%). a- In Spain, Alternaria conidia are most abundant under humid and temperate climates. a- Rural environments show higher Alternaria conidia content than urban. a- Daily minimum and variation temperature and 3 prior days rainfall increase conidia levels. a- Logit Regression equations show the occurrence of 1, 10 and 30conidia/m2 levels. a- The percentage of right predictions of the models is 78.6% [range 61.3% to 92.5%].