Pathogen seasonality and links with weather in England and Wales: a big data time series analysis

Cherrie, Mark P. C.; Nichols, Gordon; Iacono, Gianni Lo; Sarran, Christophe; Hajat, Shakoor; Fleming, Lora E.

We detected significant seasonality in 91 organisms using TBATS models at varying periodicities (91/277; 33%) (Additional file 2: Table S1 ); with varying links with weather (Additional file 3: Figure S2 ). Two k-means clusters (identified as the optimum number of k) were generated from the cross correlation coefficients with weather variables and represented groups of pathogens that had similar correlations with weather variables (Fig. 4) . The two groups were characterised by their relationship with the weather variables (Additional file 4: Table S2 ). Group 1 had mean positive correlations with higher temperature (min, mean, max), sunshine and vapour pressure; whilst the Group 2 had positive mean correlations with lower temperature variables (snow lying, ground frost, air frost), precipitation (rain days over 1 mm, rain days over 10 mm and rainfall), mean In Group 2, pathogens had highest correlations with relative humidity (n = 8) and Ground frost (n = 5) (Additional file 5: Figure S3 ). There was at least one pathogen with the highest correlation for each meteorological variable. Summary information on seasonality and links with weather, by temperature cluster group are presented in Table 1 .