Table 1.

Results of the general additive modeling (GAM) explaining the number of cases of leptospirosis per subdistrict in Mahasarakham province using a negative binomial link (theta = 0.267), with approximate significance of smooth terms. Deviance explained = 14%, REML = 2950.7, AIC = 5852 (see Figure 7).

Terms	edf	Ref.df	Chi Square	p Value
S (longitude, latitude)	17.82	29	67.44	<0.0001
Te (month, rainfall)	12.28	24	94.43	<0.0001
Te (month, flood cover)	3.36	20	42.89	<0.0001
S (population)	3.98	9	72.42	<0.0001
S (pig number)	0.87	9	5.99	0.006
S (cow number)	1.26	9	3.44	0.037
S (average slope)	0.68	9	2.11	0.043

The best GAM model selected on the basis of AIC value included all initial variables except the number of buffaloes (Table 1, Figure 7): gam(leptospirosis cases per sub-district) ~ s(longitude, latitude) + te(month, rainfall) + te(month, flood cover) +s(number of people in sub-district)+s(number of pigs) + s( number of cattle) + s(average slope).