Table 1 Comparison between different mathematical models of the logarithm of the ratio between CO2 and EW and CO2 in CTL using different explicative variables. The comparison is based on the statistical criteria AIC (Akaike Information Criterion) and BIC (Bayesian Information Criterion). Lower AIC and BIC values indicate a better fit with the experimental data. The models with random temporal effects consider that data belonging to the same treatments are not independent, and the models with fixed temporal effects consider each datapoint to be independent from the others.
Time fixed-effect | Time random-effect | |||
|---|---|---|---|---|
AIC | BIC | AIC | BIC | |
Time model | ||||
Without time effect | 404 | 412 | − 897 | − 884 |
Linear model with time | 366 | 379 | − 1255 | − 1234 |
Quadratic model with time | 358 | 375 | − 1256 | − 1226 |
Cubic model with time | 357 | 379 | − 1263 | − 1225 |
+ Co-variable (fixed-effect) | ||||
+ EW density | 47 | 73 | − 1285 | − 1242 |
+ EW category | 345 | 371 | − 1264 | − 1216 |
+ OM type | 281 | 311 | − 1261 | − 1205 |
+ Temperature | 365 | 390 | − 1105 | − 1059 |
+ Land occupation | 331 | 357 | − 1268 | − 1225 |
Interaction with density | ||||
Time density interaction | 36 | 66 | − 1360 | − 1304 |