Table 2. Best models describing nine different indicators of productivity and income (response variables).
| Response | N | ModelType a | Model | Interpretation | |
|---|---|---|---|---|---|
| Fixed effects | Random effects | ||||
| Number of colonies | 187 | LMM | Years keeping bees, Number of known beekeepers, Native vegetation, Use of vinegar, and Supplementary feeding | Main species | Beekeepers with more years of experience that know a larger number of other beekeepers, use vinegar to control parasitic flies and feed their colonies, have more colonies. The number of colonies is also higher in locations that have native vegetation within 3 Km. These effects hold across the main species kept. |
| Number of colonies of main species | 195 | LMM | Years keeping bees, and Number of known beekeepers | Main species | Beekeepers with more years of experience that know a larger number of other beekeepers, have more colonies of the principal species. These effects hold across the main species kept. |
| Number of multiplied colonies | 126 | LMM | Number of colonies, Number of known beekeepers, Supplementary feeding, and Property ownership | Main species | Beekeepers with more colonies, that know a larger number of other beekeepers, and feed their colonies, manage to multiply a larger number of colonies per year. Property owners multiply fewer colonies per year than not-owners. These effects hold across the main species kept. |
| Liters of honey produced per colony | 70 | LMM | Selective breeding | Main species | Honey production per colony is higher among beekeepers that multiply their colonies selectively. This effect holds across the main species kept. |
| Number of colonies lost b | 86 | GLMM | Inspection frequency and Honey harvest method | Individual, Main species | Beekeepers that inspect their colonies less frequently and harvest honey by flipping the boxes lose more colonies. These effects hold across the main species kept. |
| Number of colonies sold | 60 | LMM | Years keeping bees | Main species | Beekeepers with more years of experience keeping bees, sell more colonies per year. This effect holds across the main species kept. |
| Liters of honey sold | 64 | LMM | Number of colonies of main species, Years keeping bees, and Honey conservation method | Years keeping bees, and Main species | Beekeepers with more colonies of the main species, more years of experience keeping bees, and using an established honey conservation method, sell more honey per year. These effects hold across the main species kept, although the magnitude of the relationship between Years keeping bees and Liters of honey sold varies. |
| Earnings in R$ 3 | 75 | LM | Number of colonies, Labeling of honey containers, Honey conservation method and the interaction Number of colonies-Labeling of honey containers | - | Beekeepers with more colonies that label honey containers and use an established honey conservation method, have higher yearly earnings. The influence of the number of colonies on earnings is more pronounced among beekeepers that label honey containers. These effects do not vary across the main species kept. |
| Costs in R$ c | 132 | LM | Number of colonies, Education level, Feeding frequency, Age, and the interaction Feeding frequency-Age | - | Yearly costs increase with the number of colonies, the level of education, the feeding frequency, and the age of beekeepers. Older beekeepers spend more money regardless how frequently they feed their colonies. These effects not vary across the main species kept. |
The number of observations included in each model is provided (N) along with the model type, the model structure and its biological interpretation. Regression coefficients, p-values, and confidence intervals for all models are summarized in S5 Table.
a Linear model (LM), linear mixed model (LMM), or Generalized linear mixed model (GLMM).
b Number of lost colonies per year standardized by the total number of colonies kept (used as an offset). GLMM with a Poisson distribution. Overdispersion accounted for by including individual as a random effect.
c Main species kept was excluded as a random effect since its variance approached zero in a mixed model.