**

** Indicates means are significantly different at 5% level (two-tailed, two-sample t-test).

** Indicates means are significantly different at 5% level (two-tailed, two-sample t-test).

The sample was also divided into subsamples in order to examine the average number of varieties planted by agroecological characteristics. The first category corresponds to the major ecological zones in the region, Tierra Caliente (Hot Lands - below 1200 masl) and Tierra Fria (Cold Lands - above 1200 masl). The average number of varieties grown is higher at the higher elevations, due to agroecological conditions. The second category is the number of plots farmed by the household, and this is used as proxy for whether the households are matching varieties to soil conditions. The average number of plots is significantly higher for households with multiple plots, indicating that the agroecological conditions also hold at the household level. The next two categories address the quantity of land, a key constraint to the number of varieties planted. Own hectares is the total hectares owned by the household, and maize hectares is the total hectares planted by a household to maize. For the own hectares, the means are not significantly different, while for hectares planted to maize, the average number of varieties planted is higher. This indicates that within the land planted to maize and milpa, land is a constraint to planting a greater number of varieties. Finally the market integration category is proxied by comparing small towns to larger towns. Large towns are a municipal capital, on a major paved road, or have a significant commercial sector and services. The average number of varieties planted by a household is significantly higher in the small towns, indicating that when the level of market integration increases, the number of varieties decreases.

However, each of these categories is showing a change in household levels of diversity in isolation from other factors. Furthermore, the effect of each condition could have a different effect, when all other effects are held constant, ceteris paribus. For instance, the age of the household head could increase or decrease diversity, if isolated from the effects of the number of plots farmed and the agroecological zone that the household is in. Therefore, the use of categories or correlation limits the ability to test for all of the effects that could be included in a household model. A general, nested model may be needed to test the effects of individual parameters and groups of parameters on the level of diversity maintained by households.

Was this article helpful?

## Post a comment