Examination of income sources and other characteristics among agricultural colonists in Ecuador's northern Amazon reveals economic differentiation of the frontier associated with farming, off-farm work, and cattle raising. Regression models use data from a unique, representative farm-level survey in 1990. As expected, farm income is higher on large, fertile, extensively cultivated plots near roads. Evidently, greater start-up capital helps some migrants, enabling the purchase of better land, title, credit, cattle, and other assets. Off-farm work results from positive and negative factors: involvement in low-paying agricultural work (jornalero) is common among recent settlers and those with little farm income; at the same time, wealthier, better-educated, larger, households choose lucrative nonagricultural work. in turn, they hire more farmworkers, earn more farm income, and invest in pasture and cattle. A third model examines cattle income, which is more important among households with greater start-up capital, more land, a legal title, and Sierran origins (regardless of labor availability). The varied status of frontier households is thus associated with farming, off-Farm work, and cattle raising, livelihood activities that result from differentials in wealth before migrating, combined with frontier conditions and household characteristics. The settlement of Ecuador's Oriente has thus achieved very modest success in redistributing some land and a livelihood to people without land. Yet, at the same time, settlement is also providing more land to those who already had land, setting off a process of differentiation that parallels highland disparities (the "Andeanization of the Amazon") as well as colonization processes: across the globe. Jointly, rich and poor colonists are deforesting Ecuador's rich "hot spot" of biological diversity through their distinctive activities. Some policies to aid small farmers, prevent further pasture extensification, and promote more ecologically sound livelihoods in this region are suggested, based on the statistical findings and supplementary fieldwork in the 1990s.
