If a population exploits or tracks a limiting resource perfectly through time, it will be constantly under density-dependent population regulation and density-dependent natural selection. Such complete resource use underlies many theoretical formulations in ecology and population genetics, particularly the so-called logistic models (Roughgarden, 1971; Charlesworth, 1971; MacArthur, 1972; May, 1974). Often the limiting resource is food, which is equivalent to energy. Occasionally space is the limiting resource, but for plants, microbes, or sessile animals space limitation may be equivalent to energy limitation. Perfect density-dependence does not occur when predators, parasites, pathogens, demographic lags, resource fluctuations or abiotic factors rarify a population relative to its resources leaving no resource in absolute short supply. In discussions of natural selection, density-dependence is usually about the same as K-selection and density-independence is the same as r-selection, where K and r are the usual terms of the logistic equation (MacArthur and Wilson, 1967). These dichotomies of ecological and evolutionary theory are too simple. First of all, at a theoretical level, K and r are not independent. The solution of the logistic equation for either must include the other. Hence, at population densities between zero and some maximum, there is an interrelation of K and r through which we try to describe selection in such models (King and Anderson, 1971). Second, selection always operates through birth, developmental, and death events which are the components of r, hence selection for any adaptation, K-related or not, proceeds by the same mechanism as selection for rrelated adaptations. What varies is the degree to which realized r is positive or negative. The selection occurring when r is negative for substantial periods may be formally equivalent to the selection occurring when r is positive (Istock, 1970). We are left not with a simple dichotomy between density-independence and densitydependence, but with the more cumbersome continuum between them. Since many aspects of an individual's ecology can be influenced by the activities of other co-occurring, conspecific individuals (mating, nesting, cannibalism, group, foraging, etc.) even when the population is well below K, we may assert that selection is always densitydependent to some extent. What we really want to assess is the degree of density-dependence. One way to do this, though not a perfect one, is to measure how close a population is to K at any given time. Measurement of the closeness of a natural population to its K may reveal imperfect resource tracking. Experimental results presented in this paper will show highly imperfect tracking in a natural population of the pitcher-plant mosquito Wyeomnyia smithii. Small deviations from K may not seriously challenge the assumptions or even the realism of logistic models (MacArthur, 1972). Major deviations such as recorded here do. If it turns out that highly imperfect resource tracking is widespread, many of our models of intraand interspecific competition, predation, community organization and natural selection may require revision. Imagine the following specialized and oversimplified, though probably not uncom-