The sex ratio changes in Sand martin populations, described in Persson (19876), will by definition act as a control loop to the ‘reproduction amplifier’ of the population. Age structure integrates the effect over a couple of years, thereby creating the specific time-lag of three to four years. One hypothesis of this paper suggests that these changes are intentional, a true feedback from the population level or the overall sex ratio. The main advantage of a feedback is to stabilize a system against parameter variations, in the present case primarily ‘catastrophic’ winter mortality and/or reproduction failure. In consequence the observed cycle is at least the by-product of an adaptation. If numerical values of r (the intrinsic rate of natural increase) and Tare introduced into the equations defining limit values of an age-structured model, its dynamic behaviour will be confined between diverging and damped oscillations, leaving a stable limit cycle as a very probable outcome.
The disturbance creating a need for regulation may be epitomized as a variation of equilibrium levels. This variation must be dependent on African drought, and perhaps ultimately on the sunspot cycle (the second hypothesis of the paper). Regulation as an adaptation is created (and probably maintained) by means of group-selection; it has earlier been shown (Persson, 1978) that one prerequisite for group selection is at hand; the study population is virtually confined within the census area. The ‘vulnerability’ (Persson, 19873) of maximum populations complicates the picture. Additional vulnerability may result from a conflict between individual selection and group selection-young birds with high breeding fitness attempting double broods when they should not-that is injurious to perfect regulation.