An important question in the Babcock-Leighton framework is what nonlinear mechanism acts to regulate the solar cycle and prevent runaway fluctuations. Here I present observational evidence that the dominant nonlinearity may be latitude quenching (whereby flux emerges at higher latitudes in stronger solar cycles) rather than tilt quenching (whereby the tilt of active regions is reduced in stronger cycles). This is based on a historical database of individual magnetic plage regions derived from digitized Mount Wilson data. In particular, I find that proxy observations of the Sun's polar field suffice to constrain the "dynamo effectivity range" in surface flux transport simulations driven by the historical database. This in turn favours latitude quenching