The rate of signal production by social ectotherms is often temperature dependent. This has been typically attributed to an underlying thermal constraint on physiology, but there are other reasons why signal rates might be correlated to temperature. We tested 3 hypotheses. The maximal performance hypothesis: temperature limits motor activity at cold and hot temperatures, which predicts a hump-shape function between signal rate and temperature. The metabolic rate hypothesis: the available energy released by metabolism increases exponentially with temperature, which predicts an exponential increase in signal rates with temperature. The recipient availability hypothesis: the number of potential receivers varies with temperature, and signalers change their signal rates accordingly, which predicts an indirect association between signal rate and temperature. We tested these hypotheses using field data on the rate of advertisement display production by territorial Jamaican and Puerto Rican Anolis lizards from a variety of thermal environments, coupled with extensive observations on one montane population of A. gundlachi. In both cases, the slopes of display rate as a function of ambient temperature were statistically indistinguishable from slopes predicted by the performance hypothesis. Support for the other 2 hypotheses was weak or equivocal. This is the first study to test alternative hypotheses of why signal rates and temperature in ectotherms are correlated and to indicate that thermal performance curves measured in the lab can reliably predict important social behavior in the field.