Although I agree we have to be cautious, I do think sufficient (at least 1000 m) gradients can serve as altitudinal proxies. In my Arizona study area (a mountain canyon), the gradient ranges from semi-arid to temperate. See Crimmins TM, Crimmins MA, Bertelsen CD (2009) Flowering range changes across an elevation gradient in response to warming summer temperatures. Global Change Biology 15:1141-1152.
Edward, that also depends on what aspects of climate change you want to study. For instance, is it the current climate change, or the effects of past climate changes across various geographical and climatic domains? For deglaciation chronosequences in the current climate change and impact on biodiversity, space-for-time substitution experiments are being used. This is not a perfect approach, and, yes, it needs to be used cautiously but if applied properly can provide very useful results. I am using this to study the impact of melting ice in the Andes. Here are a few references about the method by various authors.