Spatial ecological patterns are usually ascribed to Turing-type reaction-diffusion or scale-dependent feedback processes, but morphologically indistinguishable patterns can be produced by instabilities in fluid flow. We present a new hypothesis that suggests that fluid convection and chill damage to plants could form vegetation patterns with wavelengths similar to 1-2 times the plant height. Previous hypotheses for small-scale vegetation pattern formation relied on a Turing process driven by competition for water, which is thought to occur in large vegetation patterns. Predictions of the new hypothesis were consistent with properties of natural grass patterns in North Carolina, contradicting the Turing hypothesis. These results indicate that similarities in pattern morphology should not be interpreted as implying similarities in the pattern-forming processes, that small-wavelength vegetation patterns may arise from mechanisms that are distinct from those generating long-wavelength vegetation patterns, and that fluid instabilities should be recognized as a cause of ecological patterns.