In studies of animal colouration it is no longer necessary to rely on subjective assessments of colour and conspicuousness, nor on methods which rely

There are many possible spatial patterns of selection and gene flow that can produce a given cline structure; the actual geography of natural selection and gene flow must be worked out before an attempt is made to explain a given natural cline in terms of a model. The results of experimental and theoretical models show that it is possible for local differentiation to evolve parapatrically in spite of considerable gene flow if the selection gradients are relatively uniform. Irregularities in environmental gradients increase the sensitivity of clines to the effects of gene flow in proportion to the increase in the differences in gene frequencies between the emigrants and the demes receiving the immigrants. It is not necessary for a sharp spatial environmental change to be present for distinct differentiation to occur. In some cases even a gentle environmental gradient can give rise to marked spatial differentiation along a genetically continuous series of demes; such environmental differences may be below the practical limits of resolution in field studies. Any asymmetry in gene flow does not lead to dedifferentiation if the environmental gradient is smooth; it merely shifts the position of the transition zone between the differentiated areas from that which would be expected if there were no asymmetry. Abrupt geographic differences in gene, genotype, or morph frequencies should not, therefore, be interpreted as evidence for environmental changes in the immediate vicinity of the steepest part of the cline; neither should they be interpreted as evidence for geographic barriers, sharp environmental differences, or sexual isolation among the differentiated groups of populations when there are no other sources of evidence for these phenomena. Gene flow may be unimportant in the differentiation of populations along environmental gradients.