Selection on allozymes has sometimes been advanced as one explanation for the low levels of population differentiation detected in avian populations by the use of enzymatic markers. Comparisons of the amount of population subdivision (estimated by FST values or analogous indices) measured by enzymatic and mitochondrial DNA (mtDNA) markers in birds were seen as evidence for this because mtDNA typically produces a more structured picture of population subdivisions. In fact, when taking into account the smaller effective population size of mtDNA, nuclear and mitochondrial markers give concordant results. Some discrepancies still exist, but I suggest that some might originate from different amounts of nuclear vs. mitochondrial gene flow due to partial reproductive isolation. Variable number of tandem repeat (VNTR) loci do not provide a dramatically different picture of population structures in birds compared to allozymes. Although more tests are needed, such as comparing the amount of genetic structure detected in the same populations with allozymes and microsatellites, the low levels of population subdivision measured with allozymes in birds seem to reflect historical and demographic processes and would not appear to result from any peculiarities of bird enzymatic loci.

We investigated the role of selection in generating and maintaining species distinctness in spite of ongoing gene flow, using two zones of secondary contact between large gull species in Europe (Larus argentatus and Larus cachinnans) and North America (Larus glaucescens and Larus occidentalis). We used the pattern of neutral genetic differentiation at nine microsatellite loci (F ST ) as an indicator of expected changes under neutral processes and compared it with phenotypic differentiation (P ST ) for a large number of traits (size, plumage melanism and coloration of bare parts). Even assuming very low heritability, interspecific divergence between L. glaucescens and L. occidentalis in plumage melanism and orbital ring colour clearly exceeded neutral differentiation. Similarly, melanism of the central primaries was highly divergent between L. argentatus and L. cachinnans. Such divergence is unlikely to have arisen randomly and is therefore attributed to spatially varying selection. Variation in plumage melanism in both transects agrees with Gloger's rule, which suggests that latitude (and associated sun and humidity gradients) could be the selective pressure shaping differentiation in plumage melanism. We suggest that strong species differentiation in orbital ring colour results from sexual selection. We conclude that these large gull species, along with other recently diverged species that hybridize after coming into secondary contact, may differ only in restricted regions of the genome that are undergoing strong disruptive selection because of their phenotypic effects.