Ideal phenotypes and mismatching haplotypes - errors of mtDNA treeing in ants (Hymenoptera : Formicidae) detected by standardized morphometry

A total of 401 nest samples of Formica lugubris Zetterstedt, F. pratensis Retzius, F. aquilonia Yarrow, F. rufa Linnaeus, and F. polyctena Forster, covering the entire Palaearctic range of these species and including 2100 individual workers, was phenotypically investigated by a system of standardized morphometry, pairwise removal of allometric variance, and canonical discriminant functions. A mitochondrial DNA fragment including the cytochrome b gene was sequenced in 148 samples from basically the same range. In the more difficult F. pratensis vs. F. lugubris case, the phenotypic system correctly determined 99.6% of all nest samples, and 95.1% with p < 0.05. In all other pairwise species discriminations any nest sample was correctly determined with p < 0.01, and three samples with hybrids F. rufa x lugubris were identified. At four localities in the Pyrenees and the Urals, 9 samples with F. pratensis phenotypes (7 of them ideal) but F. lugubris mtDNA haplotypes could be identified, resulting in 14.5% of phenotype/haplotype mismatches. A local dominance of this mismatch combination was observed at one Pyrenean and one Ural locality. There was no indication of an F. pratensis haplotype associated with an F. lugubris phenotype. One ideal F. polyctena phenotype was associated with an F. aquilonia haplotype in a sample from the Urals, and one ideal F. aquilonia phenotype was combined with an F. lugubris haplotype in a sample from central Siberia, resulting in overall phenotype/haplotype mismatch frequencies of 12.5% and 11.1%, respectively. We conclude that all these samples cannot represent actual F. hybrids but are the result of hybridizations in the past followed by unidirectional purging of the nuclear genome. Whether this process of purging worked very fast or over longer periods of population history, and whether or not it was complete or incomplete, cannot be assessed from the available information. These facts of hybridizing in two thirds of the W Palaearctic wood ant species, of extreme regional hybrid frequencies (up to 26%), of unidirectional purging of nDNA associated with mismatching mtDNA haplotypes, and of occasional achievement of local dominance of these mismatch combinations, may serve as urgent warning not to perform isolated mtDNA phylogenetic studies without a geographically and locally wide sampling basis and without control by nDNA information or reliable phenotypic determination. The latter two systems definitely have superior significance when conflicts with mtDNA indications arise. (C) 2004 Elsevier GmbH. All rights reserved.