A method for phenetic analysis of karyotype data has been developed for Leishmania populations. Measurement of size difference between chromosomes recognized by a given DNA probe in different isolates led to the formulation of a Chromosome Size Difference Index (CSDI). The method was applied to phenetic analysis of 4 sets of chromosomes – each set being recognized by a different probe – in 37 L. (Viannia) peruviana isolates sampled along a North–South transect through the Peruvian Andes and, in 11 L. (V.) braziliensis isolates from the Amazonian forest (Peru, Bolivia and Brazil). Karyotype variability was better accounted for by CSDI than by a method based on disjunctive encoding of karyotype data. CSDI evidenced the nature of relationships between L. braziliensis and L. peruviana and it provided a coherent picture of geographical and genomic differentiation among parasite populations. The latter did cluster according to their geographical origin. L. braziliensis was found karyotypically more homogeneous than L. peruviana. Within L. peruviana, Northern populations were closer to L. braziliensis than to Southern L. peruviana populations. The validity of karyotypic populations, or karyodemes, was sustained.

The evolutionary significance of chromosome size polymorphism was explored in a representative panel of 26 Trypanosoma cruzi stocks. We tested a progressive model (aCSDI) assuming that the larger the size difference between homologous chromosomes, the more divergent the parasites are. This was contrasted with a non-progressive model (Jaccard's distance), in which any chromosome size difference has the same weight. ACSDI-based dendrograms were very similar to those built-up from multilocus enzyme electrophoresis (MLEE) and random amplified polymorphic DNA (RAPD) data: structuring in 2 major lineages (T. cruzi I and T. cruzi II) and 5 small subdivisions within T. cruzi II was identical, and branching was very similar. Furthermore, a significant correlation (P<0·001) was observed between aCSDI and phenetic distances calculated from MLEE and RAPD data. In contrast, analysis of chromosome size polymorphism with Jaccard's distance generated dendrograms with relatively long branches, causing most branching points to cluster close together, which generates statistically uncertain branching points. Our results thus support a model of progressive chromosome size-variation and show that despite an extensive polymorphism, chromosomal sizes constitute valuable characters for evolutionary analyses. Furthermore, our data are consistent with the clonal evolution model previously proposed for T. cruzi.