Patterns in chromosome numbers
The main result is that all the Nymphalidae studied here have the lepidopteran modal of n=29–31 as the most common chromosome numbers. This strengthens the hypothesis put forward in the earlier papers of this series that the n=21 of the genus Heliconius (Brown et al. 1992), n=28 of Morphini, n=29 of Brassolini, n=29 of other Satyrinae (Brown et al. 2007) and the variable numbers with n=14 as the modal one of the Ithomiini (Brown et al. 2004), the quite divergent numbers of the tribes Anaeini and Preponini of Charaxinae and the multitude seen in the satyrines (Brown et al. 2007) are all derived from the lepidopteran modal of n=29–31.
White (1978, p. 74) pointed out that the mechanisms for reducing the chromosome number below n=29–31 have been far more efficient than ones leading to increases above it. There will, however, be a problem with telomeres, which have to be silenced lest they interfere with the achiasmatic meiosis of lepidopteran females. de Lesse (1967a), Suomalainen and Brown (1984), Brown et al. (2007), and Lorkovic (1990) have pointed out that there seems to be a process of concerted fusion that involves all chromosomes. It would explain why chromosome numbers that are about one half of the modal number are frequently seen among nymphalids. Nordenskiöld (1951) has observed a similar concerted halving of chromosome numbers in the plant genus Luzula that has a diffuse kinetochore structure resembling that of lepidopterans. In fact, the Ithomiini have a strong modal at n=14–16 and very few forms with n=29–31 (Brown et al. 2004). Tellervo and Danainae represent evidently the plesiomorphic condition. Accordingly, Ithomiini descend from forms that already have had their chromosome set halved. Our results show that n=14–15 has evidently arisen repeatedly from n=29–31 in different branches of nymphalid phylogeny, often without apparent intermediates. Temenis sp. shows, in addition, that n=14 may experience one more round of concerted fusion with n=7 as a result. The chromosome numbers of neotropical Nymphalidae show both stable numbers in certain taxa and apparently irregular numbers in others. Low, again stable or unstable, numbers may characterize entire subfamilies, while numbers higher than n=31 are relatively infrequent.
Numbers between the modal numbers and even fractions or multiples of them are harder to explain. Seiler (1925) observed that fragmentation gives rise to different numbers. The nearly holocentric nature of lepidopteran chromosomes that makes fragmentation feasible was not known then and he was unable to give an adequate explanation to what he saw.
In the list given by Robinson (1971), p. 589) all groups of lepidopterans other than lycaenids have a modal number of n=29–31. We have here observed a set of exceptions. If we project the chromosome numbers of Neotropical Nymphalids onto the phylogeny of Wahlberg et al. (2003) we observe the pattern seen in Fig. 1.
Libytheinae, the proposed sister group of all other Nymphalidae, have n=31–32 (this study); Danaini have n=30 as the modal number (Brown et al. 2004) and their sister group Ithomiini has numbers ranging from n=5 to 120 with a peak at n=14 (Brown et al. 2004). The two tribes of Charaxinae have quite different distributions of numbers (Brown et al. 2007). The Anaeini have a peak at n=31, followed by a descending series with many numbers in n=26 through 30 and a minor peak at n=21, all the way to n=6, while the Preponini have a peak at n=12 with a single number above n=19.
Among Satyrinae (Brown et al. 2007), the Morphini have a peak at n=28, while the Brassolini have an equally distinctive peak at n=29. The other tribes of Satyrinae have a weak modal of n=29 (Brown et al. 2007), starting with the basal groups with n=29 relatively common, followed with the first clade of Pronophilina (Peña et al. 2006) which has n=29 fixed, followed by the second clade that has an uneven distribution resembling the one of Euptychiina that has all numbers between n=6 and n=31 present at least once, with n=13 as the most common one but without any clear modal number.
The next clade is made up of Heliconiinae (including Argynnini, Heliconiini and Acraeini) and Limenitidinae. The samples for Argynnini and Acraeini are small but they seem to be almost fixed for n=31, while the Heliconiini show an evolution away from n=29–31 to a new modal number (Suomalainen and Brown 1984; Brown et al. 1992). The basal genus, Philaethria, is made up of species with n=12 up to n=88. Two of the species have n=29, which is also found in the other primitive genus, Podotricha, which again has a species with n=9. The next clades are genera with n=31, followed through a series (Neruda, Laparus) that go down from n=32 to n=19–21. The large genus Heliconius is, with the exception of the most derived, pupal mating clade, stabilized into n=21. With the exception of H. hewitsoni that has n=21, the pupal mating species represent an ascending series up to n=62.
The Limenitidinae have a strong modal n=30 (this study) and the Nymphalinae an even stronger n=31 and this is also the case in the small sample of Apaturinae that we have in this study. Finally the Biblidinae have a strong modal number of n=31 (31 out of 80 taxa) followed with n=30 and n=29. We may also note that numbers about half that are relatively common with a total of 13 counts between n=13 and n=16.
Wahlberg (2006) has estimated that the basal groups of Nymphalinae diverged at about the K/T boundary, i.e. about 65 million years ago and the age of Nymphalidae is older than 70 million years. This gives us a handle to assess whether the modal number represents a primitive condition rather than an equilibrium karyotype in the sense of White (1973) to which the chromosome number will return after having been perturbed. These two concepts need, of course, not be mutually exclusive. Many of the basal subfamilies and tribes of nymphalids have the modal n=29–31. Again, Brown et al. (2004) argued that the Ithomiini evidently descend from an ancestor that has already had the chromosome number halved to about n=14–15. The minor peak seen at about n=7–8 seen among them results from further concerted fusion of all chromosomes. Given that the nymphalid subfamilies and tribes have diverged from each other tens of millions of years ago (Wahlberg 2006), it is unlikely that there is selection that will restore n=31 once it has been perturbed. Evidently n=29–31 represents the ancestral condition of Nymphalidae.
Chromosomes in speciation
The pairing and segregation of chromosomes at meiosis is a component of fertility selection, a constituent of postzygotic isolation and speciation (Dobzhansky 1968). Chromosome number changes have been shown to give rise to reinforcement in satyrine speciation (Lorkovic 1958). The factors underlying reinforcement are being studied with molecular methods: Lukhtanov and Dantchenko (2002) and Lukhtanov et al. (2005) have studied the behavior at meiosis of the chromosomes of Lycaenidae, in particular species with extremely high chromosome numbers, again, Wolf et al. (1997) have observed meiosis in lepidopterans with low chromosome numbers. We have here an acraeine that may have an extremely high chromosome number (n=150) and we have reported both very low and as high or higher numbers also in other Nymphalids (Brown et al. 1992, 2004, 2007).
de Lesse (1966, 1967b, 1968) and de Lesse and Condamin (1962, 1965) have published chromosome numbers of African representatives of the neotropical Nymphalid groups that we report here. In general the African Nymphalinae have n=31, the Limenitidinae have a peak at n=30, like in South America; the few Biblidinae at n=31 and the single Libythea species n=31. The sample of 18 species of African Acraeinae has a peak at n=31 but nine species have numbers higher than that and one species has n=137, comparable to Abananote erinome of Bolivia (this paper). Francini (1989) has, in our opinion convincingly, demonstrated that in the study of chromosome numbers of Acraeini one should only look at early prepupae; all other stages of development yield nonreproducible results. Nevertheless, given the expertise and perspicuity of de Lesse, we think that his two similar sets of observations carry weight and should not be disregarded without rechecking.
Robinson (1971) has made an extensive compilation of worldwide chromosome numbers of lepidopterans. Virtually every group of Nymphalids from North America, Europe, Asia and Australia has a modal number at n=30–31. The only exceptions are the African Charaxines that have a peak at n=25–26 and the African Satyrines that have a modal of n=28 (Brown et al. 2007). With these two exceptions, all other Nymphalid groups with modal chromosome numbers different from the general lepidopteran modal of n=29–31 are Neotropical. Lorkovic (1990) pointed out, on the basis of a limited material, that the tropical Satyrinae tend to have lower chromosome numbers than the n=29 that characterizes them in the rest of the world. Given that there is no crossing over in the females, chromosome numbers may represent a way to adapt to tropical conditions through adjusting recombination. We doubt that the hypothesis of Lorkovic (1990) needs to be discussed further; e.g. most tropical and temperate Drosophila species lack recombination in the heterogametic sex but all have low chromosome numbers.
Modes of selection
Dobzhansky (1950) argued that in the physically mild environments of the tropics the interrelationships between competing and symbiotic species or biotic interactions in general are the agents of natural selection, while in the harsher environments of the temperate zone and beyond physical factors drive evolution. Janz et al. (2006) have shown that host plant diversification drives evolution in Nymphalidae. Mimicry is another obvious case point. Among the groups discussed here, Danainae, several Charaxinae and Satyrinae, Heliconiinae, many Nymphalinae and some Biblidinae are involved in mimicry rings either as movers or followers. Ithomiini, Charaxinae, Satyrinae and some Heliconiini are characterized through chromosomal instability. The large genus Heliconius stands out among these mimetic forms, as it is almost fixed for the new modal n=21. Gilbert (2003) shows that there is extensive between species mating that explains the striking convergence of Müllerian mimetic patterns across the genus.
Hybridization is a potential mechanism that could give rise to chromosomal changes within and among closely related species (Mallet 2007). Our material (Brown et al. 1992) examined includes certain Heliconius hybrids found in nature. Some of them are hybrids between different morphs of one heliconiine species (Eueides tales tales×E. tales pythagoras, Heliconius clysonymus clysonymus×H. clysonymus hygiana, H. sapho sapho×H. sapho chocoensis). One of them (H. cydno×H. melpomene) is a species hybrid. In all these cases the parentals of the hybrid have the same chromosome number (even though the chromosome number of E. tales pythagoras is unknown). The chromosomes seem to pair in general in the normal fashion in the hybrid meiosis, indicating that the hybrids may well be fertile. In only one of the hybrids studied by us (E. tales tales×E. tales pythagoras) all chromosomes do not pair in a part of the cells. The subspecies (morphs) and closely related species of Heliconius in general have the same chromosome number. Consequently their hybrids lack the obstacle for fertility conferred by different chromosome numbers of parent species. There is, indeed, good evidence for homoploid speciation in Heliconius: H. heurippa has originated as a hybrid between H. melpomene and H. cydno (Mavárez et al. 2006). Here the hybrid phenotype isolates the hybrids from the parent species. Consequently, between species hybridization would be a force that stabilizes the chromosome numbers. As mentioned in the introduction, chromosome number change is in general expected to give rise to reproductive isolation and reinforcement (Lukhtanov et al. 2005; Kandul et al. 2007).
Another case in point is sexual selection. The males of the pupal mating clade of Heliconius mate with the female before she has eclosed from the pupa and consequently sexual selection is relaxed. Gilbert (2003) has suggested that sexual selection is a conservative force in the evolution of Heliconius. Our chromosomal results show that once sexual selection is removed, chromosome numbers become unstable, which is certainly compatible with Gilbert's (2003) suggestion.
Wahlberg et al. (2005b) and Wahlberg (2006) have reconstructed the historical biogeography of the Nymphalinae. They conclude that the major clades have three centres of diversification, from which they have spread to the areas they now occupy. The pattern of chromosomal evolution that we have observed agrees with their conclusions: there have been several dispersal events from South America to other continents; again South America has received many taxa in particular from the Afrotropical and Nearctic regions. Interestingly, groups that may have invaded to the Neotropics from elsewhere (e.g. Argynnini and Melitaeini) seem to have retained the lepidopteran modal n=31, while at least some clades of Satyrinae, a putative Neotropical subfamily, have retained chromosomal instability and make use of it in speciation (Lorkovic 1958).