polymorphism for autosomal fusion combinations, resulting in rings of autosomes of various sizes at male meiosis. One of these localities has been surveyed in suffi cient detail to reveal it as the most chromosomally variable population identifi ed to date of this extremely variable species. At this location in the Ovens Valley, six ring forms have been identifi ed, (RIV1, RXII, RXIV, RXVI, RIV+RIV and RVI+RIV). At the other location near Tumbarumba, a RIV2 form was found that may or may not be the same as RIV1. Due to the lack of data from the Tumbarumba locality, most of the following discussion will focus on the Ovens Valley. Within the Ovens Valley, the various rings have highly overlapping distributions within a 20 km2 area, and are generally found
mixed with each other and with mII individuals at sites and in colonies. Around 50% of the individuals found at ring sites carry the mII karyomorph, and all individuals caught in the surrounding areas carry this confi guration. Interestingly, all ring forms share their X fusion confi guration with these mII populations.
Characterizing the polymorphism
The ring carrying forms of D. cancerides reported here are qualitatively
different to all previously reported chromosomal confi gurations of this species. All other chromosome multiples observed include X-autosome fusions (Sharp & Rowell 2007), and are characteristic of karyomorphs that are widespread and fi xed, with mixed sites and colonies found only in narrow contact zones. The rings, by contrast, are clearly not sex-linked, and have only been found in two small localities, and in highly mixed populations. The mixing of karyotypes at sites as well as in colonies, in addition to the lack of sex-linkage of multiples precludes the possibility that these rings represent a
fi xed system of any sort.
Floating or balanced polymorphism could explain the occurrence of rings in these populations, however these explanations are not considered likely. Rowell (1990) surveyed 20 variable allozyme loci in this species across a large geographic area (thousands of kilometres and several karyomorphs), and found only minor allele frequency differences, suggesting that this species is highly mobile and outbred (reviewed in Chapter 3). The area of the Ovens Valley in which rings have been observed is only 20 km2, and not geographically isolated, or environmentally distinct.
Rather it is in the middle of a long thin habitat strip, close to junctions with several other such strips (Figure 2 B). Therefore, local differentiation of spiders at this location is unlikely, suggesting that this polymorphism is not fl oating or balanced. Indeed, the rings of chromosomes described in this chapter seem unlikely to represent polymorphism within the mII population.
Therefore, the autosomal rings are most likely to be produced by inter- population processes of differentiation and hybridization. Due to the lack of
chromosome specifi c markers for this species, it is possible that the mII range is occupied by a mosaic of populations homozygous for different fusion combinations. Hybridization between these different mII types would result in rings of autosomes at both male and female meiosis. The size of the rings would depend on the number of fusion differences between the hybridizing populations. Sex chromosomes are predicted to only be involved in multiples of up to four (all X’s) in females, and only if the
hybridizing populations have different X fusion combinations.
Describing the Hybrid Zone
If the Ovens Valley locality is interpreted as a hybrid zone, the parental
populations must occur on either side of the zone, each occupying one end of the valley. Three aspects of the Ovens Valley hybrid zone are immediately informative regarding its structure; at around 9.5 km wide it is narrow for a large mobile outbred spider, it contains nearly coincident clines for many chromosomal fusion differences, and it is set amongst many geographic barriers. As discussed in the introduction, hybrid zones can have a variety of structures. Some of the models clearly do not fi t the available data well, such as neutral, positively heterotic or environmentally maintained clines. This is due to the narrowness of the zone, the coincidence of the clines, the presence of 50% parental types, and lack of any apparent environmental transition (Barton & Hewitt 1985; Harrison 1993). Very recent contact would explain the narrow zone and coincident clines, but would not explain the extent of diversity observed within the zone, which is greater than that expected in a simple two way hybrid zone (discussed below). The tension zone model is supported by the data however, as it can explain the width, structure, and geographic context of the hybrid zone. However, the diversity observed within the Ovens Valley suggests that this tension zone is not a simple one.
When two populations fi xed for different Robertsonian fusions hybridize, the hybrid zone is expected to contain only those fusions observed in the pure populations. The Ovens Valley hybrid zone, however, must contain at least three rearrangements in excess of those that could be carried by two fi xed parental populations. There are three ways in which this extra polymorphism could be interpreted:
- Four differentiated mII populations are required if all six ring karyotypes are interpreted as products of primary hybridization (Figure 4 A).
- Three differentiated populations are required if some karyotypes are assumed to represent backcross or later generation hybrids (Figure 4 B).
- Two differentiated populations could account for this diversity if a large ring is assumed to represent the primary hybrid product, and the remaining karyotypes were produced by a minimum of three WART type A rearrangements (or equivalent) within the zone (Piálek et al. 2005) (Figure 4 C).
The fi rst two options are plausible given the complex geography of the area, especially the intersection of several valleys in the vicinity of the observed polymorphism (Figure 2 B). The zone is some distance from these intersections,
however, and the valleys may not be large enough to support a differentiated population each. Both of these models imply that independent hybrid zones have moved so as to become coincident, which is consistent with tension zone dynamics. The third option requires WART mutations to occur within the hybrid zone. Such rearrangements have been observed in hybrid zones of other species, where they are attributed to hybrid dysgenesis (Crocker & Cattanach 1981; Hauffe & Piálek 1997; Piálek et al. 2001; Piálek et al. 2005). Hybridization experiments described in chapter 4 have been conducted to explore the plausibility of these models.
Fig. 4 Schematic representation of three hypotheses to explain the excess diversity observed in the Ovens Valley. Circles represent differentiated populations, arrows represent hybridization producing the karyotype indicated. (a) 4 population model, where each karyotype results from an F1 cross.
(b) 3 population model, where the two largest rings are F1 products, and the remaining karyotypes are produced by fusions from all three populations.
(c) 2 population model, where the largest ring is the F1 product, and the remaining rings indicate that at least three WART type A mutations are present within the hybrid zone.
(d) Schematic representation of X chromosome fusion differences between populations, if they exist, and the expected result in hybrids.
(a) (b) (c)
Although males in most other populations of this species carry large sex-linked chains of chromosomes, the capacity to accurately segregate them may have been selected for in those populations (Grüetzner et al. 2006; Sharp & Rowell 2007). This suggests that mII populations may not share this adaptation. Moreover, females of this species are not thought to carry chromosome multiples in any other population, and so are not likely to have undergone selection for this trait in any population of the species. In addition to carrying the autosomal rings described here, if X fusion differences exist between these mII forms, the hybrid females are expected to carry a chain of all four X chromosomes (Figure 4 D). These considerations suggest that malsegregation, especially in females, may be driving the selection against large multiple carrying hybrids, and resulting in tension zone dynamics. Mechanisms of reproductive isolation will be discussed in Chapter 4.
Although the above discussion has focused on the Ovens Valley, it is likely that the Tumbarumba site represents a similar system. Unfortunately this polymorphic population was discovered too late to investigate it in any detail, but preliminary data show similarities with the Ovens Valley. RIV2 and mII individuals were found mixed within the sampled colony, and the site is surrounded by other mII sites. The RIV2 could indicate that the differentiated populations here only differ by two fusions, or that this is a part of a larger hybrid zone. This could even be a part of the same hybrid zone as that found in the Ovens Valley. Geographically, Tumbarumba is less striking than, but not dissimilar to the Ovens Valley, in that it provides narrow strips and small patches of habitat restricted to valleys amongst the mountains of the GDR at the edge of the mII distribution. Historically, these mountainous areas may have provided multiple refuges during glacial maxima, from which populations could have expanded and met allowing hybridization during inter-glacial periods (Garrick et al. 2004; Sunnucks et al. 2006). This interplay of climate and geography is well known to result in the differentiation and later hybridization of isolated lineages (Avise 2000; Hewitt 2000).
Irrespective of whether divergence was allopatric or parapatric, depending on when they diverged relative to the onset of fusion saturation, these populations could have evolved in three different ways. The divergent mII populations could have independently fi xed the characteristic mII karyotype of homozygous autosomal fusions, and telocentric and metacentric X’s. Alternatively, the different mII populations could share a common mII ancestor, from which one or both have diverged via WART mutations. It is also possible that these populations diverged during fusion saturation, and so may share the earliest fusions but not later ones.
Presently, the mII population has the largest distribution of all the known karyomorphs of D. cancerides. Given the width of the Ovens Valley zone and the sparsity of sampling throughout much of the range of the mII (Figure 4, Sharp & Rowell 2007), the chances of stumbling on such zones are small. Given that two zones may have been found by this study, such diversity could be common within the mII
karyomorph. The presence of multiple populations fi xed for different homozygous fusions would provide an interesting counterpoint to the diversity of chain carrying karyomorphs of this species. The contrast between these different types of fusion saturated population may prove to be highly informative regarding the impact of chromosome fusions, and the genetics of post-mating isolation in this species.
If the polymorphism observed in the Ovens Valley and near Tumbarumba is a result of hybridization between differentiated mII forms, then this extra diversity needs to be considered in the broader context of the species. Sharp and Rowell (2007) noted the parapatric distributions of all single chain karyomorphs, and the two groups of similarly distributed double chain karyomorphs. With the addition of contiguous yet differentiated mII populations, a clear pattern emerges; karyotypic diversity is not distributed randomly across the landscape in this species, but is rather grouped according to karyomorph type. As karyomorph type is determined by X chromosome fusion confi guration, this observation suggests that X fusion confi guration has been an important factor in the evolution of this system.
In conclusion, the diversity of ring forms observed in the Ovens Valley is most parsimoniously explained as resulting from hybridization between two or more differentiated mII forms. Selection against individuals with large chromosome rings, possibly due to elevated rates of malsegregation at meiosis, has caused tension zone dynamics to develop. It is possible that the second locality at which rings were found near Tumbarumba shares similar attributes. With the addition of chromosome rings and differentiated mII populations to the list of chromosomal forms of this spider,
D. cancerides is now undoubtedly the most chromosomally diverse and complex
species known.