Journal
of Heredity, In press
Mapping a Cave Fish Genome:
Polygenic Systems and
Regressive Evolution
by
Richard
Borowsky1 and Horst Wilkens2
1 Cave
Biology Research Group, Department of Biology, 1009 Main, New York University,
Washington Square, NY 10003 USA. Tel:
(212) 998-8260; FAX (212) 995-4015.
E-mail: rb4@scires.nyu.edu.
2 Zoology Institute and
Acknowledgments: We thank M. McClelland and J. Welsh for
support and advice during the early stages of this study. Technical help was provided by B. Andiak, J.
Khlevner, and J. Starobinets. This study
was partly funded by a Research Challenge Fund grant (NYU) to RB.
Running Title: Mapping a Cave Fish Genome
We
used RAPD fingerprinting to generate anonymous DNA markers in the fish Astyanax mexicanus, a species with both
surface and cave populations. Surface
individuals are eyed and pigmented; troglobitic forms are blind and
depigmented. We hybridized surface fish
and Pachon population cave fish and produced a RAPD genomic map 1064 cM in
length (about half the total length of the genome) that was used to screen for
quantitative trait loci (QTL) for troglomorphic traits. Three QTL for reduced eye size, two for
decreased numbers of melanophores, two for condition factor and the locus for
the unifactorial trait, albinism, were mapped.
These factors account for an average of 46% of the variance of these
traits in the backcross. The results are
the first direct demonstration that troglomorphic changes in this population
are multifactorial. Two closely linked
pairs of QTL were found. Each consisted
of a regressive and a constructive trait QTL.
These close linkages are unlikely to have occurred by chance (P <
0.05 for each) and suggest that troglomorphic evolution might be facilitated by
pleiotropy or by genetic hitchhiking.
Introduction
Regressive
evolution is the reduction of a trait within a lineage over time. Common examples include loss of vision in
obligatory cave organisms (troglobites) and metabolic or structural
simplifications in parasites. Although
we generally think of regressive evolution as a minor issue in biology, its
effects are widespread. Without
regressive evolution to prune the phenotype, all species would be encumbered by
billion-year-long lists of superannuated traits.
The
changes that occur in regressive evolution are no different than those that
occur in constructive evolution: allelic frequencies and character states alter
over time. The problem is in explaining
the process. As
Troglobitic
fishes are classic examples of regressive evolution and were cited by
There
are two principal competing hypotheses to account for regressive evolution,
illustrated here by loss of visual function in troglobites. The first is that selection favors eye loss,
perhaps for reasons of organismal or neural processing economy. An alternative is that the genes controlling
the development of eyes become effectively neutral with the relaxation of
selective constraints and are free to accumulate mutations impairing their
function. In essence, the positions are selection versus neutralism. These basic possibilities and numerous
variants have been discussed with reference to the evolution of troglobites
(Culver and Wilkens, 2000).
Troglobites
also exhibit convergence for constructive changes, like increases in sensory
modalities other than vision, or increased metabolic efficiency. In contrast to regression of eyes or
pigmentation, constructive changes are more easily accounted for by natural
selection. In spite of this difference
it is important to consider both regressive and constructive changes together,
for a fuller understanding of the cave adaptation process (Culver et al. 1995,
pp. 25-26). Because traits evolve in the
context of changes in other traits it is important to understand trait
correlations in order to describe evolutionary changes adequately and to
construct testable hypotheses about mechanisms.
One
previously unexplored approach to the problems of regressive evolution and
troglobitic evolution is genetic linkage mapping. Linkage mapping and subsequent quantitative
trait loci (QTL) analysis is a direct approach to the enumeration of the genes
involved in character evolution, and a plausible first step in their eventual
isolation and cloning. Selection, as
opposed to drift, predicts patterns in the distribution of eye loss genes
within the genome, within developmental pathways, or in the types of nucleotide
substitutions involved in alterations of function. Identifying the genes involved, therefore,
will aid in testing alternative hypotheses.
In addition, linkage data allow the detection of correlations between
loci in genomic position.
The
Mexican Tetra (Astyanax mexicanus
(Filippi, 1853) (= A. fasciatus (Cuvier,
1819)) is particularly useful for studying the genetics of cave adaptation in
fishes because it has both cave and surface forms that are fully
interfertile. Pachon cave in
northeastern
Here
we test the hypothesis of multigenic inheritance directly by mapping. Using RAPDs, we produced a partial genomic
map to enumerate and identify quantitative trait loci (QTL) for troglomorphic
traits in the Pachon cave population. We
mapped QTL for eye size, pigmentation loss, and "condition factor," a
surrogate for metabolic efficiency. The
study was small in scale, but the results provide the first direct evidence
that troglomorphic traits are multifactorial.
Details of the QTL-linked RAPD markers are given elsewhere (Markers.htm).
Material and Methods
An
F1 hybrid female (Pachon X Epigean) was mated to a Pachon male to
produce a backcross progeny (BC, N = 111) for analysis. BC animals were raised in the light in
community tanks to the age of six to eight months, prior to sacrifice. At sacrifice, BC progeny were preserved in
70% ethanol, weighed to the nearest mg, and their standard lengths determined
to the nearest 0.1 mm. DNA for RAPD
analysis was prepared by proteinase k digestion of fin tissue, followed by
phenol and chloroform:IAA extractions.
RAPDs: RAPD amplifications and scoring was performed
after published procedures (Borowsky et al. 1995; Kazianis et al. 1996)
modified as follows: we used short
primers (10-11 mers) in primer pair combinations and employed Stoffel fragment
Taq for amplification. Amplimers were
labeled by direct incorporation of 33P, separated on 4%
polyacrylamide sequencing gels under denaturing conditions, and visualized by
autoradiography. Bands were scored on
the autorads by eye. Marker loci with
segregation patterns more probably 3:1 than 1:1 were discarded from
analysis. Seventy percent of the
segregating RAPD markers were suitable for mapping. An average of 18.3 + 1.8 RAPD markers with
approximate 1:1 segregation ratios were typed with each primer combination.
Quantitative
Traits: Eye size: The eye region was
dissected carefully under 10X magnification on the right side of the animal and
the eyeball rudiment exposed. The eyeball
diameter was measured optically and the residual of its size, after regression
on standard length of the fish, was the quantitative measure of eye development
used in all analyses. Melanophores: Progeny were scored as albino or non-albino
under 25X magnification. In non-albino
fish, melanophores were counted on standardized areas in three locations on the
back, near the dorsal fin. Condition
Factor: The measure of condition factor (CF) was the residual of logarithm
of weight after regression of log weight on log length.
Linkage
and QTL Analyses: Map Manager XP
(Manly 1998) was used to identify and build linkage groups (significance level
= 0.001). Each quantitative trait was
analyzed using Map Manager QT (ver. 2.6, Manly and Cudmore 1997; Manly 1998;
Manly and Olsen 1999) and its associations with mapped RAPD markers were tested
by regression analysis (Links Report module).
Markers having apparent associations with the trait were examined
serially by interval mapping, starting with the marker with the strongest
association (Interval Mapping module).
For the first association examined, no cofactors were used. For subsequent associations analyzed, markers
with previously established significance were included as cofactors. The serial examination was terminated at the
first non-significant association.
Significance levels of QTL were determined by permuting the data
(Churchill and Doerge 1994), as implemented by the Permutation Test module. Permutations for the first association
examined were made without control for variation at other marker loci. Subsequent permutations were made controlling
for loci previously found to be significant.
QTL
Contributions to Trait Variance:
Estimates of the variance contributions of individual QTL are biased
upwards when the power to detect QTL is low (Beavis 1994). We corrected for this bias for each QTL (from
Fig 15.8 of Lynch and Walsh 1998, pp. 474-476) based on the power of its being
detected, estimated from initial estimates of its effect, the number of
informative progeny, and estimates of dominance (from published data, Wilkens
1985).
Results
and Discussion
One
hundred and forty-two RAPD loci with segregation ratios close to 1:1 were
scored. Of these, 81 coalesced into 27
distinct linkage groups using the standard criterion of LOD > 3.0. The total size of the partial map is 1064
cM. Astyanax
mexicanus has a haploid number of 25 and, based only on comparison to
zebrafish and platyfish, an expected map size of about 2000 cM (Zebrafish =
2200 cM, Postlethwait et al. 1994; Platyfish = 1800 cM, Morizot et al. 1991).
We
tested for correlations between phenotypic traits within the BC progeny. Condition factor was weakly correlated with
melanophore number (r = +0.21, df = 78, P = 0.05) and more strongly with eye
size (r = +0.43, df = 107, P = 0.01).
Trait correlations could reflect developmental constraints, pleiotropy,
or coincident locations of QTL. The
positive correlation between eye size and CF could also reflect the advantage
of larger eyes in social interactions and competition for food, since the
fishes were raised in community tanks and in the light.
QTL
analysis, as outlined above, revealed seven statistically significant, putative
QTL: three for eye size, two for melanophore number, and two for CF. In addition to the mapped QTL, the locus for
albinism was found to be part of linkage group lg26. The QTL detected account for 27% of the
variance in eye size, 39% in pigmentation, and 19% in CF (Table 1). The single locus for albinism accounts for
100% of the variance in that trait.
Thus, of the four troglomorphic traits studied, the factors found
account for an average of 46% of the variation.
For the QTL alone, this figure is 28%.
In accord with our estimate that the map is half complete, these results
are consistent with previous quantitative genetic estimates of about six
factors for regression of eye size
The
data reveal two pairs of linked QTL, in linkage groups lg25 and lg4. In lg25 a factor for pigmentation is linked
to another for CF, near RAPD locus C8a.
In lg4 a factor for eye size is linked to another for CF, between RAPD
loci I7a and AB6a. Based on the number
of linkage groups and their size, the chance probability of either of these
co-occurrences is less than 5%. The
chance probability of two is much less than 1%.
It is of interest that in both cases, the linkage is between QTL for
regressive and constructive traits.
Determining if these co-occurrences are common phenomena or simply
intriguing coincidences must await further work.
Although
these apparent linkages might be spurious and derive from the positive
correlations between traits seen in the BC progeny, there is little evidence
for this interpretation. In lg25 the
linkage is between CF and pigmentation, and these traits not significantly
correlated in the progeny (r2 = 0.05). In lg4 the linkage is between CF and eye size
and, while the overall correlation of these traits is stronger (r2 =
0.21), the QTL for eye size in lg4 is the weakest of the three detected (5% of
the variance versus 8 and 14%). The
other two eye size QTL show no apparent linkage to QTL for CF.
Close
linkage of regressive and constructive QTL might arise from pleiotropy. For example, if a gene for eye size also
influences metabolism, selection on metabolism could also affect the character
state of eyes. Pleiotropy is looked upon
as a potential means by which seemingly “neutral” changes could be driven by
natural selection. Its role in troglomorphic evolution has been
reviewed by Culver (1982) and by various authors in Culver (1985).
There
is also the possibility that linkage reflects a history of genetic
hitchhiking. If, after invasion of the
subterranean environment, new mutations occur at random in loci affecting both
regressive and constructive traits, some of these might be linked. Given linkage, hitchhiking might then
facilitate the increase in frequency of those regressive mutants linked to
selected constructive mutants. This
hypothesis remains to be tested; like pleiotropy, it predicts the close linkage
of troglomorphic factors in troglobite genomes.
The two hypotheses can be distinguished because pleiotropy predicts
identity of any troglomorphic factors, while hitchhiking predicts close linkage
of regressive and constructive QTL.
Hitchhiking is a novel and potentially important mechanism for
regressive evolution. If true, it would
allow for faster evolutionary change than possible through drift alone. While hitchhiking and pleiotropy may play
roles in regressive evolution, neither mechanism excludes an important role for
drift of neutral mutations under conditions of relaxed selection, as outlined
in the introduction.
Genetic
analysis of a single family can reveal only a fraction of the variability in an
outbred population. Analysis of other
Pachon lines or cave populations of this species would presumably yield
different estimates of the total numbers and relative magnitudes of the troglomorphy
QTL. These studies are in progress. The present work, however, shows clearly that
troglomorphic traits are multigenic and that the QTL exhibit significant
linkages in the line analyzed.
References
Beavis WD, 1994. The power and
deceit of QTL experiments: lessons from comparative QTL studies. In: 49th Annual Corn and Sorghum Research
Conference.
Borowsky R, McClelland M, Cheng R, and Welsh J., 1995. Arbitrarily primed DNA fingerprinting for
phylogenetic reconstruction in vertebrates.
Mol Biol Evol 12:1022‑1032.
Culver, DC and Wilkens, H, 2000.
Critical review of the relevant theories of the evolution of
subterranean animals. In Ecosystems of
the World, Subterranean Ecosystems, Volume 30, (Wilkens, H, Culver, DC, and
Humphreys, WF, eds).
Darwin, C, 1859. On the Origin of
Species.
Kazianis S, Morizot DC, McEntire BB, Nairn RS and Borowsky RL,
1996. Genetic mapping in Xiphophorus hybrid fish: assignment of 43 AP-PCR/RAPD and isozyme
markers to multipoint linkage groups.
Genome Research 6:280-289.
Lynch M and Walsh B, 1998.
Genetics and Analysis of Quantitative Traits.
Manly KF, 1998. User's Manual for
Map Manager Classic and Map Manager QT.
[http://mcbio.med.buffalo.edu/mapmgr.html.]
Manly KF and Cudmore RH Jr, 1997.
Map Manager QT, Software for mapping quantitative trait loci. Absts of
the 11th Int. Mouse Genome Conference,
Manly KF and Olsen JM, 1999.
Overview of QTL Mapping Software and Introduction to Map Manager QT.
Mammalian Genome 10:327-334.
Mitchell RW, Russell WH and Elliott WR, 1977. Mexican eyeless characin fishes, genus Astyanax: Environment, distribution and evolution. Special Publ. Mus. Texas Tech. Univ. 12:1-89, 21 Figs.
Morizot DC, Slaugenhaupt SA, Kallman KD, and Chakravarti A, 1991. Genetic linkage map of fishes of the genus Xiphophorus (Teleostei:
Poeciliidae). Genetics 127:399-410.
Postlethwait JH, Johnson SL, Midson CN,
Talbot WS, Gates M, Ballinger EW, Africa D, Andrews R, Carl T, Eisen JS,
Horn S, Killel CB, Hutchinson M, Johnson
M and Rodriguez A, 1994. A genetic linkage map for the zebrafish. Science 264:699-703.
Weber A, Proudlove GS, Parzefall J, Wilkens H and Nalband TT, 1998.
Pisces (Teleostei). In: Encyclopaedia Biospeologica, Tome II. (C. Juberthie & V. Decu, eds.). Moulis, France:
Société de Biospéologie; 1179-1190
Wilkens H, 1985. The evolution of
polygenic systems, studied on epigean and cave populations of Astyanax fasciatus (Characidae, Pisces).
Natl. Speleol. Soc. Bull 47(2):101-108.
Wilkens H, 1988. Evolution and
genetics of epigean and cave Astyanax
fasciatus (Characidae, Pisces); support for the neutral mutation theory. Evol. Biol. 23:271-367.
Table 1. QTL for eye size, melanophore
number and condition factor detected in the experiment. Var: estimate of percentage of variance in
the trait accounted for by the QTL, after adjustment for estimation bias and
the effects of other QTL ("control").
P values for statistical significance are "experiment-wide"
and were determined by permutation tests.
Trait Var P < Control QTL
Linkage
Eye Size
lg17 14% .05 None
lg27 8%
.05 lg17
lg4 5%
.05 lg17, lg27
Pigmentation
lg25 28% .05 None
lg3 11% .05 lg25
Condition
lg25 12%
.001 None
lg4
7%
.001 lg25