Copyright 0 1997 by the Genetics Society of America
Inbreeding Depression and Inferred Deleterious-Mutation
Parameters in Daphnia
Hong-Wen Deng and Michael Lynch
Department of Biology, University of Oregon, Eugene, Oregon 97403
Manuscript received October 16, 1996
Accepted for publication May 19, 1997
ABSTRACT
DENG and LYNCH recently proposed a method for estimating deleterious genomic mutation parameters
from changes in the mean and genetic variance of fitness traits upon inbreeding in outcrossing popula-
tions. Such observations are readily acquired in cyclical parthenogens. Selfing and life-table experiments
were performed for two such Daphnia populations. We observed a significant inbreeding depression
and an increase of genetic variance for all traits analyzed. DENC and LYNCH’S original procedures were
extended to estimate genomic mutation rate (U), mean dominance coefficient (A), mean selection
coefficient (3, and scaled genomic mutational variance (VJV,). On average, 5 ? and pm/V,
(A indicates an estimate) are 0.74, 0.30, 0.14 and 4.6E-4, respectjvely. For the true values, the Qand 6
are lower bounds, and ?and p,,,/ V, upper bounds. The present a Eand pJ Kare in general concordance
with earlier results. The discrepancy between the present Sand that from mutation-accumulation experi-
ments in Drosophila (-0.04) is discussed. It is shown that different reproductive modes do not affect
gene frequency at mutationselection equilibrium if mutational effects on fitness are multiplicative and
not completely recessive.
I
NBREEDING depression has been documented in
almost all organisms that have been examined
( CHARLESWORTH and CHARLESWORTH 1987; FALCONER
1989). The magnitude of inbreeding depression has
many implications in modern evolutionary theory, such
as the evolution of self-incompatibility systems in mon-
oecious plants (LANDE and SCHEMSKE 1985; SCHEMSKE
and LANDE 1985; CHARLESWORTH and CHARLESWORTH
1987) and the evolution of dispersal mechanisms for
inbreeding avoidance in animals (SHIELDS 1982). While
it is well known that inbreeding can cause a change of
both mean genotypic value and genetic variance (CROW
and KIMURA 1970; FALCONER 1989), most empirical and
theoretical work on inbreeding related subjects has
been concentrated on the change of the mean (6.J
LANDE and SCHEMSKE 1985; SCHEMSKE and LANDE 1985;
CHARLESWORTH and CHARLESWORTH 1987; FALCONER
1989; CHARLESWORTH et al. 1990). Study of the change
of genetic variance upon inbreeding may provide some
valuable information on evolutionary processes (ROB-
ERTSON 1952; COC:KERHAM 1984a,b; GOODNIGHT 1987,
1988; WILLIS and ORR 1993; SCHULTZ and WILLIS 1995),
such as an inflated additive genetic variance upon popu-
lation bottlenecks. Moreover, joint information on the
response of mean genotypic values and genetic vari-
ances to inbreeding can be used to estimate spontane-
ous deleterious genomic mutation parameters (DENG
and LYNCH 1996a), such as the genomic mutation rate
to mildly deleterious alleles ( U) .
Corresponding author: Hong-Wen Deng, Osteoporosis Research Cen-
ter, Creighton University, 601 N. 30th St., Suite 6787, Omaha, NE
68131. E-mail: deng@creighton.edu
Genetics 147: 147-155 (September, 1997)
Estimates of U are crucial to testing theories for the
evolution of sex (MULLER 1964; KONDRASHOV 1985,
1988; CHARLESWORTH 1990), mate choice (KONDRA-
SHOV 1988; CHARLESWORTH and CHARLESWORTH 1987;
IRKPATRICK and RYAN 1991), outbreeding mechanisms
(CHARLESWORTH and CHARLESWORTH 1987), diploidy
(KONDRASHOV and CROW 1991), and the accelerated
extinction rate of small populations (LYNCH and GA-
BRIEL 1990; LYNCH et al. 1993, 1995a,b). However, few
estimates are available (CROW and SIMMONS 1983; KON-
DRASHOV 1988; CROW 1993). A direct estimation ap-
proach, the traditional mutation-accumulation experi-
ment (BATEMAN 1959; MUM et al. 1972), takes exten-
sive time and labor and is only feasible for asexual
organisms, special sexual organisms (such as Drosoph-
ila where special chromosomal constructs are avail-
able), and artificially constructed purely inbred lines.
An indirect estimation procedure in highly selfing
plants makes use of inbreeding depression data
(CHARLESWORTH et al. 1990), but it depends on an un-
known mean dominance coefficient (@ of deleterious
mutations. Estimation of x requires some more assump
tions (COMSTOCK and ROBINSON 1948; HAYMAN 1954;
MUKAI et al. 1972; JOHNSTON and SCHOEN 1995; DENG
and LYNCH 1996a). Even then, the estimate is biased,
and weighted by selection coefficients of individual mu-
tant alleles. Estimation of other parameters of spontane-
ous deleterious genomic mutations, such as the mean
selection coefficient (Q and the genomic mutation vari-
ance scaled by environmental variance ( VJ V,), is also
important for testing different evolutionary theories.
For example, estimates of 3 and s are important for

148 H-W. Deng and M. Lynch
testing the theories of the evolutionary transition from
haploidy to diploidy (PERROT et al. 1991) and of the
role of deleterious mutations in extinction of small pop-
ulations (LANDE 1994; LYNCH et al. 1995a). Stimulated
by the work of CHARLESWORTH et al. (1990), DENG and
LYNCH (1996a) developed a methodology, which uses
the data (changes of the mean and genetic variance for
fitness traits) that can be acquired from inbreeding/
outbreeding in outcrossing/highly selfing populations,
to estimate not only U, but also x, s; and VJV,.
In regularly outcrossing animals, data on the change
of genetic variance for fitness traits upon inbreeding
are not readily come by. Since genotypes normally can-
not be cloned easily, the total genetic variance cannot
be estimated without bias (FALCONER 1989). An unbi-
ased estimate of the total genetic variance requires that
clones of the genotypes be available and distributed
randomly across the experimental environment, so that
there will be no common environmental effects for
members of the same genotype and the environmental
variation will be clearly separated from the total genetic
variance in analyses of variance (ANOVA). In cyclical
parthenogens, genotypes can be preserved and repli-
cated (forming a clone) by asexual reproduction, and
inbred progeny can be constructed by mating relatives
(.g., mating among clonal members is genetically
equivalent to selfing) . Thus, outcrossed and inbred ge-
notypes can be assayed side by side in one controlled
environment, with each having multiple replicates. Un-
der such a situation, the change of the genetic parame-
ters across generations will not be confounded by tem-
poral environmental change. Performing one-way AN-
OVA, with clonal genotypes as main effects and clonal
replicates for genotypes as random effects, provides un-
biased estimates for the total genetic variance for quan-
titative traits. Thus, cyclical parthenogens are quite suit-
able for the application of DENG and LYNCH'S (1996a)
technique.
In this study, inbreeding and life-table experiments
were performed on populations of two cyclically parthe-
nogenetic species of the freshwater cladoceran, Daphnia
arenata and D. pulicam'a. Changes of the mean and ge-
netic variance were estimated and used to infer deleteri-
ous genomic mutation parameters by an extension of
DENG and LYNCH'S (1996a) procedure.
MATERIALS AND METHODS
Study organism: In nature, the life cycle of cyclically parthe-
nogenetic Daphnia species consists of several consecutive gen-
erations of parthenogenetic reproduction followed by a bout
of sexual reproduction, during which males and sexual fe-
males are produced and mate randomly (LYNCH 1983a,b; HE-
BERT 1987). Normally, the population size is effectively infi-
nitely large (LYNCH 1983a,b). In the laboratory, parthenoge-
netic reproduction can be maintained indefinitely as long as
environmental conditions remain favorable. During parthe-
nogenetic reproduction, genotypes are faithfully replicated
barring new mutation (HEBERT 1987), which makes it possible
to estimate the total genetic variance for any quantitative trait
by an appropriate experimental design (LYNCH 1985; LYNCH
and DENG 1994). Sexual reproduction can be induced reliably
in the laboratory, and the resultant resting eggs can be
hatched relatively easily (INNES 1989; DE MEESTER 1993;
LYNCH and DENG 1994; DENG 1995, 1996, 1997; DENC and
LYNCH 1996b). Generation time is -2 weeks at 20" (LYNCH
and ENNIS 1983).
Study populations and production of selfed progeny: The
experimental populations are from Amazon Park (D. arenata)
in Eugene, OR, and Dorena Reservoir (D. pulican'a) in Cottage
Grove, OR. Electrophoretic studies showed that the two popu-
lations reproduce by cyclical parthenogenesis with mating
within each population being effectively random (LYNCH and
DENG 1994; DENG and LYNCH 1996b). Detailed experimental
procedures of sampling populations, cloning and selfing ge-
notypes, determining species identity, and hatching selfed
resting eggs have been published in LYNCH and DENG (1994),
and DENG and LYNCH (199613). Briefly, populations were sam-
pled at the end of their growing season, when there were still
millions of individuals. Females sampled from the field were
isolated into individual beakers containing -200 ml of fil-
tered and aged water from the populations' source habitats
and fed with the green alga Scenedesmus. The species identity
was determined by morphological (BROOKS 1957; BRANDLOVA
1972) and biochemical (HEBERT et al. 1988, 1989; LYNCH et
al. 1989) criteria. Over a period of 2 months, the isolated
females reproduced asexually, forming cohorts of genetically
identical individuals. During this period, most clones also re-
produced sexually; ie., males were produced ameiotically, and
some asexual females switched to sexual reproduction by pro-
ducing sexual eggs meiotically. Since males and females in
each beaker are genetically identical, mating among them is
genetically equivalent to selfing. The resultant sexually pro-
duced eggs are in a diapausing form (ephippia) and hatched
by taking them through light/warm and dark/cold cycles
(LYNCH and DENG 1994; DENC 1995; DENC and LYNCH 1995b).
The hatched selfed individuals were then expanded clonally.
Life-table experiments: Two life-table experiments were
performed, one for each population, with 30 random out-
crossed parental clones and 30 random selfed offspring clones
(each from 30 different random parental clones) being used
for each population. Each genotype was replicated three times
and all were acclimated to the experimental conditions for
two generations before any measurement, ensuring that ma-
ternal and grand-maternal effects do not contribute to the
among-clone variance in the final analysis (LYNCH 1985). All
clonal replicates were randomly distributed in the experimen-
tal setting, so that the common micro-environmental effects
would not contribute to the resemblance of the clonal repli-
cates of each genotype, ensuring that genetic variance can be
unbiasely estimated. Each clonal replicate was maintained in
100 ml water (aged for at least 1 month and filtered before
use) from the populations' source habitats, with -300,000
cells of the green alga Scenedesmus per ml. For the Amazon
population, the experiment was conducted at 20" (a typical
daytime temperature during the growing season in the pop-
lation's source habitat, a seasonal pond), 12 hr:12 hr light-
dark photoperiod (typical during this population's growing
season), and the culture water was replenished every day. For
the Dorena population, the experiment was conducted at
10" (a typical temperature during the growing season in the
population's source habitat, bottom of a permanent lake), 16
hr: 8 hr lightdark photoperiod (typical during this popula-
tion's growing season), and the culture water was changed
every other day (due to the slow development of Daphnia in
this temperature). Starting from the third generation, new-
born individuals were measured daily (Amazon population)