Biology Reference
In-Depth Information
All worm burden traits were normalized using an inverse Gaussian
transformation prior to analysis. Covariates for worm burden as assessed
by Ascaris total count included sex, age, and number of days of sampling
post-albendazole treatment. Only age was significant (showing a decrease
in total worm count with age, p
0.0160). Formal testing of the variance
component models revealed that for Ascaris total worm count, the model
which included only host genes fits the data nearly as well as the most
general model which included both host genetic and worm genetic
effects. Thus, the host genetic model represents a parsimonious model
that fits as well as a more general model that included both host and
worm genetic factors. Additive genetic host factors accounted for
approximately 54% of the total variation in total worm counts. Models
that only included worm genetic factors or no genetic factors were both
rejected as being significantly different from the general model. A priori,
one might expect host genetic effects to be the primary influence on total
worm count since each worm represents an independent infection. We
would expect strong selection for infectivity in the worms since infection
must occur in order for the worm to reproduce. Therefore, worm counts
are also a function of the survival of individual worms and represent a life
history or fitness-related traits that would be expected to have low levels
of genetic variation.
The above test of worm genetic variation suffers from the relatively
small sample size (320 hosts) that was incurred by only considering hosts
for which we had also genetically assayed the collected worms. Therefore,
we also decided to examine potential spatial effects influencing host
worm burden. The inclusion of a random spatial process may also allow
inference on worm population genetic structure effects, such as inference
of isolation by distance or spatial autocorrelation effects on worm kinship
and, therefore, may be an indirect proxy for worm genetic variation. For
this analysis, we utilized data on 1108 individuals (544 males and 564
females ranging in age from 3 to 85 years).
We employed a standard exponential decay model to parameterize the
expected correlations between hosts as a function of geographic distance.
64
Because host genetic factors are explicitly accounted for by the pedigree
information, any resulting spatial component may represent either: (1)
a geographic component of exposure patterns due to local environmental
conditions such as temperature or moisture, or (2) a component repre-
senting worm population genetic differentiation resulting from an
isolation-by-distance process. In order to account for such a potential
spatial process, we modified our variance component model as follows:
U ¼ð2Fh
2
þ expðlDÞs
2
þ Ie
2
Þs
p
¼
(12.5)
where
is a matrix of geographic distances (in kilometers) between
individuals as assessed from geographic coordinates of the houses that
D
