For an individual of a species to be weaker than the other members of it's species, it must be genetically imperfect in some way, in other words, there is a variation in it's genetic code that is not in the stronger individuals.
For an individual of a species to be strong, it must be closer to genetic perfection, in other words there is less variation in it's genetic code than in the weak members.
There's something weird about talking about the amount of variation in an
individual. Variation is a property of groups. There is, for example, no variation in my height, or my eye color, or my sex. This is because there is only one of me.
Would natural selection then actually narrow the variation of species instead of broadening it, as is claimed?
That depends on circumstances. Always bear in mind that there is no such thing as "less favored" or "more favored" except with relation to an environment. When a group has an opportunity to spread between environments or between niches within the environment, then natural selection will act
differently on the subgroups in different circumstances, thus increasing the variation of the group.
For example, natural selection worked differently on bears living in the Arctic Circle to those outside of it: white fur was favored in those living against a perpetual background of snow and ice. The result was that the lineage became more diverse with the addition of a white species of bear.