At least according to the chart on page 4607, the beetles selected for low population groups [B] had lower rates of adult-on-eggs and adult-on-larvae cannibalism than the control [C], and comparable rates to beetles selected for high-population groups [A].
The table is only reporting the grand means: all As pooled together, all Bs pooled together, and so on, potentially averaging away things. The body of the paper explains why this lumping-together obscures distinct responses to group selection pressure (emphasis added):
The high group-selected populations (A) exceeded the controls (C) in all assayed components expected to contribute to population size except larval egg cannibalism. The low group-selected populations (B) also exceeded or equalled the C populations in all components assayed [ie. increased cannibalism too], yet the B [low] population maintained much lower adult numbers. This unexpected result can be explained by examining the mean of each B [low] population separately rather than the grand mean. In the B [low] treatment, there is a significant between-populations variance for five of the nine traits assayed (Table 1, column 5; p < 0.025). That is, some of the B [low] populations enjoy a higher cannibalism rate than the C controls while other B [low] populations have a longer mean developmental time or a lower average fecundity relative to the controls. Unidirectional group selection for lower adult population size resulted in a multivarious response among the B [low] populations because there are many ways to achieve low population size.
So, group selection could operate as Eliezer described: it did not always “of course” produce the nice humane ‘obvious’ solution to keeping group size low, but could produce horrible baby-eating solutions that naive wishful group-selectionists hadn’t even thought about.
And I see nothing about female larvae in particular.
I’m not sure about the targeting female larvae claim; I also don’t see where in the paper that would be implied. (Table 1 seems to imply that there wasn’t because I would expect the sex ratio entry, which is defined as starting with ‘pupae surviving adult cannibalism’, to show between-population variance if there were populations where the adults selectively cannibalized female eggs.) So Eliezer might have misread Table 1 and made an error there, or assumed that killing female eggs would be the most efficient way to suppress growth, or be referencing something else.
The table is only reporting the grand means: all As pooled together, all Bs pooled together, and so on, potentially averaging away things. The body of the paper explains why this lumping-together obscures distinct responses to group selection pressure (emphasis added):
So, group selection could operate as Eliezer described: it did not always “of course” produce the nice humane ‘obvious’ solution to keeping group size low, but could produce horrible baby-eating solutions that naive wishful group-selectionists hadn’t even thought about.
I’m not sure about the targeting female larvae claim; I also don’t see where in the paper that would be implied. (Table 1 seems to imply that there wasn’t because I would expect the sex ratio entry, which is defined as starting with ‘pupae surviving adult cannibalism’, to show between-population variance if there were populations where the adults selectively cannibalized female eggs.) So Eliezer might have misread Table 1 and made an error there, or assumed that killing female eggs would be the most efficient way to suppress growth, or be referencing something else.