I've worked it out - the effect on the recall election of voters choosing not to select a replacement candidate.
Of course, there must have been a few who voted for a replacement but not on the yes/no question, but I'm ignoring that.
Despite the call for Democrats not to vote for a replacement at all, it's impressive how desperate many Democrats were for someone, anyone, to vote for. All 9 registered Democrats on the ballot were among the top 14 of the 46 candidates in number of votes received. 35% of the votes for Democrats went to Paffrath, so there was plenty of interest in the others. (69% of the votes for Republicans went to Elder.)
Now, for total votes. Of the votes cast for replacement candidates, 68% went to Republican candidates and 28% to Democratic candidates, with the rest to 3rd party or unaffiliated candidates. That sounds pretty meek of the Democrats, but (previous caveat allowed for) 44.7% of the ballots cast chose nobody. That's a hefty abstention (in the previous recall, only 8% abstained for a replacement), and while not all of the abstainers were deliberately abstaining Democrats, added to the votes for Democrats it's 60.2% of the total ballots cast.
We can test this out by comparing it to the most pro- and anti-recall counties in the state.
The most anti-recall was, no surprise, San Francisco. 86.7% voted against the recall. Of the total voters, 17.4% voted for Democratic candidates, 15.4% voted for Republican candidates (fairly close to the 13.3% "yes" vote on the recall), and a whopping 64.3% voted for nobody.
The most pro-recall county in the state was Lassen, a lightly-populated high-desert county in the northeast part of the state, more easily accessible from Reno than anywhere else of note, threatened by but as yet not much damaged by the fires that are ravaging Plumas County to its southwest. It jumped up above even its neighbors by voting 82.9% for the recall. But of its total voters, 80.7% chose Republican replacement candidates, just 6.1% chose Democrats, and only 11.8% abstained. That could include both abstaining Democrats and anybody else who just decided not to vote on that question, but it's very small either way. (As 1.3% of Lassen's voters chose a 3rd-party or unaffiliated candidate, that + R gets within a percentage point of the "Yes" votes on the recall, thus making D + abstainers a close match to the "No" vote.)
The enormous difference between the percentages of abstainers in the extreme pro- and anti-recall counties, and the matching of them plus Democrats to "no" votes on the recall, suggest that the bulk of the abstainers were indeed deliberate anti-recall abstainers, and that those in that specific category strongly outnumbered those who chose a Democrat.
And that explains Larry Elder's apparent victory. Because while he got 46.9% of the votes cast for replacement candidates, almost matching Arnold's 48.6% in 2003, if you count his vote against the total ballots cast, candidate-choosers and abstainers alike, he got only 26%. Which looks close to a core DT loyalist vote to me. So it's a good thing the recall went down, because while Elder as the leading Republican certainly encouraged the hefty "No" votes the recall got, the Democrats brought the possibility of his governorship on themselves.