It's a combination of the tendency for centers to touch the puck in transition more than wingers, and also an artifact of the one player point per scoring play rule. Essentially, goal scorers tend to lose secondary assists they would have otherwise gotten because they scored the goal.
NHL Stats
The season I want to concentrate is the 16-17 season, when Crosby led the league in goals. You can see he had an abnormally low number of secondary assists that year. Now let's look at Crosby's power play numbers for each year.
NHL Stats
You can see Crosby's highest PPG/60 numbers correspond with his lowest PPA2/60. From what I remember the last time I discussed this, with Mike Farkas, I remember watching a Crosby goals seasonal compilation, and that year Crosby had a bunch of PP goals (like 3 or 4) where he passed to a Penguin for an entry, then beat the defense to the far post for the return pass tip-in. Normally he'd have gotten the secondary assist there, but since he scored the goal, no secondary was garnered.
From the little research I've done, it affects no more than 10% of a player's goals per year, and it's basically completely random, but if you look at that scoring per game for a particular player-season, and notice a crash in A2 rate, go and look at a compilation of that player's goals for the season, and most likely they were involved an a couple extra goals where they ought to have been credited with a secondary assist, but scored instead.
[As an aside, Ovechkin over his career has been a 0.20 A2 per game player, only over the past 3 years, he's been a 0.128 per game one. The year, he already has 11 secondary assists after 20 games (incidentally also the number of secondary assists he had in 07-08, his first peak year), and his 0.128 per game rate over the past 3 years has turned into a 0.168 per game over the past 4 years. (source here:
NHL Stats) The power of noise in secondary assist generation.]