+/- is a terrible stat, and we will all be better off when people don't use it to compare players.
I'm going to not compare Olofsson because we already looked at his stats and he's garbage.
Let's deep dive look at the defencemen you mentionned.
Reilly has a 61.01% offensive zone start percentage
Kulak has a 63.46
Fleury has a 67.07
Petry has a 49.6%
So.... Reilly, Kulak and Fleury are being sheltered about the same amount in terms of O-Zone starts, with Reilly's small sample size, the difference in Ozone starts is negligible. We see that all three are being used as third pairing guys when they play, and that Julien is sheltering his third pair (as he probably should). For comparison, Weber is at 48, Chiarot at 51 and Mete our lowest at 41 (meaning least sheltered).
Reilly has a PDO of 1.006 (close to expected, but a slight touch on the "lucky" side)
Kulak's PDO is 0.938 (very unlucky)
Petry PDO is 0.973 (moderately unlucky)
Fleury PDO is 0.977 (moderately unlucky)
The PDO explains almost entirely the differences in plus/minus. In the small sample sizes, Habs goalies have come up with a higher save percentage than the opponent for Reilly, but haven't for the other three defenders. Kulak looks so bad because really goalies have been bad when they are on the ice.
But we still can look farther... is there something any of these d are doing to create more scoring chances against their goalie than against the other team? Is that why the PDO is down? Let's look
Reilly has a High Danger Chance of 45.59% (ie the other team gets more high danger chances when Reilly is on the ice than the Habs do)
Kulak has a high Danger Chance 52.5% (habs get more high danger chances when he is on the ice then the other team does).
Fleury has a high Danger chance of 38.84 (this surprises me and is a lot concerning)
Petry 57.69 (no surprise, he is the best of these defencemen in this stat... cause Petry is Damn good, and he's doing this against tougher opposition and with shittier zone starts).
Conclusions
1) Petry is a stud and shouldn't be in this comparison. he's head and shoulders above the other D. This shouldn't be a surprise, anyone with a good eye test could have told you that before I looked at that stats. The fact that the eye test and the stats are confirming each other, gives credence to the value of the stats.
2) Fleury might be being overused as a rookie and third pair RD could also use some improvement. This isn't solely a LD problem. This surprised me a little bit. That said, he's a rookie who should get better, and his only competition is Folin right now, and he's better than Folin. We've seen the Kulak-Reilly pair where one of them was on the right side, and that was very bad, and so playing them out of position is not the answer. Fleury is still probably our best bet on the right.
3) Kulak is better than Reilly despite the plus/minus differences. The Habs control the scoring chances much better when Kulak is on the ice vs Reilly. That said Reilly is better than Olofsson, so while Mete is injured, Reilly should be considered. However the team seems to want to give Leskinen a chance. Given that Reilly is a replacement level #7 D, this isn't a bad strategy as maybe a rookie can be better than replacement level. That said if Leskinen doesn't work, you can always go back to Reilly on the third pair.
PS I didn't write down Chiarot's stats, but his stats show that he's been good. Not great, but good. (51% zone starts, 1.003 PDO, 58% HDCF)