I'd guess it's the Vr-V2 speed spread, especially if you have a constant (or near-enough) V2 at the same altitudes.
Since it's really a thrust issue, if there's a specific FADEC schedule with a steep gradient of thrust with altitude, that might explain the Vr-V2 (and hence the Vr also) having a discontinuous behaviour.
If on the other hand it's a case of the V2 ALSO varying similarly, I don't think it would be for climb gradient reasons; more likely would be that the increased Mach number for the same IAS at higher alts is causing a small but significant change to stall speeds, and hence to any speed scheduled on minimum speed ratios.