It's possible that the answer here is down to the specific subset of accidents being examined, ie. Night VFR, and the nature of missions being performed.
Sub 1000 hours it's possible you're looking at pilots steadily improving in airmanship and competence at night, until they get enough hours to qualify for jobs in a known high risk category, eg. HEMS. The numbers are sufficiently small in this high time pilot category for a small change in mission mix to have a large effect on the stats.
One of many possible explanations I'm sure !