From the original work (linked at the end of the article)
"CPR that has been towed within the North Atlantic and adjacent seas. 36% of the total number of CPR tows between 1957 and 2016 (16,725 tows) had faults logged, 4% of these faults were due to plastic entanglement and 1% were due to natural entanglement "
They only had a total of 208 cases of their devices getting entangled out of those 16,725 tows of which 52 were discarded because they did not involve plastics. Then they did a normalization across each year. I think the large variance is probably due to the small absolute numbers of total incidence and the exaggerated scale that they chose (0-5%).
It's not terrible and it's clearly labeled which is great. It's just that the total numbers are small and small numbers give rise to the appearance of large variance simply by virtue of being small. It's a data set issue and one to be aware of because in other cases it is one that can be used to misrepresent data trends.
More importantly, hopefully this data can be used to increase pressure on governments to strengthen regulations against single-use plastic items.