Recorded losses supplied most of the 20-year mean northward center shift for 59 Swedish forest plant species

We tested how recorded gains and losses changed plant latitude centers on fixed Swedish forest plots. We used 59 vascular plant species on 1,712 plots. Each pair of full-area forest visits was exactly 20 years apart, in 1999–2004 and 2019–2024. Visit days differed by at most 30 days. The equal-species mean center moved 12.47 km north. Its central 95% spatial weight interval was 6.35 to 18.81 km. An exact count split assigned 10.82 km to recorded losses and 1.65 km to recorded gains. A symmetric split and all planned selection checks retained a large positive loss term. A national reference placed the observed gain and loss counts uniformly among early absences and presences. It gave a 25.99 km mean northward shift. Observed minus reference was −13.52 km, with an interval of −31.82 to +3.76 km. This contrast changed sign with the species cutoff and visit-date rule. A reference that kept actual gain and loss counts within each county gave 11.43 km north. Its observed excess was +1.04 km, with an interval of −0.83 to +2.80 km. The intervals used 4,999 joint positive spatial block weight sets. These results describe recorded centers and local events in a selected forest panel. They do not measure migration, correct detection change, or identify a climate cause. The county reference retains all between-county changes by definition. The package contains complete compact source data, an exact placement method, all code and outputs, figures, source hashes and reproduction instructions. An independent audit matched all 8,817,270 source status cells. A fresh offline replay matched all 35 generated result and figure files.

species distributions · forest plants · Sweden · latitude centroid · recorded gains and losses · spatial uncertainty · placement reference · reproducible analysis