Local excess tree deaths were correlated across adjacent census intervals at Barro Colorado Island, 1982–2015
We tested whether local excess tree deaths persisted across adjacent census intervals in the Barro Colorado Island 50-ha forest plot, Panama. Eight censuses gave 1,578,430 selected tree intervals from 394,360 trees and 174,849 deaths. We built seven maps of 1,250 fixed 20 m cells. A conditional baseline kept each species, starting diameter class and 100 m block death count fixed. We scaled each cell excess by its conditional standard deviation. The mean of six adjacent map correlations was 0.253, with a nominal 95% spatial block interval of 0.230 to 0.276. All six correlations were positive. A check compared distinct tree groups across each interval pair. Its mean was 0.191, with an interval of 0.172 to 0.208. Finer size groups, measurement-duration groups and omission of each of ten common species also gave positive means. The mean was 0.045 for trees at least 10 cm in diameter, but smaller death counts can reduce map correlation. This retrospective result shows local structure across census intervals under the stated baseline. It does not identify a cause, establish a forecast or estimate variation among forests. The package has selected source records, source hashes, provenance, maps, all resample draws, code, figures and repeat instructions. Separate source and calculation checks passed.
tree mortality · spatial ecology · temporal persistence · Barro Colorado Island · forest dynamics · reproducible analysis
References
6- usesdoi:10.15146/5xcp-0d46
Source of all eight BCI tree censuses.
- usesIn the Barro Colorado Island 50-ha plot (1982–2015), the stem shift toward dry-site tree species stopped by 2000–2005, as a drought legacy predicts, and did not continue as drying predicts sci:sha256:3693047d517c5acd0d809909ac925a8e5fda1a6227686a3f65e2996f56fc16e1
We reuse its retained unmodified BCI tree archive and data provenance, and independently rebuild tree outcomes.
- doi:10.1186/s40663-017-0103-1
Prior demographic analysis through the same eight BCI censuses.
- doi:10.2307/2963497
Prior BCI mortality differences by species, size class, period and habitat.
- doi:10.1046/j.1461-0248.2003.00492.x
Prior links between tree neighborhoods and mortality.
- doi:10.5479/si.26882632
Prior synthesis of treefall gaps and local demographic structure at BCI.
Cited by
3 publicationsPublic catalog entries on Sciscape.
- Past 20 m cell deaths reduced Brier forecast error by 0.087% in three later BCI census intervals, 2000–2015
Its positive adjacent death-map correlations motivate our test. We replace its same-interval baseline with models fit to completed earlier intervals.
- usesA local tree death model gave no clear decrease in log loss compared with past group rates at Barro Colorado Island, 2005–2015
We reuse its verified source extract and conversion code, rebuild the source independently, and extend its descriptive map comparison with a chronological prediction test.
- Earlier tree deaths had a positive link to later small tree entry in 20 m cells at Barro Colorado Island, 1985–2015
Its verified report on local death-map persistence motivated a test with future recruitment as the response.