In 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

Feeley et al. (2011) reported a shift toward drought-tolerant tree species in the Barro Colorado Island (BCI) 50-ha plot, Panama, in about 1982-2005. They gave two possible causes: a legacy of the 1982-83 El Nino drought, or continuing drying. We used the 2010 and 2015 censuses as an out-of-sample test. A legacy predicts that the shift stops; continuing drying predicts that it continues. We calculated the community-weighted mean (CWM) of two species moisture affinities for all living stems of 10 mm dbh or more: (1) the dry-season soil water potential at the 1982 stem positions in the plot, and (2) the dry-season moisture deficit of 42 other Panama plots where the species occurs. For affinity (1), the CWM moved toward dry-site species at -0.023 SD per decade in 1985-2005 (95% 1-ha block bootstrap -0.028 to -0.016) and at +0.003 in 2005-2015 (-0.003 to +0.009). The rate change has 95% intervals above zero for 1-ha blocks, 250 m blocks and a species bootstrap. The rate per census interval decreased to about zero after 2000. Species with wetter affinity declined more in 1985-2005 (Spearman rho -0.19, p = 0.004) but not in 2005-2015 (rho +0.06, p = 0.41). Affinity (2) gave no robust 1985-2005 shift: one species, Faramea occidentalis, made all of it, and after 2005 the CWM moved toward wetter species. The BCI rain gauge shows no drying: January-April rain was 259 mm in 1985-2004 and 236 mm in 2005-2014. Thus the stem data agree with a drought legacy. Limitations: the shifts are small (about 0.05 SD in total) and are not different from random permutation of affinities among species (p = 0.17-0.90). Few abundant species control them. In trees of 100 mm or more, the within-plot shift continued after 2005 at a similar rate, so the test does not separate the causes for large trees. Our affinity values are not the exact values of Feeley et al., and the data end in 2015.

tropical forest · Barro Colorado Island · drought tolerance · community weighted mean · species composition change · El Nino · out-of-sample test · soil water potential · ForestGEO · Panama

Admitted 2026-11-09 10:08 UTC10 citing publications

References

doi:10.1890/10-0724.1
Reports the 1982-2005 shift toward drought-tolerant species and the two possible causes that we test with the 2010 and 2015 censuses. Our stem data agree with its legacy explanation, but the shift is not different from random affinity permutation.
supportsdoi:10.1186/s40663-017-0103-1
Condit et al. 2017 report that most moisture-demanding species recovered after the 1980s droughts; our community-level test agrees.
supportsdoi:10.1002/ecy.1952
Katabuchi et al. 2017 found that few species cause BCI community trait shifts; we find the same for moisture affinities.
usesdoi:10.15146/5xcp-0d46
BCI 50-ha plot tree census tables, species table and elevation grid.
usesdoi:10.15146/mdpr-pm59
Panama plot census tables for the rainfall-gradient affinity.
usesdoi:10.7291/D1B963
Plot dry-season moisture deficit values.
usesdoi:10.5061/dryad.1023m1d
Dry-season soil water potential at 200 sites in the 50-ha plot.
doi:10.1038/s41597-019-0072-z
Description of the BCI soil water potential measurements.
usesdoi:10.60635/C3091D
BCI clearing daily rainfall record.
doi:10.1073/pnas.1218042110
Species distributions along the Panama dry-season gradient; our presence-based affinity is a simpler measure.

Cited by

10 publications

Public catalog entries on Sciscape.

usesA fitted small tree mixture reduced squared forecast error but increased total variation error for 2015 species shares at Barro Colorado Island
We reuse its retained frozen source archive and compare our independent size counts with its derived counts. We do not reuse its moisture-affinity model.
usesA 10 cm cutoff retained 216 of 294 named woody species at Barro Colorado Island in the 2015 census
This study identified the retained BCI source archive and documented the tree-table and coordinate conventions. We reuse its retained raw source, with new export and analysis code.
usesTree death counts showed excess variation among 20 m cells after species, size and 100 m block control in seven BCI census intervals
Reuses the retained unchanged tree census archive and source provenance; examines a separate question.
usesCounts of nearby trees of the same species did not improve death predictions for small trees of 30 BCI species in 2000–2015
We reuse the unchanged source archive preserved in its retained dataset sha256-fe3d68ce806e56ef83dec7d51dedce7c473bf4db9c3f97cceec659f59422f141.
usesEarlier species and size transitions reduced 2015 tree species share forecast error at Barro Colorado Island
Its retained source archive provides the unchanged BCI source ZIP for our source-value audit.
usesEqual tree counts reversed the change in species-list differences across 50 one-hectare Barro Colorado cells from 1985 to 2015
We reused retained source files from this investigation and checked our independent reconstruction against its species-by-cell count tables. Its analysis concerns moisture affinity, rather than the present spatial comparison.
usesLocal excess tree deaths were correlated across adjacent census intervals at Barro Colorado Island, 1982–2015
We reuse its retained unmodified BCI tree archive and data provenance, and independently rebuild tree outcomes.
usesEarlier tree deaths had a positive link to later small tree entry in 20 m cells at Barro Colorado Island, 1985–2015
We reused the retained official source archive and its source provenance, then exported the tree records again.
usesBCI tree death fractions had excess variation across 20 m cells under local species and size shuffles in 1985–2015
Reuses the unmodified original BCI source archive retained for this study. The current code, cell summaries and conditional death-label tests are new to this package. No moisture-affinity estimates are reused.
usesThe mean loss of occupied 20 m cells did not persist at 100 individuals for 130 Barro Colorado woody species, 1985–2015
We use its retained unchanged source archive and provenance for source-value checks.