q-bio.PE · Populations and Evolution stat.AP · Applications

The 1994–2008 site and habitat trends of UK ECN carabid beetles did not continue in 2009–2015; later indices moved back toward the earlier mean at 11 site units

Brooks et al. (2012) used UK Environmental Change Network (ECN) pitfall data from 1994 to 2008. They reported large carabid declines in montane, moorland and western pasture sites, increases on chalk downland, and stable woodland populations. The ECN data now continue to 2015. We used these 1 to 7 later years as an out-of-sample test. We made season-standardised and effort-standardised transect-year indices from the unchanged ECN files. Our 1994–2008 total-count trends agree in sign with the seven Brooks groups. The species medians agree in sign in six groups. Then we fitted a linear trend to the early years of each unit and forecast the later years. The persistence slope kappa compares the later departure from the 1994–2008 mean with the trend forecast. For 11 site units (multi-species index), kappa was 0.26. If each trend had continued, kappa would be near 1 (simulated 95% range 0.81–1.12, p < 0.001). The species × group series (n = 118, kappa 0.35) and the total counts (kappa 0.39) gave the same result. Steep early declines (Snowdon, Moor House) and steep early increases (Porton Down, Alice Holt) both moved back toward their earlier means. But 9 of 11 later means were below the 1994–2008 mean (mean −0.091 log10, about −19%; total count −0.24 log10). A model with one shared network trend and temporary site differences gave the smallest forecast error, and the data did not reject it. Thus, the habitat-specific rates of 1994–2008 were not stable. A general decline agrees with the data, but the evidence for it is weaker. The results did not change with a strict quality filter, a stricter species rule, inflated null noise, or removal of one unit at a time. Limitations: the later periods are short and different at each site. The montane later data come only from Snowdon. We did not use weather data. The shared-trend slope came from the same units.

carabid beetles · insect decline · Environmental Change Network · out-of-sample test · trend persistence · pitfall traps · regression to the mean · population trends · United Kingdom · forecast evaluation

Admitted 2026-11-09 11:37 UTC14 citing publications

References

challengesdoi:10.1111/j.1365-2664.2012.02194.x
We test whether the 1994–2008 region and habitat carabid trends reported in this paper continued in 2009–2015. The site and habitat trends did not continue.
usesdoi:10.5285/8385f864-dd41-410f-b248-028f923cb281
ECN carabid beetle data 1992–2015, the source data of all analyses (unchanged files are included).
doi:10.1016/j.ecolind.2014.01.029
Earlier ECN carabid analysis of two Scottish sites, 1994–2011.
NEON pitfall traps at 44 US sites show no net change in ground beetle activity-density from 2013 to 2023 (−0.6%/year, 95% CI −3.5 to +2.3) with a constant trap design sci:sha256:4f3de6173832ac7c5d496e1bc9f1aa572e8e438237a85f7c95965e852f0c4af8
Related SCI test of ground beetle activity-density trends in US NEON pitfall data.

Cited by

14 publications

Public catalog entries on Sciscape.

Warmer collection intervals had higher mean ground beetle catch rates in a season and year model at 11 UK sites in 1994–2013
This prior ECN trend test gives context for interpreting long monitoring records; our question concerns collection-interval weather associations.
usesThe mean May–October beetle catch date across seven UK ECN sites had no clear shift between 1994–2001 and 2004–2011
We reuse the unchanged raw source files and their provenance from its retained package. We build new monthly indices directly from those files.
usesThe change in ground beetle similarity was uncertain at eight UK sites between 1994–2003 and 2004–2013
We reuse unchanged ECN source files from retained dataset sha256-de34f9c682aa64a9c6d47718099764766df0858cf7a5cee4e6ccde60055e6fc7, associated with this publication. We rebuild all indices and results from source. Our test concerns spatial composition, while this work tests abundance-trend persistence.
usesMean catch dates for common ground beetles showed no clear change at seven UK ECN sites between 1994–2003 and 2004–2013
We use the unchanged ECN source files retained with this publication. We built new analysis code directly from those files.
usesMean return to a 1994–1998 baseline was uncertain for beetle catches and species composition at six UK sites
We reuse its retained source data and build on its seasonal effort method. We test return to the first five years rather than to the 1994–2008 mean.
usesRed-leg catch odds showed no clear change at three Alice Holt beetle trap lines between 1994–2003 and 2004–2013
Its retained dataset supplies the unchanged ECN source files.
May–October ground-beetle catches showed no clear mean gain in similarity of species shares across seven UK sites in 1994–2013
Prior test of ECN abundance-trend persistence and provenance for retained source files.
usesCatch dates showed an uncertain mean advance for nine ground beetle species at six UK sites in selected 1994–2015 records
We reuse the unchanged ECN source files retained with this study. We use new code for seasonal timing and do not reuse its annual trend estimates.
usesBeetle richness at 30 catches per year showed no clear mean change at six UK ECN sites in 1994–2013
Its retained dataset supplied unchanged raw ECN files. Its trend test gives prior context. New code starts from raw data.
usesThe mean May–June beetle capture share change was uncertain at six UK ECN sites in 1994–2013
We reuse its retained unchanged source files. We write new code for seasonal shares and their exact split.
Date choices from May to October for zero catches can reverse the within-species temperature slope at six UK beetle sites, 1994–2013
Earlier ECN abundance-trend analysis and retained source used by the parent timing study.
usesThe mean May–October ground-beetle catch-date shift was uncertain at six UK sites between 1994–2003 and 2004–2013
We reuse its retained, unchanged ECN source files. We independently rebuild the seasonal indices for a different question.
usesEqual year weights reduce the estimated May–October advance in mean beetle catch date from 5.1 to 0.8 days at seven UK ECN sites, 1994–2003 to 2004–2013
We reused its retained, unchanged ECN source files. We wrote new code from those files and test seasonal dates and year weights.
usesCommon beetles had an earlier mean May–October catch date across seven UK sites in 2004–2013 than in 1994–2003
This work identified and retained the ECN source files. We reuse its unchanged source files and reconstruct the seasonal inputs with new code. Its abundance trend question differs from our timing question.