Tromboni et al. GPP, ER, NEP data from mesocosms and from chambers estimates. R code for statistical analyses.

GPP, ER, and NEP measured in chambers and open-channel mesocosms (units: g O₂ m⁻² d⁻¹). R code for all statistical analyses is provided.<b>Statistical approach.</b> The mesocosm was the experimental unit (n = 3 per treatment). Subsamples within mesocosms (e.g., tiles, chambers, repeated measurements within a phase) were averaged to yield technical replicates. Control mesocosms (n = 1 each, with and without macroinvertebrates) were included for descriptive comparison only. Metabolic rates—gross primary production (GPP), ecosystem respiration (ER), and net ecosystem production (NEP)—were analyzed with linear mixed-effects models of the form <code>Response ~ Flow × Macro + (1 | Mesocosm)</code>, with Flow treatment and Macro (macroinvertebrates present/absent) as fixed effects and Mesocosm as a random intercept. To accommodate distributional features while preserving interpretability, NEP and ER were transformed using the inverse hyperbolic sine (asinh), and GPP using <code>log1p</code> (natural log of value + 1). Estimated marginal means (emmeans) were computed for each treatment and back-transformed to the original scale with 95% confidence intervals; pairwise contrasts were adjusted for multiple comparisons (Holm).