Temporal eligibility rules alter country composition and ecoregion representation in global river water-quality monitoring networks

Temporal eligibility can reorganize a monitoring network in directions that station loss alone does not predict.This instability was characterized in Caravan-Qual Lite version 1.1 over 2000-2020 using six constituents, country labels, 426 freshwater ecoregions, seven realms, and Hill numbers q=0, 1, and 2. The primary rule retained 14,387 electrical-conductivity stations (Cohort A) and a 4.373-station intersection of the independently eligible conductivity, phosphorus, and dissolved-oxygen networks (Cohort B); overlap in eligible years or sampling dates was not required.With minimum requirements of three, five, or ten nonconsecutive eligible calendar years, dissolved organic carbon was lost in 90.7% of stations, while ecoregion q = 2 increased from 3.909 to 6.594.Electrical conductivity and total phosphorus also gained q = 2 ecoregion representation at selected thresholds, whereas dissolved oxygen declined across countries, ecoregions, and realms.Nitrate and suspended solids realm q = 2 rose slightly from three to five years before falling to ten.Thus, comparable attrition did not imply a common geographic response.Equal-size references were treated as descriptive baselines rather than exchangeability tests.In Cohort B, ecoregion q = 0 was 10.8% below its equal-size median while q = 2 was 46.1% above it; the +0.5% within-country q = 2 residual showed that the evenness excess was largely reproduced after country counts were fixed, whereas the q = 0 deficit persisted.Requiring coverage in all four calendar quarters removed 40.1% of Cohort B stations and nine ecoregions and shifted country leverage.Tempoal rules, therefore, changed country composition and FEOW representation in constituent-and order-specific ways that station retention alone could not summarize.