Reminder on data uploaded:
In many ecotoxicological tests, the measured endpoint is an integer: number of offspring per female per day, total juveniles, number of fronds (Lemna), number of emerged adults (Chironomus), etc. These are reproduction or more generally count data. A dose–response relationship is then fitted to quantify the effect of a toxicant on these totals (or derived rates).
Statistically, such data are suited to models for integers: Poisson or negative binomial GLMs (to account for overdispersion), zero-inflated or hurdle models if many zeros occur, and hierarchical models for repeated measures over time (e.g., daily reproduction per female). From these, ECx/EDx values (EC10, EC20, EC50) or BMDs (Benchmark Doses) can be estimated with confidence intervals, using approaches that respect variance increasing with the mean.
Depending on the test, the response may be: (i) a cumulative total (e.g., juveniles at day 21), (ii) a rate (offspring/female/day), (iii) a count per unit of time. Proper transformation into per-female means and accounting for parental mortality are essential to avoid bias.
OECD provides both statistical principles and test guidelines. For analysis, the Guidance Document on Statistical Analysis of Ecotoxicity Data (ENV/JM/MONO(2006)18) covers methods for ECx estimation, distribution comparisons, and modeling of quantal and count data (including overdispersion and model choice).
Benchmark Dose (BMD) approaches are promoted in several regulatory frameworks because they provide continuous effect estimation with uncertainty. For count data, Poisson/negative binomial distributions are typically used in curve fitting.
Aquatic invertebrates:
• TG 211 — Daphnia magna Reproduction Test : number of neonates per female (cumulative and/or daily).
• TG 218 / 219 — Chironomus, sediment/water toxicity test : survival, development and emergence (count of emerged adults).
Soil invertebrates:
• TG 222 — Earthworm Reproduction Test (Eisenia spp.) : number of juveniles produced.
• TG 220 — Enchytraeid Reproduction Test : juvenile counts.
• TG 232 — Collembolan Reproduction Test (Folsomia candida, etc.) : number of juveniles/frequency of reproduction.
Plants and macrophytes (integer counts):
• TG 221 — Lemna sp. Growth Inhibition Test : number of fronds (integer) and biomass; frond number is a true count endpoint suitable for GLMs.
• TG 208 — Seedling Emergence and Seedling Growth Test : emergence counts per pot/lot and early growth.
• TG 227 — Vegetative Vigour Test : number of shoots/vigour-related parameters (some recorded as integers).
Fish and vertebrates — reproduction endpoints (often counts):
• TG 229 — Fish Short-Term Reproduction Assay : fecundity (eggs produced), fertilisation/hatching (counts of eggs/larvae).
• TG 240 — Medaka Extended One-Generation Reproduction Test (MEOGRT) : fecundity, fertility, hatching — all count endpoints.
• TG 206 — Avian Reproduction Test : laying, fertility, hatching (counts of eggs and chicks).
When the endpoint is the number of cells (e.g., cell density measured by counts), the response may be treated as count data if the measurement is truly an integer (cell/colony counts).
• TG 201 — Freshwater Alga and Cyanobacteria, Growth Inhibition Test : the main endpoint is biomass or growth rate (continuous), but if expressed as raw cell counts (cells·mL⁻¹, integer), count-data GLMs or log-normal models can be compared.
• TG 221 — Lemna sp. : the number of fronds is explicitly an integer endpoint, well-suited to Poisson/negative binomial models, possibly with zero-inflation.
• Check for overdispersion → prefer negative binomial if variance > mean;
• Examine zero-inflation (e.g., no reproduction at all) → zero-inflated or hurdle models;
• Use offsets/exposures (e.g., number of females, test duration) to model rates;
• Propagate ECx/BMD uncertainty (confidence intervals via likelihood profiles or bootstrap);
• For repeated measures (e.g., daily reproduction), consider mixed models (GLMM) or aggregate into biologically meaningful totals (e.g., day 21 cumulative).
• ECHA/REACH : Guidance documents R.7 (R.7a/b/c) for aquatic/terrestrial toxicity recommend ECx and BMD approaches, particularly for reproduction endpoints.
• US EPA OCSPP 850 series : e.g., 850.1300 (Daphnid Chronic Toxicity Test — reproduction), 850.6200/6300 (terrestrial plants: emergence/vigour) — endpoints reported as counts or derived rates.
OECD — ENV/JM/MONO(2006)18, Guidance Document on Statistical Analysis of Ecotoxicity Data (methods for ECx, count data modeling, model choice).
OECD TG 211, 218/219, 220, 222, 232, 208, 221, 227, 229, 240, 206 — Test Guidelines where primary or secondary endpoints include counts .
ECHA (REACH) — Guidance R.7 (R.7a/b/c) : choice of endpoints, ECx, analysis principles.
US EPA — OCSPP 850 Series Test Guidelines : daphnids (reproduction), plants (emergence/vigour), and other count endpoints.