Independent ribbon models matched 32 zebrafish cone summaries and did not fix each ribbon's SNR

We tested what unresolved cone glutamate responses can show about vesicle release and noise at individual synaptic ribbons. We used 32 PR1 cone traces from Herzog et al. (2026). We recovered all Figure 3C means and standard deviations by unique early-trial cuts. They retained 2,892 responses. The largest joint error was 2.51e-7 response units. Mean response CV was 0.13335. Mean power SNR, defined as mean squared divided by variance, was 93.3046. For an ideal readout X=qK with unknown response per vesicle q, CV squared equals count Fano factor divided by mean count. Poisson, binomial and negative-binomial models matched the same two moments in all 32 records. These matches do not establish a full distribution fit. Two independent, equal ribbons gave mean single-ribbon power SNR 46.652. A 1%/99% allocation of mean and variance gave 0.933 and 92.372 with the same terminal moments. A sharp bound applies to the best independent ribbon, not to every ribbon. Shape checks also found serial dependence, material background variance, and limits on a simple Poisson optical model. Missing fish and recording labels prevented a measured shared-noise test. Exact pairing bounds for 496 marginal two-input sums had median minimum and maximum variance ratios of 0.184 and 1.826 relative to independence. They are not observed biological pair effects. The result concerns identification from selected optical summaries. It does not identify actual fusion statistics or single-ribbon reliability. The package preserves proofs, source hashes, compact data, source code, results, figures and repeat instructions.

cone photoreceptor · synaptic ribbon · signal-to-noise ratio · Poisson release · identifiability · glutamate imaging · zebrafish · noise covariance