Review Papers Claim by Claim

• Design: Cross-sectional field survey of 18 U-pick strawberry fields; 54 composite berry samples (3 per field: central, edge, bitten) and 16 irrigation-pond water samples during spring 2014 and winter 2015 Paper methods and sample sizes
• Major outcomes: bitten berries >> higher total coliforms and E. coli; Cryptosporidium detected on berries (overall low oocyst counts) and in 2 irrigation ponds; Salmonella only in 2 bitten berries; no E. coli O157 detected Main results summary

Interpretation & critical appraisal (evidence-first)

1. Effect sizes are large for wildlife-bitten berries: mean E. coli on bitten berries (76.7 MPN/g) is roughly 48× and 192× the edge and central means respectively, and total coliforms show similar amplification — consistent with direct fecal/wound contamination from animals or deposition of feces during biting events Indicator levels by category
2. Protozoa detected but at low concentrations: Cryptosporidium oocysts were observed on berries and in 2 irrigation ponds (0.1, 0.2 oocysts/L). Detection at low counts prevented genotyping — this weakens causal source attribution (human vs animal species), but presence still indicates fecal/animal-sourced contamination potential Cryptosporidium detection and limits
3. Water quality is poor in many ponds: 100% of pond samples positive for total coliforms and 87.5% for E. coli, with many measurements reported as >11,000 MPN/100 mL (censoring at assay upper limit) — this suggests chronic fecal loading of pond water used for irrigation Irrigation water contamination data
4. Modeling is appropriate but limited: logistic regression identified E. coli on fruit, air temperature, and irrigation Cryptosporidium as predictors of Cryptosporidium detection on fruit (coefficients and odds ratios reported). However, with 54 samples and low oocyst counts the model risks overfitting and confounding (season, rainfall, farm practices). Authors acknowledge this limitation Regression details and caveats

Limitations, biases, and blindspots

• Cross-sectional single-visit sampling prevents temporal inference (causality) and under-samples within-field spatial heterogeneity and temporal variability (rain events, visitor traffic) Cross-sectional design note
• Small n for water (16) and low oocyst counts limit pathogen source attribution and molecular confirmation (no genotyping) — therefore species (human vs animal Cryptosporidium) remain unknown Limits to genotyping
• Potential measurement censoring: the MPN upper bounds (>468 MPN/g for berries; >11,000 MPN/100 mL for water) were set to the assay limits and treated as equal to the censored value in analysis, which biases means downward or upward depending on distribution and complicates inference about extreme contamination events Censoring of MPN values
• Selection bias risk: farms volunteered/selected; visitors and farmer behaviors (harvest timing, irrigation scheduling, pest control) may differ across sites and seasons; authors did not report sampling-weight corrections or farm-level covariate adjustments beyond simple variables (temperature, E. coli, irrigation Cryptosporidium) Potential selection/confounding

Practical takeaways (evidence-weighted)

• Avoid harvesting/consuming berries with bite marks or signs of wildlife contact — these had substantially higher fecal indicator organisms and included the only Salmonella-positive berries in the study Bitten berries risk
• Irrigation pond quality appears poor; improving water sources or preventing pond water contact with edible surfaces (avoid overhead splash, use clean water for last-stage irrigation) would plausibly reduce protozoal and bacterial contamination risk (consistent with literature linking irrigation water contamination to produce contamination) Irrigation water as source

What would change the conclusions?

• Genotyping Cryptosporidium from water, wildlife feces, and berries to demonstrate matching species/genotypes would strongly support wildlife/irrigation pathways over alternative explanations (e.g., human handlers) Need for genotyping
• Longitudinal sampling across rain events and visitor-use periods would confirm the seasonal/temperature associations and clarify causality between rainfall/temperature and contamination pulses.

Suggested follow-up experiments (concise)

1. Paired molecular genotyping: collect matched samples of wildlife feces (rodent traps / scat), pond water, and berries across 10 farms over 12 months; use qPCR and sequencing to link genotypes; outcome: direct source attribution.
2. Intervention trial: in matched farm pairs, implement improved water treatment / substitution for last-stage irrigation (clean water), and enhanced edge fencing to reduce wildlife access; measure before/after changes in E. coli, Cryptosporidium prevalence on berries.