Ciprofloxacin resistance along a river: selection, dilution or fitness cost?

The pattern is a hotspot with a decay plume. Three processes compete to explain it, and the data discriminate between them only partly - which is the point of the question.

1. Selection at the source. The plant concentrates people, hospitals and antibiotic residues; resistant strains are enriched in the effluent (site A's 3 % upstream is the baseline).
2. Dilution and mixing. Effluent is diluted by the main flow, so the proportion of resistant cells falls as clean-water bacteria and volume are added. A fall from 42 % to 24 % over 500 m and 11 % by 2 km is consistent with exponential dilution.
3. Fitness cost and plasmid loss. If resistance were costly in the absence of the drug, resistant cells would decline faster than dilution alone predicts; loss of the plasmid would convert resistant cells to sensitive ones.

The median MIC column is the key. If the downstream fall were due to loss of resistance, the still-resistant cells at site D would be a low-MIC subpopulation; and indeed the median MIC does drop from 16 to 8 µg/mL between sites B/C and D. That is compatible with high-level resistance being selectively disadvantageous downstream. But the drop is small, and the fraction resistant at 2 km (11 %) is still nearly four times the upstream baseline, so resistant cells are not simply dying out. Interpreting the 3 % upstream baseline as “natural” background is also an assumption - it may itself include resistant isolates from wildlife or earlier contamination. A good answer names the confounder and proposes a normalising marker (e.g. faecal indicator load, or a conserved housekeeping gene) to separate dilution from selection.

The experiment shows a fitness cost of resistance in the absence of the drug: from 50 % resistant at day 0 to 12 % at day 30, a relative fitness disadvantage of roughly (12/88) ÷ (50/50) ≈ 0.14 in 30 days - the resistant strain has a growth/survival rate about an order of magnitude below the sensitive one under these conditions. Caveats: filter-sterilised river water lacks the natural community, protist grazing and sediment surfaces, so the measured cost is a laboratory estimate; and the starting ratio of 50 % is far above the 42 % seen in the river, so competition dynamics may differ.

The prediction follows from the cost–benefit balance:

• At 0.05 µg/mL, below the MIC of both strains, there is little or no killing, so no direct growth advantage from resistance.
• However, sub-inhibitory concentrations are well documented to select - sensitive cells suffer subtle growth inhibition and SOS induction, while resistant cells grow unimpeded. Even a small growth-rate difference maintained continuously can hold resistance at a higher equilibrium than the cost alone predicts.
• Sub-MIC concentrations may also increase plasmid transfer frequency, adding a second route to persistence.

So the expectation is that resistance will persist downstream at well above the 3 % baseline, with the exact level set by the balance between the fitness cost and the continuous low-level selection - precisely the situation the river data appear to show. A prediction to test: sample at increasing distance and check whether the resistant fraction approaches the upstream baseline or plateaus above it; a plateau implies ongoing selection or transfer.

The same gene on an IncF plasmid in E. coli and Aeromonas is direct evidence of horizontal gene transfer, most plausibly conjugation in the water or in a shared host. Its significance is that resistance is not confined to the indicator organism being counted. Risk assessment based on E. coli alone measures one host and misses the wider resistome: the plasmid could move into opportunistic pathogens, and a species that is not a faecal indicator can act as a reservoir.

Practical consequences: monitoring should be extended beyond a single indicator - for example plasmid/gene-level quantification (qPCR for blaCTX-M-15), metagenomic sampling of the community, or screening of several genera including Aeromonas and Klebsiella. A sampling design that reports only E. coli percentages will systematically understate the hazard, and comparing studies that use different indicators becomes invalid.

Management option - upgrade the plant to tertiary treatment (ozonation, UV, or membrane filtration). Evaluation: it reduces the total bacterial load and therefore the absolute number of resistant cells released, which is the quantity that matters for exposure; but it does not necessarily destroy extracellular or intracellular DNA, so gene-level pollution may persist. Ozone can also select for more tolerant organisms, and capital and energy costs for a plant serving 180,000 people are substantial. A defensible conclusion is that tertiary treatment is worth it primarily for the pathogen-load benefit, with gene-level monitoring before and after to check that the resistome actually falls rather than being redistributed - plus source control of antibiotic prescribing upstream, which reduces selection at its origin rather than at the end of the pipe.