Antibiotic Resistance: Selection on Pre-Existing Variation or a Learned Response?
In a high-school biology lab, students grow E. coli in four flasks with ampicillin. Flask 0 has no antibiotic. Flask 25 has 25 µg/mL, flask 50 has 50 µg/mL, and flask 100 has 100 µg/mL. After 24 hours at 37°C, optical density readings are 1.20, 0.90, 0.40, and 0.05. The next day they plate 0.1 mL from each flask on agar containing 100 µg/mL ampicillin. After incubation, colonies: flask 0 -> 320; flask 25 -> 210; flask 50 -> 40; flask 100 -> 2. A student concludes, 'The bacteria in flask 100 learned to survive by making the antibiotic.' Another says, 'The antibiotic caused mutations that let bacteria live.' The teacher asks the class to evaluate both claims using the data and the principles of natural selection. They have no genome data.
1. Evaluate the first claim ('learned to survive by making the antibiotic') using the data. What pattern would you expect if the claim were true, and what does the data actually show?
Read the worked answer
The claim treats resistance as a learned response acquired during exposure, as if the bacteria detect ampicillin and manufacture it or otherwise acquire the ability on demand. If that were the main explanation, we would expect survival to depend on exposure history in a directed way: cells from the high-antibiotic flask should be the most able to survive the 100 µg/mL plate, and the no-antibiotic flask should have few or no resistant colonies. The data show the opposite pattern in absolute counts: flask 0 produced 320 colonies on the 100 µg/mL plate, flask 25 produced 210, flask 50 produced 40, and flask 100 produced only 2. That pattern is not evidence of learning; it is more consistent with the high-antibiotic flasks having far fewer total cells after 24 hours, so even if resistant cells were enriched as a fraction, their absolute number was small. The claim also lacks a mechanism: 'making the antibiotic' is not supported by any measurement, and the class has no genome or chemical data. The data alone cannot prove learning; they mainly show that resistant cells existed in the no-antibiotic population.
2. Evaluate the second claim ('the antibiotic caused mutations that let bacteria live') and explain how natural selection better accounts for the data.
Read the worked answer
The second claim says the antibiotic directed the mutations. Standard natural selection predicts the opposite order: genetic variation, including resistance alleles, arises before exposure, usually by random mutation; the antibiotic then selects the variants that can grow. The key observation is that the flask never exposed to ampicillin already produced 320 colonies on a 100 µg/mL plate. That means resistant cells were present in the population before antibiotic exposure, so the antibiotic did not have to create them. The decreasing colony counts from flask 0 to flask 100 do not show that the antibiotic caused resistance; they likely reflect the much lower total population in the high-antibiotic flasks (OD 0.05 versus 1.20 in flask 0). To compare resistance, we need the fraction of resistant cells, not just the absolute colony count. Natural selection also explains why susceptible cells died at higher concentrations while resistant variants survived. The data are consistent with selection on pre-existing variation, but they do not prove that the resistance mutations were random or that no new mutations occurred during the experiment.
3. Propose a follow-up measurement that would distinguish whether resistance was present before exposure or increased after exposure. Explain what result would support each explanation.
Read the worked answer
Plate each flask on nonselective agar as well as on 100 µg/mL ampicillin, and calculate the resistant fraction: colonies on ampicillin divided by total colonies on nonselective agar, adjusting for dilution. If resistance was already present before exposure, the no-antibiotic flask should have a measurable resistant fraction, and the high-antibiotic flasks should show a similar or higher fraction after selection, even though their absolute counts are lower. If resistance was induced or increased only after exposure, the no-antibiotic flask should have a much lower resistant fraction, and the fraction should rise with antibiotic concentration. A second follow-up would sequence resistance genes from colonies isolated before and after exposure; finding the same resistance alleles in the no-antibiotic population would support pre-existing variation. A third would run a fluctuation test or replay experiment: split a sensitive population into many independent cultures, expose them to the same antibiotic, and see whether resistance appears randomly across cultures. The key is to measure frequency and timing, not just absolute colony counts.