Antibiotic and Pesticide Resistance

Resistance is natural selection observed under controlled conditions, on a schedule short enough to watch. A chemical is applied, most of the population dies, the survivors reproduce, and the population that returns is no longer susceptible.

This is not a laboratory curiosity. Murray et al. (2022), the largest assessment to date, attributed roughly 1.27 million deaths in 2019 directly to bacterial antimicrobial resistance, with around 4.95 million deaths associated with it. The US Centers for Disease Control and the World Health Organization both track it as an active public-health emergency.

Agriculture and medicine plan around the process, which is the practical test of whether it is understood.

Selection watched directly#

Alexander Fleming warned of resistance in his 1945 Nobel lecture, describing how easy it was to make microbes resistant to penicillin in the laboratory by exposing them to concentrations too low to kill them. Penicillin-resistant Staphylococcus was widespread within a decade.

The most direct visual demonstration is Baym et al. (2016), who built a two-metre agar plate divided into bands of increasing antibiotic concentration and filmed E. coli colonising it. The bacteria spread to the edge of each band, pause, and then a mutant lineage breaks through and expands into the next. Over about eleven days the population crossed a thousand-fold increase in concentration, and the lineage that reached the final band could be traced back through the recorded history of the plate.

Richard Lenski's Long-Term Evolution Experiment provides the same thing on a longer timescale with frozen ancestors available for replay, discussed in What is Evolution? and Genetic Information.

The mutations come first, and are not directed#

A common misreading is that exposure to an antibiotic causes bacteria to develop resistance. Two classic experiments established otherwise, before the structure of DNA was known.

Luria and Delbrück (1943) reasoned that if mutations arose in response to a challenge, resistant counts across parallel cultures would follow a predictable distribution; if they arose randomly beforehand, the variance would be enormous, because an early mutation produces a large clone and a late one a small clone. The observed variance was enormous. The result won a Nobel Prize and gave the mutation rate a way to be measured.

Joshua and Esther Lederberg (1952) settled it visually with replica plating: colonies were transferred from a plate never exposed to the drug onto a selective plate, and the resistant colonies could be located back on the original, unexposed plate. The resistant cells existed before the antibiotic was ever applied.

This is the mechanism as stated in What is Evolution? — random variation, non-random selection — demonstrated on a bench.

Resistance genes are older than antibiotics#

Resistance is not solely a human artefact, which is why it appears so quickly.

D'Costa et al. (2011) recovered resistance genes for beta-lactams, tetracycline, and glycopeptides from 30,000-year-old Beringian permafrost sediments, and confirmed the vancomycin-resistance gene product was structurally functional.

Bhullar et al. (2012) sampled bacteria from Lechuguilla Cave, isolated from the surface for four million years, and found strains resistant to most clinical antibiotic classes, including some entirely synthetic ones.

This is expected. Antibiotics are largely derived from compounds soil microbes use against each other, so counter-measures have been under selection far longer than medicine has existed. Clinical use takes an existing rare variation and makes it common — which is what selection is.

Novel enzymes, not only tweaks to old ones#

Much resistance works by modifying an existing protein: a changed target that the drug no longer binds, or an upregulated efflux pump.

Some cases go further. Nylonase is a bacterial enzyme that digests a by-product of nylon manufacture, a compound that did not exist before 1935. Ohno (1984) proposed it arose from a frameshift in an existing gene; later structural work indicates it derives from an existing esterase through a small number of substitutions in the active site. Either way, an enzymatic activity now exists for a substrate that did not exist a century ago.

Lenski's populations produced aerobic citrate metabolism after roughly 31,500 generations, a capacity E. coli is defined as lacking, by placing an existing transporter under a promoter active in oxygen. These cases and the argument around them are covered in Genetic Information.

The same process outside medicine#

The commercial response is the useful detail. Refuge planting — deliberately leaving unsprayed areas so susceptible individuals survive and dilute resistance alleles — is a regulatory requirement for Bt crops in the United States, and antibiotic stewardship programmes exist for the same reason. Both are evolutionary predictions applied as policy.

Common objections#

"It's still bacteria — this is microevolution, not macroevolution"

Correct, and no biologist claims otherwise. Nobody expects E. coli to become a multicellular organism in eleven days.

What resistance establishes is that the mechanism works as described: random mutation supplies variation, selection changes population frequencies, and new capacities appear. The dispute over whether accumulated change can produce larger differences is addressed by speciation, fossils, and genomes — separate lines of evidence.

The point here is narrower and worth keeping separate: the engine is observed running.

"Resistance genes already existed, so nothing new evolved"

Often true, and the Lechuguilla Cave and permafrost results confirm it directly.

Selection on pre-existing variation is not a competing explanation — it is one of the mechanisms. Darwin's account requires variation to exist before selection acts on it, which is exactly what Luria–Delbrück and replica plating showed.

De novo cases are also documented. The Baym plate captured mutations arising within the experiment, with the lineage's history recorded on film, and resistance mutations in Mycobacterium tuberculosis are routinely sequenced as they appear within individual patients during treatment.

"Resistance carries a fitness cost, so it's degeneration"

Frequently true. Many resistance mutations reduce growth rate without the drug, which is part of the rationale for cycling antibiotics.

It does not stay true. Compensatory mutations that restore fitness while retaining resistance are documented in Staphylococcus aureus, M. tuberculosis, and E. coli, which is why resistance often persists after use of a drug declines.

Fitness is also relative to environment, not absolute. A variant that grows slowly in a drug-free flask and survives in a hospital is not degenerate; it is adapted to the environment it is in. That is the whole content of the term.

"Mutations only destroy information — resistance is loss, not gain"

Some resistance is loss: a deleted porin channel that the drug used to enter through, or a disabled enzyme that used to activate a prodrug.

Much of it is not. Beta-lactamases are functioning enzymes that cleave the drug, they have diversified into thousands of documented variants with expanding substrate ranges, and each expansion is an enzymatic capability the ancestor lacked.

The underlying issue is that "information" is rarely defined precisely enough to test. That argument is treated on its own terms in Genetic Information.

What the evidence shows#

Populations change measurably under selection, on timescales short enough to record. The Baym experiment filmed the process; Lenski's has run since 1988 with the ancestors frozen and available for replay.

The variation is present before the selective agent is applied, shown by Luria and Delbrück in 1943 and by the Lederbergs in 1952, and resistance genes predate clinical medicine by millions of years in isolated environments.

The process is general rather than a peculiarity of bacteria, documented across insects, weeds, rodents, and parasites, and its predictions are built into agricultural regulation and hospital protocol.

Whether the same mechanism accounts for larger differences is argued separately in Speciation, Transitional Fossils, and Common Descent.