First distinguish source evidence from pathogen evidence
Groundwater microbial source tracking, often shortened to MST, asks whether fecal contamination might have come from humans, ruminants, birds or another host group. A result labelled HF183 qPCR is commonly used to investigate human-associated fecal material. It is not a direct count of sewage leaks, infected people, living pathogens or infectious particles. To read it correctly, identify the genetic marker, the assay, the units, the laboratory’s quality controls and the environmental setting before discussing a possible source.
The key distinction is between detecting a DNA sequence associated with a source and demonstrating a harmful living organism. Quantitative polymerase chain reaction, or qPCR, measures amplification of target nucleic acid under controlled conditions. DNA can persist after its source organism is no longer viable; conversely, sampling and extraction can fail to recover all material present. A positive HF183 result provides probabilistic source-related evidence that must be interpreted with controls and other measurements.
An immediately useful result sentence is: ‘A human-associated Bacteroides genetic marker was detected at a reported concentration in this groundwater sample by the specified qPCR assay, with the stated quality-control limits.’ It should not be compressed to ‘human sewage proved’ or ‘groundwater contains a confirmed human pathogen.’ Those stronger claims require evidence the assay does not inherently supply.
What is the actual HF183 target?
The name HF183 refers to a human-associated marker based on a region of Bacteroidales or Bacteroides 16S ribosomal RNA gene sequences. The specificity of a marker is a measured performance characteristic, not a magical property of the label. Research comparing human and animal fecal samples has found useful human associations but also demonstrated that no bacterial MST marker should be assumed perfectly host-specific under every setting.
EPA’s Method 1696.1 provides a method framework for characterizing human fecal pollution in water by the HF183/BacR287 TaqMan qPCR assay. Its specific validation context and quality-control requirements matter. A reader should verify whether a laboratory actually used that method or a related laboratory-developed assay; related designs can differ in primer sequences, calibration materials and reporting conventions. The method name is not merely a citation to add after interpreting the number.
A detection can be compatible with human fecal influence while leaving multiple possible physical sources: septic leakage, sewer infrastructure, wastewater-impacted recharge, combined drainage, or other connections. A marker does not choose among those possibilities without the location of wells, hydraulic gradients, infrastructure maps, storm history, sample timing and corroborating tracers. The translation from source-associated DNA to a specific leak is an inference requiring additional steps.
The path from groundwater to qPCR instrument
A typical molecular workflow begins with a measured water volume captured in a controlled sampling container. The laboratory may concentrate microorganisms or nucleic acids by filtration, extract DNA from the collected material, and place a known fraction of the DNA extract into individual qPCR reactions. Fluorescent signal is measured during repeated cycles of amplification. The software estimates a target quantity against a calibrated standard relationship or another validated quantification approach.
Each physical transfer introduces a denominator or recovery issue. The original water volume differs from the extract volume, and the reaction usually contains only an aliquot of the extract. If the final data table lists target copies per reaction, it cannot automatically be read as target copies per litre of groundwater. A transparent report documents the processed water volume, extraction and elution volumes, template volume per reaction, any dilution, and how these factors were applied.
The extraction process can lose material, while some groundwater constituents inhibit polymerase activity. Mineral-rich or organic-rich samples may therefore amplify less efficiently than laboratory-grade standards. A reliable protocol includes appropriate negative controls, positive controls, standards or reference material and an assessment of sample inhibition. The right result is sometimes a qualified estimate, an inhibited result or a request for reanalysis rather than an unqualified number.
Reading the amplification curve and Cq
An amplification curve displays fluorescence versus cycle number. In the exponential region, target copies increase through repeated reaction cycles. The quantification cycle, abbreviated Cq and sometimes Ct, is the cycle at which the fluorescent measurement passes a defined threshold or algorithmic criterion. Under comparable assay conditions, larger initial target quantities generally cross earlier. The Cq value is a laboratory signal descriptor, not a source-water concentration by itself.
Suppose two extracts yield Cq values of 28 and 31 using the same validated assay, run and comparable non-inhibited conditions. With ideal doubling each cycle, three cycles correspond roughly to an eight-fold difference in initial target. That is a teaching approximation. Actual amplification efficiency, calibration curve shape and sample processing factors determine the defensible ratio; comparing Cq from two different assays, instruments or extraction volumes without normalization is unreliable.
A late curve can be associated with low target quantities, nonspecific amplification, contamination or analytical noise depending on assay design and controls. The exact quality criteria must come from the laboratory’s method. An analyst cannot declare every late signal positive or every weak signal negative based only on an arbitrary universal Cq threshold. The amplification trace, controls and accepted reporting range must be read together.
Turning reaction copies into water concentration
Work through a fictional mass-balance conversion. A laboratory filters 250 mL of groundwater, elutes extracted DNA into 100 microlitres, and uses 5 microlitres of undiluted extract in one qPCR reaction. The reaction is estimated to contain 20 target copies. Assuming the extract is homogeneous and making no recovery correction, the 100-microlitre extract contains 20 × (100 ÷ 5) = 400 copies. The original 0.25 L sample then corresponds to 400 ÷ 0.25 = 1,600 estimated marker copies per litre.
That computation must not silently imply all 1,600 were living bacteria or that extraction recovered every copy in the original sample. It simply scales a reaction estimate to an equivalent water-volume basis under the given assumptions. If only half the collected material were recovered, the underlying concentration could be higher, but deriving a corrected value requires a method-appropriate recovery estimate and an explicit uncertainty model. Multiplying by an invented correction factor would reduce rather than improve integrity.
Now change the template to 5 microlitres from a ten-fold diluted DNA extract. If the qPCR result still estimates 20 copies in that diluted aliquot, an extra factor of ten applies before reconstructing the original extract, giving 16,000 copies/L under otherwise identical assumptions. Forgetting the dilution factor yields a ten-fold error. A simple dimensional audit follows copies/reaction → copies/microlitre of extract → copies per entire extract → copies per litre of original water.
The calibration line is a measurement model
A qPCR standard curve relates Cq to the logarithm of a known target quantity. Many conventional assays plot Cq on the vertical axis against log base ten of copies on the horizontal axis. A negative slope is expected because higher concentrations cross the fluorescence threshold earlier. An idealized assay doubling product in every cycle has a slope around minus 3.322 cycles for a ten-fold target change, subject to the chosen mathematical conventions.
One illustrative calibration equation is Cq = 38 − 3.32 log10(copies per reaction). At 100 copies, log10 is two and the predicted Cq is 31.36. At 1,000 copies, log10 is three and the predicted Cq is 28.04. The difference is 3.32 cycles, consistent with the assumed ideal ten-fold step. This example demonstrates the algebra; it is not the calibration curve of any actual groundwater sample or instrument.
An analyst should review accepted standards, their range, replicate behaviour and method-specific acceptance criteria. A high coefficient of determination alone cannot rescue a curve with an inappropriate efficiency or degraded reference material. Extrapolating far below the lowest validated standard can produce a precise-looking number with weak quantitative support. Report detected but not quantifiable when the assay’s accepted convention calls for it, not an imaginary exact concentration.
Efficiency and the difference between detection and quantification
Amplification efficiency is often computed from a calibration slope using E = 10 raised to the power of minus one divided by the slope, minus one. Multiplying E by 100 expresses a percentage. With a slope of minus 3.32, that formula gives approximately 100 percent efficiency. A slope of minus 3.80 gives roughly 83 percent. These values illustrate why Cq differences cannot always be translated using the naive ‘one cycle equals twice as much’ rule.
Limits of detection and limits of quantification answer different questions. A target may be seen reproducibly enough to count as detected at concentrations too low for reliable numerical quantification. The detection limit is not a universal fixed Cq across all laboratories. It depends on the method, target, matrix, accepted false-positive and false-negative risk, and other validation details. Where an assay reports DNQ, detected but not quantifiable, preserve that category.
Suppose a fictional result is ‘HF183 detected, below the quantification limit’. A dashboard designer replaces it with zero because the numerical field is empty. That destroys evidence of a detected target. Replacing it with the quantification limit as though it were measured is also misleading. Retain the detection flag, the censored numeric status and the relevant reporting limits, and analyse trends using procedures appropriate for censored observations.
Inhibition is not a mere machine warning
Groundwater can contain substances that interfere with DNA extraction or polymerase amplification. Humic substances, metals or other matrix constituents may reduce measured signal. A sample can contain target material yet produce a delayed curve or no acceptable amplification. An internal amplification control or another method-specific inhibition check asks whether that sample matrix behaved as expected compared with a control condition.
A common diagnostic exercise tests a sample extract both undiluted and diluted. Dilution decreases both the inhibitor and the target concentration. If inhibition eases, the diluted extract may yield a more coherent result than a theoretically stronger undiluted sample. But dilution also raises the effective detection limit for the original water sample. The correct calculation restores the dilution factor and states whether the new result lies within the validated quantification range.
Imagine a target-positive extract gives Cq 35 undiluted, while a ten-fold dilution gives Cq 33, rather than the expected approximately 3.3-cycle later signal. Such counterintuitive behaviour suggests the undiluted reaction was impaired or another assay issue occurred; it does not prove a numerical inhibition correction by itself. Review control response, amplification curves, reagent performance and method rules before selecting a defensible corrected interpretation.
Read negative controls as separate failure detectors
A no-template control receives the relevant reaction mixture but no sample DNA. Its role is to reveal target material introduced through reagents or setup. A method or extraction blank passes through more of the preparation process and can reveal contamination introduced while handling filters, tubes or extraction reagents. A field blank can additionally expose problems during field handling or transport. Each control interrogates a different section of the workflow.
If a no-template control is positive, it is unsafe to interpret low-level sample positives as clean independent detections without a method-specific investigation. A negative no-template control does not prove field collection was uncontaminated; that is why other blanks exist. Similarly, a positive extraction control says amplification is technically possible, but does not establish that environmental extraction losses or sample inhibition were absent.
A useful reporting matrix separates three questions: was the marker detected in the sample, did the negative controls remain acceptable, and did sample-specific inhibition checks pass? The answer can be yes/no/pass/fail/qualified rather than a single red or green icon. This gives the eventual reader a way to distinguish valid low-level results from technical failures and prevents laboratory uncertainty from disappearing in summaries.
Recovery controls and sample processing losses
Filtering a litre of water, extracting a filter into an elution tube and loading a few microlitres into the thermocycler is a sequence of physical transfers. Material can remain attached to the membrane, degrade, fail to lyse, bind to extraction reagents or be lost during cleanup. Recovery controls add known material before a defined processing stage to test a portion of that chain. Their meaning depends on exactly where they are introduced.
If a control is added after extraction, it does not test extraction efficiency, only later amplification-related performance. If a spike is added before concentration, its observed recovery may reflect collection, filtration, extraction and measurement together, although it might not mimic the native target perfectly. A single percent-recovery number is not automatically a valid correction for every environmental DNA marker. Refer to the method’s requirements and documented uncertainties.
Suppose a simulated control contains 10,000 copies and 4,000 are recovered: apparent recovery is 40 percent. A naive correction would divide an observed marker estimate by 0.4, multiplying by 2.5. That arithmetic is transparent, but using the corrected concentration as truth would require evidence that the control’s losses accurately represent native HF183 target losses. Present the measured estimate and qualified recovery context rather than applying an unjustified correction.
Host association is probabilistic, not absolute
An HF183 detection increases the plausibility of human-associated fecal material in a setting where the assay is sufficiently specific and controls are sound. It cannot by itself prove that every target sequence came from humans. Cross-reactivity with some nonhuman fecal material and differences among host populations have been evaluated in original research. Likewise, marker absence can reflect low concentration, inhibition, degradation, or sampling a time and place that missed a plume.
A strong source investigation uses multiple lines of evidence. Host-associated markers can be combined with sanitary maps, indicators of sewage influence, water-level relationships, rainfall responses, upstream/downstream samples and, where appropriate, markers associated with other animals. A second human-associated assay can help, but agreement should be evaluated with regard to shared biases rather than treated as independent certainty by default.
Bayesian reasoning clarifies the problem. The probability of a human source after a positive result depends partly on how frequently a human source was plausible beforehand, the assay’s sensitivity to genuine human contamination and its false-positive rate for other sources. A claim of ’99 percent human’ cannot be borrowed from a particular study and applied universally to an unknown groundwater well. Site priors, reference populations and assay performance must be relevant.
Groundwater is not a river sampling problem in miniature
Surface-water microbial source tracking can draw on mixing along a stream, outfalls and changing upstream contributions. Groundwater may move through matrix pores, fractures, karst conduits and imperfect well construction. Microbial markers can attach to particles, be filtered by sediments, decay at different rates from chemical tracers or arrive through a short circuit. A single groundwater well integrates unknown portions of this subsurface system, often over a screen interval rather than a single depth.
Before inferring source, note well depth, screened interval, confining layers, pumping history, nearby infrastructure, recent rainfall and hydraulic head. A human-associated signal at a shallow well after rain could reflect rapid recharge through a preferential pathway or a defect near the wellhead, among other possibilities. A detection at depth raises important questions but does not automatically confirm deep aquifer-wide transport; a poorly sealed borehole can connect zones that would otherwise be separated.
Do not interpolate HF183 copies/L across space as if it were a conservative dissolved chloride plume. Genetic material has different particle attachment, retention and decay characteristics and the sampled organisms may not track water molecules exactly. A spatial pattern is an observation about the sampled network and dates. A convincing hydrogeological interpretation needs an explicit conceptual model and field tests that could disconfirm a suspected pathway.
Design a monitoring round that can distinguish explanations
Imagine a fictional village well near several possible contamination sources. First draw the pathways: surface drainage, septic or sewer infrastructure, groundwater-flow direction, construction details and treatment components. A meaningful network may include raw wellhead samples, appropriate background wells, plausible upgradient and downgradient sites, and selected storm-condition samples. The correct arrangement depends on hydrogeology and practical access; a map of sample labels alone is not an investigation design.
Pair molecular source markers with culture indicators where appropriate, while recognising they answer different questions. A positive HF183 with no culturable E. coli could reflect differing persistence, sensitivity or timing, rather than an automatic laboratory contradiction. The converse—culturable E. coli without HF183—does not prove a purely animal source, because a human marker may be absent or undetectable under the sample and analytical conditions.
Repeat over seasons and events rather than collecting all evidence in one calm afternoon. A dry-weather baseline, a rain-event sample and a follow-up after water levels recede can ask whether contamination is episodic. Still, those events may not be directly comparable if sampling volumes or extraction approaches differ. Record method versions, dilution factors, blank results, sample inhibition and calibration changes to keep an apparent environmental trend separate from analytical drift.
Worked case: a marker rises but the well is not yet diagnosed
Suppose fictional Well A reports 200 HF183 copies/L in dry weather and 4,000 copies/L after rain, with valid comparable procedures. The second reported concentration is twenty times the first. That supports a changed marker signal at the sampled point. It does not show a twenty-fold rise in infectious pathogens, twenty times as much sewage by mass, or a specific contaminant-entry route. Those are different questions requiring different measurements.
The next review checks hydraulic conditions, possible dilution of other tracers, raw sample volume, inhibition controls and the presence of any procedural change. If the post-rain sample required DNA dilution, confirm that its factor was restored correctly. If its negative controls were clean and inhibition acceptable, the increase becomes more credible. Pairing it with wellhead inspection and event-synchronous samples at adjacent points may make one pathway hypothesis more plausible than alternatives.
Transfer problem: both reported concentrations were derived correctly, but the dry-weather sample was processed from 500 mL and the rain-event sample from 50 mL due to turbidity. Can they be compared? Potentially, if the methods remain validated, effective reporting limits and recoveries are appropriate and the result estimates were normalized correctly. The different volumes change detection sensitivity and uncertainty; they must not be hidden merely because both display copies/L.
Worked case: templates, dilutions and unit conversions
Assume 1 L of groundwater is filtered and eluted to 200 microlitres. Five microlitres of undiluted extract gives a calibrated estimate of 50 copies per reaction. The implied extract contains 50 × 200/5 = 2,000 copies, giving an apparent 2,000 copies/L before any recovery adjustment. Now suppose the laboratory uses a five-fold dilution of that extract and again measures 50 copies in five microlitres. The inferred original extract corresponds to 10,000 copies/L, not 2,000.
That factor of five must be reconciled in the report. A spreadsheet that multiplies by five twice reports 50,000, also wrong. Independent dimensional checking asks whether the calculation uses reaction copies, the undiluted-equivalent template volume, the total extract volume, and the original water volume exactly once. Volumes in microlitres and litres must be converted consistently. The method may have additional processing factors, so this simplified example does not replace its prescribed equation.
Transfer problem: a 250-mL sample is extracted into 50 microlitres, and a 2-microlitre aliquot holds 12 calibrated copies. Without dilution or recovery corrections, extract copies are 12 × (50/2) = 300, and water concentration is 300/0.25 = 1,200 copies/L. A proposed result of 24 copies/L is incorrect because it divided reaction copies by an unrelated volume. The final units should be reconstructed from the sampling chain.
How multiple markers help without becoming magic proof
Suppose human-associated HF183 and HumM2 markers are both detected, while a ruminant-associated marker is not. This pattern may strengthen a human-fecal-influence hypothesis, but its strength depends on each method’s performance, the host populations studied, the number of samples and the observed environmental context. Some marker assays share biological or sampling dependencies, so their agreement is not necessarily equivalent to two independent experiments.
Another hypothetical sample detects both human-associated and ruminant-associated markers. A simplistic dashboard forces the case into a single-source category, hiding possible mixed contributions. A better explanation states that evidence is compatible with more than one host-associated source, while recognizing that cross-reactivity, differential persistence and collection conditions could affect relative signals. The work is to test a mixture hypothesis, not to select a winner by comparing raw copy counts from distinct assays.
Do not treat ten HF183 copies as chemically equivalent to ten copies of another marker. Different organisms and genomes have different target-copy frequencies and different environmental behaviour, while qPCR assays can differ in recovery and calibration. The quantitative comparisons are most meaningful within a validated marker–method–matrix combination. Source fractions require much more than dividing unrelated marker concentrations by their sum.
When pathogen-risk language outruns the data
Gene copies are not disease-causing organism counts. Even an assay targeting a gene associated with a pathogen may detect DNA from damaged or nonviable organisms. A source-associated assay such as HF183 is an additional step removed: it reports a marker related to fecal material from a host class. Water-safety decisions still need exposure context, treatment performance, direct pathogen evidence when appropriate, and applicable authority guidance.
The U.S. Geological Survey has studied multiple pathogens and fecal markers in public-supply wells, illustrating that molecular measurements can reveal contamination missed by a narrow panel. The same research also underscores challenges in translating qPCR measurements into regulatory risk. A negative HF183 does not mean ‘safe to drink’, and a positive HF183 does not quantify the number of infections that would result from a litre of water.
An especially important communication rule is to separate observed detection, plausible source and protective recommendation. The first is an analytical statement. The second is an interpretation with uncertainty. The third is a decision that should integrate health guidance, local context and precaution. Keeping the three sentences separate avoids both exaggeration and false reassurance.
Read reporting qualifiers before drawing a map
Some qPCR tables distinguish not detected, detected but not quantifiable, quantified within range and failed because of inhibition or other control criteria. Turning all these states into numbers such as zero, half a reporting limit or the quantification limit without a declared statistical rationale can manufacture trends. A responsible data dictionary records each state separately and preserves the original analyst’s qualification text.
A marker result may be expressed as copies per litre, log10 copies per 100 millilitres or another convention. Ten to the third copies per 100 millilitres equals 1,000 copies per 100 millilitres, or 10,000 copies per litre; it is not 1,000 copies per litre. Mixing linear and logarithmic columns can create even larger errors. If a graph uses a log scale, label its tick marks and report zeros or censored values without pretending log10(0) exists.
Quality flags belong on maps as well as in the master table. A dot representing an inhibited sample should not be drawn identically to an analytically valid nondetection. A high-copy sample that was obtained outside the reliable quantification range may deserve a distinct annotation. Good visualization makes the spatial question easier to ask, while retaining enough measurement provenance that another analyst can challenge a proposed source.
Set boundaries on comparisons across laboratories
Two laboratories can use different versions of HF183 assays, extraction chemistries, concentration procedures, standard materials and inhibition checks. Even when both report copies/L, that common unit does not guarantee full analytical interchangeability. A time series spanning a laboratory transition needs bridging samples, interlaboratory comparison or at least a clearly labelled method change. Otherwise an apparent trend could reflect altered recovery or calibration rather than changed groundwater.
Within one laboratory, different batches also deserve scrutiny. Standard material degradation, changes to extraction reagent lots or altered filtration volumes can affect results. Blanks, controls, standard curves and replicate performance provide evidence about those batch effects. A reproducible trend should be robust enough to withstand these checks and, when possible, be supported by independent hydrologic or chemical observations.
A field duplicate asks how similar two samples collected close together behave under the full sampling and processing procedure. A technical qPCR replicate asks how consistent multiple reactions from an extract are. Those are different sources of variation. Treating tiny technical replicate variance as proof of reproducible groundwater conditions is a common mistake. The variance of the final environmental conclusion includes far more than the thermocycler.
Specific questions readers frequently ask
Is HF183 proof that sewage leaked from a particular pipe? No. It is a human-associated molecular fecal marker and can support a source hypothesis when quality controls and local assay specificity are appropriate. The location of a pipe or septic unit, direction of groundwater movement, changes through time and corroborating markers are still needed to narrow the physical pathway. Investigating those relationships is more informative than naming a source from one well measurement.
Does a negative qPCR result prove the absence of all human fecal material? No. Sampling is finite, marker concentrations vary, and assay sensitivity may be constrained by low water volume, dilution or inhibition. The result should be phrased as ‘not detected under these analytical conditions’ and paired with the relevant reporting limit. A robust investigation may sample again during conditions in which the suspected entry process is most active.
Can Cq 29 from one laboratory be ranked directly against Cq 31 from another? Not reliably without demonstrating comparable calibration, efficiency, extraction and water-volume factors. The cycle difference itself has meaning only within a documented assay context. Convert results to method-defined concentration estimates and compare their uncertainty, then inspect whether differences exceed technical and environmental variation.
Does qPCR tell whether the fecal organisms are alive? Standard DNA qPCR generally does not distinguish viable from nonviable source cells. Additional methods may address particular viability-related questions, but they have their own assumptions and limitations. It is safer to communicate DNA-based evidence as genetic detection rather than an equivalent culture count or pathogen dose. Parallel culture analysis provides complementary—not identical—evidence.
Is a ten-fold increase always dangerous? A ten-fold change in a genetic marker is potentially informative, but its health significance cannot be read from the ratio alone. Determine whether both measurements were valid, comparable and within quantifiable ranges, then consider exposure and possible pathogens. The responsible action may involve investigation or protective measures guided by the relevant local authority, rather than an arbitrary universal risk label.
What should a laboratory report provide? At minimum it should identify the marker and assay, sample identification, analysed volume, reported concentration and unit, reporting limits, dilution factors, relevant controls and deviations. A technical reviewer also benefits from extraction batch, standard curve quality, inhibition results and replicate information. A clear summary should make those records retrievable without burying the main finding beneath unexplained abbreviations.
Independent practice problems with worked answers
Problem one: a 500-mL groundwater sample is extracted into 100 microlitres. A 2-microlitre template aliquot is estimated at 8 copies, with no dilution. The extract contains 8 × 100/2 = 400 copies, and the sample concentration is 400/0.5 = 800 copies/L before recovery correction. A colleague reports 80 copies/L. Their answer has misplaced a factor of ten; dimensional analysis exposes the error.
Problem two: the same extract is diluted twenty-fold and the reaction quantifies 8 copies in a 2-microlitre aliquot. The undiluted-equivalent concentration is twenty times larger than the previous calculation, or 16,000 copies/L, provided the measurement is inside its validated range. That increase is an analytical dilution-factor correction, not evidence the groundwater changed between two reactions.
Problem three: a fictional calibration line is Cq = 39 − 3.32 log10(copies/reaction). If Cq is 29.04, then log10(copies) = (39 − 29.04)/3.32 = 3.00, so copies per reaction are approximately 1,000. If the Cq were 32.36, the same model gives approximately 100. The solution depends on the stated calibration; a different slope would change the inferred quantity.
Problem four: six wells are tested and two HF183 results are quantified, one is detected but below quantification, two are not detected, and one is inhibited. How many ‘confirmed zeros’ does the dataset contain? None. The two nondetections do not prove environmental zero, the below-quantification detection contains nonnumeric positive evidence, and the inhibited sample is not a valid nondetection. A full map should preserve all four result classes.
Problem five: a contamination hypothesis relies on a single detection after rainfall. Name two independent lines of inquiry. Suitable answers include controlled repeat sampling at the same point, inspection of wellhead seals, groundwater levels and gradients, a background monitoring well, evidence from infrastructure leakage assessments, or another marker with a complementary biological basis. Good answers specify the observation that would weaken the proposed hypothesis.
Problem six: two assays report 900 HF183 copies/L and 300 copies/L of a second host-associated marker. Can you infer 75 percent human source from 900 divided by 1,200? No. The target sequences do not represent comparable counts of fecal mass from two hosts; biological abundance, analytical recovery and host specificity all differ. Source apportionment requires a calibrated mixing or statistical approach, not a ratio of unrelated raw markers.
A practical reporting sentence that survives review
An example of defensible wording is: ‘A groundwater sample collected at Well W-4 on the stated date contained quantifiable HF183/BacR287 target DNA at the reported copies-per-litre estimate using the documented protocol. Negative controls were acceptable and the sample inhibition check passed. This supports considering human-associated fecal influence, but the result alone does not establish a specific sewage source or the concentration of viable pathogens.’ Every component is tied to a question the measurement can actually answer.
That wording invites the right next work rather than dictating a premature verdict. If results repeatedly rise after rainfall, examine hydraulic pathways and construction integrity. If results are inconsistent between wells, review spatial coverage, pumping and aquifer connectivity. If controls fail, resolve analytical validity before mapping the observation. If a water-use or health decision is imminent, consult qualified professionals and the applicable public-health authority; source tracking is not an alternative to protective decision-making.
Readers who want to strengthen their precision in technical language can use the eduKateSG Vocabulary Articles and Knowledge Hub, while The Gold Standard of Hydrogeology develops the physical groundwater context. The transferable skill is to separate a directly observed target signal, an interpreted source association, and a practical action. Keeping those layers distinct makes a scientific report more useful, not less decisive.
Original methods and specialist research
Original and institutional sources: US EPA, microbial source tracking Methods 1696.1 and 1697.2; USGS, pathogens and fecal markers in supply wells; USGS, probabilistic source marker analysis; USGS, spike-and-recovery controls. Worked numbers and field scenarios above are explicitly illustrative.