TL;DR
  • Decomposition "stage" is a categorical label pasted onto a gradient with no real breaks — the same body can be "Active Decay," "Early," or "Butyric Fermentation" depending on which staging system your collaborator uses.
  • Total Body Score (TBS) and accumulated degree-days (ADD) are more information-preserving alternatives; they should be the primary decomposition variables in any quantitative study.
  • In necrobiome and microbiome work, treating stage as a categorical grouping variable destroys the microbial succession gradient that gives community data its PMI-estimation power.
  • Practical fix: sample at defined ADD intervals, score TBS regionally at every time point, and use ADD and TBS as continuous covariates in your statistical models — not as factors.
  • Always report environmental context (continuous temperature logs, substrate, insect access). TBS and ADD values without environmental history are not cross-study comparable.

Every working definition of a decomposition “stage” draws a line across a gradient that has no line in it. Fresh shades into bloat. Bloat shades into active decay. The body does not receive a memo. The researcher does.

This is not a stylistic complaint. The way decomposition is measured and encoded as a study variable has direct downstream consequences for what analyses are valid, what comparisons are meaningful, and whether findings replicate across sites and researchers. The problem matters especially for microbial and necrobiome studies, where community succession is inherently time-resolved — and where using broad stage categories as grouping variables throws away exactly the information that distinguishes one community state from another.

What “Stage” Has Historically Meant

The classic conceptual framework of vertebrate decomposition identifies a series of recognizable phases. The most commonly cited scheme — fresh, bloat, active decay, advanced decay, and dry/skeletal — has five stages, though implementations vary: some systems use four, others six, and the criteria for transition differ meaningfully between them. Bass and Sledzik’s four-stage system (fresh, early, advanced, skeletonized) and Goff’s entomological staging do not map cleanly onto one another, despite describing the same biological process in the same species. There is no governing standard.

Within any of these systems, stages are assigned by observer judgment of gross morphological features: skin discoloration, bloating, skin slippage, tissue liquefaction, exposure of bone. These assessments are categorical by convention, not by the biology they represent. A body that is “mostly” in active decay but retaining some trunk integrity sits at a boundary that different observers — and different staging systems — will resolve differently. Finaughty et al. (2023) identified the lack of standardization in decomposition scoring methodology as one of the core bottlenecks limiting progress in quantitative forensic taphonomy, noting that inconsistent protocols make inter-study comparisons unreliable.1

The same continuous decomposition gradient interpreted by three published staging systems. Specimens A and B (colored dots) sit near cross-system boundary zones: Specimen A is "Bloat" under Goff's 5-stage scheme, "Early" under Bass, and "Black Putrefaction" under Vass. Specimen B maps to three different stage labels across the same three systems. The gradient has no natural lines in it — only the systems do.

The Partial Fix: Scoring Systems and Thermal Summation

The most important methodological advance toward continuous measurement of decomposition came from Megyesi, Nawrocki, and Shahrom (2005), who introduced the Total Body Score (TBS). Rather than assigning a single categorical label to the whole body, TBS separately scores the condition of the head and neck, the trunk, and the limbs, then sums the sub-scores into a composite ranging from 0 (fresh) to 35 (fully skeletonized). TBS is operationally continuous in a way that stage labels are not: two bodies at the same nominal “active decay” stage may produce TBS values separated by ten or more points, representing meaningfully different states of tissue loss and regional asynchrony.

TBS becomes most useful when paired with accumulated degree days (ADD), a thermal summation metric borrowed from agricultural entomology. ADD integrates the number of degrees above a base temperature (typically 0°C) over time, providing a temperature-adjusted proxy for elapsed PMI. The appeal is obvious: a body decomposing over a week during a heat wave is not equivalent to one decomposing over a week in cold weather, and raw elapsed time fails to capture that. Multiple groups have used TBS + ADD as a framework for PMI estimation from decomposition morphology, including Simmons et al. (2010), who measured decomposition rates every ~50 ADD in buried versus surface rabbit remains, and Dautartas et al. (2018), who applied TBS and ADD to directly compare decomposition trajectories of humans, pigs, and rabbits at the University of Tennessee’s Anthropology Research Facility.2,3

TBS and ADD represent real progress. Both are more information-preserving than categorical staging, and the TBS + ADD framework has enough uptake that results are now meaningfully comparable across some sites and studies. The Burcham et al. (2024) study in Nature Microbiology — one of the most methodologically thorough necrobiome studies to date, tracking 36 human cadavers across three climatically distinct locations — used TBS to characterize decomposition state when coordinating sampling across sites, recognizing that a shared continuous metric was necessary for any cross-site comparison to be valid.4

What These Metrics Still Get Wrong

Neither TBS nor ADD is the underlying process. Both are summaries, and both introduce assumptions that limit their accuracy.

TBS collapses regional asynchrony. Decomposition is not spatially uniform. The abdomen typically bloats and liquefies before the extremities, the face before the torso, and exposed skin before covered skin. TBS accounts for this by scoring body regions separately, but the sub-scores are then summed into a single composite — discarding the regional variation that contains biologically meaningful information. A body with high abdominal TBS and low limb TBS versus a body with moderate scores in both regions may produce the same total, but they represent different states of decomposition with different microbial community profiles and different entomological dynamics. Hyde et al. (2013) recognized this directly: rather than sampling cadavers at a single “bloat” label, they sampled at the onset of bloat and again at its end, capturing within-stage community turnover that a single categorical label would have erased.5

Simulated TBS sub-score trajectories for three body regions over accumulated degree-days. The trunk decomposes fastest, the limbs slowest. The yellow band marks the ADD range an observer would typically label "Active Decay" — but within that window, the three regions are at dramatically different sub-scores. Two bodies with the same composite TBS can have entirely different regional profiles depending on where each region falls on its own trajectory.

ADD assumes a linear thermal relationship. The model treats each degree-day as equivalent, but there are well-documented nonlinearities. Freeze-thaw cycles do not simply decelerate decomposition and then accelerate it symmetrically; the physical disruption of freezing alters tissue structure in ways that affect subsequent microbial colonization. Extreme summer temperatures can desiccate surface tissues, slowing decomposition in ways that degree-day accumulation does not predict. Pittner et al. (2020) found in a field evaluation of PMI methods that TBS + ADD performed poorly in certain seasonal conditions, particularly in advanced decomposition stages where the model’s linear assumptions diverge most from observed trajectories.6

Cumulative ADD accumulation over 14 calendar days under cool (~7°C average) versus hot (~30°C average) conditions. Colored bands show approximate stage boundaries in ADD space. After 14 days, the cool scenario reaches roughly 99 ADD — still in early Bloat — while the hot scenario exceeds 410 ADD, placing the body into Advanced Decay. Calendar time is not decomposition time.

Validation of established methods has been incomplete. Suckling, Spradley, and Godde (2015) tested the Megyesi TBS-to-ADD equation longitudinally using human remains donated to the Forensic Anthropology Research Facility at Texas State University and found statistically significant discrepancies between predicted and actual ADD across multiple donor cases — indicating that even the best-validated scoring system produces unreliable estimates when applied outside the conditions in which it was developed.7 Ribéreau-Gayon, Carter, and Forbes (2023) made this point operationally when they developed an entirely new scoring method for humid continental climates in Quebec, having found that the existing literature offered no validated method applicable to their conditions.8 The need to rebuild scoring systems for each major climate zone is itself evidence that decomposition is not a universal sequence of discrete stages but a climate-contingent continuous trajectory.

The Downstream Problem: Necrobiome and Microbiome Studies

Microbial succession during decomposition is among the most temporally structured community processes in ecology. Metcalf et al. (2013) demonstrated in a mouse model that postmortem microbial communities change in ways that are dramatic, measurable, and repeatable, allowing PMI estimation to within approximately three days across a 48-day window — a level of precision that would be destroyed if the data were collapsed into coarse stage bins.9 The reason microbiome data is useful for PMI estimation at all is that community composition at day 12 differs from community composition at day 15 in predictable ways. Treating “active decay” as a single group level conflates everything from day 10 to day 30 (depending on temperature and environment), creating within-group heterogeneity that masks the very gradients the analysis is meant to detect.

When decomposition stage is encoded as a categorical factor in a necrobiome study, a specific set of problems follows:

Arbitrary group boundaries create artificial discontinuities. The transition from “bloat” to “active decay” is assigned by an observer at a discrete decision boundary. Two samples collected one day apart — one at the last day of bloat, one at the first day of active decay — will be placed in different groups and treated as belonging to categorically different conditions, even though their communities may be near-identical. Conversely, the first day of active decay and the last will be placed in the same group, even if their communities are maximally dissimilar.

Within-stage variance is suppressed in group-level analyses. PERMANOVA, ANOVA-on-diversity, and other group-comparison methods treat each stage as a unit. The variance within stages is pooled, not modeled. If the primary axis of community variation is continuous time or degree-days, forcing that axis into discrete bins will systematically reduce statistical power and can produce misleading results if the within-group variance is large relative to between-group variance.

Cross-study comparisons become undefined. A body in “active decay” at 20°C after four days does not correspond to a body in “active decay” at 30°C after two days, even if both received the same stage label. With no common reference variable — only shared label — microbial community data from different studies at the same “stage” may represent entirely different points on the decomposition continuum.

Simulated necrobiome community structure along a decomposition gradient, rendered two ways. In the continuous view, dots are colored by ADD — the gradient structure is immediately visible and each sample is uniquely positioned. In the categorical view, dots are colored by observer-assigned stage label. Within-stage spread is large: samples at the same stage label span a wide range of community states, and the gradient that drives community turnover is invisible at the group level. Toggle between views to see how the same data reads differently depending on how decomposition state is encoded.

Belk et al. (2018) and Johnson et al. (2016) both built random forest models for PMI estimation from microbiome data, and both operationalized decomposition as a continuous variable (elapsed time or ADD) rather than a stage category — a choice that appears directly related to the predictive accuracy their models achieved.10,11 The continuous framing allows the model to learn gradients; the categorical framing destroys them.

Using TBS and ADD in Necrobiome Studies

The argument for continuous decomposition metrics is general, but the practical stakes are highest in necrobiome research, where the fundamental question is which microbial community states occur at which points along the decomposition continuum — and where the answer depends entirely on how that continuum is measured and encoded. The figure below makes the relationship between TBS, ADD, and calendar time concrete: drag the slider to move through ADD space and watch how the three body regions decompose at different rates, how the composite TBS tracks behind the fastest region, and how radically the same ADD value translates into different calendar durations depending on ambient temperature.

Interactive TBS body schematic. The slider moves through accumulated degree-days (ADD 0–500). Each body region — head/neck, trunk, and limbs — decomposes at a different rate; fill colors track the regional TBS sub-score from fresh (green) through active decay (orange/red) to skeletonized (gray). The metrics panel shows regional sub-scores, composite TBS, the observer-assigned stage label, and the equivalent number of calendar days in a cool versus hot climate at that ADD value. Note how two very different calendar durations produce the same TBS state, and how regional sub-scores can diverge substantially even at the same composite score.

Sample at defined ADD intervals, not stage transitions. Tying sample collection to stage transitions anchors your data to an observer-assigned decision boundary. Two researchers observing the same body on the same day may disagree about whether it has transitioned from bloat to active decay, and that disagreement places samples in different groups. Sampling every 50 ADD (or every 100 ADD, depending on study duration) instead produces samples at known, reproducible positions on the thermal-time continuum. This requires a temperature logger deployed at the body surface throughout the study, but the resulting data can be pooled with any other dataset that follows the same protocol, making cross-study comparison meaningful rather than label-dependent.

Record TBS sub-scores at every sampling event. TBS composite and all three regional sub-scores should be logged for each sample alongside ADD. If sampling is body-region-specific — skin swabs, soil cores adjacent to the torso, internal cavity swabs — the regional sub-score for that body area is a more relevant covariate than the composite. A trunk surface swab taken when the trunk sub-score is 7 is not equivalent to one taken when it is 3, even if the composite TBS is the same in both cases because the other regions happen to differ.

Treat ADD or TBS as continuous covariates in statistical models. In vegan’s adonis2, pass ADD or TBS as a numeric vector, not a factor. In linear models for alpha diversity, use ADD as the predictor variable. In random forest or gradient boosting frameworks, ADD and regional TBS sub-scores are natural continuous features. Encoding stage as a nominal variable in any of these contexts converts an ordered gradient into a set of unordered categories, destroying the very information that makes microbiome succession tractable as a research question.

Decision guide across the three phases of a necrobiome study. Each phase lists the common default approach (✗) and the preferable alternative (✓) with a brief rationale. The pattern is consistent: wherever stage appears as a categorical variable, replacing it with ADD or TBS — as a continuous metric in the appropriate context — produces data that is more quantitative, more reproducible, and more analytically flexible.

What Better Practice Looks Like

None of this argues for abandoning descriptive staging entirely. Stage labels are useful for rapid field communication, for records with incomplete environmental data, and for organizing literature syntheses. The problem is when categorical stage is treated as the primary independent variable in quantitative analyses that are sensitive to the distinction between ordinal bins and continuous gradients.

Report time since death whenever it is known. For studies at body farms, taphonomy facilities, or controlled experiments where time of death is known, time or thermal-summation metrics (ADD) should be the primary axis. Stage category can be reported as a descriptor but should not substitute for the quantitative variable.

Use TBS rather than stage categories for decomposition state. TBS retains meaningful variance that stage labels suppress. Where body-region-specific dynamics matter — and in microbiome studies they often do, since sampling site affects community composition — report regional sub-scores alongside the composite. Dawson et al. (2022) showed that mass loss provides a comparably informative continuous metric for studies with the infrastructure to measure it, offering another instrument-independent option.12

Model decomposition state as a covariate, not a factor. If stage must enter a statistical model, treat it as an ordinal or continuous predictor, not a nominal grouping variable. Better still, use TBS or ADD directly in regression frameworks rather than discretizing them. Dawson, Ueland, Carter, and Barton (2023) argued explicitly that forensic taphonomy should integrate continuous decomposition theory into its PMI models rather than relying on empirically derived but theoretically underspecified stage-based equations.13

Report environmental context. Decomposition state and environmental history are inseparable. Temperature records, precipitation, substrate, insect access, and season all determine where on the decomposition continuum a given TBS or ADD value actually sits. A TBS of 18 in a subtropical summer is not the same biological state as a TBS of 18 in a temperate autumn. Papers that report stage labels without environmental context cannot be meaningfully integrated into meta-analyses or cross-site comparisons.

The point is not that categorical staging is wrong — it is that it is a lossy compression of a continuous variable, and the degree of loss matters for the conclusions that can be drawn. Forensic taphonomy has been moving toward more quantitative, continuous metrics for decades. Necrobiome and decomposition microbiology should move with it.

References

  1. Finaughty DA, Pead J, Spies MJ, Gibbon VE. Next generation forensic taphonomy: Automation for experimental, field-based research. Forensic Science International. 2023;345:111616. doi:10.1016/j.forsciint.2023.111616
  2. Simmons T, Cross PA, Adlam RE, Moffatt C. The influence of insects on decomposition rate in buried and surface remains. Journal of Forensic Sciences. 2010;55(4):889–892. doi:10.1111/j.1556-4029.2010.01402.x
  3. Dautartas A, Kenyhercz MW, Vidoli GM, et al. Differential decomposition among pig, rabbit, and human remains. Journal of Forensic Sciences. 2018;63(5):1278–1287. doi:10.1111/1556-4029.13784
  4. Burcham ZM, Belk AD, McGivern BB, et al. A conserved interdomain microbial network underpins cadaver decomposition despite environmental variables. Nature Microbiology. 2024;9(3):595–613. doi:10.1038/s41564-023-01580-y
  5. Hyde ER, Haarmann D, Lynne AM, Bucheli SR, Petrosino JF. The living dead: bacterial community structure of a cadaver at the onset and end of the bloat stage of decomposition. PLoS ONE. 2013;8(10):e77733. doi:10.1371/journal.pone.0077733
  6. Pittner S, Bugelli V, Weitgasser K, et al. A field study to evaluate PMI estimation methods for advanced decomposition stages. International Journal of Legal Medicine. 2020;134(4):1361–1373. doi:10.1007/s00414-020-02278-0
  7. Suckling JK, Spradley MK, Godde K. A longitudinal study on human outdoor decomposition in central Texas. Journal of Forensic Sciences. 2016;61(1):1–8. doi:10.1111/1556-4029.12892
  8. Ribéreau-Gayon A, Carter D, Forbes SL. Developing a new scoring method to evaluate human decomposition in a humid, continental (Dfb) climate in Quebec. Journal of Forensic Sciences. 2023;68(2):476–489. doi:10.1111/1556-4029.15201
  9. Metcalf JL, Parfrey LW, González AP, et al. A microbial clock provides an accurate estimate of the postmortem interval in a mouse model system. eLife. 2013;2:e01104. doi:10.7554/eLife.01104
  10. Belk AD, Xu ZZ, Carter D, et al. Microbiome data accurately predicts the postmortem interval using random forest regression models. Genes. 2018;9(2):104. doi:10.3390/genes9020104
  11. Johnson H, Trinidad D, Guzman S, et al. A machine learning approach for using the postmortem skin microbiome to estimate the postmortem interval. PLoS ONE. 2016;11(12):e0167370. doi:10.1371/journal.pone.0167370
  12. Dawson BM, Wallman JF, Barton PS. How does mass loss compare with total body score when assessing decomposition of human and pig cadavers? Forensic Science, Medicine and Pathology. 2022;18(3):282–293. doi:10.1007/s12024-022-00481-6
  13. Dawson BM, Ueland M, Carter D, McIntyre D, Barton PS. Bridging the gap between decomposition theory and forensic research on postmortem interval. International Journal of Legal Medicine. 2023;137(6):1763–1778. doi:10.1007/s00414-023-03060-8