Deviations in experimental procedure

On December 2nd 2022, both pairs of F hands were exhumed en bloc in their original taphonomic contexts, and individually placed in clean plastic bins, as below-freezing temperatures were expected. Initially, bins were stored in the ARISTA shed, and were later transferred to the AMC morgue on December 8th to reach their target ADD due to the increasingly cold weather (outdoor air temperature of 4.4 °C vs. indoor shed temperature at 5 °C on 08/12, average outdoor air temperature of − 0.2 °C on 09/12).

Scavenging was discovered on December 2nd 2022, with the F time point 14 left hand partially exposed, and the burial pit of the right hand disturbed without the hand being exposed. No damage to the hands was observed, and the pits were promptly refilled upon discovery. Scavenging likely occurred within 4 days of its discovery, as the heavy rainfall prior would have otherwise washed away the thick layer of soil present on the exposed hand.

Donor characteristics

No significant difference in age between male and female donors was found (p > 0.05, ANOVA).

Temperature, precipitation, and qualitative observations of decomposition

Air and soil temperatures were compared during each FFR exhumation to assess discrepancies between these. Variations not exceeding 1 °C were observed between air and soil at a depth of 20 cm. Therefore, air temperature was deemed suitable in calculating ADD, with a threshold of 0 °C, as below this temperature enzymatic and bacterial activity, and therefore decomposition, are inhibited [45, 46]. In instances where air temperature data was unavailable, data from a nearby weather station were used [28]. Table 2 details ADD for each hand pair. The discrepancy in ADD between F and FFR hands at time points 14 and 24 days reflect temperature differences during the burial and exhumation periods of late September to mid-October for FFR hands, and early November to mid-December for F hands.

Table 2 ADD for each hand pair

At time point 24 days, FFR hands were exposed to 6.5 mm less total precipitation (including watering) than the corresponding F pair, while at time point 14, F hands were exposed to 26.7 mm less total precipitation than the corresponding FFR pair. Pits for FFR hands were not left to dry for more than 24 consecutive hours, while F hands were not left to dry for more than 48 h at a time, with one exception where drying lasted for 72 h. These discrepancies reflect the challenge in maintaining consistency in moisture exposure across treatment groups due to the unpredictable nature of weather and inaccuracy of forecasts. Details on precipitation have been compiled in ESM J.

Hand discoloration was observed at all time points for both FFR and F hands. At time point 14 days, FFR hands had begun turning green, with extensive slippage and some brown discoloration at the severance points. By contrast, F hands at the same time point were greyish-pink with extensive skin slippage but no brown discoloration. At time point 24 days, FFR hands displayed extensive skin slippage and further brown discoloration, while F hands showed some brown discoloration, almost complete fingernail loss on the right hand, and loosely attached fingernails on the left hand, loose thumbs due to tissue mass loss, and blue tissue on the dorsal side of the left hand, likely due to a fungus. Refer to ESM K for detailed visual observations.

Fluorescence spectroscopy

Fluorescence signals in the wavelength range for PROT were saturated in all F skin and one F muscle sample at time point 0, and one F muscle sample at time point 14 days. An ND filter was used to measure these samples again. Three extreme outliers from the adipose dataset, one from the muscle dataset, and two from the skin dataset were removed prior to statistical analysis. An overview of these outliers is shown in ESM L.

Decomposition over time by tissue type and treatment

Relationship significance between time points for each tissue type within each treatment group, and corresponding scatter plots with lines of best fit are summarized ESM M. The best fit (highest R² possible for both treatment groups concurrently) trend line was obtained when applying a quadratic function to the datasets. R² and quadratic coefficient values are displayed in Table 3.

Table 3 R² and quadratic coefficient sign (+ or −) for each treatment group, by parameter and tissue typeSkin

PROT-FOX ratios also did not yield a clear significance pattern over time in FFR skin, while all comparisons in the F group resulted in significant differences. A trend line with a negative quadratic coefficient provided a poor fit for the FFR dataset. PROT-FOX ratios increased variably until time point 6 in FFR skin, followed by a decrease at time point 14, and remained relatively unchanged at time point 24 compared to time point 14. PROT-FOX ratios decreased over time in the F group, while a trend line with a positive quadratic coefficient resulted in a very good fit for this group.

Adipose

The ANOVA test produced significant differences between PROT AUCs for every time point comparison in adipose, except for the time point 0 to time point 6 comparison in the FFR group, and time points 14 to 24 in the F group. Quadratic trend lines displayed a moderate fit for the FFR dataset, and very good fit for the F dataset. In both instances, the quadratic coefficient was positive, with PROT declining over time in both groups.

Significant differences in FOX AUCs emerge from time point 14 in FFR adipose, and from time point 24 in F adipose. Trend lines with positive quadratic coefficients provided a moderate fit for each treatment group. An initial decrease in FOX was observed in both treatment groups, and was more marked in the FFR group, followed by an increase.

PROT-FOX ratios did not display a clear significance pattern in FFR adipose, but significant differences were present at all time point comparisons in the F group. A trend line with a positive quadratic coefficient provided a good fit for the F dataset, while the trend line for the FFR group showed a very poor fit and a negative quadratic coefficient. Similarly to F skin, PROT-FOX ratios decreased over time in F adipose, and displayed a slight increase until time point 14 in FFR samples, followed by a decrease at time point 24.

Muscle

Only comparisons to time point 24 produced significant differences between PROT AUCs in FFR muscle, while all comparisons in the F group yielded significant differences. A trend line with a positive quadratic coefficient produced a poor fit for the FFR dataset, and a trend line with a negative quadratic coefficient resulted in a moderate fit for the F dataset. PROT levels appeared to decrease until time point 14 in FFR muscle, followed by an increase, while in F samples, a decrease over time was observed.

FOX AUCs showed no clear pattern of significance over time in FFR muscle, but significant differences were present for comparisons to time point 24 in F samples. A trend line with a negative quadratic coefficient resulted in a very poor fit for the FFR dataset, while a trend line with a positive quadratic coefficient produced a good fit for the F dataset. Similarly to F adipose, FOX levels in F muscle displayed an initial decrease followed by an increase. On the other hand, FOX displayed a slight increase until time point 6, followed by a decrease in FFR samples.

PROT-FOX ratios displayed an inconsistent pattern of significance over time in FFR muscle, while significant differences emerged from time point 24 in F muscle. A trend line with a positive quadratic coefficient provided a very poor fit for FFR PROT-FOX ratios, which displayed a slight increase over time. A trend line with a negative quadratic coefficient yielded a moderate fit for the F dataset, with ratio values significantly decreasing at time point 24.

Fresh vs. fresh-frozen

ANOVAs and corresponding post hoc tests were conducted to identify significant differences between F and FFR samples at corresponding time points for the same tissue type. Significant differences between treatment groups were found for at least two parameters (PROT AUCs, FOX AUCs and/or PROT-FOX ratios) across tissue types, except for the comparison at time point 14 in skin samples, where no significant difference was found. Only one non-significant relationship was found in adipose. An overview of these results is displayed in ESM N.

Fluorophores in excitation-emission matrices

The endogenous fluorophores expected to be observed in samples, as displayed in Table 1, required revision due to discrepancies between expected and observed fluorophores. Consequently, the literature search was extended. Table 4 provides an overview of endogenous fluorophores potentially contributing to the fluorescence peaks observed in EEMs.

Table 4 Overview of endogenous fluorophores from literature potentially contributing to observed EEM peaks

Peak E was included in fluorophore peak monitoring despite no suitable endogenous fluorophore having been identified via a preliminary literature search, as it was observed in most EEMs and could potentially be an indicator of sample state or degree of decomposition. Lipofuscins/lipofuscin like lipopigments/ceroids, as described by [13], could have been a contributor to peak E, but this could not be substantiated with other scientific literature [10, 12]. Peaks A and C correspond to PROT and FOX peaks respectively. Peak D’s range is based on a single measurement at time point 0 in FFR adipose. Furthermore, six additional peaks observed in one or two EEMs each were excluded from analysis (see ESM O).

Decomposition over time by tissue type and treatment

Patterns of occurrence in time-dependent peaks, referring to peaks appearing either earlier or later in the decomposition process, were monitored. An overview of fluorophore peak occurrences is shown in ESM P. See Figs. 2 and 3 for the normalized EEMs of F and FFR tissue over time.

Fig. 2

Normalized EEMs for F skin, adipose, and muscle at time points 0, 14, and 24 days (n = 1 for each EEM)

Fig. 3

Normalized EEMs for FFR skin, adipose, and muscle at time points 0, 2, 6, 14, and 24 days (n = 1 for each EEM)

In skin, no peak showed time dependence in FFR samples, while two displayed time dependence in F samples. Peak B was only present at time point 24 in F skin, while peak E occurred at time points 0 and 14, with no discernible difference in fluorescence intensity between these time points.

One fluorophore peak displayed time dependent patterns in FFR adipose: peak E. This peak appeared only at time points 14 and 24 in FFR adipose, and showed a ~ 60% decrease in (fluorescence) intensity between these time points. On the other hand, no peak showed time dependence in the F adipose group.

Unlike in adipose, F muscle exhibited more time-dependent peaks than FFR samples. Peak E showed opposing patterns of occurrence, appearing only in undecomposed FFR muscle, and exclusively at time point 24 in F muscle. Peaks B, C, and D also showed time dependence in F muscle, with the same pattern of occurrence as peak E.

Fresh vs. fresh-frozen

A fluorophore peak can help distinguish between F and FFR samples if a peak is consistently found across all time points in one group, and is absent from the other.

In skin, peak A was consistently found across all time points in both treatments, while no peak was consistently absent across samples in either treatment group.

In adipose, peaks B and E were the only peaks consistently found in all samples in the F group, while peaks A and B were consistently present in all samples in the FFR group. Peak C was the only peak to be absent across all time points in F adipose. No one peak was present at all time points in one treatment group while being absent from all time points in the other group.

Similarly to skin samples, peak A was consistently present at all time points in both F and FFR muscle. No peak was consistently found in one treatment group while being absent from all samples in the other group.

Multiple freeze-thaw cycles

The effect of one vs. two F-T cycles on PROT and FOX fluorescence was examined after samples from time point 2 in the FFR group were re-frozen following initial measurements, and later had to be measured again. Graphs illustrating the fluorescence intensity ratio of PROT and FOX before and after a second F-T cycle for each tissue type are displayed in ESM Q.

PROT levels tended to be less impacted by a second F-T cycle compared to FOX. Average levels of PROT and FOX appeared to be less affected by F-T cycles overall in skin samples, with PROT ratios varying from being 20% lower to 3% higher, while FOX fluctuated between being 3% and 27% lower. Additionally, both fluorophores tended to display higher fluorescence intensities following re-freezing.