Modern Alnus pollen was extracted from catkins of Alnus trees growing near the River Mersey in Greater Manchester, NW England in February 2020. Modern Calluna pollen and Sphagnum spores were collected from fresh Sphagnum moss from Rindle Moss, near Manchester during October 2022. Pollen and spores were separated from other plant tissues by sieving and washes with type-1 deionised water. Microscopic inspection confirmed the presence of the protoplasm in the pollen and spores. The modern palynomorphs were washed by pyrophosphate (Na₄P₂O₇) and centrifugated in sodium polytungstate (SPT), and then oven-dried (105°C, 8 min), before FTIR analysis; in this way, a consistent pretreatment was ensured between modern and sub-fossil palynomorphs. As described below for sub-fossil samples, we performed checks to ensure that sodium pyrophosphate and SPT were not influential in the results.

The sub-fossil spore and pollen samples were obtained from a 360 cm peat core (HM20), which was collected using a Russian corer from an ombrotrophic raised bog at Holcroft Moss, NW England (53°26′09.4”N, 2°28′31.4”W, 23 m.a.s.l.). Peat samples were taken at seven depth ranges (42–47 cm, 81–89 cm, 121–129 cm, 161–169 cm, 201–209 cm, 241–249 cm, and 281–289 cm). Sample ages were estimated by Bayesian age-depth modelling in the R package “Bchron” (Haslett and Parnell, 2008) incorporating 11 AMS radiocarbon dates and a core-top date constraint (Table 1); the resulting sample ages span the last ca. 2,600 years (Table 2). The full age model is shown in Supplementary Figure 1. Samples of 21–30 g were taken from each depth range, and were further divided into 20 subsamples for chemical treatment experiments (~1.5 g of peat material was used in each treatment group) including control samples.

For acid and alkali treatments, peat samples were placed in 50 ml centrifuge tubes with either 10% (v/v) hydrochloric acid (HCl) or 10% (w/v) potassium hydroxide (KOH) and placed in the water bath at 90°C for treatment times of 5, 10, 30, 60, 120, and 210 min. For acetolysis, peat samples were firstly dehydrated in glacial acetic acid prior to addition of Erdtman (1960) acetolysis mixture (9 parts acetic anhydride and 1 part concentrated 18.4 M sulphuric acid). Samples were placed in a hot water bath at 90°C for treatment times of 1, 3, 5,10, 30, 60, and 120 min. Immediately on removal from the water bath, the reaction was stopped by topping up the samples with glacial acetic acid. Following all treatments, samples were centrifuged and washed with type-1 deionised water (3 cycles). Different treatment times were chosen to cover a range of typical treatment applications, as well as to extend into much longer than typical treatment times. Control samples received only water washes.

Pollen grains of Alnus and Calluna and Sphagnum spores were subsequently isolated from each sample (control and different treatments). Palynomorphs were first concentrated by micro-sieving (Heusser and Stock, 1984) using mesh sizes of 63 and 32 microns (retaining the intermediate fraction <63 μm and > 32 μm). Then, sodium pyrophosphate (Na₄P₂O₇) washes (Bates et al., 1978) were used help disaggregate the samples prior to density separations using sodium polytungstate (SPT; Nakagawa et al., 1998) following the method outlined in Bell et al. (2019). In this study, we used density floats at 1.6 s.g. and 1.2 s.g, retaining the intermediate fraction with density < 1.6 s.g. and > 1.2 s.g. Lastly, the concentrates were given multiple washes with type-1 deionised water to remove any traces of SPT and to clean the residues.

To investigate any potential impacts on FTIR spectra by Na₄P₂O₇ and SPT used during procedures above, we also isolated one additional sample for each taxon directly acquired from the peat cores without the use of either Na₄P₂O₇ or SPT. These samples were washed in type-1 deionised water only before being sent for FTIR scanning. Spectra of pure Na₄P₂O₇ and SPT (Supplementary Figure 2) were also analysed. We did not find any impact on pollen spectra of the use of Na₄P₂O₇ and SPT in the sample preparation or potential for distortion of the critical UAC peak at ~1,516 cm⁻¹ (Supplementary Figure 3). In the spectra for sodium pyrophosphate, we identified a peak occurring at ~1,084 cm⁻¹ (related to phosphates), and in the SPT, a peak occurring at ~1,616 cm⁻¹, which were within the fingerprint region. There were also additional peaks occurring below ~900 cm⁻¹, outside the fingerprint region. The peak at 1616 cm⁻¹ was already present in pollen and spores (i.e., pollen with only water washes), while there were also peaks occurring near to (but not at) ~1,084 cm⁻¹ in pollen and spores. Comparing the water washed pollen spectra to the sodium spectra, we could see no observable difference in the spectra, and no impact on the UAC peak at 1516 cm⁻¹ (Supplementary Figure 3). At the two identified peaks that were present in the sodium (1,616 cm⁻¹ and 1,084 cm⁻¹), again, there are no observable influences (e.g., peak height seems unaffected in the treated pollen).

A droplet of pollen concentrate was placed onto a well slide using a plastic Pasteur pipette and viewed using a Leica inverted optical microscope (OM) with built-in digital camera. Individual pollen grains and spores were then hand-picked from these residues using a mouth operated pipette (Mensing and Southon, 1999). Pollen and spores were transferred to zinc selenide (ZnSe) microscope slides, with each taxon placed in a separate region of the slide. These slides were then dried in an oven at 105°C for 8 min prior to FTIR analysis.

Samples were analysed using a Perkin Elmer Spotlight 400 Fourier Transform Infrared (FTIR) Imaging System equipped with continuum IR microscope (using a Reflachromat 15 × objective) and fitted with MCT-A liquid nitrogen-cooled detector. FTIR analysis was performed on approximately 30 individual pollen grains and spores for each sample. After comparing spectra qualities from our samples under different aperture sizes (available in Supplementary Figure 4), we used transmission mode with an aperture size of 30 μm × 30 μm for Alnus, and 40 μm × 40 μm for Calluna and Sphagnum. Supplementary Figure 5 shows the common palynomorph grain sizes for three taxa in this study, and indicates the aperture sizes we used covered either most or entire grains/spores. Background spectra on a blank region of ZnSe window were generated before each sample run and subtracted from pollen and spore spectra. To investigate an adequate number of scans, we compared spectral noise with different numbers of scans (Supplementary Figure 6), and a value of 16 was selected, which had adequate noise removal and clear peak information. Also, the Spotlight 400 system employs bi-directional scanning which means we process more data points per scan reducing the noise. Both background and pollen/spore spectra were measured using 16 scans at a resolution of 4 cm⁻¹ in the range between 4,000 and 650 cm⁻¹. In total, we studied 420 samples (20 control or treatment types x 7 depths x 3 taxa). These 420 samples each comprising a mean of 26 ± 6 spores or pollen grains were analysed using FTIR constituting a total body of >11,000 FTIR measurements on individual palynomorphs. When results are averaged across the seven depths, results for each treatment type are based on ~190 individual palynomorphs for each taxon. The numbers of spectra analysed for this study in each sample and taxon are provided in Supplementary Table 1.

Data processing was carried out using R v4.1.2 (R Core Team, 2021), following a spectral-averaging approach as used in recent FTIR pollen studies by Zimmermann (2018) and Jardine et al. (2021). To reduce the effect of noise, individual spectra were first smoothed using a Savitzky–Golay filter using a 2^nd degree polynomial with a window size of 11 using the “prospectr” package (Stevens and Ramirez-Lopez, 2020). All individual smoothed spectra were also checked to remove anomalous spectra, which may result where the FTIR aperture has not been well aligned to the palynomorph or if extraneous matter was adjacent or adherent to the palynomorph. The wavenumber position of six peaks of interest (~2,920 cm⁻¹, ~1700 cm⁻¹, ~1,600 cm⁻¹, ~1,510 cm⁻¹, ~1,160 cm⁻¹, ~1,100 cm⁻¹; see Table 3) was checked manually to ensure peaks occurred at their expected location. Spectra where any three of these peaks were undetectable were deemed not to represent pollen or spores and were excluded from the analysis, resulting in 8% of individual spectra being removed across the entire dataset based on these criteria. As non-negligible impacts from Mie-type scattering effects on FTIR spectra from biological materials were reported by previous studies (Baker et al., 2014; Zimmermann, 2018; Diehn et al., 2020) and can be observed in the raw spectra in our analysis (e.g., Figure 1A), we applied extended multiplicative signal correction (EMSC) on individual palynomorph spectrum, using ‘EMSC’ package in R (Liland, 2021), incorporating third order polynomial effects. A baseline corrected mean spectrum of each sample for each taxon was used as reference spectrum in this model (e.g., Figure 1B). A third order polynomial baseline was applied here for the reference spectrum using the ‘baseline’ package in R (Liland et al., 2010). After running the EMSC, mean spectra for each sample (shown in Supplementary Table 1) were calculated to minimise noise level and furthermore offset Mie-type scattering effects (e.g., Figures 1C,D), resulting in 420 EMSC corrected, averaged spectra (henceforth, simply “mean spectrum/a”) for further analysis. We experimented with more advanced ME-EMSC (Mie extinction – extended multiplicative signal correction) methods (Solheim et al., 2019) but did not observe any improvement over the standard EMSC for our samples. A workflow example of the spectral processing described above is presented in Figure 1. Each mean spectrum was normalised to its mean absorbance to make spectra comparable among samples. The mean spectra were used for Principal Components Analysis (PCA) to illustrate clustering patterns of different groups of spectral information, using the ‘ordr’ package (Brunson, 2022). The mean spectra were also used for subsequent peak area calculations.

For the detailed analysis of UAC peaks (~1,516 cm⁻¹), peak areas were calculated using a shoelace formula, with a fixed peak range between 1,540 and 1,492 cm⁻¹, similar to Prosr et al. (2016). Peak area offers several advantages over peak height as the unit of measurement, specifically for broadened or overlapped peaks (Prosr et al., 2016). To understand changes in the UAC peak area with treatments, ordinary least squares linear regression was performed on UAC peak area vs. log₁₀ transformed treatment time, due to the mathematically skewed distributions of treatment times. Regression analysis was performed by averaging each treatment type across all core depths, thus minimising any potential influence of variability in humification (decomposition) of the peat material on pollen and spore resilience. Since the control sample had an effective treatment time of 0, we added 1 to all treatment times for the log regression analysis (since log of 0 results in infinity). Mean UAC peak area difference (ΔUAC) between the control sample (no treatment) and the treated samples was calculated. We extracted the slope value of the regression as an indicator of the magnitude of UAC peak reduction across the entire range of treatment times. We also calculated the percentage change in UAC peak area between the control sample and treated sample for each time step to measure the magnitude of peak area change relative to the control (no treatment) sample.

Figure 2 shows mean spectra of the three taxa investigated from modern (fresh) and control (sub-fossil palynomorphs, no chemical treatment) samples. Overall, there are strong similarities between the modern and sub-fossil material, especially for Calluna and Sphagnum, while the modern and sub-fossil Alnus samples show greater differences, associated with noisier modern spectra for this taxon. In Sphagnum, there is a notable peak occurring at ~3,008 cm⁻¹ (Figure 2, grey shaded bar “i”) in modern spores that is not present in the sub-fossil spores. Additionally, a peak occurring at ~1740 cm⁻¹ (grey shaded bar “ii”) that is evident in all three taxa in modern spores and pollen (albeit weakly so for Calluna), is not evident in the sub-fossil material. Several differences are observed in peaks near 1,600 cm⁻¹ (grey shaded bars “iii-a,b,c”). The UAC peak at ~1,516 cm⁻¹ (grey shaded bar “iv”) is more evident in the sub-fossil samples, consistent with an origin in sporopollenin. Finally, the broad O-H peak (grey shaded bar “v”) is generally reduced in the sub-fossil material relative to the modern.

Figure 3 shows spectral differences between the control (untreated sample) and treated samples for different chemical treatments and treatment durations. HCl treatment does not substantially change the spectra for any of the taxa even at long treatment times, particularly in the fingerprint region (1800–1,000 cm⁻¹). Nevertheless, spectra from the HCl-treated samples for all taxa do show some reduction of the hydroxyl OH peak at ~3,300 cm⁻¹, and we also observe increases in peak height at ~2,850 and ~ 2,930 cm⁻¹ for Alnus at all treatment times, and Calluna and Sphagnum at 210 min treatment.

KOH treatment results in modest changes to the spectra, with spectral changes for all taxa in the fingerprint region. As with HCl, there were reductions to the hydroxyl OH peaks in all taxa, and similarly, increases in peak height for ~2,850 and ~ 2,930 cm⁻¹. The UAC peak at ~1,516 cm⁻¹ is reduced for all treatment times for Alnus and Calluna, but is largely unaffected in Sphagnum for treatment times under 60 min, which is reduced by longer treatment.

Acetolysis treatment shows the most significant spectral changes across all three taxa and all treatment times, even after 1 min of treatment. As with HCl and KOH, the hydroxyl OH peak at ~3,300 cm⁻¹ is reduced for all treatment times for Alnus and Calluna, whereas for Sphagnum, significant reduction occurs after 3 min. In contrast, the peaks at ~2,850 and ~ 2,930 cm⁻¹ are reduced across all three taxa and all treatment times for acetolysis. The UAC peak ~1,516 cm⁻¹ is reduced at all treatment times for Alnus and Calluna, and for treatment times greater than 3 mins for Sphagnum. A strong peak appeared at ~1,140 cm⁻¹ in Alnus and Calluna, and at ~1,200 cm⁻¹ in Sphagnum after being treated for more than 1 min, which was not present in the untreated (control) samples. The ether phenol C-O peak at ~1,100 cm⁻¹ was reduced in all taxa for all treatment times.

The PCA biplot (Figure 4) shows spectral differences among samples between taxa chemical treatments and treatment times. Samples treated with HCl were most similar to the control samples for all taxa and PC scores on both axes are close to the control groups. Similarly, there was a relatively small distance between KOH-treated and control samples. Acetolysed samples plot farther from the control samples and show considerably greater dispersion. PC1 explains 39.5% of the variance and primarily reflects impacts from different treatment chemicals (negative PC1 scores for HCl and KOH treated samples, positive scores for acetolysed samples). PC2 explains 22.3% of the variance and mainly reflects inter-taxon differences, discriminating between all Calluna samples (negative PC2 scores) and the other taxa (positive PC2 scores), and also explaining some variance resulting from treatment time, with generally higher PC2 scores for longer treatment times. The analysis also shows the strongest influence of treatment time for acetolysis treated samples, especially notable for Sphagnum, while there is less effect for HCl and KOH (samples more strongly clustered).

Table 4 reports the average change in UAC peak area as a percentage of the control sample values (∆UAC) for the three taxa. ∆UAC values are visualised in Figure 5, showing the standard deviation across samples from the seven depths (error bars) accompanied by the regression line with 95% confidence interval for the regression of ∆UAC onto log-transformed treatment time. Regression statistics are reported in Table 5. In general, we observe that chemical pretreatments tend to reduce the UAC peak area, although the pattern is not entirely consistent across all treatments and taxa. Regression models for reduction in UAC peak area with time are significant for KOH treatment on Calluna and acetolysis on Sphagnum at the 95% confidence level (p ≤ 0.05) and for acetolysis on Alnus, acetolysis on Calluna, and KOH treatment on Sphagnum at the 90% confidence level (p ≤ 0.1; Table 5). In contrast, a significant increase in UAC peak area is modelled for HCl treatment on Sphagnum (p ≤ 0.05). Broadly, the magnitude of UAC changes is greater for KOH than HCl, and for acetolysis than KOH, reflected in the slope of the regression models (Figure 5; Table 5).

In detail, HCl treatments yielded variable impacts on UAC peak areas, with positive relationships (across the entire treatment period) for Calluna and Sphagnum, and a negative relationship for Alnus. Sphagnum yielded a significant regression result (p ≤ 0.05) but weak slope coefficient, while the relationships of the other two taxa were insignificant (Table 5). Overall, the total impact of HCl treatment, even after very long treatment times, was modest; ∆UAC values after 210 min were − 9.6%, +14.2%, +18.9% for Alnus, Calluna and Sphagnum, respectively (Table 4).

For KOH, there were negative relationships for all three taxa. The relationships for Calluna and Sphagnum were significant at p ≤ 0.05 and p ≤ 0.1, respectively, but insignificant for Alnus. The detailed pattern of change over time is specific to each taxon: Alnus displays a marked step-like reduction in ΔUAC after 10 min (−47.8%); Calluna, a similar marked reduction is noted after 60 min (−61.4%); and a progressive strong reduction for Sphagnum between 60 (−22.3%) and 120 min (−60.1%; Table 4). Thus a limited impact is detected from KOH on UAC peak areas in pollen and spores within short treatment time (e.g., ≤10 min for Alnus, and < 30 min for Calluna and Sphagnum), and increasing impact of KOH between 10 and 120 min for Alnus, and 60 and 120 min for Calluna and Sphagnum. Extended treatment (210 min) resulted in a strong reduction in UAC peak areas for all taxa: −29.7, −43.0% and − 60.1% for Alnus, Calluna and Sphagnum, respectively (Table 4).

Acetolysis treatments resulted in the strongest negative relationships between ∆UAC and treatment time, significant for Alnus (p ≤ 0.1), Calluna (p ≤ 0.1) and Sphagnum (p ≤ 0.05). Slope values for Sphagnum samples treated by acetolysis had higher negative values than the other two taxa, indicating the greatest degree of UAC reduction during the whole treatment period. For Alnus and Calluna, acetolysis impacts on UAC peak area were notable after 1 min (−67.3% and − 72.1%, respectively), while Sphagnum showed some resilience after 1 min followed by stronger changes from 3 mins onward. Extended treatment (120 min) resulted in a very strong reduction in UAC peak areas for all taxa: −76.9, −69.4% and − 88.6% for Alnus, Calluna and Sphagnum, respectively (Table 4).

In the comparison between modern and sub-fossil spectra (Figure 2), visible peaks of C=O at ~1740 cm⁻¹ (corresponding to lipids or ester in palynomorphs; Watson et al., 2007), and amide I and amide II at ~1,650 cm⁻¹ (iii-a) and ~ 1,550 cm⁻¹ (iii-c; contributed by protein; Sowa et al., 1991; Lahlali et al., 2014) are observed in modern spectra of three taxa, while not for sub-fossil ones. These differences are likely the result of palynomorph degradation and oxidation from fresh (modern) samples to sub-fossil products in the environment. Also, protoplast contents in nuclei of fresh palynomorphs are important sources of proteins (Ueda and Tanaka, 1994), but they could be rapidly destroyed by bacteria, fungi, oxidation and hydrolysis, all probable impacts suffered by sub-fossil materials (Traverse, 2007). Similarly, a study on Ambrosia pollen showed that the cellulosic intine layer of the pollen wall contains antigenic proteins which are lost on contact with a humid surface (Howlett et al., 1973). For the peak at ~3,008 cm⁻¹ in modern palynomorph spectra, interpretations in other studies on this peak signal attributed it to C=C bonds (Depciuch et al., 2018; Kyselka et al., 2018), and disappearance of this peak in sub-fossil Sphagnum spores may represent losses in degree of carbon unsaturation for this taxon. Van Bilsen et al. (1994) found loss of lipids in pollen grains after storage, resulting in loss of unsaturation of grain chemistry, reinforcing the disappeared peak of both C=C (~3,008 cm⁻¹) and C=O (~1740 cm⁻¹) in modern spectra (Figure 2 “i” and “ii”). These features may explain the insignificant amide peaks we observe in sub-fossil spectra. Since up to 60% of the N content of fresh pollen may be water soluble (Greenfield, 1999), changes in pollen chemistry may be very rapid (hours to days) on deposition in the environment. This may explain the stronger amide peaks detected in the modern Alnus sample, for which pollen was collected directly from catkins, as compared to the modern Calluna and Sphagnum samples, where palynomorphs were scavenged following flowering/sporulation from the surface of living sphagna. Overall, the absence of peaks at 3008 cm⁻¹ and ~ 1740 cm⁻¹ in sub-fossil spectra represents a clear discriminator between modern and sub-fossil pollen and spores, likely related to the loss of pollenkitt, intine and protoplasm in the aerobic peatbog surface layers. The absence of these spectral features is also evident in the untreated control samples from the core, thus confirming that their absence is not due to chemical treatment. This is true for the sub-fossil samples at all depths, including the youngest samples at 44.5 cm (approximately 600 cal years BP), consistent with complete “transformation” to sub-fossil status.

Spectra treated by HCl have few observable changes, but peaks attributed to O-H bonds at ~3,300 cm⁻¹ decrease with longer treatment time for all taxa (Figure 3), and peaks at ~2,930 and ~ 2,850 cm⁻¹ (both attributed to C-H bonds) increase after being treated by HCl. They could be the result of hydrophobicity increases on the palynomorph surface. Noma et al. (2018) found increased hydrophobicity for spores treated with HCl, by negatively charging the spore coat. Meanwhile, Yang et al. (2009) compared FTIR spectra for methyl-modified silica films with different levels of hydrophobicity, and found weaker O-H bond peaks and stronger C-H bond peaks adjacent to the wavenumbers in our results.

For KOH, both the decreased O-H ~ 3,300 cm⁻¹ peaks and increased C-H 2930/2850 cm⁻¹ peaks are also observed (Figure 3), but with different reasons. The decreases of O-H peaks could be from decomposition of humic acid. Müller et al. (2021) found negative mean depletion of carbon isotope values (δ¹³C) for Alnus pollen grains treated by KOH, and attributed this to decreases of more derivatives of humic acids. Considering that strong peaks at around 3,300 cm⁻¹ are observed in humid acid infrared spectra (Yonebayashi and Hattori, 1989), decreases in humic acid could explain the weakened O-H peak. The increases of C-H peaks at ~2930/2850 cm⁻¹ could come from hydrolytic degradation of carbon chains and branching in sporopollenin. Sammon et al. (2000) found increased C-H peaks at ~2930/2850 cm⁻¹ for FTIR spectra of KOH-treated polyethylene, and attribute these changes to change in chemical (electronic) environment occurring during hydrolytic degradation. Sporopollenin also has substantial ester groups (Li et al., 2018), enabling similar hydrolytic degradation processes when treated with KOH. Also, sporopollenin treated by KOH had been found to be similar to IR spectra with K salt glucuronic acid, indicating an oxidised aromatic ring (Traverse, 2007), and this may support the weakened UAC-related signal at ~1,516 cm⁻¹. To deal with this issues, spectral deconvolution with Voigt profiles is usually applied to isolate single peak signals from overlapped peaks (Bandosz et al., 1994; Sadat and Joye, 2020), while an alternative method on peak area calculation of “Tangential integration” has been also been tested to be comparable with the Voigt profile deconvolution (Wieser et al., 2022). We ran both methodologies to isolate UAC 1516 cm⁻¹ peak area, and get comparable results as well.

Samples treated by acetolysis showed several significant changes in spectral peaks (Figure 3). Peaks representing O-H bonds (3,300 cm⁻¹), C-H bonds (2,925 and 2,850 cm⁻¹), aromatic C=C bonds (1,516 cm⁻¹) and ether phenol C-O bonds (1,100 cm⁻¹) were weaker, while peaks belonging to aryl ether C-O bonds (1,040 cm⁻¹) and phenol C-O/C-H bonds (1,150 cm⁻¹) became stronger. Generally speaking, these variations in bond peaks suggest fundamental element gain and loss tendencies for acetolysis treatment on sporopollenin: losses for carbon functional groups and enrichment for oxygen ones (Ahlers et al., 2000). Moreover, weakening of peaks at ~1,516 cm⁻¹ and ~ 1,100 cm⁻¹ was also identified by Dominguez et al. (1999) who observed that unsaturation and ether bonds were oxidised into carboxylic acids. Meanwhile, strengthening of two bonds correlated with phenol and aryl ether is also supported by evidence from Watson et al. (2012) for substitutions of aromatic rings by hydroxyl group, resulting in increased intensity of phenol O-H and C-O bonds. Meanwhile, considering that the peak around 1,150 cm⁻¹ was also correlated with S=O stretch vibrations (Oh and Baiz, 2019), the elevated intensities here could also be the results of newly-formed sulfoxide bond from acetolysis treatment. Traverse (2007) notes the uptake of S by palynomorphs during acetolysis. Overall, the spectral changes point to quite profound impacts of acetolysis on sub-fossil palynomorph chemistry, associated with size and colour (thermal maturation) changes (Supplementary Figure 5).

Considering Euclidean distances in the PCA biplot as an approximation of total spectral changes (Figure 4), all three taxa were fairly resilient to chemical treatment with HCl, moderately resilient to KOH treatment, and least resilient to acetolysis treatment (reflected in PC1). We also observe that taxon-specific differences in the FTIR spectra are retained after chemical treatment (reflected in PC2), for all treatment types and treatment times. This indicates that it should be possible to discriminate between taxa on sub-fossil material following chemical treatment subject to careful application of a systematic set of protocols, which is essential for chemotaxonomic applications of FTIR on pollen (Julier et al., 2016; Diehn et al., 2020). Nevertheless, the evident displacement of samples in the PCA following any of the treatments emphasises the need for careful attention to the impacts of processing methods, especially where these may not be parallel between reference and ancient specimens.

Our findings reveal that UAC peak areas are impacted by chemical treatments. Despite variability between treatments and taxa, and fairly high dispersion of results across samples drawn from different depths in the core, a generally consistent pattern of loss of UAC peak area is observed for KOH and acetolysis treatments. Furthermore, in several cases —KOH treatment on Calluna and acetolysis treatments on all taxa— the results conform to a log-decay model (i.e., regression models significant at 95% or 90% confidence level), highlighting a distinctive pattern of rapid loss of UAC peak area during the first minutes of treatment time followed by progressive reduction over longer treatment times. This model has been selected empirically and could be further refined through the development of physico-chemical models for different treatments.

For changes to the UAC peak area, different reactions to acetolysis treatment were observed among three taxa investigated (Figure 5; Tables 4, 5). Sphagnum showed strong resilience to treatment after one minute, but eventually demonstrated the strongest change in UAC peak areas after prolonged treatment. The thicker wall (exine) of Sphagnum spores (Brown et al., 2015), may give them increased initial protection against chemical treatment compared to Alnus and Calluna pollen grains. However, the eventual magnitude of the UAC changes in Sphagnum spores, greater than for Alnus or Calluna, are consistent with a previously reported vulnerability of Sphagnum spores to degradation by acetolysis of 10 min duration (Charman, 1992).

Our study reveals that chemical treatment can have a significant impact on the chemistry of pollen and spores even under short duration treatment times. This result has important implications for UV-B reconstructions based on UACs in ancient pollen and spores (Seddon et al., 2019). Importantly, our findings have been obtained on sub-fossil palynomorphs, so chemical impacts of the selected treatments cannot be not solely related to the loss of pollenkitt, intine or protoplasm, which are not preserved in sub-fossil grains and spores. Therefore, the progressive reduction in peak area of the aromatic C=C UAC peak at ~1,516 cm⁻¹ with treatment time can only be explained by loss of UACs from within the sporopollenin itself as a result of damage caused to the palynomorphs from the chemical treatment. Not only were strong changes observed with short acetolysis times, KOH treatment also resulted in the reduction of the UAC peak areas albeit at longer treatment times and to a slightly lower overall magnitude (Table 5). In contrast, HCl treatment does not yield a consistent direction of change in the UAC peak areas and only modest proportional changes in UAC peak area. Thus cautious application of HCl, or even KOH for short treatment times (up to 5 min), may not overly impact the UAC peak. We would suggest that for chemical analysis of palynomorphs, researchers should make cautious selection of the pre-treatment protocols and that careful control and consistency when using HCl and KOH treatments are needed, while further investigation of the specific effects of chemical treatments on different taxa are required. This could include, for example, investigation of other chemical treatments that are frequently used in standard pollen preparation techniques, such as hydrofluoric acid.

Due to significant reductions in the UAC peak areas as well as taxon-specific effects (different magnitude and pace of spectral changes), we recommend that acetolysis is avoided for all palynomorph chemistry analysis in palaeoecological studies on pollen and spore chemistry. Since UAC peak areas can be strongly reduced after just 1 min of treatment, even efforts to implement a systematic protocol could risk introducing differential changes to UAC peak areas down core as a result of small changes in exposure time between samples to the acetolysis treatment. In addition, long treatments may introduce additional features to the spectra making peak detection more difficult.

Overall, as we demonstrate in this study, the use of physical separation methods including micro-sieving and density separation is preferable for FTIR analysis of pollen and spores, since it does not impact on UAC peak detection or quantification. However, we recognise this method may not be suitable for all sediment types, and further research is required for pollen extraction from non-peat or non-lake sediments. We recognise that researchers may wish to carry out standard pollen analysis (identification and counting) in addition to chemical analysis of pollen on the same samples. We suggest a practical approach would involve taking two sub-samples from the sample, so that one can be prepared in the standard way (applying chemical digestions as required to maximise the quality of the residues for counting), while the other is prepared using physical separation methods and minimising the use of harsh reagents.