C stocks

We measured vegetation and soil C stocks from samples taken at the beginning and end of the project. We also up-scale plot measurements to the landscape scale based on Ground Penetrating Radar (GPR) derived peat depth transects. Care was taken to not affect bulk density during sampling of peat sections, bulk density is a crucial parameter not only for the carbon stock measurement (see Garnett et al. 1998) but also when modelling C dynamics (see Kennedy et al., 2008 and Heinemeyer et al., 2010). We also installed permanent peat rods with (datum) surface marker plates to determine long-term peat growth and short-term peat shrinkage/expansion due to environmental changes (i.e. due to water table and thus peat moisture changes). This will include the total peat depth, which is quite often not considered as measurements are limited to a certain depth such as 30 m or 1 m, but total peatland C stocks and models should consider the total peat depth (see Heinemeyer et al., 2010). This has become particularly evident in global C stock estimates (Tarnocai et al., 2009).

In August 2012 we took one manual peat depth sample down to the bedrock with a combination of a 1 m box corer and a D-shaped Russian peat corer for depth below 1 m (kindly provided by the YPP). This was done at 50 cm distance from the individual temperature and water table plots for all monitoring plots. We sampled 5 cm slices with minimal bulk density impact at the following depth intervals if depth allowed: 0-5; 10-15; 20-25; 40-45; 80-85; 120-125; 150-155 cm. If peat depth was deeper we also sampled at the very bottom of the peat column. We also recorded the main vegetation coverage percentages at the time of coring (i.e. % Calluna, sedge, rush, grass, moss, Sphagnum and bare ground). The same locations were surveyed again for peat depth in 2016 using GPR surveys.

We further manually recorded the peat depth along the same paths of the DMS automated GPR survey (see section on peat pipes). At ~25 locations in each sub-catchment peat depth was assessed manually using metal rods pushed to the bedrock and heather coverage was recorded at those locations from around 5 m2.

Peat profile sample for SOC determination

Peat core sections for bulk density and SOC determination

Manual peat sampling alongside GPR surveys

Manual peat depth survey at Whitendale

Preliminary SOC stocks across all sites

Bulk density values per depth layer across all sites

Preliminary peat depths across all sites

Peat shrinkage / expansion impacts on C stocks:

(see Morton & Heinemeyer 2019)

In the past, it has been claimed that an increase in temperature led to a huge loss in soil organic carbon (SOC) losses, especially from peat soils. However, this study (Bellamy et al., 2005) only measured the top 15 cm and did not consider changes in SOC stocks due to peat surface shrinkage and expansion in relation to natural changes in peat moisture. This is important particularly when only assessing SOC changes in surface layers, susceptible to such shrinkage/expansion; the same peat layer (e.g. 15 cm)could therefore contain more C than when expanded. We assessed this potential impact on 'apparent' changes in SOC stocks in a combined laboratory and field study (Morton & Heinemeyer 2019).

We used the below peat rods (for changes in peat surface height) and peat cores (for changes in bulk density) and found considerable changes in peat shrinkage/expansion of about 2 cm on average but up to 8 cm when including particularly dry periods (summer 2018) across our monitoring plots (including mown and burnt plots) and across the wider catchment scale at locations with different plant functional types (PFT; i.e. Calluna, Eriophorum and Sphagnum), which clearly related to peat moisture (i.e. water table depth) changes (but not to management):

Marking a 20 cm mark

Marking a 20 cm marker position on the steel rod

Hammering peat rod into place

Hammering the rod into place to the 20 cm marker

Ready in position

Steel rod with marker position visible

With marker disk for protection

Steel rod with marker disk for protection

Shrinkage Expansion NMW
Shrinkage Expansion PFT

Notably, those changes in peat surface related to changes in bulk density (of about 0.2 g cm3), which caused considerable changes in estimated (apparent) SOC stocks over the surface layer; we then compared the impact of such a potential bulk density change on the detection of 'apparent' changes in SOC stocks in the data provided in the Bellamy et al., 2005 study; we found that the reported changes in SOC could have been a result purely of changes in bulk density due to shrinkage and expansion (i.e. reflecting dry vs. wet moisture at the time of sampling vs. resampling, respectively, which (as was the case for bulk density) was not been reported on by Bellamy et al., 2005):

Shrinkage Expansion bulk density changes
Shrinkage Expansion apparent SOC changes

We therefore recommend particular caution when measuring SOC stocks in surface peat layers and in the conclusions drawn from such studies (unless it can be shown that such peat physical changes did not occur).

Peatland carbon stocks and burn history:

(see Heinemeyer et al., 2018)

We also assessed charcoal impacts on peat physical properties and long‐term carbon accumulation and storage in relation to past burn frequency in surface peat layers (see Heinemeyer et al., 2018). For this we took three peat cores at each of the sites near a burnt plot and assessed carbon content, bulk density and charcoal concentrations. We did this at a fine scale (0.5 cm sections over the top ~25 cm) using the spheroidal carbonaceous particle (SCP) dating technique. The study assessed two hypotheses:

1. While burning decreases SOC input (loss from litter combustion), it increases bulk density (i.e., higher charcoal content).

2. Peat C accumulation relates positively to higher burn frequencies as determined by charcoal layers and largely in relation to increased bulk density

Notably, the study did not compare burnt to unburnt sites and did not assess net C accumulation, the hypotheses did not require this test and it would be near impossible to find such paired sites (long-term burnt versus long-term unburnt under the same climatic and environmental conditions). Furthermore, the cores were taken from flat areas (slope <5 degrees) and the sites are not subject to any deep drainage (drainage is only documented to have only started at two of the sites in the 1970s with drainage ditches located at a distance of at least 15 m from the sample location and had infilled over time). Notably, drainage impacts from drainage ditches in blanket bog are known to be very limited (reducing water tables by only a few cm after a few metres away from the ditch: see Holden et al., 2004; Luscombe et al., 2016; Wilson et al., 2010). However, the Whitendale site does not have any past ditches but some 'natural' gullies, also at a distance of about 15 m distance from the sample location. Anyhow, our study aim was unrelated to any drainage impacts, simply focusing on the hypotheses around physical charcoal impacts on carbon storage.

We found unexpected but interesting results:

Peat cores C BD charcoal
Peat core correlations charcoal

We observed clear changes in the peat profile (above left) for bulk density, C content and SCP counts (i.e. dating tool), highlighting the advantages of high spatial sampling (0.5 cm sections). We also observed positive relationships between bulk density, C content and C accumulation versus charcoal counts per peat layer (above right).

Peat core C accumulation
Peat core burn frequencies

We also found that C accumulation over the specific age/depth layers was different between periods and sites (above left), and we also described past burn frequencies based on the charcoal counts (above right), which indicated overall burn frequencies were around 25 years (decreasing from Mossdale to Nidderdale to Whitendale), but they were more frequent in recent times (ranging from 22 years at Mossdale to 16 years at Nidderdale to only 13 years at Whitendale).

Peat core C accumulation comparisons

Notably, our C accumulation data were compared to those available in the literature (above Table), which revealed good overall agreement with other C accumulation rates, not indicating any lower values due to burning. Importantly, we clarified various caveats and site context dependent interpretation constraints in the discussion of our publication. Our actual C flux measurements indicate that such high (cumulative) C accumulation is possible within a burn rotation (notwithstanding heather beetle damage at two of the burnt sites, there is so far a clear indication for rapid C uptake, especially at Nidderdale) and that charcoal might have an important role in explaining such peat core C storage observations. This includes both, direct C storage as charcoal and indirect impacts on decomposition processes. However, this does not imply that burning is better for C storage, particularly as topographic aspects such as erosion (e.g. POC export was only assessed in stream overall, including various vegetation and habitat conditions across the entire catchment) have nor been assessed in detail yet.

However, our study has been criticised by several people (Evans et al., 2019 ; Young et al., 2019). Whilst we appreciate the need and indeed benefit of scientific debate, particularly in the case of unexpected or controversial findings, we provided a detailed rebuttal to the Evans et al. comment (Heinemeyer et al., 2019), outlining various aspects around the study's scope (i.e. no generalisation to other biomes, no applicability to unmanaged fires), no requirement for control sites or wider catchment-scale assessments, difficulties in comparing C stock versus C flux approaches, time scales needed for flux assessments and dating techniques (no overall or clear impact of burning on SCP-dating). Notably, we already addressed most of those concerns in the original paper. Moreover, a recent study by Flannagan et al. (2020) highlights two of our findings around potentially positive impacts of low-intensity burns on long-term C storage via charcoal by-passing litter decomposition and slowing down decomposition processes. Anyhow, we summarised our arguments in a final conclusion section:

"In conclusion, our defense clarifies the misunderstandings and misconceptions held by Evans et al. in relation to the scope and objectives of our study. We do, however, agree with Evans et al. that our findings have clear limitations. But we would also highlight that most of the criticisms made by Evans et al. are based on issues which we previously addressed in our paper (such as the lack of an unburnt control, other dating tools and wider catchment and site assessments). We would also argue that our study provides a vital addition to the prescribed burning evidence base, albeit in a very narrow context of UK grouse moor management on blanket bogs under specific climatic and environmental conditions. Our study will hopefully stimulate funding bodies to support further (and specifically long‐term) work so that the many remaining research gaps can be addressed - this is vital if we are to implement environmentally sound and scientifically robust land‐use policies."

We also would like to clarify important misconceptions and misrepresentations made by Young et al. (2019). In our discussion in Heinemeyer et al. (2018) we already highlighted the following (which clarifies and deals with most of their criticism, so we do not fully understand why it was not considered):

    1. “However, the functional role of charcoal is still little understood (Pingree & DeLuca, 2017) and SOC models do not include the here observed burning impacts on soil properties (i.e., bulk density), C compounds (i.e., charcoal) and thus long‐term C storage.”
    2. “Moreover, our findings highlight that these changes have potentially important implications on C cycling via eco‐hydrological feedbacks, for example on water‐holding capacity due to changes in BD, but also via soil biota, potentially affecting microbial communities and decomposer activity (Lehmann et al., 2011) due to so far unknown interactions.”
    3. “In fact, mean C accumulation rates (2015–1950) of 3.2 t CO2 ha−1 year−1 (87 g C m−2 year−1) were very similar to the 3.8 t CO2 ha−1 year−1 as reported previously by Evans et al. (2014) for unburnt management based on data presented by Garnett et al. (2000).”
    4. “C accumulation rates in these studies are generally much higher during the most recent periods (about 50–100 g C m−2 year−1), reflecting highly undecomposed peat, whereas long‐term accumulation rates for older layers are about 30 g C m−2 year−1.”
    5. “However, the conclusions reached here are based on a C‐stock inventory which could be different compared with using a C‐flux approach.”
    6. “The major disadvantages of the C‐stock approach are that it relies on uncertain dating techniques (particularly when using only one dating tool, such as SCPs, as in our study) and considers sections of peat separately, which ignores incorporation of surface C into deeper sections through roots and changes in decomposition rates over time.”

We have some more points of concern with the Young et al. study:

    • We found it surprising that Young et al. (2019) did not also mention/include the (similar) peat surface C accumulation study by Garnett et al. (2000), but maybe this is because it showed a more 'convenient' and 'expected' C loss? Notably, it was the only (and also a peat surface) study used by Evans et al. 2014 in their review on blanket bog management impacts on ecosystem services.
    • The climate in Young et al. is much dryer (although only net rainfall and not total precipitation data are shown - which is odd in itself) than at two of our and the Moor House (Hard Hill) sites (the latter is also much colder), so drainage & moisture impacts are unlikely comparable.
    • Overall, modeled water tables are not shown but likely were much lower than in wetter blanket bogs. What about seasonal changes and overall effect? Ideally a model versus data validation would have been provided with detailed information on the impact of drainage at what distance to the drain.
    • Water table drainage reductions in blanket bogs are known to be very low (few cm) in general and do not seem to extend far (only over a couple of metres) from ditches: e.g. see Holden et al., 2004; Luscombe et al., 2016; Wilson et al., 2010.
    • Our sites did not compare land uses, they compared 'like with like' (all were burnt). For hypotheses reasons we only compared between comparable sites and not to 'natural' (which we all know is nearly impossible to find - certainly not next to burnt sites under the same climatic and environmental conditions)! Notably, our surface C accumulation rates show high variability as burn management leads to peaks from charcoal inputs (high Corg and BD) and low periods due to reduced NPP after burn management (i.e. slow regrowth of vegetation; see Heinemeyer & Swindles, 2018 for a modelling study).
    • Spheroidal Carbonaceous Particle (SCP) distribution is a robust peat cohort dating tool (which was also used by Garnett et al. (2000). We are open about its limitations and previously explained issues with other dating methods (Heinemeyer et al.'s response to Evans et al., 2019). Another study at Mossdale (McCarroll et al., 2017) found very similar SCP patters (multi-peak) and better surface age agreement than for 14C data; their peat age at 25 cm is also approximately 1700 (as was estimated in our study).
    • Charcoal impacts on carbon accumulation are context specific (i.e. controlled burns on UK blanket bog) and net C accumulation is not referred to in our paper (we specifically highlight that our data only represent C accumulation over a measured depth and clearly highlighted potential issues from not assessing deeper layers - but how anyone can know what carbon could or should have been accumulated or what potential past climatic or management reasons were remains a mystery to us; even measurements over the entire depth would not completely address such uncertainty).
    • The 'artifact' C loss is not necessarily true - it depends on the conditions in deeper layers. If these deeper layers are water logged and there is no other significant erosion/loss from deeper down, then it is quite 'safe' to assume the surface C accumulations are indicative. However, a clear confirmation of no C loss is the steadily increasing %Corg content in our peat cores at all three sites (i.e. if decomposition occurred then %Corg should show a depletion period) and the very high and stable (and even higher than those in Young et al. reported for temperate bogs; cf Fig 1: 300 - 200 years of peat age = 17 - 25 cm) C accumulation rates in the bottom section of our peat cores. However, there is an indication in our cores for a potentially short period of drainage reduced C accumulation rates (and only at the two drained sites, Nidderdale & Mossdale) at around 9 - 12 cm depth (i.e. 1900 - 1870) corresponding to drains added in the 1970s (assuming a 5 - 10 cm drop in water tables this would have affected peat of ~60 - 100 years earlier) - but this period is included in our calculations (and crucially also relates to a decline in %Corg during that period). However, in our surface cores the %Corg 'dip' at around 10 cm depth (below right) occurred after the onset of the observed decline in C accumulation rates at around 6 cm (below left), which was gradual at the undrained site (Whitendale). Interestingly, our 'dip' in C accumulation rates to around 20-30 gC m-2 yr-1 aligns very nicely with the rate reductions shown in Young et al (Fig. 2).
C accumulation rates vs depth &amp; age
Corg  vs depth &amp; age

In fact, our sites show very similar C accumulation rates (actually, higher at depth) and similar behaviour to the sites shown by Young et al. (2020) for temperate sites (Fig. 1; see below left). Variability near the surface is likely related to: species specific litter inputs (i.e. decomposability); vegetation changes (roots/leaves); herbivory (i.e. heather beetle) and climatic (e.g. drought) impacts; burning and charcoal layer impacts. Moreover, our sites show no indication of deep drainage impacts compared to Young et al. (Fig. 2b; see below right), C accumulation rates at depth are actually higher than predicted by the model. However, a slight drop in C accumulation rates at around 9-12 cm depth (i.e. 1900-1870) was observed, and likely corresponds to drains in the 1970s (assuming a 5-10 cm drop in water table depth affected peat of ~60-100 years earlier). Images were reproduced by kind permission from the journal (Nature; Scientific Reports) under the Creative Commons Attribution 4.0 International License (CC-BY) and overlayed with our site data (Nidderdale, Mossdale, Whitendale) based on peat depth / age profile information (SCPs) aligned with the peat age given by Young et al. (2020).

Comparison to Young et al (1)
Comparison to Young et al (2)

We can also show that there is no visual evidence of deeper drainage impacts on C accumulation rates over the entire peat core profile (see below); neither bulk density nor %Corg indicate any such changes (i.e. bulk density remaining fairly constant and %Corg is overall increasing).

BD &amp; Corg vs entire peat core depth

Moreover, drainage ditches naturally infill over time (often quite quickly); there are no detailed methods given on the hydrological & decomposition impacts and at our sites drainage has only been documented to have started in the 1970s and our sample sites were about 15 m away from any ditch or gully.

The comparison to C flux assessments overlooks (as does Evans et al., 2019) the long time scales needed (although this was already acknowledged by Evans et al., 2014) to adequately and robustly capture management (plant regrowth) and recovery (disturbance) - hardly any such long-term studies exist. However, our NEE flux data suggest >25 years are needed, but likely longer (specifically if also considering methane fluxes, overlooked by both Evans et al. and Young et al). This also relates to the surface C accumulation rates being noise and lower than predicted by the model - our sites are rotationally burnt, which decreases NPP C inputs and then has slow recovery rates due to slow plant regrowth (see Heinemeyer & Swindles, 2018).

Finally, the Young et al. study is purely based on a model simulation (without any validation; we therefore suggest it should only be used cautiously and only to formulate hypotheses which should then be tested in the field), for only one type of land management (drainage, their model does not include any burning processes - but they criticise studies specifically concerned with burning only - ignoring charcoal and bulk density impacts) and (it seems from their schematic) only in a raised bog context (but this study is on blanket bog), yet they make general conclusions about interpreting recent carbon and peat additions within all types of bog; we actually provided a similar model study assessing grouse moor management and associated drainage impacts for Moor House, which included some model validation and comparison to paleo-ecological water table reconstructions (see Heinemeyer & Swindles, 2018) and clearly shows that we are aware of drainage issues in relation to C accumulation and GHG emissions.

Overall, the Young et al study is a nice study, which confirms a well-known potential issue of deep drainage affecting peat decomposition and thus net C accumulation. However, based on the above reasons we question why this directly applies to our study and we recommend the title should read "Potential for misinterpretation..." and it should not directly criticise studies such as ours out of context, unless it can provide robust and validated criticism of direct relevance; specifically as their model study has not been validated, assumes long-term and deep drainage, whilst not including crucial processes applicable to burning.