The carbon sequestration potential of Scottish native woodland

Woodland creation sequesters carbon and contributes to climate change mitigation. Most previous assessments of the carbon sequestration of new UK woodlands have focused on tree planting, little is known about the scale of the potential contribution from natural regeneration. We used a Potential for Native Woodland Model to make the first estimate of carbon sequestration by large-scale native woodland expansion through natural regeneration in Scotland. We estimate native woodland could expand to cover an additional 3.9 million hectares of the Scottish uplands removing an average of 6.96 million tons of CO2 per year. This represents 35%–45% of the carbon removal target for UK woodlands that has been suggested by the UK Committee on Climate Change. Expanding woodlands to just 10% of this potential would double existing native woodland and could provide a multitude of benefits, including carbon removal equivalent to approximately 4% of this target. The next few decades are critical in terms of climate change mitigation, therefore further work is now required to improve these estimates and better constrain this potentially large contribution.


Introduction
Conservation and restoration of forests, wetlands and grasslands can provide one third of the actions needed to hold global temperature increase below 2°C and prevent disastrous climate change (Griscom et al 2017). Many countries have included such natural climate solutions (also known as nature-based solutions) in their Nationally Determined Contributions submitted in support of the Paris Agreement on Climate (Grassi et al 2017). There is the potential for new woodland in the UK to sequester carbon and contribute to emission reduction targets (Valatin 2012). The UK Committee on Climate Change (UKCCC) has advised that to reach net-zero emissions by 2050, the UK must create 30,000-50,000 hectares of forest per year, sequestering 15-20 million tons of CO 2 per year (Mt CO 2 yr −1 ) (UKCCC 2019).
In the UK, new woodland creation is achieved primarily by tree planting with much less focus on natural regeneration, which can also create new native woodlands (Spracklen et al 2013). Consequently, little is known about the carbon sequestration potential of woodland expansion through natural regeneration. O'Neill et al (2020) estimated that woodland creation on European sheep pasture was economically viable through natural regeneration for a carbon price of $4 per ton of CO 2 (tCO 2 ) but required $55/tCO 2 if woodland creation occurred through tree planting. A better understanding of the carbon removal potential of natural regeneration is therefore crucial to assess its role in climate change mitigation efforts.
Scotland has a large potential for woodland expansion, and the country accounted for 80% of new tree planting in the UK in 2019/2020 (Forestry Statistics 2020). Recent analysis by Sing and Aitkenhead (2020) provides an update to work commissioned by the Woodland Expansion Advisory Group (WEAG 2012, Sing et al 2013, estimating that 2.96 million hectares (Mha) of land are ecologically suitable for woodland expansion, with a further 0.54 Mha constrained to varying degrees by national designations and policies, showing that in total up to 45% of Scotland's land has the potential for woodland expansion.
Large areas of the Scottish uplands would be suitable for woodland expansion through natural regeneration, and this is the preferred method of expanding semi-natural and native woodland in the UK (Forestry Commission 1994). Advantages of natural regeneration over tree planting include best matching of species to site, structural heterogeneity of resulting woodlands, and conservation of local genetic stock (Peterken 1996). Additionally, natural regeneration does not involve the same level of ground disturbance as planting, which can lead to loss of carbon from the soil (Friggens et al 2020). Current land-use practices mean regeneration of native woodlands is often limited and a natural tree line (the physical limit above which trees cannot survive) is unable to establish across the majority of the country's uplands. Amongst other factors, the historic depletion of woodland has left long distances to seed sources in some areas and practices such as grazing by livestock (Speed et al 2010) and deer (Bunce et al 2014), and repeated muirburn (rotational burning of heather and grassland) suppress sapling growth. Modelling by Tanentzap et al (2013) found that deer browsing and ground vegetation cover were the main factors influencing birch regeneration, over seed availability. Current research into woodland expansion rarely accounts for land that could support scrub-type woodland up to a naturally established tree line, and often excludes montane and other scrub habitats on highly exposed land. Montane scrub is severely depleted in Scotland, but benefits of this habitat include unique wildlife, slope stability, water runoff retention and reduced downstream flooding, and reduced windthrow of adjacent forest plantations (Scott 2000). Initial estimates of current climatic factors by Hale et al (1997, cited in Gilbert andCosmo 2003, p.178) suggest that 609,400 hectares of land could support woodland above the commercial timberline, with a further 322,300 hectares unsuitable for tree growth but with potential for tall shrub species. Changes in climate may mean that this area increases in the future.
Here we make a first estimate of the potential carbon removal that would be achieved by large-scale expansion of native woodlands in the Scottish uplands through natural regeneration. First, we determined the land available for woodland to re-establish across Scotland if conditions allowed (i.e. if a seed source were present, existing vegetation allowed germination and seedling growth, and browsing was sufficiently low), up to its climatically determined extent. Next, we used this information to estimate the carbon removal and storage potential of this woodland. We specifically consider the potential across a range of ecosystems, including upland and montane habitats. Following preliminary calculations, we make recommendations as to how estimates could be improved and what further work would be required to achieve this.

Methodology
Our approach involved two components. First, we assessed the potential area of land in Scotland that could become new native woodland. Second, we assessed the carbon uptake and storage potential of this woodland.

Potential extent of native woodland
In the first step we defined the extent of potential woodland cover across Scotland and then determined the type of native woodland that is likely re-establish, given the right conditions.
To do this, we applied the Potential for Native Woodland Model (NWM), created by the Macaulay Institute and SNH (Towers et al 2004). The NWM was developed as a planning tool to aid expansion of native woodland across Scotland. Using national scale soil and landcover data, the NWM predicts potential National Vegetation Classification (NVC) woodland types that would be expected under current soil and vegetation conditions, down to a 1:50,000 scale. It should be noted that the potential for different woodland types is based on current conditions and is likely to vary with climate change.
The NWM encompasses 5.3 Mha, approximately two thirds of the country, covering upland mainland Scotland but excluding the central and eastern lowlands, where modified soils mean that it is more difficult to predict appropriate native woodland types. Current woodland is not accounted for in the NWM, therefore areas of existing woodland, determined using the 2018 National Forest Inventory (NFI) Woodland Map Scotland, were removed from the analysis. All areas categorised as woodland in the NFI were excluded, except for those classed as 'failed' areas of plantation, where woodland is not currently present. The outputs of the NWM are categorized into 58 woodland types. Inland water, built-up land, developed rural land and land unsuitable for tree/scrub growth are also identified. The woodland types may be single or interchangeable NVC classes, and in some cases mosaics of different NVC classes are predicted. The full range of NWM outputs is shown in table 1.

Carbon storage calculations
Canopy cover Due to the complexity of interaction between vegetation, soil and climate, canopy cover in naturally regenerating woodland can vary widely and it is likely that a mosaic of denser and more sparsely wooded areas would occur naturally across the landscape. It is therefore important to account for this in calculations of carbon sequestration in native woodlands.
To determine the percentage canopy cover for the woodland types predicted by the NWM, each component part of the woodland types was assigned a canopy cover value, as given in Towers et al (2004). Although woodland in the UK is defined as land under stands of trees with a canopy cover of at least 20% (Forestry Statistics 2020), the NWM incorporates areas with canopy cover as low as 10% in order to include the potential for open woodland and scattered trees/scrub. Types W4a, W6-W11 and W16-W19 were assigned 80% canopy cover; W4 (with open ground) and Sc1, Sc3, Sc6 and Sc7 were assigned 30% canopy cover; and Sc2, Sc4, Sc5 and Sc8 were assigned 10% canopy cover. Inferring mosaic composition proportions from the NWM, we then calculated the percentage canopy cover for each of the woodland types predicted by the NWM. Values for canopy cover are shown in table 1.

Carbon sequestration
We estimated carbon sequestration based on data from a study by the Scottish Forest Alliance (SFA), which modelled above-ground carbon sequestration at 12 native woodland sites across Scotland . Established through a combination of planting and natural regeneration, these sites were predominantly upland, with nutrient poor soils, focusing on NVC types W17 (upland oak/birch with bilberry), W18 (Scots pine with heather), W11 (upland oak/birch with bluebell/wild hyacinth), W7 (alder/ash with yellow pimpernel), W9b (upland ash with birch/rowan/aspen) and W4 (birch with purple moor grass). An average carbon sequestration total of 84 tons of carbon per hectare (tC ha −1 ) for mature woodland (based on total values given in table 1  We assumed this value represented an 80% canopy cover mature woodland. To provide carbon sequestration for different woodland types predicted by the NWM, we used the canopy cover for each woodland type to scale carbon sequestration. That is, for Sc2 woodland with canopy cover of 10% we scaled by 10% over 80% giving a carbon sequestration of 10.5 tC ha −1 (84 tC ha −1 ×10/80). To calculate average annual sequestration rates, we assumed that mature woodlands take 100 years to develop and, as a simplification, that carbon uptake is linear over this period. In reality, carbon uptake will be slow during the early years of regeneration but would also peak at higher values than the average rate we calculate here.

Results
There is potential for new native woodlands across 3.9 Mha, roughly 74% of the area of our analysis. Figure 1 shows the potential woodland area for the main NWM types. The woodland type covering the largest area is peatland with scattered trees/scrub (Sc5), which covers just under 490,000 ha. The NVC woodland type covering the greatest area is W11 (upland oak/birch woodland with bluebell/wild hyacinth), which is predicted to cover nearly 258,000 ha. Other NWM types which cover significant areas as part of woodland mosaics include W4 (birch with open ground), W17 (upland oak/birch with bilberry) and W18 (Scots pine with heather). Open woodland and scattered trees/scrub (canopy cover 30%) accounts for 1.9 Mha, roughly 50% of the total woodland area. Figure 2 reports carbon sequestration by the different woodland types and shown in detail in table 1. Total carbon sequestration is dominated by W11 (11% of total carbon sequestered), W18 (8%) and W18/W17 mosaic (7%). Scattered trees and scrub on peatlands cover the largest area, equivalent to 12% of total native woodland cover but only contribute 3% of carbon storage due to the low canopy cover and low assumed carbon storage per area. Open woodlands (canopy cover 30%) account for 50% of woodland area, but only 22% of total carbon sequestered.
Montane habitats, defined by the NWM as types Sc1, Sc2, Sc3, Sc4, Sc7 and Sc8, cover a total of 353,000 ha (9% of the model area) and collectively sequestered 8.22 Mt C, or 4% of the total for all native woodland.
Total carbon sequestration potential across all NWM predicted model outputs is 190 Mt C (table 1), which equates to 696 Mt CO 2 . Based on the broad assumptions that a woodland takes 100 years to mature and that uptake of carbon is linear during that time, our calculations suggest an average carbon sequestration potential of 1.90 Mt C yr −1 , which equates to 6.96 Mt CO 2 yr −1 .

Discussion
If native woodlands expanded to the potential estimated by the NWM, they would cover an additional 50% of Scotland's land area, making a major contribution to Biodiversity Action Plan priority habitats (figure 3). Woodlands composed of native species currently cover only 5.8% of Scotland (Forestry Statistics 2020), with semi-natural woodland covering only 4% of Scotland (Bunce et al 2014). Allowing native woodlands to expand to an additional 10% of their potential area (0.4 Mha) would therefore represent a doubling of native woodlands in Scotland. Such an expansion would also make a substantial contribution towards existing plans for woodland expansion from 18% to 25% of land area by 2050 (Thomas et al 2015). Forestry Strategy 2019-2029 calls for an increase in the annual woodland creation target from 10,000 to 15,000 ha per year by 2025, including 3000-5000 ha of native woodland (The Scottish Government 2019). To achieve expansion to 10% of their potential area by 2050 would involve creation of 13,000 ha of native woodland per year. Although this rate is comparable to that set for overall planting targets, natural regeneration requires fewer resources than planting, allowing greater  scope for large scale woodland expansion. Consequently, natural regeneration may augment planting targets as a method of woodland establishment in areas where it is appropriate and feasible.
A large fraction of the potential native woodland would consist of open woodland and scattered trees rather than closed canopy woodland. The average canopy cover of the potential woodlands is 60%. The definition of woodland in the UK is land under stands of trees with a canopy cover of at least 20% (Forestry Statistics 2020). About one fifth of the potential woodland (0.7 Mha) in our study has a canopy cover less than this and so would not be classified as woodland under this definition. Only one third of the woodlands (1.2 Mha) consists of closed canopy cover woodlands (80% canopy cover) and half of the woodland area (1. Previous analysis focusing on woodland expansion by tree planting has suggested around 3.5 Mha of land is available across the whole of Scotland (Sing and Aitkenhead 2020). In the context of natural regeneration, we estimate 3.9 Mha are available across two thirds of the same modelled area. Under the UK Forestry Standard (Forestry Commission 2017), areas of peatland and other low productivity land are deemed inappropriate for tree planting due to net carbon losses from deep peat and poor timber yields on certain soils or climate conditions. In contrast, allowing woodland to expand naturally means that trees are likely to establish in such areas, although they may be sporadic, stunted and the timber of little economic value. By removing these areas from their analysis, Sing and Aitkenhead calculated 1.23 Mha to be unsuitable for woodland creation, compared with only 0.2 Mha across two thirds of the same area by the NWM. This is reflected in the high proportions of low canopy cover woodland and scrub predicted by the NWM, that may grow on poor quality or exposed land.
We estimate that 3.9 Mha of native woodland could be established in the Scottish uplands with a potential carbon sequestration of 696 Mt CO 2 over a 100 year period, equivalent to an average removal of 6.96 Mt CO 2 yr −1 . This carbon sequestration is equivalent to 35%-45% of the carbon removal targets through woodland creation suggested by the UKCCC for achieving net-zero emissions in the UK. Under an assumed afforestation rate of 10,000 ha per annum, Scottish forests would remove less than 4 Mt CO 2 yr −1 between now and 2050 (UKCCC 2020). Expansion of native woodlands to 10% of their potential across the Scottish uplands could result in an additional 0.7 Mt CO 2 yr −1 removal, demonstrating the substantial opportunity for increased native woodlands to contribute to carbon removal. Carbon removal is dominated by W11 and W18 which together account for 20% of potential carbon uptake. Prioritising actions to areas that support these woodland types (oak and Scots pine woodlands, figure 3) would maximise carbon uptake per woodland area (tC ha −1 ) of new woodlands. Overall, our analysis shows large-scale expansion of native woodlands through natural regeneration has significant potential to deliver climate mitigation in Scotland. Future work is required to assess the potential for native woodland creation in upland areas across the UK.

Limitations and recommendations
Our analysis contains a range of simplistic assumptions. The amount of carbon stored in a woodland is highly variable, and will depend on factors such as species composition, soil, and former land use (Ostle et al 2009). Therefore, each woodland type predicted by the NWM will have a different carbon storage potential. Currently our estimates do not fully capture this complexity. We apply one carbon storage value  scaled by the assumed canopy cover of different woodland types, to broadly account for differences in carbon storage for the different woodland compositions predicted by the NWM. Our calculations also assume linear carbon uptake, whereas in reality accumulation rates are likely to vary significantly at different stages of maturity. To improve on this approach, we would require carbon sequestration and storage values specific to each of the main NVC classes used by the NWM, covering a range of variables such as soil type, age and species composition. This information is not currently available for Scottish native woodlands, therefore further work will be needed to gather data for this purpose. Our analysis can be used to prioritise woodland types for detailed carbon measurements, namely W11 (upland oak/birch) and W18 (Scots pine).
Our estimates do not account for changes in soil carbon as woodlands establish, which could be substantial (Matthews et al 2020). Afforestation on deep peat, especially when drainage occurs before tree planting, results in large carbon emissions from loss of soil organic matter (Morison et al 2010). It is known that tree planting on organo-mineral soils can also lead to loss of soil carbon on decadal scales (Friggens et al 2020), particularly if tree planting includes substantial ground preparation. The change in soil carbon under naturally regenerating woodlands, particularly composed of open woodland and scattered trees, is not well understood. As natural regeneration does not involve the same level of soil disturbance as planting, it has the potential to become a net carbon sink much sooner than planted trees, which may have to offset soil carbon losses for several decades after planting. This could be critical in the next few decades of climate change mitigation.
We have assumed the maximum hypothetical expansion of native woodland in the Scottish uplands. The NWM does not model the extent of woodland potential in lowland and urban parts of the country. Future work needs to assess how additional constraints impact the available area for woodland cover. Constraints may include presence of agricultural land; land designations such as national parks or national scenic areas; areas protected for conservation or historic value; areas of deep peat soils; and other areas of potential land-use conflict (Sing et al 2013, Thomas et al 2015. Accounting for some of these restrictions reduces the land suitable for woodland creation across the whole of Scotland to 2.96 Mha (Sing & Aitkenhead 2020). Creation of open woodland and scattered trees may be possible in areas deemed unsuitable for woodland creation. More work is needed to understand the constraints and barriers to woodland creation (Burton et al 2019) and how these vary for different woodland types, treescapes and woodland creation mechanisms. Research by Pollock et al (2015) has shown that birch woodlands in Scotland can regenerate with livestock grazing present, if there is sufficient good quality forage present. Silvopasture and farm woodlands also have the potential to store greater carbon than grassland pasture, whilst still supporting livestock (Beckert et al 2016). Some conservation areas, such as heathland, are currently protected from being converted to woodland, however small increases in shrub cover, scattered trees and woodland patches may not negatively impact and could potentially increase the biodiversity of these habitats (Fuller and Calladine 2014).
The potential for natural regeneration to achieve woodland expansion on timescales suitable to contribute to near-term climate targets needs to be better understood. Although woodland can regenerate with deer present, provided their impact is low , Tanentzap et al 2013, high browsing pressure from sheep and deer are a major constraint on tree regeneration in the Scottish uplands, with deer numbers increasing over the last few decades (Albon et al 2017). In many parts of the Scottish uplands, a lack of seed sources and competition from grasses may mean natural regeneration would be very slow even when grazing is reduced (Bunce et al 2014). In places where seed source is lacking, direct seeding may be a viable method of establishing native woodland at some sites (Willoughby et al 2019). Modelling of Scottish upland birch woodlands by Tanentzap et al (2013) shows how regeneration rates can be predicted using the variables of deer browsing pressure, adult tree sizes and locations, and ground favourability. Their analysis has the potential to predict above-ground carbon storage in response to changes in these variables. For example, they suggest that in a landscape with 8000 adult trees and 80% substrate favourability, browsing an average of 10% of trees could lead to the storage of at least 60 tC after 30 years. This represents substantially faster uptake of carbon than we assume, suggesting our estimates may be conservative. Greater understanding of the amount of carbon stored and rate of sequestration in different native woodland types, combined with such analysis, could prove a powerful tool for future land management decision making, particularly in relation to the government's net-zero targets.
Expansion of native woodlands would have many benefits in addition to carbon sequestration and climate mitigation. Native woodlands support and increase biodiversity (Scridel et al 2017), reduce flooding through increasing water use and infiltration and slowing overland flow (Nisbet and Thomas 2006, Jackson and Wheater 2008, Dadson et al 2017, improve water quality and provide physical and mental health benefits through providing opportunities for recreation (Ward Thompson et al 2005). Future work needs to characterise the full array of benefits provided by expansion of native woodlands.

Conclusions
Improving our understanding of carbon storage in naturally regenerated native woodlands is crucial to achieving the UK government's targets for woodland expansion and net-zero carbon emissions. By combining spatial data on the potential for native woodland across the Scottish uplands with carbon sequestration estimates based on Scottish native woodlands sites, our analysis shows that there is the potential for 3.9 Mha of new native woodland sequestering 6.96 Mt CO 2 yr -1 . Expanding woodlands to just 10% of this potential would double existing native woodland in Scotland and could provide a multitude of benefits, including carbon removal equivalent to approximately 4% of the UKCCC's target for all UK woodlands. By considering natural regeneration, alongside tree planting, there is the potential for more ambitious woodland creation targets, although factors such as grazing, seed availability and ground disturbance, as well as constraints around land use and policy, will need to be taken into account. Further work is now needed on how variables such as species assemblage, age, soil type or the method of woodland creation affect sequestration, as this will improve our understanding of the current and potential future carbon storage of these ecosystems.