Impacts of sampling and site factors on measuring nutrient concentrations
The sampling method in this study may have had some effect on these results as samples were taken in stands in the second half of their rotation. For base cations, it is known that concentrations of some species may decrease with increasing stem diameter as cation binding capacity can be negatively related to distance from pitch (Momoshima and Bondietti, 1990). Sampling only in the later stages of a rotation may therefore have had a negative effect on the concentrations of base cations measures in stem wood. However, loggings in the yield tables and in practise for the largest extend take place in the late thinnings and final felling, so sampling was in accordance with this.
Another aspect of the method in this study that has an effect on the results is the fact that most of the harvesting in the yield tables take place in the late thinnings and final felling. This has an effect on the average proportions of tree compartments used to calculate the nutrient concentrations in stems. In particular the proportions of heartwood—with relatively low-nutrient concentrations—increase with increasing tree size, as observed in this study and by Ojansuu and Maltamo (1995), thus having an negative impact on the average nutrient concentrations of stems. Both aspects may to some extend explain the lower concentrations of base cations and P in this study compared to the data of Jacobsen et al. (2003), which also include younger stands, but it does not provide an explanation for the differences of N.
The mass weighted mean concentrations in branches in our study are affected by the fraction fine branches, branch bark and branch wood we used. The fractions are based on the idea that branches are harvested only at final felling, having a relatively large maximum diameter, thus having a large proportion of (coarse) branch wood and a relatively small proportion of branch bark and fine branches compared to branches harvested in thinnings of the first half of the rotation. For beech and common oak, the proportions of branch bark mass compare well to the proportions that can be derived from the data by André et al. (2010): 9–15% for common oak and 7–10% for beech for larger trees. Haygreen and Bowyer (1982) however mention 21% bark in branches > 2.5-cm diameter for white pine and red maple. Data from Duvigneaud and Denaeyer-De Smet (1970) show the relation between branch diameter and bark mass proportion for common oak, which ranged from 13% (15–25-cm diameter) to 32% (1–7 cm). The average was 27% bark, clearly higher than in this study. Increasing the proportions of branch bark (+50%) and fine branches (+25%) in this study would lead to more similar levels of N compared to data in Jacobsen et al. (2003), while the differences for P, Ca, K and Mg get smaller, although they are still notable (Appendix Table 12). For branches—maybe more than for stems—the mix of compartments has a large effect and so has the sampling of branch wood. When comparing concentrations in branches, it is not always clear what parts of branches have been taken as a sample. It may be a whole branch of which the size is not known or it may be some part of a branch. In some cases, foliage is included in branch biomass and concentrations (Erikson and Rosen, 1994).
We suggest additional measurements of the proportions of different compartments of branches and stems and for branch and foliage biomass expansion factors as well to improve the calculations. This type of information is not widely available, but is highly relevant for nutrient budget calculation.
The nutrient concentrations in wood cannot be explained by differences in soil properties at the investigated sites. As we showed, there is no clear difference in concentrations between poor or moderately poor sandy soils. However, soils of all sampled sites all had low base saturations, mostly < 10%. In that aspect, they can all be categorised as poor on base cations, although at the moderate poor sites input from weathering will be larger. Bijlsma et al. (2020) found in Dutch forests clearly higher base cation levels in common oak stems on (rich) clay soils, compared to sandy soils.
Nutrient concentrations and the impacts of atmospheric deposition
Comparison of nutrient concentrations with literature data
Compared to the concentrations from Japanese larch, Norway spruce, Douglas fit and Scots pine in eight stands in the Netherlands by Kofman (1983), N concentration in this study is indeed mostly higher and they tend to be lower for P, Ca, K and Mg, even though tree samples by Kofman (1983) were relatively young: mostly < 50 years and DBH < 20 cm. Similarly, the comparison with data published by Jacobsen et al. (2003), based on measurements between 1958 and 2002, except for the N concentration in branch wood (Tables 3). In stems, the N concentrations in our study appear 25–52% higher. Concentrations of P, Ca and K are 5–82% lower, except for Ca in Norway spruce (Table 2), with differences being larger for conifers than for broadleaves. For Mg, however, concentrations are lower in four tree species and higher in three tree species (Table 2). Comparison with international data from De Vries et al. (1990) shows similar results. Comparison of stem bark nutrient concentrations with Jacobsen et al. (2003) shows smaller differences in N concentrations and mostly lower concentrations of P, Ca, K and Mg in this study (Appendix Table 13) as well. For stem wood (under bark), N concentrations are clearly higher, while P, Ca, K and Mg are lower in this study (Appendix Table 14). The differences in N concentrations are even larger than in total stems. Remarkable are the higher concentrations in stem wood for P in Scots pine and Mg in Norway spruce and Douglas fir in this study (Appendix Table 14). In branches, the concentrations of P, Ca and K and to a lesser extent of Mg are also lower than those reported in Jacobsen et al. (2003), but surprisingly N concentrations are mostly up to 39% lower too (except for Douglas fir, Table 3).
For sapwood and heartwood, Jacobsen et al. (2003) does not provide data, and individual references are not abundant. Comparing with averaged data provided by Duvigneaud and Denaeyer-De Smet (1968), Duvigneaud and Denaeyer-De Smet (1970) (South of Belgium) and Mussche et al. (1998) (West of Belgium) and Lévy et al. (1996) (North of France) however show the same pattern of higher average N concentrations and lower average concentrations of P, Ca, K and Mg in this study, compared to references (Appendix Table 15). Some of the references have lower P or base cation concentrations compared with this paper, and some have almost as high N concentrations. Nutrient concentration data in these references originating from Belgium, however, may have also been affected by high N deposition, in particular the data by Mussche et al. (1998) in the West of Belgium. Sampling years may also be a factor affecting the differences.
When comparing with data provided by Wright and Will (1957) and Häsänen and Huttunen (1989) for sapwood and heartwood of Scots pine, we see on average higher N concentrations and lower concentrations of P, Ca and Mg in this study and for K only in sapwood. Penninck et al. (2001) noted similar differences when comparing oak heartwood nutrient concentrations from acidic soils from the Netherlands and the central Belgium with richer soils in France.
Part of the differences for stem wood may be related to differences in factors such as stand age, since the sources in Jacobsen et al. (2003) provide data on a mixture of younger to older stands. Nutrient concentrations tend to decrease as stands grow older (Jacobsen et al., 2003; Augusto et al., 2000; Augusto et al. 2008), what can partly be related to the mix of compartments evolving with age. However, this aspect does not explain the large differences with reported literature values and the most plausible explanation is the impact of elevated atmospheric deposition of nitrogen, reflected in higher N contents and lower P, Ca, Mg and K contents in view of ongoing soil acidification. The impact on P availability is in line with Prietzel et al. (2020) who found a generally low P status of Scots pine needles in Germany. Lower nutrient concentrations of P, Ca, Mg and K may be influenced by increased growth due to increased N availability, causing a dilution in P, Ca, Mg and K contents, combined with ongoing soil acidification. Despite the low availability of P, Ca, Mg and K, the average increment of forests in the Netherlands appear to slightly higher (in order of magnitude 15%) compared to that of most other countries in Europe (Forest Europe, 2015). Apparently, the limited availability is not reflected in growth but in dilution of nutrient contents in stem wood and heartwood.
Atmospheric deposition most likely affects nutrient concentrations
A long-term monitoring at fixed locations is needed in order to asses unequivocally the impact of elevated N and acid deposition on nutrient concentrations in forest compartments. The ICP level II forest sites is such a monitoring network, but unfortunately, this only includes the assessment of nutrient contents in foliage and not in stems and branches. This monitoring network indicates clear changes in N, P and base cations in foliage response to deposition changes (e.g. Schmitz et al., 2019; Du et al., 2021). Circumstantial evidence for those changes may however be obtained from a comparison with nutrient concentrations in stems and branches of Dutch forests measured in the past (Kofman, 1983) and in other countries (Jacobsen et al., 2003), both experiencing lower N and acid input levels.
Our hypothesis that current nutrient concentrations would differ from those found in the past in the Netherlands and in other countries, experiencing lower N deposition levels, was clearly confirmed in this study. The results are in line with literature, showing that multiple decades with high N deposition, leading to N enrichment of the soil, may cause higher N contents in woody tissues (Balster et al., 2009; Saurera et al. 2004) and to soil acidification causing lower P, Ca, K and Mg depletion of the soil (De Vries et al. 2019), which may lead to lower contents in woody tissues (Bondietti et al., 1989). The results underline the importance of using region-specific and up-to-date nutrient concentrations. Using data from literature or from past measurements may lead to over or underestimation of sustainable harvest levels at least for The Netherlands. It may have resulted in mostly unneeded restrictions of wood harvest as the concentrations of Ca, K, Mg and P tend to be higher in literature compared to the data in this study.
Implications of changes in carbon to nitrogen stoichiometry on forest carbon sequestration
The carbon to nitrogen ratio in woody compartments is an important indicator that is used in so-called stoichiometric scaling approaches in models that assess global scale impacts of N deposition on forest carbon sequestration. Considering that carbon presents 50% of the wood biomass, the C/N ratio can be derived from the measured N concentrations. The stoichiometric scaling approach is based on the assumption that C to N ratios in forest biomass and soils are constant. The C–N response ratio, defined as the additional mass unit of sequestered C per additional mass unit of N deposition, is calculated by multiplying: (i) the fraction of external N inputs that is retained in the forest ecosystem with (ii) the fraction of retained N allocated to different forest compartments (woody biomass, non-woody biomass, and soil), and (iii) the C to N ratio of each compartment (De Vries et al., 2014; Du and de Vries, 2018). A fixed C to N ratio is sometimes not assumed, in the so-called flexible stoichiometric approaches, since N deposition is known to affect foliar N concentrations and thus the C/N ratios in foliage. However, in those approaches, the effect on woody C/N ratios is generally not included, also since information on impacts is lacking.
This study however shows that N concentrations in stems were app. 25–50% higher than those published in literature overviews. Considering that the average N contents in broadleaves varied from 2.1 to 3.0 g kg-1, this implies C/N ratios ranging from 170 to 240, while is normally near 1.5 to 2.1 g kg-1, being C/N ratios ranging from 240 to 330. Similarly, the average N contents in coniferous trees are all near 1.6 g kg-1 (C/N ratios near 300), while is normally ranging from 1.0 to 1.2 g kg-1, implying C/N ratios ranging from 400 to 500). So a tree can apparently adjust its C/N content in wood in regions with prolonged high N input due to luxury consumption. The Netherlands encountered more than 40 years of elevated N deposition, on average near 40–50 kg N ha-1 year-1 around 1980 to near 25 kg N ha-1 year-1 at present, dominated by ammonia deposition (Van Pul et al., 2018). This effect in regions with prolonged high N input is in line with other approaches such as the assessment of growth responses to experimental N addition and field N gradient studies, showing that there is a flattening of the growth response to a plateau near 15–30 kg N ha-1 year-1 and a reversal above that level (De Vries et al., 2014). This effect is mainly due to soil acidification, implying reduced P, Ca, K and Mg availability, which is also reflected in the P, Ca, K and Mg concentrations in this study, which were app. 5–80% lower than those published in literature overviews (Jacobsen et al., 2003).
This effect should preferably be included in global scale carbon sequestration models even though it is likely only relevant in small parts of the world, since about 90% of the global forests receive an N deposition below 15 kg N ha-1 year-1 (Schwede et al., 2018), likely implying a constant C/N ratio for most forest. This is in line with a comparison of model predicted and measured site-averaged (n = 22) ecosystem carbon (C) changes resulting from nitrogen (N) fertilization, showing the best comparisons when using a flexible C/N ration in leaves and roots, linked to NPP, but a constant ratio for the woody parts, linked to NEP (Meyerholt and Zaehle, 2015).
Nutrient removal and the impacts of management measures
In this study, we presented the nutrient removal for the considered tree species, i.e. Scots pine, Douglas fir, Norway spruce, Japanese larch, common oak, beech and silver birch, considering both whole-tree harvesting and stem wood removal only, by using the average growth rates of these tree species on relatively nutrient poor sandy soils in the Netherland. N removals through harvest of 5–11 kg N ha-1 year-1 are relatively small compared to N deposition, which is typically over 1700 mol or 24 kg ha-1 year-1 (Hoogerbrugge et al. 2019). For S, deposition has decreased from 80 kg ha-1 year-1 in the 1980s to app. 15 kg ha-1 year-1 currently, which is still clearly larger compared to potential removals of 0.3–0.7 kg ha-1 year-1 through harvest. Removals of P range from 0.18 to 0.35 kg ha-1 year-1 for stem harvest up to 0.51 kg ha-1 year-1 when including harvest of branches at median growth levels. This exceeds the P inputs of app. 0.25 kg ha-1 year-1 (De Vries et al., 2021). For the base cations, at median growth levels, removals are lower than the inputs of weathering and deposition. Removals of K (being 1.2–3.7 kg ha-1 year-1) in most cases are lower but in some cases (namely whole-tree harvest of Douglas fir, Norway spruce and beech) approach the inputs of weathering and deposition of app. 4 kg ha-1 year-1 (De Vries et al., 2021). The removals of Ca (1.3–7.4 kg ha-1 year-1) are lower than inputs (9 kg ha-1 year-1
, De Vries et al., 2021) and removals of Mg (0.32–1.07 kg ha-1 year-1) are strongly lower than inputs (6 kg ha-1 year-1
, De Vries et al., 2021). However, taking into account leaching, outputs of K and Ca may exceed inputs, in particular of common oak, Beech and Norway spruce.
We found that the harvest of logging residues at final felling increased nutrient removals with 20% (Ca for Scots pine) to 128% (P for Norway spruce) for coniferous species. This is in the low range of the findings of Raulund-Rasmussen et al. (2008), who noted that the nutrient removal may vary considerably depending on growth model, biomass equations and nutrient concentrations in different tree compartments. For broadleaved species, the additional removals were only 10% (Ca for common oak) to 25% (P for silver birch) higher, due to relative high-nutrient contents in stems, and no removal by foliage. Besides, less than half of the stem volume (42%) for broadleaved species beech and common oak is harvested at final felling with branches, meaning that a large part of branch biomass over one rotation remains in the stand at thinnings. For the coniferous species on the other hand, app. 50–55% of the stem wood volume is harvested at final fellings. For common oak and beech, the applied biomass expansion factors for branch wood (0.16) are lower than data by Baritz and Strich (2000), who give a biomass expansion factors for branch wood of broadleafed species of 0.24. Duvigneaud and Denaeyer-De Smet (1970) and André et al. (2010) give biomass expansion factors up to 0.30 for branches < 7 cm. We thus may have used too low biomass expansion factors for these species, and the effect of whole-tree harvest compared to stem only harvest may be larger. The biomass expansion factor for coniferous species given by Baritz and Strich (2000), however, is 0.14, which is lower compared to those used in this study (0.16–0.20).
Removals averaged over one rotation are in line with removals given by Raulund-Rasmussen et al. (2008) for Norway Spruce and Scots pine and with Palviainen and Finér (2012) for Norway spruce, Scots pine and silver birch. Nitrogen removals trough stem only harvest reported by Raulund-Rasmussen et al. (2008) e.g. are up to 10 kg ha-1year-1, similar to our results. But they are based on different average biomass removals. When correcting for total biomass removal for Norway spruce to match Raulund-Rasmussen et al. (2008), we see mostly higher N removals and lower Ca, Mg—even more notable—P removals in this study as compared to Raulund-Rasmussen et al. (2008) (excluding the data for the Rääkkylä site which are very low for all nutrients). Removals of K, based on the data in this study, were higher for Scots pine and lower for Norway spruce. Correcting for total biomass removal rates is however not straight forward as differences in rotation length affect tree compartment proportions (André et al. 2010) and thus average stem nutrient concentrations and average branch biomass removals.
Forest managers have several options to mitigate the effects of nutrient removal trough harvest. Addition of nutrients (fertilisation) is an obvious way to mitigate the effects of nutrient removals through wood harvest. Fertilisation was common practice in the late nineteenth and first half of the twentieth century, but nowadays, it is hardly practised anywhere in Dutch forests. The use of slow release base cation fertilisers (rock powder) is now evaluated, to avoid any unwanted effects on pH, accelerated mineralisation of organic matter and loss trough leaching (De Vries et al., 2019). The costs for this type of nutrient additions are relatively high and is only advised at sensitive sites with vitality issues.
Leaving branches in the forest for 6 months before removal will retain part of the nutrients in the forest. This may be effective for K, which may leach for 40–80% from needles and 30–40% from branches (Palviainen et al., 2004). However, P, Ca and Mg hardly leach from needles or branches in that time span (Palviainen et al., 2004; Staaf and Berg, 1982), but depending on species, 24–42% (Lehtikangas, 1991) of the needles may fall off in 4 months. From the perspective of logistic and forest road maintenance, this option is not preferred. Limiting the harvest of branch wood is still suggested as the favourable way to avoid depletion of P and base cations on nutrient poos sites.