Jukuri, open repository of the Natural Resources Institute Finland (Luke) All material supplied via Jukuri is protected by copyright and other intellectual property rights. Duplication or sale, in electronic or print form, of any part of the repository collections is prohibited. Making electronic or print copies of the material is permitted only for your own personal use or for educational purposes. For other purposes, this article may be used in accordance with the publisher’s terms. There may be differences between this version and the publisher’s version. You are advised to cite the publisher’s version. This is an electronic reprint of the original article. This reprint may differ from the original in pagination and typographic detail. Author(s): Wei He, Päivi Mäkiranta, Petra Straková, Paavo Ojanen, Timo Penttilä, Rabbil Bhuiyan, Kari Minkkinen & Raija Laiho Title: Fine-root production in boreal peatland forests: Effects of stand and environmental factors Year: 2023 Version: Published version Copyright: The Author(s) 2023 Rights: CC BY 4.0 Rights url: http://creativecommons.org/licenses/by/4.0/ Please cite the original version: He, W., Mäkiranta, P., Straková, P., Ojanen, P., Penttilä, T., Bhuiyan, R., Minkkinen, K., & Laiho, R. (2023). Fine-Root Production in Boreal Peatland Forests: Effects of Stand and Environmental Factors. Forest Ecology and Management 550, article id 121503. https://doi.org/10.2139/ssrn.4527353 Forest Ecology and Management 550 (2023) 121503 0378-1127/© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Fine-root production in boreal peatland forests: Effects of stand and environmental factors Wei He a,b, Paivi Makiranta b, Petra Strakova b,c, Paavo Ojanen a,b, Timo Penttila b, Rabbil Bhuiyan a,b, Kari Minkkinen a, Raija Laiho b,* a Department of Forest Sciences, University of Helsinki, Helsinki, Finland b Natural Resources Institute Finland (LUKE), Helsinki, Finland c Department of Ecosystem Biology, Faculty of Science, University of South Bohemia, Ceske Budĕjovice, Czechia A R T I C L E I N F O Keywords: Fine roots Forestry-drained peatland Ingrowth core Fourier transform infrared (FTIR) spectroscopy A B S T R A C T Fine roots are an important component of ecosystem carbon (C) cycling in boreal forests and peatlands. We aimed to estimate fine-root production (FRP) for a range of peatland forests, examine the patterns in, and develop models for estimating, the relationships between FRP and tree stand characteristics as well as envi- ronmental conditions. Fine-root production of 28 drained boreal peatland forest sites in Finland, representing different site types and soil water-table conditions, was measured using the ingrowth-core method. Total FRP and FRP of conifers decreased from south to north but long-term mean annual temperature sum and precipitation alone did not significantly explain this trend. Tree stand basal area predicted FRP better than any other stand variable alone, explaining 16 % of the variation in stand-level total FRP. Basal areas of pine and spruce correlated positively with the FRP of conifers. Total FRP varied considerably among the site types and, with the exception of the most fertile site type, decreased with decreasing fertility. A model that included stand basal area and site type accounted for 47% of the variation in stand-level total FRP. Total FRP was generally higher with a deeper water- table level (WT). Together, WT and basal area explained 25 % of the variation in stand-level total FRP. Most FRP occurred in the top 20-cm layer comprising 76–95% of total FRP. The most fertile site type showed lower FRP in deeper layers than the other site types. These results can be used for estimation of FRP with forest inventory data. 1. Introduction Peatland ecosystems are the carbon (C) hotspots of our planet, covering only 3 % of the global land surface but storing over 30 % of the global soil C pool in their peat deposits (Page et al., 2011; Xu et al., 2018; UNEP 2022). A critical component of the C cycle in boreal forests and peatlands is fine-root production (FRP), which may be equal to or larger than aboveground production (Persson, 1983; Gower et al., 2001; Kor- rensalo et al., 2018; Makiranta et al., 2018). Fine roots may represent a small part of the total forest biomass, but they play a vital role in nutrient and water absorption and are key contributors to soil C due to their fast turnover (Laiho et al., 2003;Clemmensen et al. 2013; Pandey et al., 2023). A small change in FRP can thus affect the ecosystem C sink. Fine-root production may be more sensitive to environmental and biotic changes and respond more strongly to disturbances than fine-root biomass (FRB) (Yuan and Chen, 2012). In addition, FRP may respond differently from FRB to gradients in environmental conditions (Lampela et al., 2023). Generally, however, estimates of root-mediated C fluxes in peatlands are still scarce, even though there is indication that they may play an important role in the C budget (Murphy and Moore, 2010), also following climate and/or land use change, e.g., to forestry (Ojanen et al., 2014, Bhuiyan et al., 2023). A considerable proportion of peatlands has been drained for forestry, particularly in the European part of the boreal zone. In Finland, peat- lands (undrained and drained) make up 35 % (9.1 million ha) of the forest land area, of which around 4.9 million ha are drained peatland forests (Kulju et al., 2023). Drawdown of the soil water-table level (WT) can significantly alter vascular plant productivity and species compo- sition (Weltzin et al., 2000; Laiho et al., 2003). These changes may have considerable impacts on ecosystem C and nutrient cycling through changes in C and nutrient inputs to soils by aboveground litter, as well as root production and turnover (Laiho, 2006). Our understanding of the * Corresponding author. E-mail address: raija.laiho@luke.fi (R. Laiho). Contents lists available at ScienceDirect Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco https://doi.org/10.1016/j.foreco.2023.121503 Received 25 July 2023; Received in revised form 11 October 2023; Accepted 14 October 2023 Forest Ecology and Management 550 (2023) 121503 2 response of peatlands to environmental change is based primarily on aboveground plant dynamics (e.g., Laine et al., 1995; Sarkkola et al., 2005; Strakova et al., 2010; McPartland et al., 2020), and far less is known about the belowground component (e.g., Finer and Laine, 1998; Bhuiyan et al., 2017; Malhotra et al., 2020). This gap in our knowledge concerning the belowground world largely stems from methodological holdbacks. Separating fine roots from soil and distinguishing between live and dead roots as well as the species identification of fine roots are arduous; especially so when it comes to peat soils that solely consist of plant remains, including roots, at various stages of decay (Sjors, 1991). If the relationships between FRP and more easily measurable variables could be identified, it would make an important contribution to C modeling and reporting. Fine-root production in boreal forest ecosystems has been found to correlate with climatic variables, such as latitude, temperature, and precipitation (Vogt et al., 1995; Gill and Jackson, 2000; Yuan and Chen, 2010; Finer et al., 2011). Moreover, soil properties such as soil nutrient regime can also affect FRP (Nadelhoffer 2000; Yuan and Chen, 2012), and these effects on FRP may be specific to stand type or species. Due to the lack of information concerning peatlands, it is still common to generalize patterns observed in mineral soil forests in ecosystem and earth models. Since the major constraints for ecosystem functioning - temperature, moisture, and nutrient regimes - as well as the physical, chemical, and biological properties of peat soils fundamentally differ from those of mineral soils (Westman and Laiho, 2003; Paivanen and Hånell, 2012), the patterns of FRP likely differ as well. Boreal forest ecosystems in northern regions have a cold climate that likely restricts the availability of nitrogen (N) and thus fine root growth (Nadelhoffer, 2000; Rasse, 2002). In contrast to mineral soils, peat soils have a higher nitrogen (N) content but less mineral nutrients (Westman and Laiho, 2003; Paivanen and Hånell, 2012). Therefore, especially at the most N-rich sites, the availability of N should be favourable, but higher FRP may be required to explore for mineral nutrients such as phosphorus (P) or potassium (K) that are often scarce in peat soils (Westman and Laiho, 2003). The limited observations in peatlands available so far suggest that total FRP is greater in more nutrient-rich sites (Laiho et al., 2014; Bhuiyan et al., 2017). However, the FRP pat- terns of individual species may differ from the pattern in total FRP. For example, Scots pine may produce fewer fine roots in peatland forests when the soil nutrient regime is better (Finer and Laine, 1998, 2000; Bhuiyan et al., 2017). In peatlands, even when drained, we also need to consider the WT, which largely determines the soil volume where aer- obic processes can take place, and is thus a major factor controlling ecosystem structure and function in these sites (Murphy et al., 2009; Murphy and Moore, 2010; Peltoniemi et al., 2021). Stand variables such as tree species, plant functional types (PFTs) making up the ground vegetation, FRB, and stand basal area also affect FRP (Tingey et al., 2005, Hendricks et al., 2006, Finer et al., 2011; Yuan and Chen, 2010). In peatlands, different species or PFTs may have different rooting patterns and rooting depths (Ruseckas, 2000; Bhuiyan et al., 2017; Proctor and He, 2019). Fine-root production correlates positively with FRB in boreal forests (Li et al., 2003; Chen et al., 2004). Based on a compilation of tree FRB (diameter < 2 mm) data from 95 Finnish forest stands, Lehtonen et al. (2016) developed models for estimating FRB of boreal forests and found that stand basal area was a better predictor of FRB than any other stand variable alone. However, it is still unclear whether this predictor can be utilized to estimate FRP. In this study, we examined FRP and its depth distribution at 28 study sites to fill the knowledge gap in the root production of drained boreal peatland forests. We used an ingrowth core method that has been modified (Laiho et al., 2014) to minimize disturbance to the soil and roots, and determined FRP by PFT using Fourier Transform Infrared (FTIR) spectroscopy method developed by Strakova et al. (2020). Since determination of FRP is laborious and costly, and FRP may be related to aboveground plant characteristics (Murphy et al., 2009, 2010) we sought for patterns in, and developed models for estimating, the relationships between FRP and aboveground stand parameters plus environmental and climatic conditions. Our hypotheses were that (1) FRP increases with increased mean annual temperature sum and pre- cipitation; (2) nutrient-rich sites show higher FRP; (3) sites with deeper WT have higher FRP; (4) basal area is the major determinant of FRP; (5) FRP of various species and PFTs show different depth distributions. 2. Materials and methods 2.1. Study sites The 28 forestry-drained peatland forest sites used in this study (Table 1) were a subset of the sites used by Ojanen et al. (2010, 2013) to quantify soil greenhouse gas emissions. We selected sites with a peat layer of at least 40 cm. The sites were situated between 60 and 67N, covering a wide range of variation in climatic conditions, site types and stand characteristics. Site types of these drained peatland forests ranged from the most fertile to the nutrient poorest (e.g., Westman and Laiho, 2003; Fig. 8): Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (MT II and I), and Dwarf shrub type (DsT). Both Vaccinium types were divided into two classes according to pre-drainage site type and characteristics: class II sites were sparsely treed or totally treeless before drainage, while class I sites were forested. Class II sites have on average more nitrogen (N) but less po- tassium (K) in the surface peat than class I sites. Based on peat N content, we rank the class II sites higher in the fertility gradient that their type I counterparts (Table 1). All stands included both trees that were on the site before drainage and those that appeared following drainage. These two classes are not always easily separated due to different growth dy- namics, which makes estimating stand age challenging, even misleading. That is why we do not include stand age as a parameter, as the values would not be comparable among our sites. Long-term temperature and rainfall data for the study sites for the period 1983–2015 were obtained from a database with a 10 km  10 km grid of daily weather data from the Finnish Meteorological Institute (FMI). The mean annual temperature sum (sum of daily mean temper- atures exceeding 5 C in degree days (dd)) varied between 1032 and 1424 dd. The mean annual precipitation varied between 506 and 617 mm. Stand, soil, and WT data (Tables 1, 2) were the same as used by Ojanen et al. (2010, 2013) and Lehtonen et al. (2016). The stand basal area varied from 4.7 to 39.1 m2/ha (average 19.1 m2/ha). The growing stock stem volume varied from 15.9 to 300.9 m3/ha (average 125.3 m3/ ha). The C: N (carbon: nitrogen) ratios for the topmost 20 cm peat layer varied from 19.6 to 43.8 (average 28.1). The mean WT of the sites varied Table 1 Mean characteristics of the study sites by site type  standard error. Soil bulk density (BD; g cm 3), calcium, phosphorus, and potassium concentrations (Ca, P, K; mg kg 1), and C:N ratios are for the 0–20 cm peat layer. Soil water-table level (WT; cm) was measured over May–October. Site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). n ˆ number of study sites. Site types n BD C:N Ca P K WT HrT 4 0.093  0.003 23  0.18 6719  2762 1166  132 504  29 28  2.5 MT(II) 5 0.130  0.005 28  0.22 3023  887 895  160 432  114 50  3.3 MT(I) 3 0.157  0.001 31  1.92 4527  147 943  318 479  41 46  0.2 VT(II) 7 0.125  0.001 25  0.44 2908  734 823  125 334  36 30  2.1 VT(I) 3 0.100  0.005 32  0.56 1746  105 1051  201 567  72 24  1.5 DsT 5 0.064  0.001 40  0.39 2345  162 526  37 491  55 25  1.4 W. He et al. Forest Ecology and Management 550 (2023) 121503 3 between 7 and 70 cm below soil surface during the snow-free period. 2.2. Ingrowth cores We applied the ingrowth core method according to the recently developed methodological guidelines for peat soils (Laiho et al., 2014; Bhuiyan et al., 2017, 2023). The novel type of corer-installer reduces disturbance during installation by pushing the holes for the cores in the peat instead of cutting and removing the peat. The diameter of the ingrowth cores was kept small to reduce the volume of ‘unnatural’ root- free substrate introduced in the soil. The cores were made of polyester fabric with a mesh size of about 1 mm  1 mm. Four peat substrates were chosen to mimic the soil quality of the study sites (Table 3): deep-horizon sedge (Carex) peat – to avoid abundant presence of live sedge roots – was used for the MT II and VT II sites, swamp peat consisting mostly of Sphagnum and tree remains for the MT I and HrT sites, and Sphagnum peat for the VT I and DsT sites. Carex-Sphagnum peat was used for one VT II site. The initial diameter of the filled core was 3.2 cm; however, post-incubation diameter was used in the calculations as the pressure of the surrounding soil was expected to modify the core diameters soon after the installation. The effective length was planned to be 50 cm with an additional tail that remained unfilled and above ground, to facilitate locating the cores after incuba- tion. The first test sets of similar cores have been analyzed earlier (Laiho et al., 2014; Bhuiyan et al., 2017). The results showed faster colonization by tree and shrub roots than in traditional ingrowth cores (Finer and Laine, 2000; Murphy et al., 2009), and indicated that 2-year incubation is long enough for estimating fine-root production (Bhuiyan et al., 2017). The ingrowth cores were installed between October 15th and November 27th, 2013, and recovered after two years in late November 2015. In all, 15 ingrowth cores were installed at each site. During re- covery of the cores, a long sharp knife was employed to carefully cut around the cores to separate any aboveground plant components attached to or growing through the cores, and to cut the root systems, especially rhizomes and any hard lateral expansion, to avoid risk of pulling out roots from the cores. The soil surface level was marked in the cores at recovery. In some sites, all cores were not found at the time of recovery, and some had been pulled out from the soil by some curious animals. The number of cores per site thus finally ranged from 7 to 13. After recovery, the cores were wrapped in plastic foil and stored frozen ( 20 C) until further processing. In the laboratory, the cores were removed from the freezer and defrosted overnight in a refrigerator. The defrosted cores were cut into five 10-cm segments. The top and bottom diameter of each segment was measured from two perpendicular directions. All above-ground biomass that was still attached to the cores was removed. Any roots found out- ward of a core segment were cut off along the fabric surface. Then, the roots inside the core were gently separated from the peat and washed clean with water. Estimation of living and dead roots was based on colour, elasticity and toughness (Makkonen and Helmisaari, 1999; Tufekcioglu et al., 1998; Bhuiyan et al., 2017). The roots were oven- dried to constant mass at 30 C and weighed. Fine-root production was estimated by transforming the total mass of the fine roots (including living and dead roots) extracted from the ingrowth cores to represent g m 2 year 1 using the post-incubation core diameter and dividing the value by the incubation time (2 years) (Bhuiyan et al., 2017; Yuan and Chen 2012). 2.3. Fourier Transform Infrared spectroscopy To prepare the samples, dried roots from each segment were powdered using an oscillating ball-mill. Subsequently, FTIR spectra of the powdered samples were acquired using a Bruker VERTEX 70 FTIR spectrometer (Bruker Optics, Germany) equipped with a horizontal diamond ATR sampling accessory. The powdered samples were directly placed on the diamond crystal with a diameter of 1.8 mm. To ensure uniform distribution and contact between the sample and crystal, a MIRacle high-pressure digital clamp was utilized. Each spectrum con- sisted of 65 averaged absorbance measurements within the range of 4000 to 650 cm 1, with a resolution of 2 cm 1. The measured data was collected using OPUS software. Table 2 Mean stand basal area (m2/ha) and stand stem volume (m3/ha) of the study sites by site type  standard error. Site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). Site types Stand basal area Stand stem volume Deciduous Spruce Pine Total Deciduous Spruce Pine Total HrT 21.0  1.2 7.4  0.7 0.6  0.2 29.0  1.6 163.0  11.8 48.8  5.5 3.5  1.0 215.4  15.3 MT(II) 12.5  1.0 5.9  0.9 7.3  0.6 25.7  1.4 93.2  7.7 32.8  4.6 59.4  6.4 185.4  11.1 MT(I) 4.1  0.2 17  0.5 0 21.1  0.5 24.6  1.4 121.5  8.6 0 146.1  8.8 VT(II) 4.7  0.4 0.6  0.1 11.7  0.8 16.9  1.0 25.3  2.4 1.7  0.3 68.4  5.1 95.5  6.7 VT(I) 2.1  0.5 0 11.7  1.0 13.8  1.5 13.7  3.7 0 75.7  8.4 89.4  11.6 DsT 0 0 11.0  1.0 11.0  1.0 0 0 54.1  6.5 54.1  6.4 Table 3 Calcium (Ca), phosphorus (P) and potassium (K) concentrations in the three peat types used in the ingrowth cores, mg kg 1  standard deviation, after incuba- tion. Depth indicates the 10-cm segments into which the cores were divided. Site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). Site type Depth (cm) Peat type Ca K P HrT, MT (I) 0–10 Sphagnum and tree remains 2965  424 279  96 295  11 10–20 2825  1219 184  88 260  2 20–30 3643  2190 182  63 260  17 30–40 4337  2250 126  42 254  11 MT(II), VT(II) 0–10 Carex 2810  1059 234  62 862  173 10–20 2357  1082 163  42 822  159 20–30 2422  1140 155  45 827  157 30–40 2461  1428 134  19 830  146 VT(I), DsT 0–10 Sphagnum 2584  1360 399  187 436  117 10–20 1822  453 257  107 390  96 20–30 1790  335 178  77 358  84 30–40 1928  454 151  66 360  82 W. He et al. Forest Ecology and Management 550 (2023) 121503 4 2.4. Statistical analyses 2.4.1. Effect of climatic, environmental and stand variables on fine-root production The effects of climatic, stand, and environmental variables such as mean annual precipitation (P), temperature sum (Tsum), latitude (L), tree stand stem volume (V), tree stand basal area (G) and stand basal area of tree species including Scots pine (GP), Norway spruce (GS) and deciduous trees (GD), site type (ST), C:N ratio of topmost 20 cm peat layer (CN), grouping of sites to nutrient rich (HrT, both MTs) and nutrient poor (VTs, DsT) (SG) as an alternative to specific site types, and average soil water-table level (WT) on FRP were tested with linear mixed-effect models using the ‘lme4′ package (Bates et al., 2015) in R (R Core Team; R version 3.5.3; RStudio version 1.2.1335). We started by fitting a model with no predictors, a ‘null model’: FRP ˆ a ‡ csite ‡ ε (1) where a is a fixed effect intercept, csite is the random effect factor of the site, accounting for correlation between observations within each site, and ε is sampling error. Next, we fitted the models using each of the variables in turn as a single dependent variable. The models are pre- sented in Table 4, where bi is the fixed effect parameter of variable i. Different combinations of predictors were then tested by adding them one-by-one into a model version having intercept and the single variable that had the best predictive power, tree stand basal area. This test included predictors with a significant effect on FRP based on models 2–11 (Table 4), where bGST is a fixed effect parameter related to site- type specific parameter for stand basal area. P values and R2 values were achieved from package ‘lmerTest’ (Kuznetsova et al., 2017). We used analysis of variance (ANOVA) and Akaike Information Criterion (AIC) to select the best model. As the FRP on HrT sites was surprisingly low (see Results and Discussion), we further ran a second set of models excluding the HrT sites from the data (Table 5). 2.4.2. Depth distribution of fine-root production The vertical distribution of FRP was modeled using the asymptotic equation proposed by Gale and Grigal (1987) and Jackson et al. (1996) for root biomass depth distribution. The equation Y ˆ 1 - βd (2) models the cumulative root fraction (Y) as a function of soil depth (d, cm) with β as the depth distribution parameter. The β values were calculated individually for each ingrowth core using SYSTAT software. Higher β values indicate a greater proportion of roots at deeper soil depths, while lower values indicate a shallower distribution of roots (Jackson et al., 1996). We then applied the linear mixed models for β values using the ‘lme4′ package in R, as described above for FRP, to analyze the response of the root depth distribution (in terms of coeffi- cient β) to the climatic and stand variables. 2.4.3. Plant functional type contributions derived from Fourier Transform Infrared spectra The FTIR data underwent preprocessing steps including Savitzky- Golay smoothing (second polynomial order, 11 smoothing points), baseline correction, mean normalization, and second derivative trans- formation (Savitzky-Golay derivative, second polynomial order, 15 smoothing points). The FTIR calibration models for the main plant functional types (PFTs) of northern peatlands (Strakova et al., 2020) were used to predict the mass proportions of roots of the main PFTs: (1) graminoids; (2) forbs; (3) ferns; (4) shrubs and birch (Betula pubescens); (5) conifers, in the root samples manually separated from the ingrowth cores. The forb Rubus chamaemorus was included in shrubs, as its fine roots do not differ from shrub roots chemically (Strakova et al., 2020). During the calibration and validation of the models, any predicted values of the dependent variable that were less than 0 % were set to 0 %, while values predicted to be greater than 100 % were set to 100 %. Data analyses were performed using the Unscrambler 10.3 (Camo Process AS; Oslo, Norway) package. The effects of environmental and stand vari- ables on FTIR-derived FRP by plant functional types were tested simi- larly to the total FRP with linear mixed-effect models using the ‘lme4′ package in R. 3. Results 3.1. Total fine-root production Individual core FRP values ranged from 5 to 800 g m 2 y-1 with a mean of 118 g m 2 y-1. Individual site FRP values ranged from 30 to 473 with a mean of 120 g m 2 y-1. FRP varied considerably among the site types (Fig. 1), being at its lowest, 48  7 g m 2 yr 1, in the HrT sites, and at its highest, 234  27 g m 2 yr 1, in the MT II sites. Excluding the HrT sites, FRP decreased with decreasing fertility of the site type. The very low FRP in the most fertile HrT sites was an unexpected result and we cannot fully rule out a bias (see Discussion). Because of that, we are showing model results both with and without HrT data (Table 4 versus Table 5). Mean annual precipitation had a non-significant relation to FRP (model 2 in Table 4 and 5), as did mean annual temperature sum (model 3 in Table 4 and 5). Latitude correlated negatively with FRP as expected (model 4 in Table 4 and 5, p ˆ 0.025 and 0.029, respectively), indicating that FRP decreases from south to north along the studied geographic range. Tree stand stem volume (model 5 in Table 4 and 5, p ˆ 0.053 and < 0.001, respectively) and basal area (model 6 in Table 4 and 5, p ˆ 0.006 and < 0.001, respectively) correlated positively with FRP. At the stand level, tree stand stem volume alone explained 8 % of the variation in FRP (Fig. 2a), while tree stand basal area explained 16 % (Fig. 2c). When we excluded the most fertile sites (HrT), tree stand stem volume alone explained 24 % of the variation in FRP (Fig. 2b), and the tree stand basal area explained 34 % (Fig. 2d), and mean FRP was 130 g m 2 y-1. Tree stand basal area was the best individual predictor for FRP (Tables 4 and 5). When the basal area was divided into three groups (pine, spruce and deciduous trees), the linear relationship between FRP and stand basal area was significant for pine (p < 0.05). Fitness of the group-level model (model 7) was not as good as that of total basal area (model 6) based on both AIC and ANOVA comparisons (Table 6). When we excluded the most fertile sites (HrT), the outcome was reversed (Table 6). The C:N ratio of the topmost 20 cm peat had a non-significant rela- tion to FRP at the stand level (model 9 in Tables 4 and 5). When we excluded the most fertile sites (HrT), FRP was significantly higher in nutrient-rich than in nutrient-poor sites (model 10 in Table 5, p ˆ 0.010). The depth of the WT correlated positively with FRP (Fig. 3, model 11 in Table 4 and 5, p < 0.001). We observed that site type (model 12 in Table 4 and 5, p < 0.001 and ˆ 0.029, respectively) was significant when added into model with tree stand basal area. Site type and basal area together were the best pre- dictors for FRP, explaining 47 % of the variation in stand-level FRP (Fig. 4). Also, WT (model 13 in Table 4, p ˆ 0.007) was significant when added into model with basal area. Together, WT and basal area explained 25 % of the variation in stand-level FRP. When we incorpo- rated WT into the model with site type and basal area (model 14), we discovered that the model was not improved (Table 6). Similarly, in- clusion of factors such as latitude (model 15) did not enhance the model (Table 6). 3.2. Fine-root production by plant functional types Woody species (shrubs and trees) accounted for most of the FRP, on average 83 % of the total FRP in HrT, 94 % in MT II, 90 % in VT I, 81 % in VT II, and 72 % in DsT (Fig. 5). The remaining FRP was accounted for by herbaceous species (graminoids, forbs, and ferns). Fine-root production W. He et al. Forest Ecology and Management 550 (2023) 121503 5 Table 4 Models for predicting fine-root production (FRP; g m 2 year 1) and ß value using the full data set (including HrT sites), with parameter estimates (a, bi), Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and P values (P). Parameter a is the fixed effect intercept, bi is the fixed effect parameter of variable i (mean annual precipitation (P; mm year 1), temperature sum (Tsum; dd), latitude (L; ), tree stand stem volume (V; m3/ha), tree stand basal area (G; m2/ha), stand basal area of tree species including (pine (GP; m2/ha), spruce (GS; m2/ha) and deciduous trees (GD; m2/ha), site type (ST), C:N ratio of topmost 20 cm peat layer (CN), grouping of sites to nutrient rich or poor (SG), and soil water-table level (WT; cm)). Models 1, 16 and 19 are the null model. Bolded values indicate statistical sig- nificance at p < 0.05. Model Formula a bi AIC BIC P 1 FRP ˆ a ‡ csite ‡ ε 120.15 – 3164.1 3174.8 – 2 FRP ˆ a ‡ bpP ‡ csite ‡ ε 88.25 0.56 3165.6 3180.0 0.913 3 FRP ˆ a ‡ bTsumTsum ‡ csite ‡ ε 128.47 0.21 3162.8 3177.2 0.070 4 FRP ˆ a ‡ bLL ‡ csite ‡ ε 1.45  103 1.89  10-4 3161.0 3175.4 0.025 5 FRP ˆ a ‡ bVV ‡ csite ‡ ε 72.45 0.38 3162.3 3176.7 0.053 6 FRP ˆ a ‡ bGG ‡ csite ‡ ε 36.16 4.43 3158.4 3172.8 0.006 7 FRP ˆ a ‡ bGPGP ‡ bGSGS ‡ bGDGD ‡ csite ‡ ε 28.62 bGP ˆ 5.44 bGS ˆ 5.75 bGD ˆ 3.65 3161.9 3183.5 0.044 8 FRP ˆ a ‡ bSTST ‡ csite ‡ ε 86.17 bHrT ˆ -38.45 bMT(II) ˆ 155.36 bMT(I) ˆ 36.42 bVT(II) ˆ 31.80 bVT(I) ˆ -1.66 bDsT ˆ 0 3157.7 3186.5 0.011 9 FRP ˆ a ‡ bCNCN ‡ csite ‡ ε 140.37 0.71 3165.4 3179.8 0.765 10 FRP ˆ a ‡ bSGSG ‡ csite ‡ ε 99.76 47.65 3164.2 3178.6 0.956 11 FRP ˆ a ‡ bWTWT ‡ csite ‡ ε 12.94 3.22 3152.6 3167.0 <0.001 12 FRP ˆ a ‡ bGSTG ‡ csite ‡ ε 30.04 bGHrT ˆ 0.65 bGMT(II) ˆ 8.70 bGMT(I) ˆ4.22 bGVT(II) ˆ 4.74 bGVT(I) ˆ 4.44 bGDsT ˆ 4.81 3138.7 3171.0 <0.001 13 FRP ˆ a ‡ bGG ‡ bWTWT ‡ csite ‡ ε 4.37 bG ˆ 2.04 bwt ˆ 2.60 3153.0 3171.0 0.007 14 FRP ˆ a ‡ bGSTG ‡ bWTWT ‡ csite ‡ ε 18.19 bGHrT ˆ 0.31 bGMT(II) ˆ 7.70 bGMT(I) ˆ 3.23 bGVT(II) ˆ 4.18 bGVT(I) ˆ 3.98 bGDsT ˆ 4.40 bWT ˆ 0.72 3139.7 3175.7 0.410 15 FRP ˆ a ‡ bGSTG ‡ bLL ‡ csite ‡ ε 5.392  102 bGHrT ˆ -1.490  10-1 bGMT(II) ˆ 7.773 bGMT(I) ˆ 2.694 bGVT(II) ˆ 3.401 bGVT(I) ˆ 3.307 bGDsT ˆ 2.895 bL ˆ -6.941  10 -5 3139.8 3175.7 <0.001 16 FRP of conifers ˆ a ‡ csite ‡ ε 41.540 – 924.8 932.3 – 17 FRP of conifers ˆ a ‡ bLL ‡ csite ‡ ε 6.998  102 9.341  10-5 922.7 932.7 0.044 18 FRP of conifers ˆ a ‡ bBPGP ‡ bBSGS ‡ csite ‡ ε 1.384 bGP ˆ 2.331 bGS ˆ 5.443 917.4 929.9 0.003 19 ß value: ˆ a ‡ csite ‡ ε 0.89 – 686.9 676.3 – 20 ß value: ˆ a ‡ bpP ‡ csite ‡ ε 0.88 8.74  10–6 684.9 670.7 0.970 21 ß value ˆ a ‡ bTT ‡ csite ‡ ε 0.95 4.86  10-5 685.6 671.3 0.429 22 ß value ˆ a ‡ bLL ‡ csite ‡ ε 0.85 5.92  10-9 685.0 670.7 0.841 23 ß value ˆ a ‡ bVV ‡ csite ‡ ε 0.91 1.52  10-4 687.7 673.5 0.098 24 ß value ˆ a ‡ bGG ‡ csite ‡ ε 0.92 1.21  10-3 687.5 673.3 0.114 25 ß value ˆ a ‡ bGPGP ‡ bGSGS ‡ bGDGD ‡ cS,i ‡ ε 0.90 bGP ˆ 2.13  10-4 bGS ˆ -1.50  10 -3 bGD ˆ -1.15  10 -3 685.6 664.3 0.197 26 ß value ˆ a ‡ bSTST ‡ csite ‡ ε 0.91 bHrT ˆ -0.095 bMT(II) ˆ -0.004 bMT(I) ˆ -0.047 bVT(II) ˆ -0.056 bVT(I) ˆ -0.178 bDsT ˆ 0 701.4 672.9 <0.001 27 ß value ˆ a ‡ bCNCN ‡ csite ‡ ε 0.83 2.22  10-3 689.7 675.4 0.030 28 ß value ˆ a ‡ bNN ‡ csite ‡ ε 0.91 2.94  10-2 688.6 674.4 0.056 29 ß value ˆ a ‡ bWTWT ‡ csite ‡ ε 0.88 0.16  10-4 685.1 670.9 0.712 30 ß value ˆ a ‡ bCNSTCN ‡ csite ‡ ε 0.85 bCNHrT ˆ -1.68  10-3 bCNMT(II) ˆ -2.45  10 -3 bCNMT(I) ˆ -5.07  10 -4 bCNVT(II) ˆ -1.74  10 -3 bCNVT(I) ˆ -1.23  10 -3 bCNDsT ˆ -1.45  10 -3 699.0 666.9 <0.001 W. He et al. Forest Ecology and Management 550 (2023) 121503 6 of ferns was minor at all sites, on average, 2 %. Fine-root production of graminoids was highest in the DsT sites, 17 %, whereas FRP of forbs was highest in the HrT sites, 10 %. Separation of FRP by depth and plant functional type indicated that shrubs and trees distributed the majority of their FRP in the top 20 cm of the soil profile whereas herbs distributed the majority of their root production from 30 to 50 cm below the soil surface (Fig. 6). Few correlations were found between PFT-level FRP and the climatic and stand variables. Latitude correlated negatively with FRP of conifers (model 17 in Table 4 and 5, p ˆ 0.042 and 0.025, respectively), indi- cating FRP of conifers decreases from south to north. No such trend existed in FRP of shrubs and birch. Pine and spruce basal areas corre- lated positively with the FRP of conifers (model 18 in Table 4 and 5, p ˆ 0.003 and < 0.001, respectively). Table 5 Models for predicting fine-root production (FRP) and ß value using data without the HrT sites, with parameter estimates (a, bi), Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC) and P values (P).Parameter a is the fixed effect intercept, bi is the fixed effect parameter of variable i (mean annual precipitation (P; mm year 1), temperature sum (Tsum; dd), latitude (L; ), tree stand stem volume (V; m3/ha), tree stand basal area (G; m2/ha), stand basal area of tree species including (pine (GP; m2/ha), spruce (GS; m2/ha) and deciduous trees (GD; m2/ha), site type (ST), C:N ratio of topmost 20 cm peat layer (CN), grouping of sites to nutrient rich or poor (SG), and soil water-table level (WT; cm)). Models 1, 16 and 19 are the null model. Bolded values indicate statistical significance at p < 0.05. Model Formula a bi AIC BIC P 1 FRP ˆ a ‡ csite ‡ ε 132.22 – 2731.9 2742.2 – 2 FRP ˆ a ‡ bpP ‡ csite ‡ ε 101.89 0.05 2733.9 2747.7 0.928 3 FRP ˆ a ‡ bTsumTsum ‡ csite ‡ ε 152.42 0.24 2731.1 2744.9 0.098 4 FRP ˆ a ‡ bLL ‡ csite ‡ ε 1.58  103 2.06  10-4 2728.9 2742.7 0.029 5 FRP ˆ a ‡ bVV ‡ csite ‡ ε 47.48 0.77 2722.6 2736.4 <0.001 6 FRP ˆ a ‡ bGG ‡ csite ‡ ε 0.89 1.77  10-5 2715.0 2728.7 <0.001 7 FRP ˆ a ‡ bGPGP ‡ bGSGS ‡ bGDGD ‡ csite ‡ ε 35.92 bGP ˆ 2.57 bGS ˆ 4.24 bGD ˆ 11.76 2708.1 2728.8 <0.001 8 FRP ˆ a ‡ bSTST ‡ csite ‡ ε 123.56 bMT(II) ˆ 117.99 bMT(I) ˆ 0 bVT(II) ˆ -5.58 bVT(I) ˆ -39.06 bDsT ˆ -37.39 2728.4 2752.5 0.021 9 FRP ˆ a ‡ bCNCN ‡ csite ‡ ε 224.24 3.08 2732.6 2746.3 0.248 10 FRP ˆ a ‡ bSGSG ‡ csite ‡ ε 99.74 99.64 2727.3 2741.1 0.010 11 FRP ˆ a ‡ bWTWT ‡ csite ‡ ε 20.56 3.27 2722.0 2735.8 <0.001 12 FRP ˆ a ‡ bGSTG ‡ csite ‡ ε 30.55 bGMT(II) ˆ 8.68 bGMT(I) ˆ 4.20 bGVT(II) ˆ 4.72 bGVT(I) ˆ 4.41 bGDsT ˆ 4.78 2712.2 2739.7 <0.001 13 FRP ˆ a ‡ bGG ‡ bWTWT ‡ csite ‡ ε 14.10 bG ˆ 5.71 bwt ˆ 1.37 2714.7 2731.9 0.132 14 FRP ˆ a ‡ bGSTG ‡ bWTWT ‡ csite ‡ ε 19.00 bGMT(II) ˆ 7.83 bGMT(I) ˆ 3.35 bGVT(II) ˆ 4.26 bGVT(I) ˆ 4.05 bGDsT ˆ 4.48 bWT ˆ 0.65 2713.6 2744.6 0.442 15 FRP ˆ a ‡ bGSTG ‡ bLL ‡ csite ‡ ε 5.221  102 bGMT(II) ˆ 7.813 bGMT(I) ˆ 2.752 bGVT(II) ˆ 3.456 bGVT(I) ˆ 3.545 bGDsT ˆ 2.972 bL ˆ -6.711  10 -5 2.713.6 2744.5 <0.001 16 FRP of conifers ˆ a ‡ csite ‡ ε 45.45 – 830.1 837.3 – 17 FRP of conifers ˆ a ‡ bLL ‡ csite ‡ ε 8.525  102 1.145  10–4 827.12 836.64 0.025 18 FRP of conifers ˆ a ‡ bGPGP ‡ bGSGS ‡ csite ‡ ε 11.110 bGP ˆ 1.675 bGS ˆ 6.077 819.97 831.88 <0.001 19 ß value: ˆ a ‡ csite ‡ ε 0.90 – 662.7 652.3 – 20 ß value: ˆ a ‡ bpP ‡ csite ‡ ε 0.86 6.74  10-5 660.8 647.1 0.726 21 ß value ˆ a ‡ bTT ‡ csite ‡ ε 0.99 7.18  10-5 663.1 649.3 0.122 22 ß value ˆ a ‡ bLL ‡ csite ‡ ε 0.80 1.46  10-8 661.0 647.3 0.564 23 ß value ˆ a ‡ bVV ‡ csite ‡ ε 0.91 9.36  10-5 662.0 648.3 0.249 24 ß value ˆ a ‡ bGG ‡ csite ‡ ε 0.91 7.36  10-4 662.0 648.2 0.255 25 ßvalue ˆ a ‡ bGPGP ‡ bGSGS ‡ bGDGD ‡ cS,i ‡ ε 0.92 bGP ˆ -1.45  10-3 bGS ˆ -2.82  10 -3 bGD ˆ 1.14  10 -3 663.8 643.3 0.068 26 ß value ˆ a ‡ bSTST ‡ csite ‡ ε 0.87 bMT(II) ˆ 5.15  10-2 bMT(I) ˆ 0 bVT(II) ˆ 3.11  10 -2 bVT(I) ˆ 2.94  10 -2 bDsT ˆ 4.69  10 -2 662.1 638.1 0.117 27 ß value ˆ a ‡ bCNCN ‡ csite ‡ ε 33.40 0.78 661.3 647.6 0.437 28 ß value ˆ a ‡ bNN ‡ csite ‡ ε 0.90 3.14  10-3 660.7 647.0 0.819 29 ß value ˆ a ‡ bWTWT ‡ csite ‡ ε 0.91 2.12  10-4 661.0 647.4 0.535 W. He et al. Forest Ecology and Management 550 (2023) 121503 7 3.3. Depth distribution of total fine-root production Overall, most (76–95 %) of the FRP took place in the uppermost 20 cm of the peat profile (Table 7). The asymptotic depth distribution model (Equation 2) fitted well to the data (Fig. 7). The climatic variables did not have significant relationships with the ß value (Table 4, p > 0.05). Of the stand variables only site type (Model 26 in Table 4, p < 0.001) and the C:N ratio of topmost 20 cm peat layer (Model 27 in Table 4, p ˆ 0.03) had significant relationships with ß value. The most fertile HrT sites showed smaller proportion of FRP in deeper layers than the other sites, while the depth distribution of FRP in the nutrient poorest DsT sites was deeper than the other sites (Fig. 7). Correspond- ingly, ß value correlated positively with the C:N ratio of topmost 20 cm peat layer (Fig. 8) indicating that roots are produced proportionally deeper with decreasing fertility. Site type and soil C:N ratio together explained 20 % of the variation in the ß value (Figs. 8, R2m ˆ 0.201). Water-table level had a non-significant effect on the ß value (Model 29 in Table 4). When we excluded the HrT sites, neither climatic nor stand variables had a significant effect on the ß value (Table 5). 4. Discussion 4.1. Fine-root production in peatland forests To our knowledge, this is the first comprehensive survey of fine-root production (FRP) in boreal peatland forests, covering a wide north- –south range as well as different tree stand structures, and soil nutrient and water-level regimes. We found a strong correlation between basal area and total FRP at the stand level. Stand basal area alone explained 16 % of the variation in FRP. Finer et al. (2011) examined the re- lationships between environmental and stand variables and FRP in forests at the stand or tree level. They discovered that stand basal area explained 28 % and FRB as much as 53 % of the variation in the FRP for trees at the tree level. Less variation in FRP could be explained at the stand level. Lehtonen et al. (2016) used data from 95 forest stands (both mineral-soil and peatland forests) to develop models for estimating FRB (diameter < 2 mm) of boreal forests, and found that stand basal area predicted also FRB better than any other stand variable alone. As trees grow in diameter and add to their basal area, they require more re- sources to support their growth and maintenance, and this is often achieved by increasing the allocation of resources to the root system. However, the strength of this relationship can vary depending on site- specific factors such as soil nutrient and hydrological regimes, as well as tree species composition. The relationship between stand basal area and FRP at the MT (I) sites was not consistent with the other site types, but this is probably just due to the small variation in the basal area not facilitating reliable estimation of the relationship for this site type. Adding site type, a general descriptor of site nutrient regime, to the stand basal area model increased the coefficient of determination to 47 %. The soil nutrient regime thus clearly influences FRP; however, the directions of the responses have varied among studies in different eco- systems. Recent FRP studies from stands on mineral soils have shown total FRP (pine and understorey) to decrease with increasing site fertility (Ding et al., 2021). For peatlands, total FRP has previously been observed to be higher in more nutrient-rich and floristically diverse sites than in nutrient-poor sites (Finer and Laine, 2000; Bhuiyan et al., 2017). In contrast, a recent study by Lampela et al. (2023) reported that FRP was generally higher in the nutrient-poor, pine-dominated sites than the nutrient-rich, spruce-dominates sites. These contrasting findings high- light the complexity of FRP dynamics and suggest that factors such as tree species composition and their associated nutrient requirements may influence root productivity. Our extensive data yet generally indicates decreasing FRP with decreasing site fertility, apart from HrT. Unexpectedly, we observed the lowest FRP for the most fertile site type, HrT. Apart from that, production decreased with decreasing site type fertility. We have two potential explanations. First, we observed that the peat used in the cores of the HrT sites contained clearly less phosphorus (P) than the ambient soils of the sites (Tables 1 and 2). Thus, the lowest FRP observed for the most fertile HrT sites may simply mean that roots avoided the cores in these sites, where P is often a growth- limiting nutrient (Paivanen and Hånell, 2012). Second, the HrT sites were relatively wet (Fig. 3), which can both limit the availability of oxygen to plant roots retarding FRP, but also increase nutrient avail- ability with the inflowing water leading to less need for FRP. Verifying the critical mechanism would require a specific further study. In addition to the soil nutrient regime, the growth of tree roots on drained peatlands may be limited intermittently due to a lack of oxygen, and several studies have shown a positive correlation between the depth of the soil water-table level (WT) and FRP (Lieffers and Rothwell, 1987; Finer and Laine, 1998; Murphy et al., 2009; Murphy and Moore, 2010). When WT is deeper, air-filled pore volume of the rooting zone is greater, which promotes root growth (Boggie, 1972). Also in our study sites, total FRP was generally higher with a deeper WT, possibly as a result of a greater volume of aerated soil. However, stand basal area and WT depth generally have a positive correlation, due to both better site conditions for tree growth when the WT is deeper and the biological drainage through evapotranspiration increasing with increasing stand basal area (Sarkkola et al., 2005). This hampers determination of the primary factor leading to higher FRP with deeper WT. Our results indicated that FRP in peatland forests decreases with increasing latitude but does not correlate with either mean annual temperature sum or precipitation. It is possible that latitude, as a mea- sure of geographical location, encompasses a range of environmental conditions that are not captured by mean annual temperature or pre- cipitation alone. Stand characteristics were also earlier found to explain a greater proportion of the variation in FRP than environmental factors (Finer et al., 2011). So, also the possible effect of climate, as well as WT depth, seems to be largely explained by stand characteristics. In comparison to earlier studies in drained peatland forests, our total FRP estimates fluctuated within a wider range. Finer and Laine (1998, 2000) reported annual FRP (88 to 239 g m 2 yr 1, diameter < 2 mm) for Fig. 1. Mean fine-root production, g m 2 yr 1  standard error, by site type: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). Site types are here ordered based on their surface peat N content, decreasing from HrT to DsT (Westman and Laiho 2003). Different letters denote statistically significant differences between site types. W. He et al. Forest Ecology and Management 550 (2023) 121503 8 three drained peatland forests using sequential coring and ingrowth cores over a 3-year period. The DsT sites from that study yielded FRP (88 g m 2 y-1) similar to ours (85 g m 2 y-1). Bhuiyan et al. (2017) re- ported a production of 244, and 561 g m 2 for fine roots derived from ingrowth cores installed for 2 years in a DsT site, and a MT (II) site, respectively. Other estimates were based on 1-year production data, e.g., 118 g m 2 yr 1 for a DsT (Laiho et al., 2014), 216 g m 2 yr 1 for a MT (II) (Laiho et al., 2014) similar to our 2-year mean (234 g m 2 yr 1), and 78 g m 2 yr 1 for a DsT site (Murphy et al., 2009). These differences may result from our study covering a higher number and wider geographic range of sites, as well as different tree stand structures and development stages. 4.2. Fine-root production by plant functional type Visually distinguishing between the roots of different tree species and plant functional types can be difficult and occasionally arbitrary, especially when multiple tree roots grow in the same layer and resemble one another in terms of form and color. As a result, misidentification of Fig. 2. Relationship between tree stand stem volume and fine-root production (a, with HrT sites; b, without HrT sites), and tree stand basal area and fine-root production (c, with HrT sites; d, without HrT sites). Open circles represent HrT Sites. Solid line depicts the fitted linear regression lines, with their 95 % confi- dence intervals indicated by the shaded areas. R2m describes the proportion of variance explained by fixed effects. R2c represents the proportion of variance explained by both fixed and random effects. Table 6 Comparisons of the linear mixed models. * Probability of the lower order model is better than that of the latter model, P indicates P values in ANOVA compar- ison; Δ AIC indicates difference of AIC values. Compared models Δ AIC P with HrT sites Model 6 and 7 3.5 0.804 Model 6 and 12 19.7 1.671  10-5 * Model 6 and 13 5.4 6.596  10-3 * Model 12 and 14 1 0.319 Model 12 and 15 1.1 0.343 without HrT sites Model 6 and 7 6.9 4.384  10-3 * Model 6 and 12 2.8 2.87  10-2 * Model 6 and 13 0.3 0.132 Model 12 and 14 0.4 0.443 Model 12 and 15 1.4 0.439 W. He et al. Forest Ecology and Management 550 (2023) 121503 9 roots may occur. Our study is the first to quantify the FRP and its depth distribution by different tree species and plant functional types in peatland forests using FTIR spectroscopy. In peatlands where the carbon cycle is dominated by fluxes mediated by vascular PFTs: sedges, shrubs, and/or trees (Sjors, 1991; Saarinen, 1996; Laiho et al., 2003; Jauhiainen et al., 2005; Bubier et al., 2006), such analysis provides a further insight into ecosystem structure and function. Herbaceous PFTs are to a large extent replaced by trees and shrubs at drained sites (Lampela et al., 2023). Our results, quite logically, indicated that shrubs and trees ac- count for most of the FRP in peatland forests. 4.3. Vertical distribution of fine-root production Most of the total FRP occurred in the top 0–20-cm of the peat soil and declined significantly at subsequent depths, which is in accordance with previous observations (Ruseckas, 2000; Murphy and Moore, 2010; Bhuiyan et al., 2017; Lampela et al., 2023). The majority of the FRP from shrubs and trees was similarly distributed in the top 20 cm of the soil Fig. 3. Relationship between soil water-table level and total fine-root production (a, with HrT sites; b, without HrT sites). Open circles represent HrT Sites. Solid line depicts the fitted linear regression lines, with their 95 % confidence intervals indicated by the shaded areas. R2m describes the proportion of variance explained by fixed effects. R2c represents the proportion of variance explained by both fixed and random effects. Fig. 4. Relationship between tree stand basal area and fine-root production for the different site types separately. Site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). R2m describes the proportion of variance explained by fixed effects. R2c represents the proportion of variance explained by both fixed and random effects. Solid lines depict the fitted linear regression lines, with their 95 % confidence intervals indicated by the shaded areas. Fig. 5. Arithmetic means of FTIR-derived fine-root production by plant func- tional type in different site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). W. He et al. Forest Ecology and Management 550 (2023) 121503 10 profile, whereas herbs distributed most of their root production from 30 to 50 cm below the soil surface. For woody trees and shrubs that do not form aerenchymous roots, the extent of the oxic soil layer above the WT logically plays a critical role in determining the distribution of fine roots (Schenk and Jackson, 2002; Murphy and Moore, 2010). Better nutrient availability in the upper soil layers may also play a role (Jobbagy and Jackson, 2000; Murphy and Moore, 2010). Many graminoids, on the other hand, have aerenchymous roots and may produce substantial root biomass even under waterlogged conditions and deep into the anoxic soil layers (Bernard et al., 1988; Saarinen, 1996; Proctor and He, 2019). Somewhat surprisingly, our results suggested that WT has no effect on the depth distribution of FRP in peatland forests. Similarly, Heikur- ainen (1955) and Paavilainen (1966) reported that the average depth of the root systems increases only little when forested peatlands are drained, even though the WT drops deeper. Yet, it seems that at the nutrient-poor sites (VT II and I, DsT), a higher proportion of FRP occurring at depth may be needed to scavenge enough mineral nutrients. Although shrubs and trees are shallow-rooted, they still produce a Fig. 6. The proportions of FTIR-derived fine-root production by plant functional types in different depths in different site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). In HrT, no fine-root production was observed below 30 cm. W. He et al. Forest Ecology and Management 550 (2023) 121503 11 considerable amount of fine-root biomass (also Finer and Laine, 2000; Iversen et al., 2018). 5. Conclusions This study presents a set of models that can be used with forest in- ventory data to quantify FRP in boreal peatland forests. Stand basal area, site type, and WT were the variables best predicting FRP in our drained peatland forests. Even though the FRP models were based on a large dataset representing well the variation in basal area, fertility and cli- matic conditions of boreal drained peatland forests, half of the variation in FRP remained unexplained. More detailed reporting of stand and environmental characteristics in forthcoming studies could increase the predictive power of FRP models and improve our understanding of the C cycle in boreal peatland forests. To avoid biased results in further studies, the ingrowth cores should preferably be filled with peat from the study sites or alternatively, the main nutrient contents of the standard peats should be extensively compared with the peat of the study sites in advance, even though both options increase the amount of labor involved. Author contributions Raija Laiho designed the research with contributions from Paivi Makiranta, Paavo Ojanen, Timo Penttila and Kari Minkkinen. Wei He performed Fourier Transform Infrared (FTIR) spectroscopy analyses under supervision of Petra Strakova. Data analyses were done by Wei He. Wei He led the writing under supervision of Raija Laiho, Paivi Makiranta and Paavo Ojanen. All authors participated in material preparation, data collection and analysis, commented on the manu- script, and approved the submitted version. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Data availability The data are available at Zenodo, https://doi.org/10.5281/ zenodo.8182401 Acknowledgements This study was funded by the Academy of Finland (289116, 289586). WH was supported by the China Scholarship Council (202007960010). Partial financial support during the reporting phase was received from the LIFE financial instrument of the European Union, project LIFE18 CCM/LV/001158 LIFE OrgBalt, and from the Academy of Finland (348431). References Bates, D., Machler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models using lme4. J. Stat. Soft. 67, 1–48. https://doi.org/10.18637/jss.v067.i01. Bhuiyan, R., Minkkinen, K., Helmisaari, H.S., Ojanen, P., Penttila, T., Laiho, R., 2017. Estimating fine-root production by tree species and understorey functional groups in Table 7 Mean fine-root production, g/m 2( |-) yr 1  standard error, by depth and site type. Site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). Depth (cm) HrT MT(II) MT(I) VT (II) VT(I) DsT 0 to 10 38.8  5.7 132.7  18.3 88.6  12.3 69.2  6.8 55.2  9.0 49.0  5.0 10 to 20 7.7  1.7 48.4  7.8 26.9  11.4 24.4  3.0 18.7  3.7 14.8  1.8 20 to 30 2.1  0.8 28.5  3.4 6.9  1.5 13.1  1.6 6.8  1.4 9.4  1.2 30 to 40 0.4  0.3 18.3  3.5 5.8  1.4 7.8  1.1 3.7  0.6 8.4  1.2 40 to 50 0.2  0.1 10.2  3.5 2.7  2.3 5.0 ‡ 1.1 1.4  0.4 4.0  0.8 Fig. 7. Mean cumulative fraction of fine-root production for different site types. Error bars are standard error of the mean. The lines represent the asymptotic regressions to the data using the equation x ˆ 1–ß d, where ß de- scribes the relative proportion of root located at depth. Fig. 8. Relationship between ß values (ingrowth-core level) and the C:N ratio of the topmost 20 cm peat (in situ) for the different site types. Site types: Herb-rich type (HrT), Vaccinium myrtillus types II and I (MT II and I), Vaccinium vitis-idaea types II and I (VT II and I), and Dwarf shrub type (DsT). R2c represents the proportion of variance explained by both fixed and random effects. Solid lines depict the fitted linear regression lines, with their 95 % confidence intervals indicated by the shaded areas. R2m describes the proportion of variance explained by fixed effects. W. He et al. Forest Ecology and Management 550 (2023) 121503 12 two contrasting peatland forests. Plant and Soil 412, 299–316. https://doi.org/ 10.1007/s11104-016-3070-3. Bhuiyan, R., Makiranta, P., Strakova, P., Fritze, H., Minkkinen, K., Penttila, T., Rajala, T., Tuittila, E.S., Laiho, R., 2023. Fine-root biomass production and its contribution to organic matter accumulation in sedge fens under changing climate. Sci. Total Environ. 858, 159683 https://doi.org/10.1016/j.scitotenv.2022.15968. Boggie, R., 1972. Effect of water-table height on root development of Pinus contorta on deep peat in Scotland. Oikos 23, 304–312. https://doi.org/10.2307/3543168. Bubier, J.L., Moore, T.R., Crosby, G., 2006. Fine-scale vegetation distribution in a cool temperate peatland. Botany 84, 910–923. https://doi.org/10.1139/b06-044. Chen, W.J., Zhang, Q.F., Cihlar, J., Bauhus, J., Price, D.T., 2004. Estimating fine-root biomass and production of boreal and cool temperate forests using aboveground measurements: a new approach. Plant and Soil 265, 31–46. https://doi.org/ 10.1007/s11104-005-8503-3. Clemmensen, K.E., Bahr, A., Ovaskainen, O., Dahlberg, A., Ekblad, A., Wallander, H., Stenlid, J., Finlay, R.D., Wardle, D.A., Lindahl, B.D., 2013. Roots and associated fungi drive long-term carbon sequestration in boreal forest. Science 339, 1615–1618. https://doi.org/10.1126/science.1231923. Ding, Y., Leppalammi-Kujansuu, J., Salemaa, M., Schiestl-Aalto, P., Kulmala, L., Ukonmaanaho, L., Nojd, P., Minkkinen, K., Makita, N., Zeleznik, P., Merila, P., 2021. Distinct patterns of below-and aboveground growth phenology and litter carbon inputs along a boreal site type gradient. For. Ecol. Manag. 489, 119081 https://doi. org/10.1016/j.foreco.2021.119081. Finer, L., Laine, J., 1998. Root dynamics at drained peatland sites of different fertility in southern Finland. Plant and Soil 201, 27–36. https://doi.org/10.1023/A: 1004373822354. Finer, L., Laine, J., 2000. The ingrowth bag method in measuring root production on peatland sites. Scandinavian J. Forest Res. 15, 75–80. https://doi.org/10.1080/ 02827580050160493. Finer, L., Ohashi, M., Noguchi, K., Hirano, Y., 2011. Fine root production and turnover in forest ecosystems in relation to stand and environmental characteristics. For. Ecol. Manag. 262, 2008–2023. https://doi.org/10.1016/j.foreco.2011.08.042. Gale, M.R., Grigal, D.F., 1987. Vertical root distributions of northern tree species in relation to successional status. Can. J. Forest Res. 17, 829–834. https://doi.org/ 10.1139/x87-131. Gill, R.A., Jackson, R.B., 2000. Global patterns of root turnover for terrestrial ecosystems. The New Phytologist 147, 13–31. https://doi.org/10.1046/j.1469- 8137.2000.00681.x. Gower, S.T., Krankina, O., Olson, R.J., Apps, M., Linder, S., Wang, C., 2001. Net primary production and carbon allocation patterns of boreal forest ecosystems. Ecological Applications 11, 1395–1411. https://doi.org/10.1890/1051-0761(2001)011[1395: NPPACA]2.0.CO;2. Heikurainen, L., 1955. Structure of Scots pine root systems in a pine swamp and effect of draining on the structure (in Finnish with English abstract). article id 7466 Silva Fenn. 65. https://doi.org/10.14214/aff.7466. Hendricks, J.J., Hendrick, R.L., Wilson, C.A., Mitchell, R.J., Pecot, S.D., Guo, D., 2006. Assessing the patterns and controls of fine root dynamics: an empirical test and methodological review. Journal of Ecology 94, 40–57. http://www.jstor.org/stable/ 3599608. Iversen, C.M., Childs, J., Norby, R.J., Ontl, T.A., Kolka, R.K., Brice, D.J., McFarlane, K.J., Hanson, P.J., 2018. Fine-root growth in a forested bog is seasonally dynamic, but shallowly distributed in nutrient-poor peat. Plant and Soil 424, 123–143. https:// doi.org/10.1007/s11104-017-3231-z. Jackson, R.B., Canadell, J., Ehleringer, J.R., Mooney, H.A., Sala, O.E., Schulze, E.D., 1996. Oecologia 108, 389–411. https://doi.org/10.1007/BF00333714. Jauhiainen, J., Takahashi, H., Heikkinen, J.E., Martikainen, P.J., Vasander, H., 2005. Carbon fluxes from a tropical peat swamp forest floor. Global Change Biology 11, 1788–1797. https://doi.org/10.1111/j.1365-2486.2005.001031.x. Jobbagy, E.G., Jackson, R.B., 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecological Applications 10, 423–436. https:// doi.org/10.1890/1051-0761(2000)010[0423:TVDOSO]2.0.CO;2. Korrensalo, A., Kettunen, L., Laiho, R., Alekseychik, P., Vesala, T., Mammarella, I., Tuittila, E.-S., 2018. Boreal bog plant communities along a water-table gradient differ by their standing biomass but not by their biomass production. J. Vegetation Sci. 29, 136–146. https://doi.org/10.1111/jvs.12602. Kulju, I., Niinisto, T., Peltola, A., Raty, M., Sauvula-Seppala, T., Torvelainen, J., Uotila, E., Vaahtera, E. 2023. Metsatilastollinen vuosikirja 2022 - Finnish Statistical Yearbook of Forestry 2022. Natural Resources Institute Finland. Available at http:// urn.fi/URN:ISBN:978-952-380-584-2. Kuznetsova, A., Brockhoff, P.B., Christensen, R.H., 2017. lmerTest package: tests in linear mixed effects models. J. Stat. Soft. 82, 1–26. https://doi.org/10.18637/jss.v082.i13. Laiho, R., 2006. Decomposition in peatlands: Reconciling seemingly contrasting results on the impacts of lowered water levels. Soil Biology and Biochemistry 38, 2011–2024. https://doi.org/10.1016/j.soilbio.2006.02.017. Laiho, R., Vasander, H., Penttila, T., Laine, J., 2003. Dynamics of plant-mediated organic matter and nutrient cycling following water-level drawdown in boreal peatlands. Global Biogeochemical Cycles 17, 1053. https://doi.org/10.1029/2002GB002015. Laiho, R., Bhuiyan, R., Strakova, P., Makiranta, P., Badorek, T., Penttila, T., 2014. Modified ingrowth core method plus infrared calibration models for estimating fine root production in peatlands. Plant and Soil 385, 311–327. https://doi.org/10.1007/ s11104-014-2225-3. Laine, J., Vasander, H., Laiho, R., 1995. Long-term effects of water level drawdown on the vegetation of drained pine mires in southern Finland. J. Appl. Ecol. 32, 785–802. https://doi.org/10.2307/2404818. Lampela, M., Minkkinen, K., Strakova, P., Bhuiyan, R., He, W., Makiranta, P., Ojanen, P., Penttila, T., Laiho, R., 2023. Responses of fine-root biomass and production to drying depend on wetness and site nutrient regime in boreal forested peatland. Front. For. Glob. Change 6, 1190893. https://doi.org/10.3389/ffgc.2023.1190893. Lehtonen, A., Palviainen, M., Ojanen, P., Kalliokoski, T., Nojd, P., Kukkola, M., Penttila, T., Makipaa, R., Leppalammi-Kujansuu, J., Helmisaari, H.S., 2016. Modelling fine root biomass of boreal tree stands using site and stand variables. For. Ecol. Manag. 359, 361–369. https://doi.org/10.1016/j.foreco.2015.06.023. Li, Z., Kurz, W.A., Apps, M.J., Beukema, S.J., 2003. Belowground biomass dynamics in the Carbon Budget Model of the Canadian Forest Sector: recent improvements and implications for the estimation of NPP and NEP. Can. J. Forest Res. 33, 126–136. https://doi.org/10.1139/x02-165. Lieffers, V.J., Rothwell, R.L., 1987. Rooting of peatland black spruce and tamarack in relation to depth of water table. Canad. J. Bot. 65, 817–821. https://doi.org/ 10.1139/b87-111. Makiranta, P., Laiho, R., Mehtatalo, L., Strakova, P., Sormunen, J., Minkkinen, K., Penttila, T., Fritze, H., Tuittila, E.-S., 2018. Responses of phenology and biomass production of boreal fens to climate warming under different water-table level regimes. Global Change Biol. 24 (3), 944–956. https://doi.org/10.1111/gcb.13934. Makkonen, K., Helmisaari, H.S., 1999. Assessing fine-root biomass and production in a Scots pine stand-comparison of soil core and root ingrowth core methods. Plant and Soil 210, 43–50. https://doi.org/10.1023/A:1004629212604. Malhotra, A., Brice, D.J., Childs, J., Graham, J.D., Hobbie, E.A., Vander Stel, H., et al., 2020. Peatland warming strongly increases fine-root growth. Proc. Natl. Acad. Sci. USA 117, 7627–17634. https://doi.org/10.1073/pnas.2003361117. McPartland, M.Y., Montgomery, R.A., Hanson, P.J., Phillips, J.R., Kolka, R., Palik, B., 2020. Vascular plant species response to warming and elevated carbon dioxide in a boreal peatland. Environ. Res. Lett. 15, 124066 https://doi.org/10.1088/1748- 9326/abc4fb. Murphy, M., Laiho, R., Moore, T.R., 2009. Effects of water table drawdown on root production and aboveground biomass in a boreal bog. Ecosystems 12, 1268–1282. https://doi.org/10.1007/s10021-009-9283-z. Murphy, M.T., Moore, T.R., 2010. Linking root production to aboveground plant characteristics and water table in a temperate bog. Plant and Soil 336, 219–231. https://doi.org/10.1007/s11104-010-0468-1. Nadelhoffer, K.J., 2000. The potential effects of nitrogen deposition on fine-root production in forest ecosystems. The New Phytologist 147, 131–139. https://doi. org/10.1046/j.1469-8137.2000.00677.x. Ojanen, P., Minkkinen, K., Alm, J., Penttila, T., 2010. Soil–atmosphere CO2, CH4 and N2O fluxes in boreal forestry–drained peatlands. For. Ecol. Manag. 260, 411–421. https:// doi.org/10.1016/j.foreco.2010.04.036. Ojanen, P., Minkkinen, K., Penttila, T., 2013. The current greenhouse gas impact of forestry-drained boreal peatlands. For. Ecol. Manag. 289, 201–208. https://doi.org/ 10.1016/j.foreco.2012.10.008. Ojanen, P., Lehtonen, A., Heikkinen, J., Penttila, T., Minkkinen, K., 2014. Soil CO2 balance and its uncertainty in forestry-drained peatlands in Finland. For. Ecol. Manag. 325, 60–73. https://doi.org/10.1016/j.foreco.2014.03.049. Paavilainen, E., 1966. On the effect of drainage on root systems of Scots pine on peat soils. Commun. Inst. for. Fenn. 61, 1–110. http://urn.fi/URN:NBN:fi-metla-2012071 71093. Page, S.E., Rieley, J.O., Banks, C.J., 2011. Global and regional importance of the tropical peatland carbon pool. Global Change Biology 17 (2), 798–818. https://doi.org/ 10.1111/j.1365-2486.2010.02279.x. Paivanen, J., Hånell, B., 2012. Peatland ecology and forestry–a sound approach. University of Helsinki Department of Forest Sciences Publications 3. 267 p. ISBN 978-952-10-4531-8. Pandey, R., Bargali, S.S., Bargali, K., Karki, H., Kumar, M., Sahoo, U.K., 2023. Fine root dynamics and associated nutrient flux in Sal dominated forest ecosystems of Central Himalaya. India. Front. for. Glob. Change 5, 1064502. https://doi.org/10.3389/ ffgc.2022.1064502. Peltoniemi, K., Adamczyk, S., Fritze, H., Minkkinen, K., Pennanen, T., Penttila, T., Sarjala, T., Laiho, R., 2021. Site fertility and soil water-table level affect fungal biomass production and community composition in boreal peatland forests. Environmental Microbiology 23, 5733–5749. https://doi.org/10.1111/1462- 2920.15368. Persson, H.Å., 1983. The distribution and productivity of fine roots in boreal forests. Plant and Soil 71, 87–101. https://doi.org/10.1007/BF02182644. Proctor, C., He, Y., 2019. Quantifying wetland plant vertical root distribution for estimating the Interface with the anoxic zone. Plant and Soil 440, 381–398. https:// doi.org/10.1007/s11104-019-04079-w. Rasse, D.P., 2002. Nitrogen deposition and atmospheric CO2 interactions on fine root dynamics in temperate forests: a theoretical model analysis. Global Change Biology 8 (5), 486–503. https://doi.org/10.1046/j.1365-2486.2002.00481.x. Ruseckas, J., 2000. Root abundance of pine, spruce, birch and black alder in the peat soils. Baltic Forestry 6, 10–15. https://balticforestry.lammc.lt/bf/PDF_Articles/2000 -6[2]/Root%20abundance%20of%20Pine-Spruce-Birch%20and%20Black%20Alder %20in%20the%20peat%20soils.pdf. Saarinen, T., 1996. Biomass and production of two vascular plants in a boreal mesotrophic fen. Canad. J. Bot. 74, 934–938. https://doi.org/10.1139/b96-116. Sarkkola, S., Hokka, H., Laiho, R., Paivanen, J., Penttila, T., 2005. Stand structural dynamics on drained peatlands dominated by Scots pine. For. Ecol. Manag. 206, 135–152. https://doi.org/10.1016/j.foreco.2004.10.064. Schenk, H.J., Jackson, R.B., 2002. Rooting depths, lateral root spreads and below- ground/above-ground allometries of plants in water-limited ecosystems. J. Ecol. 480–494. https://www.jstor.org/stable/3072232. Sjors, H., 1991. Phyto-and necromass above and below ground in a fen. Ecography 14, 208–218. https://doi.org/10.1111/j.1600-0587.1991.tb00654.x. W. He et al. Forest Ecology and Management 550 (2023) 121503 13 Strakova, P., Anttila, J., Spetz, P., Kitunen, V., Tapanila, T., Laiho, R., 2010. Litter quality and its response to water level drawdown in boreal peatlands at plant species and community level. Plant and Soil 335, 501–520. https://doi.org/10.1007/s11104- 010-0447-6. Strakova, P., Larmola, T., Andres, J., Ilola, N., Launiainen, P., Edwards, K., Minkkinen, K., Laiho, R., 2020. Quantification of Plant Root Species Composition in Peatlands Using FTIR Spectroscopy. Front. Plant Sci. 11, 597. https://doi.org/ 10.3389/fpls.2020.00597. Tingey, D.T., Phillips, D.L., Johnson, M.G., Rygiewicz, P.T., Beedlow, P.A., Hogsett, W.E., 2005. Estimates of Douglas-fir fine root production and mortality from minirhizotrons. For. Ecol. Manag. 204, 359–370. https://doi.org/10.1016/j. foreco.2004.09.010. Tufekcioglu, A., Raich, J.W., Isenhart, T.M., Schultz, R.C., 1998. Fine root dynamics, coarse root biomass, root distribution, and soil respiration in a multispecies riparian buffer in Central Iowa. USA. Agrofor. Syst. 44, 163–174. https://doi.org/10.1023/A: 1006221921806. UNEP 2022. Global peatlands assessment: The state of the world’s peatlands. Main Report. Global Peatlands Initiative. United Nations Environment Programme, Nairobi. 418 p. https://www.unep.org/resources/global-peatlands-assessment- 2022. Vogt, K.A., Vogt, D.J., Palmiotto, P.A., Boon, P., O’Hara, J., Asbjornsen, H., 1995. Review of root dynamics in forest ecosystems grouped by climate, climatic forest type and species. Plant and Soil 187, 159–219. https://doi.org/10.1007/BF00017088. Weltzin, J.F., Pastor, J., Harth, C., Bridgham, S.D., Updegraff, K., Chapin, C.T., 2000. Response of bog and fen plant communities to warming and water-table manipulations. Ecology 81, 3464–3478. https://doi.org/10.1890/0012-9658(2000) 081[3464:ROBAFP]2.0.CO;2. Westman, C.J., Laiho, R., 2003. Nutrient dynamics of peatland forests after water-level drawdown. Biogeochemistry 63, 269–298. https://doi.org/10.1023/A: 1023348806857. Xu, J., Morris, P.J., Liu, J., Holden, J., 2018. PEATMAP: Refining estimates of global peatland distribution based on a meta-analysis. Catena 160, 134–140. https://doi. org/10.1016/j.catena.2017.09.010. Yuan, Z.Y., Chen, H.Y.H., 2010. Fine root biomass, production, turnover rates, and nutrient contents in boreal forest ecosystems in relation to species, climate, fertility, and stand age: literature review and meta-analyses. Critical Reviews in Plant Sciences 29, 204–221. https://doi.org/10.1080/07352689.2010.483579. Yuan, Z.Y., Chen, H.Y.H., 2012. Effects of disturbance on fine root dynamics in the boreal forests of Northern Ontario, Canada. Ecosystems 16, 467–477. https://doi.org/ 10.1007/s10021-012-9623-2. W. He et al.