Forest floor chemistry under seven tree speciesalong a soil fertility gradient

Lars Vesterdal and Karsten Raulund-Rasmussen

Abstract: Forest floor chemistry, i.e., C/nutrient ratios, pH, and element contents, were determined in stands of twodeciduous species and five conifer species replicated at seven sites along a soil fertility gradient. There were consistentdifferences between forest floors of the tree species. Lodgepole pine (Pinus contortaDougl.) forest floors had highestC/nutrient ratios, lowest pH, and the greatest element contents, whereas oak (Quercus roburL.) forest floors had lowC/nutrient ratios and the lowest element contents of all species. Differences in forest floor C/nutrient ratios, pH, andelement contents between sites of low nutrient status and sites of intermediate to high nutrient status were also great.Forest floor pH was related to mineral soil pH, and C/P, C/Ca, and C/K ratios were related to mineral soil nutrientconcentrations. Forest floor C content was negatively related to most mineral soil fertility variables and was closestrelated to texture, pH, and concentrations of P and Ca. The C content of lodgepole pine and oak forest floors tended tobe less affected by the soil fertility gradient. The results suggest that C storage and immobilization of nutrients inforest floors may be managed along an extensive soil gradient by selection of the proper tree species.

Résumé: La chimie de la couverture morte, c’est-à-dire les rapports C/nutriments, le pH et le contenu en éléments, aété étudiée dans des peuplements de deux espèces décidues et cinq espèces conifériennes dans sept sites le long d’ungradient de fertilité du sol. Il y avait des différences cohérentes entre les couvertures mortes des différentes espècesd’arbres. Les couvertures mortes du pin lodgepole (Pinus contortaDougl.) avaient les plus hauts rapports C/nutriments,les plus faibles pH et les plus forts contenus en éléments tandis que les couvertures mortes du chêne (Quercusrobur L.) avaient de faibles rapports C/nutriments et les plus faibles contenus en éléments de toutes les espèces. Lesdifférences dans les couvertures mortes, pour les rapports C/nutriments, le pH et le contenu en éléments, entre les sitesà faible statut nutritif et les sites à statut nutritif intermédiaire à élevé étaient également fortes. Le pH de la couverturemorte était relié au pH du sol minéral, et les rapports C/P, C/Ca et C/K étaient reliés aux concentrations en nutrimentsdu sol minéral. Le contenu en C de la couverture morte était négativement relié à la plupart des variables de la fertilitédu sol et était le plus étroitement relié à la texture, au pH et aux concentrations de P et Ca. Le contenu en C descouvertures mortes de pin lodgepole et de chêne tendait à être moins affecté par le gradient de fertilité du sol. Lesrésultats suggèrent que le stockage de C et l’immobilisation des nutriments dans les couvertures mortes peuvent êtregérés le long d’un fort gradient de sol en sélectionnant l’espèce d’arbre appropriée.

[Traduit par la rédaction] Vesterdal and Raulund-Rasmussen 1647

The forest floor plays an important part in nutrient cy-cling, as nutrients are released during litter decompositionand rendered available to plants. Forest floor mass reflectsthe relationship between the rate of litter production and therate of litter decomposition (Olson 1963), and the forestfloor mass indicates the amount of nutrients immobilized informs unavailable to plants. A buildup of forest floor con-taining immobilized nutrients may be harmful for forest re-

generation and productivity in the long term. Müller (1879)defined two morphological types of forest floors, mull andmor, which were associated with fertile soils and nutrient-poor soils, respectively. Later studies have also suggestedthat forest floor mass, element contents, C/nutrient ratios,and pH differ according to soil type within a single tree spe-cies (Ovington 1954; Florence and Lamb 1974; Staaf 1987).Forest floors of nutrient-poor soils tend to have a greatermass and have fewer nutrients relative to carbon comparedwith forest floors of nutrient-rich soils. This may be due tothe influence of soil nutrient status on litter quality (Lukum-buzya et al. 1994; Sanger et al. 1996) and the influence ofsoil properties on the decomposer community (Schaefer andSchauermann 1990). Forest floor mass, element contents,C/nutrient ratios, and pH may also differ significantly be-tween various tree species on the same soil type (Ovington1954; France et al. 1989; Son and Gower 1992; Muys et al.1992), probably because of inherent differences in litterquality among tree species (Kiilsgaard et al. 1988; Nordén1994). Studies of tree species differences within single siteshave led to a belief that certain tree species tend to developrelatively thick, acid forest floors (mor), whereas others de-velop relatively thin and less acid forest floors (mull). The

Can. J. For. Res.28: 1636–1647 (1998) © 1998 NRC Canada

1636

Received December 12, 1997. Accepted July 29, 1998.

L. Vesterdal.1 Unit of Forestry, The Royal Veterinary andAgricultural University, Hørsholm Kongevej 11, DK-2970Hørsholm, Denmark.K. Raulund-Rasmussen.Department of Forest Ecology,Danish Forest and Landscape Research Institute, HørsholmKongevej 11, DK-2970 Hørsholm, Denmark.

1Author to whom all correspondence should be addressed.Present address: Department of Forest Ecology, DanishForest and Landscape Research Institute, HørsholmKongevej 11, DK-2970 Hørsholm, Denmark. email:lav@fsl.dk

question remains, however, to what extent such an effect oftree species on forest floors will persist along more exten-sive soil gradients. Few studies have explored the effects oftree species on forest floors along an extensive gradient insoil properties, as this requires even-aged stands of differenttree species replicated along a soil gradient, and many treespecies are not found naturally or cultivated along extensivesoil gradients. A large-scale regional study by Muys andLust (1992) indicated a strong influence of tree species onforest floor type, while the influence of soil texture wassmaller. In contrast, Raulund-Rasmussen and Vejre (1995)reported that forest floor C/nutrient ratios, pH, and elementcontents of four tree species differed considerably between asandy and a loamy soil, while the effect of tree species wasweaker. The difference among tree species appeared to besmaller at the sandy site than at the loamy site indicatingthat tree species effects may not be consistent along soilgradients.

In the present study we examined forest floor chemistry instands of seven tree species replicated at seven sites along agradient in mineral soil nutrient status. The study focused onthe following aspects of forest floor chemistry: (i) C/nutrientratios and pH, and (ii ) element contents, and the aim was toexplore the combined influence of tree species and soilproperties on forest floors. The following hypotheses wereaddressed: (i) forest floor C/nutrient ratios, pH, and elementcontents differ consistently among various tree species alongan extensive gradient in soil fertility; (ii ) C/nutrient ratiosand pH in the forest floor are related to nutrient concentra-tions and pH in the mineral soil; and (iii ) the C content offorest floors is negatively related to mineral soil fertilityvariables.

Sites and tree speciesEven-aged monoculture stands of seven tree species replicated

at seven sites were used in the investigation. The tree species werebeech (Fagus sylvaticaL.), oak (Quercus roburL.), Norway spruce(Picea abies(L.) Karst.), Sitka spruce (Picea sitchensis(Bong.)Carr.), Douglas-fir (Pseudotsuga menziesii(Mirb.) Franco), grandfir (Abies grandisLindl.), and lodgepole pine (Pinus contortaDougl.). The stands were 30 years old. The seven sites distributedthroughout Denmark were chosen from a series of tree species trialsestablished by the Danish Forest and Landscape Research Institutein 1964–1965 (Holmsgaard and Bang 1977) to obtain gradients insoil properties. The Christianssæde(54 47 11 22° ′ ° ′N E), , Holstein-borg(54 14 11 29° ′ ° ′N E), , and Frederiksborg(55 58 12 21° ′ ° ′N E), sitesare located in eastern Denmark and are the most nutrient rich, withsoils developed from loamy weichselian till (Mollic Hapludalf,Oxyaquic Hapludalf, and Typic Argiudoll, respectively (SoilSurvey Staff 1992)). At Christianssæde, the parent materialwas calcareous (28% CaCO3 in the C horizon). Among the west-ern sites, Løvenholm(56 28 10 32° ′ ° ′N E), and Tisted Nørskov(56 47 10 01° ′ ° ′N E), are intermediate with respect to nutrient statusand are both Typic Haplumbrepts. These soils are developed fromweichselian till with textures between loamy sand and sand. Thesites Lindet(55 08 8 53° ′ ° ′N E), and Ulborg(56 17 8 26° ′ ° ′N E), are themost nutrient-poor with soils (Typic Quartzipsamment and TypicHapluhumod, respectively) developed from a parent material ofsandy till deposited during the Saale glaciation. Differences inmineral soil nutrient status among sites appear from the selectedsoil properties in Table 1.

All seven trials are planted on almost level ground, and the indi-vidual tree species plots are approximately 0.25 ha. The sites arerather similar climatically, although the annual precipitation variesslightly. Mean annual temperatures are 7.5–7.7°C for Lindet,Ulborg, Tisted Nørskov, Løvenholm, and Frederiksborg and 8.4°Cfor Holsteinborg and Christianssæde. Mean annual precipitationranges from 610 to 890 mm (Holsteinborg≈ Christianssæde≈Løvenholm < Tisted Nørskov < Frederiksborg≈ Lindet < Ulborg).All climate data are from the Danish Meteorological Institute. Thegradients in temperature and precipitation do not coincide with thegradient in soil fertility, and effects of soil properties were conse-quently not regarded as confounded by effects of macroclimate. Atfour of the sites (Christianssæde, Frederiksborg, Løvenholm,Tisted Nørskov), the land was used for agriculture prior to affores-tation. At Holsteinborg and Lindet the sites were previously for-ested (beech forest and oak coppice, respectively), whereas thearea at Ulborg was former heathland. Before planting, all sites butHolsteinborg were cultivated by ploughing, and at these six sitesforest floors are known to originate from the present stands(Holmsgaard and Bang 1977). Understory shrubs were absent inthe stands, and except for a moss layer in some of the sprucestands at the nutrient-poor sites, grasses and herbs covered theground only in the oak stands (Lange 1993).

Forest floor samplingAn area-based sampling of the forest floors was carried out dur-

ing winter 1994–1995. Ten subsamples per stand were randomlycollected by using a 25 × 25 cm frame. Forest floor depth wasmeasured at the four sides of the subsample pits. Subsamples weredried at 60°C, the herbaceous litter, cones or fruits, twigs, andlarger roots were removed, and the remaining material wasweighed (±1 g). The 10 subsamples per stand were ground andpooled to one sample for chemical analysis.

Chemical analysesGround samples of forest floor material were analysed for

pH (1:10, organic matter : 0.01M CaCl2), total carbon, and nitro-gen (dry combustion; LECO CHN 1000 Analyzer). A subsamplefrom each stand was digested with concentrated nitric acid in a mi-crowave oven, and the digests were subsequently analyzed for totalphosphorus (flow injection analysis) and total calcium, potassium,and magnesium by flame atomic absorption spectroscopy (AAS).

Mineral soil samples from genetic horizons in one central soilpit at Holsteinborg (in the Sitka spruce stand), Løvenholm, andTisted Nørskov (both in the Norway spruce stand) were analysedfor pH (1:2.5, soil : 0.01M CaCl2); particle size distribution (com-bined pipette and sieve method: clay, <2µm; silt, 2–20µm; finesand, 20–200µm; coarse sand, 200–2000µm); total C (dry com-bustion and weighing of evolved carbon dioxide); total N(Kjeldahl); P (flow injection analysis following extraction with0.1 M sulfuric acid for 2 h); and exchangeable Ca, Mg, and K byAAS after extraction with 1 M NH4NO3. Soil data forChristianssæde (samples from the Norway spruce stand) are fromRaulund-Rasmussen and Vejre (1995), and soil data for Ulborg(from neighbouring Scots pine (Pinus sylvestrisL.) stand of thesame age), Lindet (from the Norway spruce stand), and Fre-deriksborg (from the beech stand) are from Raulund-Rasmussen(1993). Nutrient concentrations weighted by horizon depth werecalculated for two soil depth strata (0–50 and 50–100 cm) to quan-tify nutrient status. Mineral soil pH and texture (percent clay, silt,fine sand, and coarse sand) were represented by horizon-weightedvalues for the two soil depth strata. Mineral soil data were notfrom the same tree species, but the relatively young stands wereconsidered to have influenced soil properties little in the soil depthstrata as compared to the differences in soil properties among sites.

© 1998 NRC Canada

Vesterdal and Raulund-Rasmussen 1637

StatisticsAll statistical analyses were carried out using SAS statistical

software (SAS Institute Inc. 1993). Effects of tree species on forestfloor C/nutrient ratios, pH, bulk density, and element contents weretested by one-way ANOVA with site as the block factor. Duncan’smultiple range test was used to compare means. General linear

models (Proc GLM) were used to explore the relationships be-tween soil nutrient concentrations and forest floor C/nutrient ratiosand relationships between forest floor pH and soil pH. The classvariable tree species was included in the linear models to test ifthere were consistently different levels in C/nutrient ratios and pHamong tree species along the soil fertility gradient. Relationships

© 1998 NRC Canada

1638 Can. J. For. Res. Vol. 28, 1998

Depth(cm)

Clay(%)

Silt(%)

Finesand(%)

Coarsesand(%)

TotalC (%)

Total N(mg·g–1)

ExtractableP (mg·kg–1)

Exchangeable cations(cmol+·kg–1)

pH Ca2+ Mg2+ K+

Christianssæde (CHR), Mollic Hapludalfa

A1 0–5 3.8 12 10 49.1 28.9 2.8 2.1 110 4.6 0.6 0.12A2 5–25 5.2 12 11 50.2 26.8 1.5 1.7 110 9.3 0.4 0.10Bt 25–50 6.2 15 18 43.6 23.4 0.4 0.5 240 12.2 0.6 0.18Btg 50–73 7.5 18 20 38.5 23.5 380 21.7 0.5 0.12Ckg 73–110 7.7 9 16 52.6 22.4 0.3 0.07Frederiksborg (FRE), Typic Argiudollb

A 0–30 4.9 16.3 23.7 50.3 9.7 0.77 0.93 126 2.84 0.30 0.15Bt1 30–50 5.3 22.6 21.7 51.9 3.8 0.16 0.33 112 4.23 0.51 0.18Bt2 50–85 5.4 24.0 29.0 44.4 2.6 0.06 0.24 182 4.02 0.72 0.18Bt3 85–120 5.5 21.8 28.2 47.5 2.6 0.02 230 4.04 2.01 0.19C 120– 5.1 25.2 25.1 46.3 3.3 0.02 263 3.72 2.87 0.20Holsteinborg (HOL), Oxyaquic HapludalfA 0–20 3.7 11.0 11.0 54.3 23.7 0.96 0.96 61 1.34 0.27 0.08AE 20–55 4.5 9.5 10.0 53.1 27.4 0.55 0.71 84 5.21 0.69 0.09E 55–85 5.4 10.5 6.0 74.1 9.4 0.04 0.22 99 4.67 0.57 0.10Bt 85–120 6.1 28.0 21.0 39.5 11.5 0.14 0.37 181 15.53 0.97 0.19C 120– 6.3 18.0 15.0 48.7 18.3 0.02 221 9.80 0.71 0.16Lindet (LIN), Typic Quartzipsammentb

A 0–6 2.9 2.5 5.8 45.0 46.8 4.97 2.70 10 0.26 0.37 0.15E 6–18 3.9 3.9 6.2 46.3 43.6 0.79 0.34 6 0.02 0.02 0.01Bhs1 18–22 3.7 9.4 7.4 40.2 43.0 1.24 0.55 9 0.04 0.04 0.03Bhs2 22–60 4.2 6.5 6.5 47.1 40.0 0.96 0.47 37 0.02 0.02 0.03Bs 60–80 4.4 4.8 2.5 48.7 44.0 0.14 38 0.01 0.01 0.02C 80–95 4.5 1.3 1.3 53.5 43.9 0.04 25 ndc nd 0.022C 95– 4.4 3.0 1.2 12.7 83.1 0.04 36 0.01 0.01 0.03Løvenholm (LOV), Typic HaplumbreptA 0–27 4.1 3.5 12.0 36.8 47.7 0.94 0.70 119 0.47 0.06 0.06Bw 27–55 4.8 2.5 6.5 49.0 42.0 0.40 0.32 161 0.48 0.05 0.04BC 55–70 4.7 4.0 9.0 48.7 38.3 0.08 0.23 223 0.21 0.03 0.04Cg 70– 4.5 3.5 10.5 43.6 42.4 0.06 123 0.78 0.07 0.07Tisted Nørskov (TIS), Typic HaplumbreptAp1 0–5 3.5 5.0 9.0 54.2 31.8 3.77 1.95 110 1.67 0.24 0.12Ap2 5–21 4.1 2.5 8.5 57.7 31.3 1.68 1.00 76 0.82 0.07 0.03Bw1 21–51 4.3 4.0 13.0 45.9 37.1 0.59 0.47 66 0.23 0.02 0.03Bw2 51–88 4.2 5.4 12.3 41.4 41.0 0.10 0.11 31 0.38 0.03 0.06C 88– 4.4 1.0 0.5 47.8 50.7 0.09 35 0.08 0.01 0.04Ulborg (ULB), Typic Hapluhumodb

A 0–18 2.7 2.1 3.0 27.8 67.1 10.74 3.30 22 0.39 0.67 0.23E 18–30 3.4 0.5 2.2 8.8 88.6 0.36 0.13 6 0.03 0.01 ndBh 30–34 3.5 10.9 3.8 17.1 68.2 6.98 2.40 24 0.20 0.13 0.09Bhs 34–40 4.1 6.2 3.6 16.3 73.9 3.62 1.30 20 0.06 0.04 0.04Bs 40–60 4.4 2.9 3.0 28.2 65.9 0.12 0.10 16 0.02 nd 0.01BC 60–100 4.5 1.3 2.1 31.8 64.8 0.03 10 0.01 nd ndC 100– 4.6 0.9 2.5 23.6 72.9 0.01 11 0.01 nd nd

Note: Site abbreviations are given in parentheses. Soil classification is according to Soil Survey Staff (1992).aRaulund-Rasmussen and Vejre (1995).bRaulund-Rasmussen (1993).cnd, not detected (below detection limit).

Table 1. Soil data for the seven sites.

between mineral soil properties and forest floor C contents weresimilarly tested. Tree species means were compared using Tukey’sstudentized range test. C/Ca ratios were logarithmically trans-formed prior to statistical analysis to normalize and homogenizevariances. In case of a significant interaction effect (tree species ×soil property), slopes of the linear models for tree species werecompared by Student’st test. Intercorrelation between the selectedmineral soil properties within the two depth strata (n = 7) was ex-amined by calculating Pearson correlation coefficients (ProcCORR).

Forest floors under the seven tree speciesTree species influenced (p < 0.001) C/nutrient ratios

(g C/g nutrient), pH, and bulk density (Table 2). Lodgepolepine forest floors had the highest C/nutrient ratios of all treespecies. Among the remaining species, grand fir had thehighest forest floor C/N ratios and Sitka spruce had thehighest forest floor C/P ratios. C/Ca ratios were fairly high

© 1998 NRC Canada

Vesterdal and Raulund-Rasmussen 1639

C/N C/P C/Ca C/K C/Mg pHCaCl2

Bulkdensity(kg·m–3)

SpeciesLodgepole pine 35.2a 674a 264a 805a 753a 3.48d 79bSitka spruce 28.7bc 530b 94bc 533b 648ab 3.99c 86abNorway spruce 26.4c 462c 77cd 412cde 480cd 4.26bc 109aDouglas-fir 25.6c 452c 114b 462bc 546bc 3.98c 94abGrand fir 31.4b 434c 58de 438bcd 482cd 4.46ab 109aBeech 26.8c 465c 48e 337de 396d 4.63a 55bOak 27.5c 440c 55de 315e 398d 3.97c 36cSitesUlborg 24.8b 523b 137ab 453bc 419c 3.62b 88abLindet 24.0b 586a 161a 506ab 620a 3.60b 105aHolsteinborg 29.9a 485bc 106bc 453bc 525abc 4.17a 84abcTisted Nørskov 31.2a 453c 106bc 566a 574ab 4.23a 80abcChristianssæde 30.5a 498bc 51d 437bc 549ab 4.27a 61cLøvenholm 30.1a 434c 65cd 492abc 542abc 4.41a 71bcFrederiksborg 30.8a 478bc 86cd 394c 473bc 4.47a 79abc

Note: Tree species means or site means in a column with the same letter are not significantly different (p > 0.05) based on ANOVA and Duncan’smultiple range test.

Table 2. Average values for C/N, C/P, C/Ca, C/K, and C/Mg ratios (g C/g nutrient), pH, and bulk density in the forest floors of seventree species and seven sites.

Forestfloorvariable

Soilproperty

Linear model: Forest floor variable = tree speciesi + b(soil property)

Hypothesis 1(b = 0)a Hypothesis 2 (LP = SS = ... = O)

r2 %rsoil2 p p Tree species differencesb

pH pH0–50 0.59 32 <0.001 (+) <0.001 a (B), ab (GF,NS),abc (SS),bc (DF,O), b (LP)pH50–100 0.45 11 0.057 <0.001 a (B,GF,NS),ab (SS,DF,O),b (LP)

C/N N0–50 0.45 12 0.053 <0.001 a (LP), ab (GF,SS),b (O,B,NS,DF)C/P P0–50 0.65 11 0.007 (–) <0.001 a (LP), b (SS,B,NS,DF,O,GF)

P50–100 0.59 2 0.266 <0.001 a (LP), b (SS,B,NS,DF,O,GF)log (C/Ca) Ca0–50 0.66 17 <0.001 (–) <0.001 a (LP), b (DF), bc (SS,NS,GF,O),c (B)

Ca50–100 0.64 14 0.003 (–) <0.001 a (LP), b (DF), bc (SS,NS,GF,O),c (B)C/K K0–50 0.80 8 <0.001 (–) <0.001 a (LP), b (SS),bc (DF), bcd (GF,NS),cd (B), d (O)

K50–100 0.76 4 0.027 (–) <0.001 a (LP), b (SS),bc (DF,GF,NS),c (B,O)C/Mg Mg0–50 0.57 4 0.169 <0.001 a (LP), ab (SS),bc (DF,GF,NS),c (O,B)

Mg50–100 0.57 2 0.259 <0.001 a (LP), ab (SS),bc (DF,GF,NS),c (O,B)

Note: Hypothesis 1 is that there is no relationship between the forest floor variable and the soil property (i.e., the slopeb = 0). Hypothesis 2 is thatthere is no effect of tree species (i.e., the same level in the forest floor variable along the gradient in the soil property). Tree species differencesareshown according to a decrease in the respective forest floor variables.r2, coefficient of determination;%rsoil

2 , percentage of model variation attributed tothe soil property;b, slope;p, probability value. Tree species: LP, lodgepole pine; SS, Sitka spruce; NS, Norway spruce; DF, Douglas-fir; B, beech;GF, grand fir; O, oak.

aSigns of significant regressions are given in parentheses.bTree species in parentheses with the same letter are not significantly different (p > 0.05) based on Tukey’s studentized range test.

Table 3. Linear models explaining forest floor pH and C/nutrient ratios by variation in mineral soil properties (depth strata 0–50 and50–100 cm) and by tree species.

in forest floors of Douglas-fir and Sitka spruce, whereasthose of beech, oak, and grand fir had relatively low C/Caratios. Norway spruce forest floors had intermediate C/Caratios. C/K and C/Mg ratios varied rather similarly amongthe tree species. Forest floor C/K ratios were higher in Sitkaspruce than in Norway spruce, beech and oak, and C/Mg ra-tios were higher in Sitka spruce than in forest floors of Nor-way spruce, grand fir, beech, and oak. Forest floor pH wassignificantly higher under beech than under all other treespecies but grand fir, and lodgepole pine forest floors werethe most acidic of all tree species. Between the two ex-tremes, forest floor pH values in grand fir and Norwayspruce stands were relatively high whereas Sitka spruce,Douglas-fir, and oak forest floors had lower pH values. For-est floor bulk densities were lowest in the deciduous speciesand highest in the spruces, Douglas-fir, and grand fir.Lodgepole pine forest floors were intermediate, although thebulk density was not significantly lower than in Douglas-firand Sitka spruce.

There were no significant interactions between tree spe-cies and soil properties in the linear models, and tree species

explained a great part of the variation in forest floorC/nutrient ratios and pH (Table 3, hypothesis 2 rejected).Tree species differences were thus consistent along the gra-dient in mineral soil nutrient status, and differences were al-most similar irrespective of soil depth stratum.

Tree species strongly affected the contents of C, N, P, Ca,Mg, and K in forest floors (p < 0.001 exceptp < 0.01 forMg content) (Fig. 1). Forest floor C contents were higher inlodgepole pine than in all other tree species, and among theremaining species, forest floors of Norway spruce and Sitkaspruce had higher C contents than those of beech, grand fir,and oak. Oak had a lower forest floor C content than allother tree species but grand fir. Douglas-fir had intermediateC contents and did not differ significantly from either thespruces or beech. The contents of N and P in the forestfloors exhibited a pattern fairly similar to that of C.Lodgepole pine differed less from the other coniferous spe-cies in N, P, K, and Mg contents than in C content, andbeech forest floors had relatively high contents of K andMg. The pattern among tree species in forest floor Ca con-tent was very different from the pattern in C content. Forest

© 1998 NRC Canada

1640 Can. J. For. Res. Vol. 28, 1998

0

5

10

15

20

Mg

·h

a-1

0

5

10

15

20

0

200

400

600

800

0

200

400

600

800

kg·h

a-1

0

10

20

30

40

kg·h

a-1

0

10

20

30

40

Carbon

Nitrogen Nitrogen

Phosphorus Phosphorus

Carbon(by species) (by site)

(by species) (by species)

(by species) (by site)

O GF B DF SS NS LP

d

cdc bc

b b

a

FRE LOVCHR TIS HOL LIN ULB

b bb

b b

aa

O GF B DF SS NS LP

d

cd

bc bc

ab ab a

FRE LOVCHR TIS HOL LIN ULB

b b b b b

aa

O GF B DF SS NS LP

e

dcd bcd

ababc a

FRE LOVCHR TIS HOL LIN ULB

bb b

b b

a

a

Fig. 1. The average C, N, P, Ca, Mg, and K contents (C in Mg·ha–1; other elements in kg·ha–1) in forest floors of seven tree speciesand seven sites. Bars with the same letter are not significantly different (p > 0.05) based on ANOVA and Duncan’s multiple range test.Error bars are 1SE. Tree species abbreviations: O, oak; GF, grand fir; B, beech; DF, Douglas-fir; SS, Sitka spruce; NS, Norwayspruce; LP, lodgepole pine. Site abbreviations are given in Table 1.

floors of oak, grand fir, lodgepole pine, and Douglas-fir hadlow contents of Ca, whereas beech and Norway spruce for-est floors had large contents of Ca.

The tree species had accumulated different amounts of Calong the gradient in soil properties, and the influence oftree species was consistent for all soil properties and bothsoil depth strata in the linear models (Table 4, hypothesis 3rejected). Tree species differences exhibited the same patternas in the ANOVA (Fig. 1) although there was some indica-tion of interaction between tree species and soil properties(Table 4, hypothesis 2).

Forest floors along the soil fertility gradientSites affected pH and C/N, C/P, and C/Ca ratios very

strongly (p < 0.001), whereas C/Mg ratios, C/K ratios, andbulk density were less affected (p < 0.05) (Table 2). Ulborgand Lindet had high C/P and C/Ca ratios, whereasLøvenholm and Tisted Nørskov had low C/P ratios, andLøvenholm, Christianssæde, and Frederiksborg had lowC/Ca ratios. The site differences for C/N ratios were oppo-site, as Ulborg and Lindet had significantly lower C/N ratios

than the other sites. C/K ratios were highest at TistedNørskov and lowest at Frederiksborg, and C/Mg ratios werehighest at Lindet and lowest at Ulborg. Forest floor pH val-ues were significantly lower at Ulborg and Lindet than at theother sites. The differences in bulk density were smalleramong sites than among tree species. However, bulk densi-ties were higher at Lindet than at Løvenholm and Christians-sæde, and forest floors at Ulborg had higher bulk densitiesthan forest floors at Christianssæde.

Forest floor pH was positively related to soil pH(0–50 cm), and C/P, C/Ca, and C/K ratios in the forest floorswere negatively related to concentrations of P, Ca, and K, re-spectively, in the soil (Table 3, hypothesis 1 rejected). Thecorrelations were fairly similar for the two soil depth strataexcept for pH and C/P ratios, which were only significantlycorrelated with properties in the upper soil stratum. The neg-ative correlation between forest floor C/N ratios and soil Nwas not quite significant (p = 0.053), and C/Mg ratioswere notsignificantly correlated with soil Mg in the upperand lower soil depth strata (p = 0.169 andp = 0.259,respectively).

© 1998 NRC Canada

Vesterdal and Raulund-Rasmussen 1641

0

50

100

150

200

kg·h

a-1

0

50

100

150

200

Magnesium Magnesium

0

10

20

30

40

50

kg·h

a-1

kg·h

a-1

0

10

20

30

40

50

Calcium Calcium

0

10

20

30

40

0

10

20

30

40Potassium Potassium

(by species)

O GF B DF SS NS LP

c

bc

a

bc

ab

a

bc

(by site)

(by species)

O GF B DF SS NS LP

cbc

ab

abc

aa a

(by site)

FRE LOVCHR TIS HOL LIN ULB

bb

b

bb

b

a

FRE LOVCHR TIS HOL LIN ULB

c c c cc

b

a

(by species)

O GF B DF SS NS LP

cc

ab

bc

aba

ab

(by site)

FRE LOVCHR TIS HOL LIN ULB

b bb

b

b

a

a

Fig. 1 (concluded).

Forest floor element contents were also strongly affectedby sites (p < 0.001 exceptp < 0.01 for Ca content) (Fig. 1).Ulborg and Lindet had by far the highest forest floor C, N,P, Mg, and K contents, while Ca content was significantlyhighest only at Ulborg. Forest floor C content was related tomineral soil properties in the linear models (Table 4, hypoth-esis 1 rejected). Logarithmic transformation of some soilproperties gave the best fit. Carbon contents decreased withincreasing mineral soil pH and increasing concentrations ofP, Ca, K, and Mg, but C contents were positively related toC and N concentrations in the mineral soil. Forest floor Ccontent was only related to K and Mg concentrations in thelower soil stratum. Carbon content was also correlated withsoil texture, as the C content decreased with increasing per-cent clay, silt, and fine sand and increased with increasingpercent coarse sand. The effect of tree species interacted sig-nificantly with some of the soil properties (Table 4, hypothe-sis 2 rejected). The C contents of lodgepole pine forestfloors were not significantly related to P concentrations(Fig. 2), C concentrations, and percent silt in the upper soilstratum nor were the C contents related to percent silt or Ca,K, and Mg concentrations in the lower soil stratum. Carboncontents in forest floors of Norway spruce, Sitka spruce,grand fir, Douglas-fir, and beech were all similarly and sig-nificantly affected by these soil properties, and there weresignificant differences (p < 0.05) between the slopes in lin-ear models for these tree species and the slope in the linear

model for lodgepole pine (as in Fig. 2 for soil P concentra-tion). The C content of oak forest floors was intermediatelyaffected by soil properties in case of interaction, and C con-tent was only significantly correlated with soil P concentra-tion (0–50 cm) and percent silt (50–100 cm).

The soil concentrations of P, Ca, K50–100, Mg50–100, C, andalso pH0–50, percent silt, and percent coarse sand were prop-erties that explained a great part of the variation, togetherwith tree species. As expected, some of these soil propertieswere intercorrelated (Table 5). Within both soil strata, pHwas correlated with P and Ca, and K was correlated with Mgand percent clay. Soil pH was also correlated with texturalproperties in the upper soil stratum. Within the lower soilstratum, both K and Mg were very closely correlated withtextural properties (clay, silt, and coarse sand), and Ca and Pwere also intercorrelated. Soil C and N concentrations andpercent coarse sand tended to be negatively correlated withthe other properties.

Forest floors under the seven tree speciesThe tree species had different pH and C/nutrient ratios in

the forest floor irrespective of the gradient in mineral soilproperties. These consistent tree species differences may beattributed to a strong influence of genetically determined lit-ter quality on forest floor C/nutrient ratios and pH. The

© 1998 NRC Canada

1642 Can. J. For. Res. Vol. 28, 1998

Soilproperty

Depth(cm)

Linear model: C content = tree speciesi + bi(soil property)

Hypothesis 1(b = 0)

Hypothesis 2(bLP = bSS= ... = bO)

Hypothesis 3(LP = SS = ... = O)

r2 p p p

pH 0–50 0.68 <0.001 (–) <0.00150–100 0.52 0.016 (–) <0.001

N conc. 0–50 0.58 <0.001 (+) <0.001log (C conc.) 0–50 0.82 <0.001 (+) 0.025 <0.001log (P conc.) 0–50 0.91 <0.001 (–) <0.001 0.003

50–100 0.72 <0.001 (–) <0.001log (Ca conc.) 0–50 0.67 <0.001 (–) <0.001

50–100 0.82 0.026 (–) 0.034 <0.001log (K conc.) 0–50 0.47 0.151 <0.001

50–100 0.85 <0.001 (–) 0.006 <0.001log (Mg conc.) 0–50 0.45 0.422 <0.001

50–100 0.82 <0.001 (–) 0.020 <0.001% clay 0–50 0.55 0.004 (–) <0.001

50–100 0.59 <0.001 (–) <0.001log (% silt) 0–50 0.85 <0.001 (–) 0.033 0.012

50–100 0.89 <0.001 (–) 0.002 0.002% fine sand 0–50 0.64 <0.001 (–) <0.001

50–100 0.55 0.004 (–) <0.001% coarse sand 0–50 0.68 <0.001 (+) <0.001

50–100 0.67 <0.001 (+) <0.001

Note: Hypothesis 1 is that there is no relationship between forest floor C content and the soil property (i.e., the slopeb =0). Signs of significant regressions are given in parentheses. Hypothesis 2 is that the slopes of linear models are the same forall tree species (i.e., no interaction). Hypothesis 3 is that there is no effect of tree species (i.e., the same level in forest floorC along the gradient in the soil property). If soil properties did not interact significantly with tree species (i.e., hypothesis 2was confirmed) linear models were tested without the interaction term.r2, coefficient of determination;b, slope;p, probabilityvalue. Tree species abbreviations are as in Table 3.

Table 4. Linear models explaining forest floor carbon content by variation in mineral soil properties (depthstrata 0–50 cm and 50–100 cm) and by tree species.

forest floors under the two deciduous species had low C/nu-trient ratios, whereas the conifer species showed great varia-tion in both C/nutrient ratios and pH. Differences betweensome of the tree species in C/N and C/Ca ratios are in agree-ment with reported differences in litter N and Ca concentra-tions (Gloaguen and Touffet 1982; Kiilsgaard et al. 1988),thereby emphasizing the link between litter and forest floorquality. However, differences in forest floor nutrient statusamong tree species may not only reflect differences in litternutrient status. The initial nutrient status of litter has proba-bly been mediated to some extent by variable decompositionrates and nutrient dynamics (Ovington 1954; Rustad 1994).The forest floor C/nutrient ratios and pH values presentedsuggest that forest floors of different tree species offer vari-able conditions for decomposer organisms, and that tree spe-cies may have implications for soil development because ofdifferences in forest floor element contents and acidity(Binkley 1995).

The deciduous species had less compact forest floors thanthe conifers as indicated by the bulk densities, and amongthe conifers, lodgepole pine with its larger needles had lesscompact forest floors than Norway spruce and grand fir.Alban (1982) found no such differences in bulk densityamong tree species with different foliage litter morphology,and differences in forest floor bulk density may also reflectdifferences in forest floor morphology. For instance, mull-like forest floors usually have a looser structure than mor-like forest floors. However, it was observed that forest floor

bulk density appeared to be more connected with differencesin foliage litter morphology.

Forest floors of the seven tree species had very variableelement contents (Fig. 1), indicating that tree species mayhave a considerable effect on nutrient immobilization in for-est floors. The differences among tree species for the mostpart corroborate differences reported by others for the samespecies or genera (Ovington 1954; Nihlgård 1971; Alban1982; Wilson and Grigal 1995). The coniferous species ex-hibited a great range in forest floor C, N, P, and Ca contents,and results do not support the common conception that coni-fer stands store more elements in forest floors than decidu-ous species. For instance, grand fir stored C, N, and P inforest floors within the range of the deciduous species. Thepine and spruce species tended to have greater forest floor Ccontents than Douglas-fir and grand fir, and these findingscorrespond reasonably well with other studies in the samespecies (Ovington 1954; Gloaguen and Touffet 1980) al-though Eriksson and Rosén (1994) found no significant dif-ferences in forest floor element contents between Norwayspruce and grand fir.

The consistent tree species differences in forest floor Ccontent along the soil gradient (Table 4) suggest that differ-ences among these seven tree species apply to soils of differ-ent nutrient status. The interaction between tree species andcertain soil fertility variables (e.g., P concentration; Fig. 2)was not strong enough to change this general pattern amongthe seven tree species. However, the indication of interactionimplies that differences among other tree species along ex-tensive soil gradients could possibly be more inconsistentthan differences among the species studied. This emphasizesthe necessity of conducting forest floor studies along soilgradients to support generalizations about the influence oftree species. Our results for the most part confirmed the hy-pothesis that tree species influence forest floor chemistryconsistently along a gradient in soil fertility, and the resultsthus corroborate views on influence of tree species based onstudies conducted at single soil types or at soil types of moresimilar nutrient status (Gloaguen and Touffet 1980; Alban1982; Son and Gower 1992). This points to the conclusionthat nutrient immobilization in forest floors may be managedat both nutrient-rich and nutrient-poor soils by selection ofthe proper tree species.

Forest floors along the soil fertility gradientForest floor C/nutrient ratios, pH, and bulk density varied

among sites (Table 2). Bulk densities tended to be highest atthe nutrient-poor sites where the forest floor morphologywas mor-like, but only some of the nutrient-rich sites withmore mull-like forest floors had clearly lower bulk densities.The variation in forest floor C/nutrient ratios and pH amongsites indicated that these forest floor variables might be in-fluenced by the mineral soil nutrient status, and some of theforest floor variables were significantly related to mineralsoil properties (Table 3). The correlation between forestfloor pH and mineral soil pH tended to be weaker withincreasing soil depth. Similarly, forest floor C/P and C/K ra-tios were more closely correlated with P and K concentra-tions in the 0–50 cm than the 50–100 cm soil stratum. Theupper soil stratum constituted the main rooting zone forSitka spruce at Frederiksborg, Ulborg, and Lindet (Pedersen

© 1998 NRC Canada

Vesterdal and Raulund-Rasmussen 1643

10 100Soil P, 0-50 cm (mg · kg )-1

0

5

10

15

20

25

LP

DF

B

O

SS

NS

GF

Symbol

Cco

nte

nt(M

g·h

a)

-1

SpeciesSlope

differenceNorway spruce (NS)Beech (B)Grand fir (GF)Sitka spruce (SS)Douglas-fir (DF)Oak (O)Lodgepole pine (LP)

aaabababbcc

Fig. 2. Linear relationship between soil P concentration(0–50 cm) and forest floor carbon content for the seven treespecies. Lines with the same lowercase letter do not havesignificantly different slopes (p > 0.05) based on a general linearmodel (Table 4) and Student’st test.

1993) and probably influenced litter nutrient status and sub-sequently forest floor nutrient status the most. The signifi-cant correlations between mineral soil Ca, K, and Pconcentrations and forest floor C/Ca, C/K, and C/P ratios, re-spectively, may be attributed to increased litter Ca, K, and Pconcentrations with increasing availability of these nutrientsin the soil (Perala and Alban 1982; Boerner 1984). Forestfloor C/N ratios were apparently not solely determined bymineral soil N concentrations, and N dynamics during de-composition (Rustad 1994) may also be influencing C/N ra-tios.

The differences among sites in forest floor element con-tents were marked (Fig. 1), and the two sites with the mostnutrient-poor soils usually had the largest element contents.Forest floor C contents decreased with increasing mineral

soil fertility as indicated by the negative correlations withpH, exchangeable bases, and percent clay and silt (Table 4).This association between mineral soil fertility variables andforest floor C contents was also indicated in other studies(Staaf 1987; Vesterdal et al. 1995). An exception in thepresent study was the positive correlation between forestfloor C content and soil N concentration. This positive cor-relation was probably due to the fact that soil N concentra-tions were highest at the two sites that had the largest Ccontents and yet were least fertile in other respects. It maybe concluded, however, that high soil N concentrations werenot associated with low forest floor C contents. The lower Ccontent at the more fertile sites might partly be a result oftranslocation of organic matter from the forest floor to themineral soil (e.g., as a result of faunal activity). However,

© 1998 NRC Canada

1644 Can. J. For. Res. Vol. 28, 1998

C N P Ca K Mg pH Clay SiltFinesand

0–50 cmN 0.87

0.010P –0.62 –0.31

0.141 0.504Ca –0.31 0.18 0.72

0.492 0.704 0.068K –0.03 0.30 0.48 0.69

0.944 0.509 0.270 0.088Mg –0.10 0.33 0.38 0.79 0.78

0.829 0.474 0.405 0.035 0.038pH –0.63 –0.26 0.91 0.84 0.65 0.51

0.126 0.576 0.004 0.019 0.112 0.244Clay –0.46 –0.13 0.52 0.68 0.86 0.72 0.76

0.299 0.777 0.236 0.101 0.013 0.067 0.047Silt –0.66 –0.42 0.66 0.49 0.67 0.41 0.80 0.87

0.108 0.345 0.107 0.263 0.098 0.360 0.030 0.011Fine sand –0.93 –0.76 0.43 0.32 0.02 0.19 0.54 0.49 0.63

0.002 0.047 0.336 0.486 0.966 0.685 0.213 0.266 0.127Coarse sand 0.84 0.58 –0.59 –0.52 –0.48 –0.44 –0.76 –0.83 –0.91 –0.88

0.018 0.174 0.164 0.229 0.279 0.318 0.045 0.021 0.005 0.00950–100 cmCa 0.89

0.007K 0.55 0.39

0.205 0.389Mg 0.46 0.35 0.96

0.294 0.435 0.001pH 0.93 0.98 0.47 0.45

0.002 <0.001 0.285 0.307Clay 0.57 0.48 0.97 0.99 0.57

0.180 0.279 <0.001 <0.001 0.186Silt 0.67 0.44 0.92 0.86 0.55 0.89

0.098 0.319 0.004 0.013 0.204 0.007Fine sand 0.29 0.34 0.58 0.50 0.31 0.52 0.27

0.523 0.450 0.169 0.257 0.498 0.231 0.557Coarse sand –0.61 –0.50 –0.98 –0.93 –0.56 –0.96 –0.86 –0.71

0.144 0.252 <0.001 0.003 0.187 <0.001 0.013 0.073

Note: Values in boldface are significant (p < 0.05).

Table 5. Correlation matrices with Pearson correlation coefficients, and withp values given in italic (n = 7) for mineralsoil nutrient concentrations, pH, and texture in the depth strata 0–50 cm and 50–100 cm.

the positive correlation between mineral soil C concentrationand forest floor C content indicated that low C contents atsome sites were not solely explained by incorporation of Cin the mineral soil.

Some of the soil fertility variables in this study werehighly intercorrelated, but soil texture, Ca and P concentra-tions, and pH (0–50 cm) appeared to be important variablesin relation to forest floor C content. These soil propertiesmay have important nutritional and environmental effects onthe diversity and activity of decomposer organisms, which inturn affect C accumulation (Schaefer and Schauermann1990; Raubuch and Beese 1995). The forest floor variablespH and C/P ratio were correlated with soil pH and soil P, re-spectively, suggesting that forest floor quality might also bemore favourable for free-living decomposer organisms innutrient-rich soils. Concentrations of K and Mg in the uppersoil stratum were not related to forest floor C contents, andthe relationships between forest floor C content and concen-trations of K and Mg in the lower soil stratum might be dueto strong intercorrelation with percent clay and silt (Table 5).It may be questioned to what extent soil properties in the50–100 cm depth stratum are of ecological significance toforest floor C accumulation. Obviously, this soil stratum hasa limited direct influence on forest floors if both roots anddecomposer organisms are mainly present in the upper soilstratum. The lower soil stratum is more representative of thesoil parent material than the upper stratum, and the correla-tions suggest an indirect relationship between parent mate-rial characteristics and forest floor chemistry. It must bestressed that, although the linear models explained a signifi-cant part of the variation encountered in the forest floors,they were not able to explain the actual processes involvedor to identify single soil variables as the main explanatoryvariables. However, the linear models do indicate that forestfloor C contents were, for the most part, negatively relatedto mineral soil nutrient status as initially hypothesized.

Forest floors of lodgepole pine and oak tended to be lessaffected by soil P (Fig. 2). This was the general pattern oflodgepole pine and oak forest floor C contents when mineralsoil variables interacted significantly with tree species(Table 4, hypothesis (2)). The fairly constant forest floor Ccontent in lodgepole pine stands suggests that soil-inducedchanges in litter quality or differences in the decomposercommunities along the soil gradient were unable to mediatethe influence of inherent litter quality on C storage in thistree species. Oak forest floors had the lowest C contentsalong the gradient, and lack of correlation with some of thesoil fertility variables indicates that C contents tended to in-crease less with decreasing nutrient status than in forestfloors of beech, Douglas-fir, grand fir, and the spruces. Theinherent properties influencing C contents in lodgepole pineand oak appear to be fairly strong. Thus, forest floor C con-tent in some tree species may vary less along a gradient insoil fertility than C content in other tree species.

Large differences in element contents were found alongthe soil fertility gradient, although stands were only 30 yearsold. However, some forest floors in the present study maynot be in steady state yet (i.e., annual decomposition in theforest floor did not equal annual litter production) but in anaccretion phase. Decomposition rate constants for beech atFrederiksborg and Ulborg based on the present forest floor C

contents and litterfall C contents (J. Bille-Hansen, personalcommunication) indicated that the time required to achieve95% of steady state would be approximately 6 and 50 years,respectively (Olson 1963). Therefore, steady-state conditionsmay not have been attained after 30 years at sites (or in treespecies) with the slowest decomposition rates. Even greaterdifferences in forest floor C content along the soil fertilitygradient and among tree species may therefore be expectedto develop in the future.

The results supported the hypothesis that forest floors ofdifferent tree species exhibit consistent differences in C/nu-trient ratios, pH, and element contents along a gradient insoil nutrient status. Lodgepole pine forest floors had thehighest C/nutrient ratios, the lowest pH, and the greatest ele-ment contents, whereas oak forest floors had low C/nutrientratios and the lowest element contents of all species. Differ-ences in forest floor C/nutrient ratios, pH, and elementcontents between sites of low nutrient status and sites of in-termediate to high nutrient status were also great. In agree-ment with our hypothesis, forest floor pH was related tomineral soil pH, and C/P, C/Ca, and C/K ratios were relatedto mineral soil nutrient concentrations. Furthermore, forestfloor C content decreased with increasing mineral soil nutri-ent status and was closest related to properties such as tex-ture, pH, and concentrations of P and Ca. While C/nutrientratios and pH in forest floors of the seven tree species werenot differently related to soil properties, the C contenttended to be less affected by soil fertility variables in lodge-pole pine and oak. The results suggest that C storage andimmobilization of nutrients in forest floors may be managedalong an extensive soil gradient by selection of the propertree species.

The study was financially supported by The EuropeanCommission, grant No. 93.60.DK.004.0. We would like tothank Lena B. Troelsen for carrying out most of the analysesand Peter Theis Hansen, Mikkel Eriksen, Mads Jensen, andEgon Madsen for assistance with the field work. Thanks arealso extended to Cindy Prescott for helpful comments on aan earlier version of the manuscript.

Alban, D.H. 1982. Effects of nutrient accumulation by aspen,spruce, and pine on soil properties. Soil Sci. Soc. Am. J.46:853–861.

Binkley, D. 1995. The influence of tree species on forest soils: pro-cesses and patterns.In Proceedings of the Trees and Soil Work-shop, 28 Feb. – 2 Mar. 1994, Canterbury, N.Z.Edited byD.J.Mead and I.S. Cornforth. Agron. Soc. N.Z. Spec. Publ. No. 10.Lincoln University Press, Canterbury, N.Z. pp. 1–33.

Boerner, R.E.J. 1984. Foliar nutrient dynamics and nutrient use ef-ficiency of four deciduous tree species in relation to site fertil-ity. J. Appl. Ecol.21: 1029–1040.

Eriksson, H.M., and Rosén, K. 1994. Nutrient distribution in aSwedish tree species experiment. Plant Soil,164: 51–59.

© 1998 NRC Canada

Vesterdal and Raulund-Rasmussen 1645

Florence, R.G., and Lamb, D. 1974. Influence of stand and site onradiata pine litter in South Australia. N.Z. J. For. Sci.4: 502–510.

France, E.A., Binkley, D., and Valentine, D. 1989. Soil chemistrychanges after 27 years under four tree species in southern On-tario. Can. J. For. Res.19: 1648–1650.

Gloaguen, J.C., and Touffet, J. 1980. Vitesse de décomposition etévolution minérale des litières sous climat atlantique. I. Le hêtreet quelques conifères. Acta Oecol.1: 3–26.

Gloaguen, J.C., and Touffet, J. 1982. Évolution du rapport C/Ndans les feuilles et au cours de la décomposition des litièressous climat atlantique. Le hêtre et quelques conifères. Ann. Sci.For. 39: 219–230.

Holmsgaard, E., and Bang, C. 1977. Et træartsforsøg med nåletræer,bøg og eg. De første 10 år. Forstl. Forsøgsvaes. Dan. No. 35.pp. 159–196.

Kiilsgaard, C.W., Greene, S.E., and Stafford, S.G. 1988. Nutrientconcentrations in litterfall from some western conifers with spe-cial reference to calcium. Plant Soil,102: 223–227.

Lange, M. 1993. Macromycetes under twelve tree species in tenplantations on various soil types in Denmark. Opera Bot.120:1–53.

Lukumbuzya, T.K., Fyles, J.W., and Côté, B. 1994. Effects ofbase-cation fertilization on litter decomposition in a sugar mapleforest in southern Quebec. Can. J. For. Res.24: 447–452.

Müller, P.E. 1879. Studier over Skovjord som Bidrag tilSkovdyrkningens Theori. I. Om Bøgemuld og Bøgemor paaSand og Ler. Tidsskr. Skovbrug,3: 1–125.

Muys, B., and Lust, N. 1992. Inventory of the earthworm commu-nities and the state of litter decomposition in the forests of Flan-ders, Belgium, and its implications for forest management. SoilBiol. Biochem.24: 1677–1681.

Muys, B., Lust, N., and Granval, P. 1992. Effects of grasslandafforestation with different tree species on earthworm communi-ties, litter decomposition and nutrient status. Soil Biol.Biochem.24: 1459–1466.

Nihlgård, B. 1971. Pedological influence of spruce planted on formerbeech forest soils in Scania, south Sweden. Oikos,22: 302–314.

Nordén, U. 1994. Leaf litterfall concentrations and fluxes of ele-ments in deciduous tree species. Scand. J. For. Res.9: 9–16.

Olson, J.S. 1963. Energy storage and the balance of producers anddecomposers in ecological systems. Ecology,44: 322–331.

Ovington, J.D. 1954. Studies of the development of woodland con-ditions under different trees. II. The forest floor. J. Ecol.42:71–80.

Pedersen, L.B. 1993. Nutrient cycling in Sitka spruce, Norwayspruce and beech stands in Denmark. [In Danish with Englishabstract and summary.] Forskningsser. nr.1-1993. Danish Forestand Landscape Research Institute, Lyngby.

Perala, D.A., and Alban, D.H. 1982. Biomass, nutrient distributionand litterfall in Populus, Pinus and Picea stands on two differ-ent soils in Minnesota. Plant Soil,64: 177–192.

Raubuch, M., and Beese, F. 1995. Pattern of microbial indicators inforest soils along an European transect. Biol. Fertil. Soils,19:362–368.

Raulund-Rasmussen, K. 1993. Pedological and mineralogical char-acterization of five Danish forest soils. For. Landscape Res.1:9–33.

Raulund-Rasmussen, K., and Vejre, H. 1995. Effect of tree speciesand soil properties on nutrient-immobilization in the forest floor.Plant Soil,168–169: 345–352.

Rustad, L.E. 1994. Element dynamics along a decay continuum ina red spruce ecosystem in Maine, USA. Ecology,75: 867–879.

Sanger, L.J., Cox, P., Splatt, P., Whelan, M.J., and Anderson, J.M.1996. Variability in the quality ofPinus sylvestrisneedles andlitter from sites with different soil characteristics: lignin andphenylpropanoid signature. Soil Biol. Biochem.28: 829–835.

SAS Institute Inc. 1993. SAS/STAT user’s guide, release 6.07 edi-tion. SAS Institute Inc., Cary, N.C.

Schaefer, M., and Schauermann, J. 1990. The soil fauna of beechforests: comparison between a mull and a moder soil.Pedobiologia,34: 299–314.

Soil Survey Staff. 1992. Keys to soil taxonomy. 5th ed. SMSSTech. Monogr. No. 19. Pocahontas Press, Blacksburg, Va.

Son, Y., and Gower, S.T. 1992. Nitrogen and phosphorus distribu-tion for five plantation species in southwestern Wisconsin. For.Ecol. Manage.53: 175–193.

Staaf, H. 1987. Foliage litter turnover and earthworm populationsin three beech forests of contrasting soil and vegetation types.Oecologia,72: 58–64.

Vesterdal, L., Dalsgaard, M., Felby, C., Raulund-Rasmussen, K.,and Jørgensen, B.B. 1995. Effects of thinning and soil propertieson accumulation of carbon, nitrogen and phosphorus in the for-est floor of Norway spruce stands. For. Ecol. Manage.77: 1–10.

Wilson, D.M., and Grigal, D.F. 1995. Effects of pine plantationsand adjacent deciduous forests on soil calcium. Soil Sci. Soc.Am. J. 59: 1755–1761.

© 1998 NRC Canada

1646 Can. J. For. Res. Vol. 28, 1998

Depth(cm)

C(Mg·ha–1)

Nutrient content (kg·ha–1)

Site Species N P Ca K Mg pH

CHR GF 0.6 2.85 72 6.5 60 7.1 5.3 4.88B 3.2 2.57 93 5.4 108 7.1 6.2 5.02LP 3.8 16.11 431 23.8 169 24.1 22.7 3.58DF 2.1 4.02 155 8.6 68 11.3 8.7 4.00O 2.4 0.72 26 1.6 27 2.5 1.8 4.13NS 1.9 6.33 249 15.5 162 19.4 14.9 4.42SS 3.5 12.76 422 22.4 206 19.3 14.3 3.87

FRE GF 0.6 1.88 48 4.2 60 4.9 4.1 5.00B 3.0 2.92 101 6.3 101 9.9 9.4 5.05LP 3.9 16.67 434 22.2 47 20.3 21.0 3.46DF 0.8 2.34 85 7.5 56 8.7 5.5 4.39

Table 1A. Forest floor depth, element contents, and pH for the 49 stands. See Table 1 for site abbreviations and Table 3 for treespecies abbreviations.

© 1998 NRC Canada

Vesterdal and Raulund-Rasmussen 1647

Depth(cm)

C(Mg·ha–1)

Nutrient content (kg·ha–1)

Site Species N P Ca K Mg pH

O 1.9 1.51 55 3.2 45 5.9 4.9 4.27NS 1.0 4.54 166 12.0 107 18.5 13.7 5.02SS 1.7 7.21 263 13.9 117 14.9 10.5 4.10

HOL GF 0.7 3.26 102 7.6 88 9.8 8.4 4.62B 1.3 3.43 135 7.1 50 9.8 6.4 4.62LP 4.3 20.88 595 34.4 66 27.2 28.8 3.51DF 2.4 7.46 288 18.7 53 17.6 14.5 3.89O 1.1 2.25 72 5.0 48 5.8 6.1 4.20NS 2.6 11.09 394 22.6 156 23.8 22.1 4.27SS 2.2 7.44 238 14.0 121 16.7 11.6 4.08

LIN GF 3.2 12.68 585 23.8 118 22.9 25.6 3.46B 4.1 15.20 672 25.2 148 47.1 37.6 3.73LP 5.7 18.25 517 25.0 90 24.6 25.8 3.36DF 1.9 11.43 540 19.8 55 22.9 15.4 3.54O 2.3 8.19 401 15.9 61 21.0 13.2 4.04NS 4.1 20.79 887 35.3 102 46.1 31.2 3.34SS 3.4 16.56 716 30.0 99 28.4 23.5 3.71

LOV GF 0.8 2.38 72 6.5 76 4.6 4.1 4.92B 1.9 3.50 131 11.4 166 17.9 12.1 5.28LP 4.0 14.90 446 22.7 103 17.4 17.9 3.57DF 1.9 5.85 214 15.2 61 10.7 9.3 4.41O 1.5 1.08 37 2.4 26 3.7 2.5 3.36NS 1.4 5.97 230 14.4 124 12.7 17.1 4.85SS 2.2 6.68 194 14.7 88 11.8 9.7 4.50

TIS GF 0.6 2.69 100 8.3 60 7.6 6.6 4.75B 2.5 3.86 115 8.4 128 8.1 8.8 4.89LP 4.8 18.19 504 26.8 60 16.9 20.0 3.51DF 2.5 9.48 342 21.0 52 15.2 15.8 3.93O 1.1 0.73 20 1.8 17 2.1 1.9 3.98NS 2.3 9.14 311 22.8 213 18.8 14.7 4.38SS 2.9 7.43 257 16.3 81 12.4 11.4 4.15

ULB GF 4.2 14.95 540 30.9 142 28.4 29.4 3.61B 7.9 19.26 866 41.6 316 53.8 50.2 3.82LP 6.5 16.16 533 26.4 37 23.1 27.3 3.36DF 4.3 15.92 685 28.0 214 31.2 35.6 3.70O 3.5 6.10 301 16.9 107 25.1 23.3 3.79NS 5.2 19.78 797 35.8 222 45.0 42.9 3.53SS 4.7 20.69 818 33.3 152 52.9 75.4 3.52

Table 1A (concluded).