Received 22 June 1999; received in revised form 17 December 1999; accepted 18 February 2000

Transcription

1 Soil Biology & Biochemistry 32 (2000) 1227± Changes to mineral N cycling and microbial communities in black spruce humus after additions of (NH 4 ) 2 SO 4 and condensed tannins extracted from Kalmia angustifolia and balsam r R.L. Bradley a, *, B.D. Titus b, C.P. Preston b a DeÂpartement de Biologie, Faculte des Sciences, Universite de Sherbrooke, Sherbrooke, Que., Canada J1K 2R1 b Paci c Forestry Centre, 506 West Burnside Road, Victoria, BC, Canada V8Z 1M5 Received 22 June 1999; received in revised form 17 December 1999; accepted 18 February 2000 Abstract Mechanisms responsible for conifer growth ``check'' on cutovers invaded by Kalmia angustifolia L. in central Newfoundland were studied by examining e ects of added Kalmia and balsam r (Abies balsamea (L.) Mill) condensed tannins on black spruce humus N dynamics and microbial community development over 10 weeks using microcosms. Because of the silvicultural implications, interactions of tannins with fertiliser N, applied as (NH 4 ) 2 SO 4, were also studied. Both tannin types signi cantly reduced ±N leaching, whereas only Kalmia tannins reduced NO 3 ±N leaching, and then only from non-fertilised humus. Tannins did not signi cantly a ect mineral N leaching from fertilised humus. Fertiliser N increased gross N mineralisation rates such that the increase in actively cycling N was many times greater than the increase in N leaching due to fertiliser N addition. Gross N mineralisation rates were higher in fertilised humus amended with tannins, suggesting possible toxicity of tannins on microbes at high N concentrations. Recovery of added tannins in leachate and in post-treatment humus samples was low. Net anaerobic N mineralisation decreased with tannin additions but increased with fertiliser N additions. There were few signi cant treatment e ects on microbial properties derived from humus respirometry. Microbial biomass and basal respiration rates of all treatments declined by 30 and 37% respectively, indicating a general loss of available C during the experiment. The ratio of C mic -to-n mineralised as well as the nutrient de ciency index was lowest in humus amended with Kalmia tannins, suggesting higher microbial N de ciency in this treatment. Utilisation rates of various C sources by microbial communities showed distinctive patterns between pre-treatment and post-treatment humus samples, but did not reveal distinctive patterns among di erent treatments. Overall, results suggested that (1) condensed tannins decreased mineral N cycling abiotically by binding to and sequestering organic N sources, (2) fertiliser N counteracted negative e ects of condensed tannins on humus N cycling, (3) microbial communities were N limited, which prevented abundant leaching of fertiliser N while maintaining fertiliser N in an active pool, and (4) the physiology and functional diversity of soil heterotrophic communities were controlled by C availability but were una ected by tannin or fertiliser N additions. Further work is needed to determine the ecological importance of Kalmia tannins, relative to tannins produced by other plants, in reducing humus N availability on spruce cutovers Elsevier Science Ltd. All rights reserved. Keywords: Kalmia angustifolia; Condensed tannins; Humus N cycling; Microbial community; Microbial biomass; 15 N isotope dilution; Respirometry; Biolog microplate 1. Introduction * Corresponding author. Tel.: ; fax: address: robert.bradley@courrier.usherb.ca (R.L. Bradley). Since the 1960s, the role of the ericaceous shrub Kalmia angustifolia L. (hereafter referred to as Kalmia ) in reducing growth of regenerating black spruce (Picea mariana (Mill.) B.S.P.) seedlings following stand dis /00/$ - see front matter Elsevier Science Ltd. All rights reserved. PII: S (00)

2 1228 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227±1240 turbance in central Newfoundland, Canada, has been investigated (e.g., Peterson, 1965). A number of mechanisms, other than competitive uptake of primary resources such as water and nutrients, may be responsible for Kalmia-induced growth ``check'' of spruce seedlings. Firstly, Kalmia may suppress primary root growth of young black spruce seedlings by producing phytotoxic compounds (Peterson, 1965; Zhu and Mallik, 1994). Leaf and litter extracts of other ericaceous and related forest plants have also been shown to have inhibitory e ects on germination and early development of conifer seedlings in other parts of the world (e.g., Calluna vulgaris (L.) Hull (Read, 1984), Empetrum hermaphroditum Hagerup (Nilsson et al., 1993) and Vaccinium myrtillus L. (Gallet, 1994)). Secondly, Kalmia may cause spruce growth check abiotically through in uences on humus nutrient cycling by production of tannins (Kuiters, 1990). Tannins are phenolic compounds with the ability to form stable cross-links with proteins and other compounds. Many forest plants produce foliar tannins that are exported in litterfall to the forest oor where they may in uence litter decomposition rates, humus formation, N cycling and ultimately plant nutrition (Handley, 1954). Tannin-induced N de ciency has been suggested, although never proven, as a cause of conifer seedling growth check on sites dominated by Kalmia (Bradley et al., 1997a). Thirdly, Kalmia may produce substances that inhibit microbial or mycorrhizal communities responsible for soil nutrient cycling and plant uptake (Bradley et al., 1997c; Yamasaki et al., 1998). However, the evidence for this is contradictory and mechanisms by which plant secondary metabolites control microbial dynamics are unclear. For example, Boufalis and Pellissier (1994) applied a ``phytotoxic'' phenolic mixture to soil and found that O 2 consumption by free-living ectomycorrhizal fungi either increased or decreased, depending on the concentration of the mixture. Shafer and Blum (1991) found that phenolic acids, which suppressed cucumber seedling growth, were readily metabolised by soil microorganisms, sometimes without detectable changes in microbial community structures. Schimel et al. (1996) separated secondary metabolites from Populus balsamifera L. into two fractions, low molecular weight phenolics and tannins, and found that the former stimulated soil respiration whereas the latter inhibited it. Kalmia-induced growth check of conifer seedlings is site-speci c and appears to be inversely related to site fertility (Bradley et al., 1997c). Damman (1971) reported that Kalmia humus did not mineralise detectable amounts (i.e., <1 mg 100 g 1 dry wt.) of N after incubation at 218C for 100 days. Field trials have shown that N fertilisation reduced the competitive ability of Kalmia in a jack pine (Pinus banksiana Lamb.) forest (Prescott et al., 1995). Titus et al. (1993) have shown that Kalmia leaf tannin concentrations decreased with N fertilisation. However, the relative importance of fertiliser N and Kalmia tannins in controlling humus N availability and microbial dynamics is largely unknown. Our aims were to examine the isolated e ects of Kalmia tannins on N dynamics and microbial community development. Condensed tannins from leaves of Kalmia were isolated, puri ed and added to humus from a black spruce cutover that had not yet been invaded by Kalmia. The e ects of Kalmia tannins were compared with those isolated from twigs and foliage of balsam r (Abies balsamia (L.) Mill), a conifer commonly found in association with black spruce in Newfoundland. Interactions between Kalmia or r tannins and fertiliser N were also tested by addition of fertiliser N applied as (NH 4 ) 2 SO Materials and methods 2.1. Forest oor material A large (ca. 20 kg fresh weight) bulked sample of humus was collected in the fall of 1996 from a 5-year old black spruce cutover near Middleton Lake (498 03' N, ' W) in central Newfoundland where Kalmiainduced growth check of black spruce seedlings is common. Humus was taken from the area on a cutover where Kalmia was not yet established, but was expected to invade based on rapid vegetative spread of the shrub from surrounding areas. The humus was sieved (6 mm) to remove roots and coarse debris and stored at 28C until used. Initial chemical characteristics of humus subsamples are presented in Table Preparation of Kalmia and balsam r tannins Puri ed condensed tannins were prepared from Table 1 Selected chemical properties of black spruce humus collected in central Newfoundland from a recent cutover with no Kalmia angustifolia present ph (humus:water, 1:10) 3.8 KCl extractable ±N (mg g 1 ) 4.6 KCl extractable NO3 ±N (mg g 1 ) 0.0 Bray extractable P (mg g 1 ) 20.1 Total N (mg g 1 ) 9.0 Total P (mg g 1 ) 0.6 Total K (mg g 1 ) 0.6 Total Ca (mg g 1 ) 2.4 Total Mg (mg g 1 ) 1.0 Gravimetric moisture content (%) 220

3 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227± green foliage of Kalmia and from tips (ca. 5 cm in length) of twigs plus needles of balsam r collected in central Newfoundland, as outlined in Preston et al. (1997). General structural information on the tannins was obtained from solution 13 C NMR spectroscopy (Czochanska et al., 1980; Ayres et al., 1997). The Kalmia tannin was almost all procyanidin units, with over 90% cis stereochemistry and a very short average chain length of 2±3 units. The balsam r tannin was almost completely prodelphinidin units, with 75% cis stereochemistry and an average chain length of 3±4 units. The smaller chain length for Kalmia was consistent with its small response to the proanthocyanidin assay described below, which does not detect the chain-terminating units Experimental units, tannin and fertiliser N additions, and periodic leaching of humus Experimental units consisted of 30 1-l plastic Buchner funnels (Bel-Art Products, Pequanoc,) equipped with fritted (`Fritware') plastic lter plates. Humus (72 g dry wt. equivalent) was gently tamped into each funnel, the tops were sealed with polyethylene lm to prevent desiccation and the funnels were incubated in the dark at 258C for 14 wk. At the same time as humus was added to the funnels, ve subsamples (ca. 500 g fresh wt.) of bulked humus were set aside in covered containers for determination of pre-treatment microbial community characteristics (described below). Funnels were leached every 2 weeks for 14 weeks with two consecutive 200 ml aliquots of 10 mm CaCl 2 solution (Stanford and Smith, 1972). After addition of each aliquot, humus was left for 5 min to ensure ionic equilibrium, and then suction ltered at 60 kpa. Humus in each funnel was then re-soaked for 5 min with 200 ml of N-free mineral nutrient solution (Stanford and Smith, 1972) to displace excess CaCl 2 and restore favourable microbial growth conditions. Excess nutrient solution was again removed after 5 min by suction ltration. The 14-wk incubation began with a 4-wk conditioning period to allow microbial communities to stabilise, followed by a 10-wk experimental period (referred to as T = 0±10 wk). Treatments were applied to humus ve times during the experiment, on the day following bi-weekly CaCl 2 leaching events (i.e., at T =0,2,4,6 and 8 wk). The six treatments were: (1) control (50 ml H 2 O), (2) Kalmia tannins (431 mg in 50 ml H 2 O), (3) balsam r tannins (431 mg in 50 ml H 2 O), (4) fertiliser N (678.5 mg (NH 4 ) 2 SO 4 in 50 ml H 2 O), (5) Kalmia tannins + fertiliser N (431 mg tannins mg (NH 4 ) 2 SO 4 in 50 ml H 2 O), and (6) balsam r tannins + fertiliser N (431 mg tannins mg (NH 4 ) 2 SO 4 in 50 ml H 2 O). The six treatment solutions were each applied to ve replicate funnels by making numerous small injections in the top half of the humus with a large-tipped Eppendorf pipette. The cumulative treatment applications were equivalent to a total addition of 3% tannins and/or 1% mineral N (wt/wt) to humus in each funnel, except in the control treatment. Howard and Howard (1993) found that the initial N content of various tannin±protein complexes was approximately 10%, of which a highly variable amount (6±85%) could be re-mineralised during a subsequent 1- to 2- week incubation. Based on these stoichiometric relationships, ±N was added to humus at a rate equivalent to 33% of added tannin weight to ensure that e ective N binding would likely be compensated by ±N additions in ammonium sulfate fertiliser Chemical analyses of leachates and humus Subsamples of leachate from each funnel were retained at T = 2, 4, 6, 8 and 10 wk. An aliquot was analysed colorimetrically for ±N and NO 3 ±N concentrations using a Technicon auto-analyser, while another aliquot was analysed for total dissolved N (TDN) concentration after oxidation with persulfate (D'Elia et al., 1976). Dissolved organic N (DON) concentration was then calculated as the di erence between TDN and DIN (i.e., DIN ˆ ±N NO 3 ±N concentrations. Condensed tannins in leachates (at T = 2,4,6,8 and 10 weeks) and humus (at T = 10 wk) were analysed colorimetrically after hydrolysis with butanol/ HCl using the proanthocyanidin assay (Preston, 1999; Lorenz et al., in press) Post-treatment humus N dynamics Two days following the nal leaching date (i.e., T = 10 wk, hereafter referred to as post-treatment ), a subsample of humus (5±8 g) was removed from each funnel to determine moisture content, and a second subsample (3±4 g fresh wt.) was analysed for mineralisable N by anaerobic incubation (Waring and Bremner, 1964). Post-treatment gross ±N transformation rates were measured by isotope dilution (Hart et al., 1995). Four humus subsamples (ca. 8 g dry wt.) from each funnel were placed in 500 ml Mason jars and 3 ml of aqueous 15 NH 4 2 SO 4 solution (282.9 mg l 1 at 99 atom% 15 N was uniformly distributed through the humus in each jar by making numerous small injections with a hypodermic needle. At 15 min after injection, duplicate samples were extracted in 200 ml of 1 N KCl, shaken and ltered (Whatman no. 5 lter paper). The remaining jars were sealed and kept at 258C for 24 h and extracted in the same manner. All extracts were analysed for ±N concentration and, based on these ±N concentrations, aliquots (5 to

4 1230 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227± ml25 ml) of each extract were di used onto acidi- ed glass micro- bre disks (Brooks et al., 1989, as modi ed by Bradley and Fyles, 1996) and encapsulated in Sn sleeves (Europa Scienti c, Franklin, OH). Diffused samples were analysed for atom% 15 N analysis by continuous ow mass spectrometry. Gross production and consumption rates of ±N were calculated using zero-order equations derived by Kirkham and Bartholemew (1954). It was assumed that isotope addition would not bias ammoni cation rates (hereafter referred to as gross mineralisation rates) in the short-term but might cause an overestimation of gross ±N consumption rates due to enrichment of reactant pools (Hart et al., 1995) Soil respirometry Basal respiration rate (B ) and microbial biomass (MB ) of the ve pre-treatment humus samples were determined prior to application of treatments (T = 0 wk). B and MB of the 30 treated samples were determined post-treatment at T = 10 wk. A second set of post-treatment humus samples from each funnel was kept for 4 wk at 258C, and MB was re-measured at T = wk. A third set of post-treatment humus samples from each funnel was used to determine the energy de ciency index (EDI ) and the nutrient de- ciency index (NDI ) of microbial communities (Bradley and Fyles, 1995a). B was determined by weighing humus samples (ca. 8 g dry wt.) in 190 ml gas sampling jars, ushing the headspace with ambient air for 5 min, sealing jars with air-tight lids equipped with rubber septa and sampling aliquots of air in the headspace with a needle and syringe after 2 h. Air samples were analysed for CO 2 concentrations using a Hewlett±Packard model 5890 GC (Hewlett±Packard, Avondale, PA) equipped with an FID and methanizer, and using N 2 as carrier gas. MB was estimated by converting substrate-induced respiration rates to microbial biomass of organic residues using equations in Beare et al. (1990). Ground (65 mm) glucose, Difco nutrient broth powder, yeast extracts and talc (as inert ller) were mixed in a ratio of 5:7:3:86 (Bradley and Fyles, 1995a). This glucose + nutrient mixture (250 mg) was dispersed into tared humus samples (ca. 8 g dry wt.) using a kitchen handmixer with one beater. Samples were then transferred into 190 ml gas sampling jars and left uncovered for 100 min. After reaching maximum initial substrateinduced respiration rates (Anderson and Domsch, 1978), each sample was ushed for 5 min with ambient air, sealed for 30 min, and air from the head space was analysed for CO 2 concentration using a GC. EDI and NDI were calculated from respiration rates induced by addition of glucose only (G, as glucose:talc = 4:25), basal respiration rates of humus samples (B, as above), and respiration rates induced by addition of glucose + nutrients (GN, as above): EDI ˆ G B =GN 100% unitless NDI ˆ GN G =GN 100% unitless Ambient temperature and atmospheric pressure were recorded at each measurement, and ambient CO 2 concentration was measured several times daily. For each sample, ambient CO 2 concentration was subtracted from sampled CO 2 concentration and the di erence was adjusted according to Ideal Gas Laws and centered at 228C using Q 10 = Potential C source utilisation patterns The functional diversity of microbial communities was characterised in pre-treatment bulk humus (n = 5) and in post-treatment humus samples from each funnel using the Biolog GN microplating system (Biolog, Hayward, CA). Moist humus samples (ca. 10 g) were weighed (21 mg) into polyethylene bottles containing glass beads (3-mm in diameter), mixed with 100 ml of 0.1% Na-pyrophosphate solution and shaken at high speed on a benchtop reciprocating shaker for 15 min. The resulting suspensions were diluted 100-fold in deionized water as initial tests had shown optimal colour development occurring at this dilution. Aliquots (150 ml) of each diluted suspension were added to each of 96 wells in duplicate Biolog plates. Ninety- ve of the wells contained redox-sensitive tetrazolium dye and a unique C source, whereas one control well contained dye only. Biolog plates were incubated at 258C and colour formation in each well was read as light absorbance (590 nm) after 20, 24, 28, 44, 48, 52, 68, 72 and 76 h using an automated platereader (Biolog Microstation and software; Biolog, Hayward, CA). Well absorbance values of duplicate plates were averaged. At each reading, absorbance value of control wells were subtracted from absorbance values of 95 wells containing C substrates. Average well color development (AWCD) of each plate was then calculated at each reading to determine the incubation time (T 0.75 ) corresponding to AWCD = 0.75 absorbance units. The average number of substrates used at T 0.75 (i.e., substrate richness, or SR) was calculated for each sample using the majority-rules decision (Goodfriend, 1998) whereby substrate use was con rmed if dye reactions in at least three out of ve replicate suspensions were positive (i.e., >0.25 absorbance units). Utilisation rates of various chemical classes of C compounds were noted for each treatment. Substrate diversity (SD) was calculated (Zak et al., 1994) from: 1 2

5 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227± SD ˆ X p i ln p i where p i is the ratio of colour development of the ith well to the sum of colour development in the plate. The evenness of substrate use across positive wells (i.e., substrate evenness, or SE) was calculated using the relationship: SE ˆ SD=log SR 2.8. Statistical analyses The e ects of tannin or fertiliser N additions on cumulative leaching of ±N, NO 3 ±N, DON and tannins, on post-treatment humus N dynamics, on humus residual tannin concentrations and on humus respiration measurements were tested statistically using one-way ANOVA. Signi cantly di erent treatment means were evaluated using Duncan's multiple range test. The e ects of treatments on ±N and NO 3 ±N concentrations in leachates over the treatment period were analysed using repeated measures ANOVA (T = 2±10 wk). Both multivariate and univariate models were used to test the e ect of time and its interaction with treatments (i.e., within subject di erences). To attain sphericity of variance±covariance matrices, the six treatments were divided into non-fertilised or fertilised groups, which were then analysed separately for tannin e ects. This division of treatments was also necessary to obtain fewer treatments than repeated measurements within each statistical test, which is a criterion for repeated measures ANOVA (Potvin et al., 1990). The univariate approach used the adjusted 3 4 Huynh and Feldt (1976) F-statistic to test within-subject hypotheses. The e ects of tannins within each date were tested using Duncan's multiple range test. Corrected absorbance values (T 0.75 ) of 95 substrate wells in each Biolog plate were centered and normalised [i.e., (Absorbance AWCD)/s], and principal component analysis (PCA) was performed on transformed data to explore e ects of treatments on patterns of substrate use by microbial communities. Unless otherwise stated, P values of 0.05 or less were considered statistically signi cant. 3. Results Results of one-way ANOVA tests for the e ect of treatments on leachate N concentrations, post-treatment humus N dynamics, post-treatment respiration measurements, and post-treatment humus tannin concentrations are given in Table 2, and details are presented below Nitrogen and tannins in leachate Both Kalmia and r tannins signi cantly (P < 0.001) reduced total extractable ±N among non-fertilised treatments (Table 3). Kalmia tannins signi cantly (P < 0.01) reduced total extractable NO3 ±N among nonfertilised humus as well as total extractable ±N and NO3 ±N among fertilised treatments. The magnitude of tannin e ects is re ected by between subjects P-values given in Table 4. Tannins did not a ect DON concentrations. Adding fertiliser N to spruce humus increased cumu- Table 2 Results of one-way ANOVAs testing treatment e ects on various measurements of microbial and N dynamics in black spruce humus Measurement df F-value Prob > F Cumulative leachate N ±N NO3 ±N DON Post-treatment humus N Gross N mineralisation rate Gross ±N consumption rate Anaerobic mineralisation rate Post-treatment respirometry Basal respiration rate EDI NDI Microbial biomass (T = 10 weeks) Microbial biomass (T = 14 week) Post-treatment humus tannin concentration Fig. 1. Changes to ±N and NO 3 ±N concentrations in leachates from unfertilised [(a) and (b)] and fertilised [(c) and (d)] treatments; data points represent control (w), Kalmia tannin (t), r tannin (q), fertiliser N (*), Kalmia tannin + fertiliser N (T), and r tannin + fertiliser N (Q) treatments; treatment means, on the same graph within single dates, denoted by di erent lowercase letters di er signi cantly (Duncan's multiple range test, P < 0.001, P < 0.01, P < 0.05, ns = not signi cant).

6 1232 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227±1240 Table 3 E ect of treatments on mean cumulative leaching of ±N, NO 3 ±N and DON over 10 weeks (SE in parentheses, n =5) Treatment ±N (mg Ng 1 soil) NO 3 ±N (mg Ng 1 soil) DON (ng N g 1 soil) Control 75.1 (3.3) 2.51 (0.06) 196 (9) Kalmia tannins 51.2 (1.6) 2.21 (0.04) 173 (9) Fir tannins 51.8 (3.1) 2.40 (0.04) 184 (17) Fertiliser N (19.6) 5.09 (0.05) 121 (8) Kalmia tannins + fertiliser N (14.4) 4.75 (0.10) 198 (18) Fir tannins + fertiliser N (32.7) 4.96 (0.09) 157 (14) lative leaching of ±N by an order of magnitude and cumulative leaching of NO3 ±N by a factor of 2 (Table 3). The N-fertilisation induced increase in extractable ±N and NO 3 ±N began at T =2wk and remained throughout the trial (Fig. 1). Lower amounts of DON leached from humus amended with fertiliser N, but DON extracted from all treatments was low relative to DIN (Table 3). Repeated measures ANOVA revealed signi cant (P < 0.001) time e ects in mean extractable ±N concentrations in non-fertilised humus (Table 4 Ð within subjects). Fig. 1(a) shows a decreasing trend in ± N leaching from the three non-fertilised treatments. Mean extractable NO3 ±N concentrations also varied signi cantly (P < 0.001) over time in both non-fertilised and fertilised humus (Table 4 Ð within subjects), but there was no clear trend (Fig. 1(b) and (d)). The e ect of tannin additions on extractable ±N and NO3 ±N concentrations also varied signi cantly (P < 0.05) over time in non-fertilised humus (Table 4 Ð Table 4 Repeated measures ANOVAs correlating tannin additions to ±N and NO 3 ±N concentrations in leachate from non-fertilised ( F ) and fertilised (+F) humus Ionic species Fertiliser Source of variation df MS F-value Prob > F Between subjects F Tannins Error F Tannins Error NO3 F Tannins Error NO3 +F Tannins Error Within subjects (univariate test) F Time Time tannins Error (time) F Time Time tannins Error (time) NO3 F Time Time tannins Error (time) NO3 +F Time Time tannins Error (time) Within subjects (multivariate test) df Wilk's l F-value Prob > F F Time 4, Time tannins 8, F Time 4, Time tannins 8, NO3 F Time 4, Time tannins 8, NO3 +F Time 4, Time tannins 8,

7 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227± Fig. 3. E ect of tannins and fertiliser N additions on post-treatment microbial biomass (vertical bars =1 SE, n = 5). Fig. 2. E ect of tannins and fertiliser N additions on post-treatment measurements of (a) gross N mineralisation rate, (b) gross ±N consumption rate, and (c) anaerobic net N mineralisation (vertical bars = 1 SE, n = 5). univariate test). The multivariate approach gave a more conservative estimate (i.e., P = and , respectively) of these same interactions. More speci cally, Kalmia and r tannins both reduced ± N leaching from non-fertilised humus during T = 4± 10 wk (Fig. 1(a)). Kalmia tannins signi cantly reduced NO3 ±N leaching from non-fertilised humus at T =10 wk (Fig. 1(b)). Tannins were not detected in any of the leachate samples Post-treatment humus N dynamics Addition of fertiliser N increased gross N mineralisation rates by 1±2 orders of magnitude (Fig. 2(a)). Gross N mineralisation rates were signi cantly higher in fertilised humus amended with Kalmia or r tannins than in fertilised humus without added tannins. Gross N mineralisation rates of all treatments were higher than corresponding gross ±N consumption rates. These di erences were large in fertilised treatments (from 701 mg 1 d 1 for fertilised treatment to 1277 mg 1 d 1 for the fertilised + r tannin treatment) and small in non-fertilised treatments (from 7 mg 1 d 1 for Kalmia tannin treatment to 15 mg 1 d 1 for the control). Gross ±N consumption rate was signi cantly higher in fertilised humus amended with r tannins than in other treatments (Fig. 2(b)). Anaerobic N mineralisation rates were signi cantly higher (ca. 4) in humus from fertilised treatments than in humus from non-fertilised treatments (Fig. 2(c)). Both Kalmia and r tannins signi cantly reduced anaerobic N mineralisation rates within nonfertilised treatments Humus tannins Post-treatment condensed tannin concentrations were signi cantly (P < 0.05) higher in humus that had been amended with Kalmia tannins, either with or without fertiliser N (Table 5). However, only 1% of the amount of tannin added was recovered in these two treatments. Table 5 Average post-treatment condensed tannin concentrations in black spruce humus amended with condensed tannins or fertiliser N; means followed by a di erent lowercase letter di er signi cantly by Duncan's multiple range test (P < 0.05, n = 5); the percent recovery is based on control treatment value subtracted from average concentration of each treatment Treatment mg tannins g 1 humus % recovery Control 385 b Kalmia tannins (3%) 734 a 1.16% Fir tannins (3%) 480 b 0.32% Fertiliser N (1%) 389 b 0.01% Kalmia tannins (3%) + fertiliser N (1%) 690 a 1.02% Fir tannins (3%) + fertiliser N (1%) 462 b 0.26%

8 1234 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227±1240 Table 6 E ect of treatment duration on average basal respiration rate and microbial biomass of all treatments (SE in parentheses, n = 30; ND = not determined) Measurement Pre-treatment (T = 0 week) Post-treatment (T = 10 weeks) Post-treatment (T = 14 weeks) Basal respiration (mg CO 2 ±C g 1 h 1 ) 4.66 (0.13) 3.25 (0.26) ND Microbial biomass (mg C mic g 1 ) 2.72 (0.08) 1.71 (0.07) 1.72 (0.07) Table 7 E ect of treatments on average number of positive wells in Biolog plates (SR), on the equivalency of colour development among positive wells (SE), and on the calculated index of functional diversity (SD) of microbial communities; all values measured at AWCD 0.75 are unitless; values in the same column followed by di erent lowercase letters are signi cantly di erent (Duncan's multiple range test, P < 0.05, n = 5) Treatment Substrate richness (SR) Substrate evenness (SE) Substrate diversity (SD) Pre-treatment 75 a 0.88 a 4.03 a Post-treatment Control 54 b 0.78 a 3.57 b Kalmia tannins 55 b 0.78 a 3.56 b Fir tannins 54 b 0.77 a 3.50 b Fertiliser N 58 b 0.79 a 3.62 b Kalmia tannins + fertiliser N 58 b 0.79 a 3.61 b Fir tannins + fertiliser N 50 b 0.76 a 3.47 b 3.4. Respirometry and C source utilisation patterns There were no signi cant treatment e ects on posttreatment microbial measurements derived from humus respirometry except for a relatively weak (P = 0.02) treatment e ect on MB (Table 2). Duncan's multiple range test revealed higher MB in humus amended with Kalmia tannins than in other treatments (Fig. 3). Di erences in MB among all treatments were small, however, in comparison to the 30% decline in average Fig. 4. Principal component analysis of C source utilisation patterns by microbial communities in black spruce humus prior to, and after, tannin and fertiliser N additions; PCA axes 1 plus 2 accounted for 39% of variance in data set; data points are same as described in Fig. 1; ve pre-treatment samples denoted by `P'. MB of all treatments between T = 0 and 10 wk (Table 6). Similarly, the average B of all treatments declined by 37% between T = 0 and 10 wk. There was no di erence between average MB of all treatments at T = 10 wk and 14 wk. There was a large error term associated with NDI measurements. Consequently, NDI did not di er signi cantly among treatments, although treatment averages varied by as much as 100%. The highest NDI value was for humus amended with Kalmia tannins. NDI was consistently higher in non-fertilised treatments compared to corresponding (i.e., paired) fertilised treatments (data not shown). In contrast, average EDI values of all treatments varied by less than 10%. PCA ordination based on C source utilisation patterns by microbial communities showed distinctive patterns between pre-treatment and post-treatment humus samples (Fig. 4). Post-treatment samples had abovemedian scores for PCA axis 1, in contrast to pre-treatment samples that had below-median scores. PCA did not reveal distinctive patterns in post-treatment substrate use among di erent treatments. PCA axes 1 and 2 explained 31% and 8%, respectively, of the variance in the data set. Substrate richness (SR) and substrate diversity (SD) were signi cantly higher in pre-treatment samples than in post-treatment samples (Table 7). The average decrease in positive wells from pre-treatment to posttreatment was due to lower utilisation rates of carbohydrates (12 fewer wells), amino acids (8 fewer wells), amines and amides (4 fewer wells) and carboxylic acids (4 fewer wells). Utilisation rates of other classes of C

9 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227± compounds (esters, polymers, alcohols, aromatics, brominated and phosphorylated chemicals) were identical in all pre- and post-treatment plates. 4. Discussion 4.1. Leachate chemistry The addition of fertiliser N to spruce humus resulted in a marked and sustained increase in extractable ±N and NO 3 ±N during the 10-wk study. However, the net increase in DIN leached was less than 10% of the fertiliser N added. Therefore, the spruce humus had a high potential for rapid immobilisation and subsequent slow release of added mineral N. Although a basic tenet of soil ecology is that microbial communities are primarily limited by available C (Smith and Paul, 1990), Hart and Stark (1997) have shown that heterotrophic communities in some mature conifer forest soils can be N-limited as well. Such a feature would predispose the black spruce ecosystem from which humus was gathered in the present study to mitigate losses of mineral N following fertiliser N application. Addition of Kalmia and r tannins signi cantly reduced ±N leaching from non-fertilised humus. However, such e ects were not detected in fertilised treatments, probably because positive e ects of fertiliser N additions were greater than (and therefore overwhelmed) negative e ects of tannin additions. In this study, fertiliser N was applied at a rate much higher than prescribed by current silvicultural practices whereas tannin addition rates were comparable to tannin concentrations commonly measured in litter and humus (Lorenz et al., in press). If it is assumed that the principal mode of action of tannins was to bind abiotically with nitrogenous substrates in humus, then there would be a threshold in N fertility beyond which the sequestering of potentially mineralisable N by tannins becomes insigni cant. This is supported by eld observations that Kalmia-induced growth check of black spruce seedlings is greater on lower quality sites (Titus et al., 1995), and that application of mineral N fertilisers alleviates growth check in many problematic forest sites dominated by ericaceous shrubs such as Calluna vulgaris (Taylor, 1991), Gaultheria shallon (Prescott et al., 1996) and Kalmia angustifolia (Prescott et al., 1995). Factors controlling NO3 ±N dynamics depend on the nature of microbial communities involved in producing and consuming this anion. While nitri cation in agricultural soils is primarily attributed to autotrophic micro-organisms, heterotrophic nitri ers may be more prevalent in some forest soils (Hart et al., 1997; Papen and von Berg, 1998). The immediate and sustained increase in NO3 ±N leaching after fertiliser-nh 4 addition suggests that the spruce humus we studied had a signi cant potential for autotrophic nitri cation. However, high net nitri cation rates in fertilised humus could also have resulted from lower microbial assimilation rates of NO3 ±N due to higher NH 4 ±N pools (Hart et al., 1994). Determining whether ±N pools increased gross NO3 ±N production rates or decreased gross NO3 ±N consumption rates was beyond the scope of our study, but there is evidence from non-fertilised treatments that ±N supply was not the sole factor controlling NO3 ±N production. The fact that only Kalmia tannins signi cantly reduced NO3 ±N production in non-fertilised humus, and that this e ect occurred only on the nal leaching date (T = 10 wk) suggests (1) that nitri cation was not controlled by ±N concentrations alone, and (2) that Kalmia and r tannins did not a ect nitri cation pathways in a similar manner. The possibility that Kalmia tannins inhibited nitri er populations cannot be excluded since a decreasing trend in NO3 ±N leaching was observed in the Kalmia tannin treatment, relative to other treatments, during the 10-wk study. For example, some studies have suggested that phenolic compounds including tannins, phenols and volatile terpenoids are capable of inhibiting soil nitri er populations (e.g., Lohdi and Killingbeck, 1980; White, 1988; Paavolainen et al., 1998). However, the higher post-treatment MB measured in the Kalmia tannin treatment suggests that Kalmia tannins may have been a growth substrate for some microbes. Therefore, the apparent inhibitory action of Kalmia tannins on nitri cation may have been caused by increased NO3 ±N consumption due to higher available C in this treatment. Future studies should include 15 NO 3 ±N dilution assays to verify these hypotheses regarding factors controlling net production of NO3 ±N in tannin-rich humus. Northup et al. (1995) proposed that tannin-rich plant communities conserve N by decreasing soil mineral N pools and minimising N leaching. Based on correlative evidence, Northup et al. (1995) hypothesised that the preponderance of DON in soil solution is attributable to the presence of tannin-protein complexes which delay the onset of N saturation. However, this was not supported by our study, as the addition of fertiliser N decreased DON leaching, and addition of tannins had little e ect on DON. DON is often considered to be the major vector for N movement through forest oors (Qualls et al., 1991; Hedin et al., 1995), but was present in low concentrations in leachate relative to DIN. There are two possible explanations: (1) pre-treatment leaching (i.e., the conditioning period in which 800 ml of leachate were discarded) removed DON from humus solution, and a more prolonged incubation would be needed to

10 1236 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227±1240 replenish these DON pools than to replenish DIN pools; or (2) the 10 mm CaCl 2 extractant, proven to be e cient in removing mineral N, was too weak to displace large molecular weight DON molecules of higher adsorptive capacity Humus N dynamics Although the increase in mineral N leaching was modest relative to fertiliser N additions, the corresponding increase in gross N mineralisation rates in post-treatment humus was relatively high. Di erences between daily gross N mineralisation and daily gross consumption rates in fertilised treatments, extrapolated over 2 wk, were 10 times greater than biweekly fertiliser N addition rates. However, such extrapolation is tenuous since gross ±N consumption probably obeyed rst order kinetics and were implicitly related to the size of ±N pools at the time of measurement (Bradley et al., 1997c). It is therefore likely that gross ±N consumption rates were highest immediately after fertiliser N addition and lowest on the date coinciding with measurement, which would explain the large di erences between the two process rates. Regardless of time-scale bias, the results of 15 ±N dilution assays strongly supported the assumption that microbial communities were N de- cient and that a major portion of added ±N was used to increase and maintain actively cycling microbial N pools. One of our aims was to determine the e ect of Kalmia and r tannins on microbial development. Results from the few studies that have examined the e ects of tannins (as opposed to other phenolic compounds) on microbial growth are contradictory, perhaps because tannins from di erent origins can vary in molecular size and structure. For example, Benoit et al. (1968) studied tannins extracted from wattle (species not speci ed, but likely Acacia spp. or mimosa) and concluded that ``the principal e ect of tannins on microbial development is not that of toxicity'', whereas Schimel et al. (1996), using tannins extracted from balsam poplar, concluded that ``tannins act as a general microbial inhibitor''. Some authors reported no evidence for decomposition of tannins during leaf decay (Scho eld et al., 1998) whereas others reported that certain soil organisms preferentially degrade condensed leaf tannins (Gamble et al., 1996). Thus, it is di cult to speculate whether Kalmia and r tannins might be toxic compounds or growth substrates for soil microorganisms. However, it is known that toxic compounds can increase gross N mineralisation rates in humus, whereas readily metabolisable substrates such as starch can reduce gross N mineralisation rates (Schimel et al., 1992). In our study, tannins did not signi cantly a ect gross N mineralisation rates in non-fertilised humus, and therefore it can be inferred that tannins were neither toxic nor easily metabolised by soil microbes when ±N concentrations were low. However, tannins did increase gross N mineralisation rates in fertilised humus, suggesting a toxic e ect to microbes when ±N concentrations are high. This observation is reminiscent of results of Bradley et al. (1997c) in which signi cantly higher gross N mineralisation rates occurred in fertile humus planted with Kalmia than with other seedlings, but there were no signi cant changes in gross N mineralisation rates in poor humus planted with the same species. Thus, a parallel may be drawn between the presence of Kalmia root systems and the addition of Kalmia tannins to humus. Their similar e ects suggest that Kalmia root systems release tannins or related compounds into soil, and that these may be toxic to some micro-organisms. High mineral N fertility, and by implication high actively cycling microbial N pools, appear necessary for tannins to signi cantly increase gross N mineralisation rates. Tannins possibly increase microbial death and turnover rates, with a portion of microbial N that is released being prone to form recalcitrant complexes with tannins. Howard and Howard (1993) found substantial binding between protein and polyphenols released from freshly fallen litter of various tree and shrub species, and that these stable complexes signi cantly reduced subsequent N mineralisation rates. Thus, tannin±protein binding may o set the potential for higher N leaching due to higher gross N mineralisation rates. As with ±N concentrations in leachates, anaerobically mineralised N (AMN) rates were signi cantly higher in fertilised humus than in non-fertilised humus, and addition of tannins signi cantly reduced AMN rates within non-fertilised humus. AMN rates have been correlated with soil N availability in forest ecosystems (Keeney, 1980; Powers, 1980) and are thought to re ect the activity of microbial N pools (Myrold, 1987). AMN rates therefore provided further evidence that fertiliser N alleviated microbial N de ciencies whereas tannins limited microbial access to organic N Tannins Little is known about the fate of litter tannins, except that there is a large and rapid decline (around 80%) in the rst year of decomposition (reviewed in Lorenz et al., in press). However, tannin concentrations in humus are generally less than 0.5%. Where an unusually high amount was found (3±4%), such as in the black spruce sites studied by Lorenz et al. (in press), it appeared to be caused more by environmental factors (climate, fauna) rather than by unusually high amounts of tannin in the litter. We found low recoveries of added tannin in the

11 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227± humus after incubation (ca. 1%), and no detectable tannin in leachates. There are no reports of similar recovery data in the literature, although Scho eld et al. (1998) indicated that tannins leached from willow leaves were tightly sorbed to mineral soil and could not be detected by a variety of approaches. It is possible that the tannins may have become tightly bound to organic matter, including proteins, or had undergone chemical reactions with organic matter, or had been microbially metabolised. The last possibility seems less likely, based on our other data from this study Microbial communities If readily-metabolisable C substrates were applied to humus at a rate comparable to the rate of tannin additions used in this experiment (i.e., 3% wt/wt over 10 wk), MB would probably increase (Bradley and Fyles, 1995b; Bradley et al., 1997b). In our study, the average post-treatment MB of all treatments declined signi cantly from pre-treatment values, therefore r and Kalmia tannins were probably not readily metabolised by soil microbes. Experimental conditions likely decreased available C in humus because (1) the humus was maintained at a higher and more constant temperature than in the eld, thus favouring rapid consumption of available C sources, and (2) there were no chronic inputs of organic substrates, such as in throughfall and rhizodeposition under eld conditions, to replenish labile C pools. Therefore, lower mineral N leaching from nonfertilised humus treated with tannins was most likely due to biochemical interference of tannins with potentially-mineralisable N sources as opposed to higher microbial immobilisation rates induced by higher C supply. Kalmia tannins had a lower molecular weight than r tannins which may have resulted in their being partially metabolised, which would explain why posttreatment MB was slightly higher in humus treated with Kalmia than with r tannins. However, this may not be of importance for two reasons. Firstly, the di erence between post-treatment MB in humus amended with Kalmia tannins and the average MB of the other ve treatments was about one-third of the average decline in MB of all six treatments over the 10-wk treatment period (i.e., 0.32 vs mg C mic g 1 ). Secondly, average MB of the six treatments remained constant for 4 wk following post-treatment (i.e., T = 14 wk), and there were no signi cant di erences in MB among treatments at this later date. The e ect of Kalmia tannins on MB was therefore relatively weak and ephemeral. Benoit et al. (1968) demonstrated that considerable amounts of readilydecomposable carbohydrates were contained in crude tannin preparations and that tannin decomposition rates decreased markedly as the purity of these tannins increased. However, use of NMR to characterise tannins before and after preparation ensured that the Kalmia and r tannins we used were free of impurities. Di erences in MB among treatments were small relative to di erences in N cycling rates, suggesting that MB was controlled by C availability whereas N cycling rates depended on microbial N pools and possibly rates of microbial death and turnover. MB is a static measurement, which does not re ect microbial turnover rates. An apparent increase in microbial N with fertiliser addition without a concomitant increase in MB corroborates the premise that microbial communities were initially N de cient. The ratio of C mic - to-n mineralised was lowest in humus amended with Kalmia tannins, suggesting higher microbial N de- ciency in this treatment. Thus, di erences in NDI among treatments, although statistically non-signi cant, were nonetheless consistent with other observed e ects of fertiliser N and tannins on N cycling and microbial dynamics. Respirometry measurements are useful indicators of the ecophysiological status of microbial communities, but do not indicate whether treatments caused fundamental changes to community structure. Biolog GN microplates, on the other hand, are commonly used as proximate indicators of the functional diversity within microbial communities, based on utilisation patterns of di erent C sources (Garland, 1996a, 1996b). Although Biolog assays do not identify or distinguish taxonomic groups within microbial communities, C source utilisation patterns re ect di erences in their metabolic capacities. Since the number of viable cells in each well a ects colour development (Garland, 1996a), adjustment of samples to equivalent inoculum cell densities has been recommended (Haack et al., 1995). The decision to use equivalent cell densities should, however, be determined by the nature and scope of the investigation. In this study, inoculum cell density was not standardised because (1) such a manoeuvre can bring about its own experimental artefact, (2) identical inoculum cell densities can result in 20% di erences in AWCD (Garland, 1996a), and more importantly (3) the aim was to determine changes in metabolic capacity of microbial communities within a single humus type, on a per unit humus weight basis, and therefore changes in microbial densities as well as in community structure were implicit in the treatment e ects that were investigated. For these reasons, plates were compared by adjusting the activity in each well to AWCD 0.75, as other studies have done (Garland and Mills, 1991; Zak et al., 1994; Garland, 1996b; Grayston et al., 1998). Post-treatment microbial communities were less diverse (i.e., lower SD) and contained fewer catabolic enzyme systems (i.e., lower SR) than pre-treatment mi-

12 1238 R.L. Bradley et al. / Soil Biology & Biochemistry 32 (2000) 1227±1240 crobial communities. The decrease of SD and SR were due to lower capacities to metabolise simple as opposed to more complex C substrates. One possible explanation is that labile C, which decreased due to experimental conditions (as con rmed by MB data), created a selective pressure favouring microbial communities that metabolised more resistant C substrates. Fungi specialised in the breakdown of complex substrates may have increased relative to bacteria during the study period. Since Biolog assays strongly underestimate fungal activity because fungal hyphae are extracted ine ciently from humus (Zak et al., 1994), lower values of SR and SD at post-treatment could therefore re ect an increase in fungal-to-bacterial ratios during the 10-wk experiment. Our results emphasise the importance of distinguishing between factors that a ect C supply, and consequently N cycling, and exogenous factors such as fertiliser N or some condensed tannins, that a ect N cycling only. The experimental period resulted in a lower C supply which, in turn, had more e ect on functional diversity of microbial communities than did the experimental treatments. Other Biolog studies have shown that microbial communities can change after exposure to plant roots (Garland, 1996b), perhaps because of C that is supplied by rhizodeposition. In our study, tannin and fertiliser N additions signi cantly modi ed humus N cycling but did not a ect C source utilisation patterns by microbial communities during the experiment. This gives rise to an important question regarding changes in site quality due to tannin or fertiliser N additions: To what extent does functional diversity of microbial communities play a role in controlling humus N dynamics? Since soil organic N is thought to mineralise stoichiometrically with soil organic C (McGill and Cole, 1981), and since individual microbial populations metabolise speci c substrates according to speci c kinetic pro les, the microbial functional diversity should, in theory, be linked to humus N cycling characteristics. However, the complexity of microbial communities probably ensures that similar substrates can be metabolised at similar rates by di erent taxa of micro-organisms, or that many micro-organisms can subsist in the absence of optimal nutrient supply Implications of ndings We have demonstrated that, under laboratory conditions, condensed tannins isolated from Kalmia and balsam r can reduce mineral N availability in black spruce humus. The lack of recovery of tannins in leachate and treated humus, the general decrease in MB over the treatment period and the lack of change in C source utilisation patterns by microbial communities suggest that tannins decreased mineral N cycling by binding to and sequestering organic N sources rather than because they were toxic to or easily metabolised by microbial communities. In order to assess the ecological importance of Kalmia tannins, relative to tannins produced by r or other plants, in reducing humus N availability on spruce cutovers, further work is needed to determine (1) how tannins move from Kalmia to humus, (2) the exact nature and persistence of Kalmia tannin±protein complexes, and (3) possible uptake of tanned substrates by Kalmia and other plants. Since the experiment was performed on humus taken from a single site, the e ects of tannins and fertiliser N should be tested over a wide range of humus forms and site qualities before drawing general conclusions. Nevertheless, the formation of Kalmia tannin±protein complexes in humus may explain the large reductions in net N mineralisation on some Kalmia heaths (Damman, 1971) and the reduction of foliar N concentrations in black spruce seedlings growing in proximity to Kalmia on some cutovers (Yamasaki et al., 1998). Where microbial communities occupying black spruce humus are also N limited, adding fertiliser N should not only reduce spruce growth check in the short term, but should also mitigate microbial N de ciencies, increase active N pools and prevent rapid leaching loss of N. The importance of increased microbial N uptake independent of C-supply needs further evaluation in terms of restoring site fertility. Acknowledgements The authors wish to thank Karen Hogg and Kevin McCullough for co-ordinating laboratory operations as well as Sarah Riecken, Carol Cutworth and Martin Smith for technical assistance. We are thankful to Dr. Bernard Colin for advice on statistical analyses. Isotope enrichment analyses were carried out by continuous ow mass spectrometry at the Stable Isotope Laboratory, University of Saskatchewan; chemical analysis of the initial bulked humus was carried out at the MacMillan Bloedel Laboratory, Nanaimo, BC; all other analyses were carried out at the Paci c Forestry Centre, Victoria, BC. The study was funded by a grant from the Science Council, Forest Renewal of British Columbia as well as Post-Doctoral Fellowship from the Natural Sciences and Engineering Science Council of Canada. References Anderson, T.H., Domsch, K.H., A physiological method for the quantitative measurement of microbial biomass in soil. Soil Biology & Biochemistry 10, 215±221.