Effects of forest harvest on soil nutrients and labile ...

2 downloads 0 Views 386KB Size Report
the procedure developed by Kenward and Roger (1997). Model diagnos- ...... Science 176, 914–915. Martin, C.W., Noel, D.S., Federer, C.A., 1984. Effects of ...
Catena 122 (2014) 18–26

Contents lists available at ScienceDirect

Catena journal homepage: www.elsevier.com/locate/catena

Effects of forest harvest on soil nutrients and labile ions in Podzols of southwestern Canada: Mean and dispersion effects Stephanie Grand a,⁎, Robert Hudson b, Les M. Lavkulich a a b

Soil Water Air Laboratory, University of British Columbia, 2357 Main Mall, Vancouver, BC V6T1Z4, Canada Vancouver Forest Region, BC Ministry of Forests, 2100 Labieux Road, Nanaimo, BC V9T6E9, Canada

a r t i c l e

i n f o

Article history: Received 24 December 2013 Received in revised form 23 April 2014 Accepted 4 June 2014 Available online xxxx Keywords: Forest harvest Soil nutrients Soil pH Exchangeable ions Inorganic nitrogen

a b s t r a c t Forest harvest can disrupt biogeochemical cycles with consequences for regenerating forest nutrition and drainage water quality. Few studies have examined harvest impacts on soil chemistry in conifer forests which have not been significantly affected by acid deposition, as found on the Canadian West Coast. This study investigates the effects of conventional clear-cutting on soil chemistry in Podzols of Roberts Creek Study Forest (southwestern Canada). We measured forest floor composition, soil pH and salt-extractable ions concentrations in undisturbed forested stands (control plots), stands harvested 2 to 5 years prior to sampling (cleared plots) and stands harvested 8 to 15 years prior to sampling (regenerating plots). We focused on the effects of forest harvest both on mean (differences in average values) and dispersion parameters (differences in variance between treatments). We found that forest floors of harvested plots had lower phosphorus and potassium concentrations than control plots. In the mineral subsoil, exchangeable K was however higher in harvested than in control plots. This suggests that some of the K lost from the forest floor was preferentially retained in mineral horizons, possibly due to sorption to poorly crystalline and free aluminum and iron mineral phases. The subsoil of harvested plots was slightly more acidic than control plots. In contrast to classic studies of forest harvest impacts conducted in the U.S. Northwest, we did not measure pronounced acidification, a loss of base cations or an increase in exchangeable Al, most likely due to the much lower prior acid deposition load at our sites. The most notable harvest effect was a large increase in the variability of inorganic N concentration. This suggests an increase in micro-heterogeneity of post-harvest nutrient availability which has implications for the nutrition of regenerating vegetation, nutrient leaching potential as well as our ability to detect harvest-induced changes. © 2014 Elsevier B.V. All rights reserved.

1. Introduction

1.1. The N cycle

Forest harvesting has the potential to significantly alter site biogeochemical cycles. Biocycling elements are directly affected by the removal of biomass, while elements predominantly under inorganic control are indirectly affected by chemical (e.g. pH) and physical (e.g. profile disturbance) changes in the soil. Changes in soil labile or exchangeable element pools are of particular concern because these pools are immediate sources of nutrients for plants and affect forest regeneration and productivity. Exchangeable ions also equilibrate rapidly with soil solution composition, so that changes in ions prevalence on exchange complexes act as indicators of changes in soil solution composition, which in turn influences drainage water composition (Sudduth et al., 2013).

Because of its role in northern forest nutrition and its potential impact on water quality, N is one of the most widely studied nutrient in forest ecosystems. In cut forests, increases in mineralization rates (Spielvogel et al., 2006), lack of plant uptake (Burns and Murdoch, 2005) and lack of canopy interception (Klopatek et al., 2006) can combine to increase N availability, with a peak often reported at 3 to 5 years following harvest (Bradley et al., 2001; Dahlgren and Driscoll, 1994). One of the factors that complicates understanding and predicting N response to forest harvest is that sitespecific interactions between environmental conditions, biological variables and substrate quality greatly influence N dynamics (Grenon et al., 2004; Schimel and Bennett, 2004). The increase in N leaching from cut forests is variable, ranging from non-existent to more than 50 times the baseline leaching rate in experiments controlling vegetation regrowth (Likens et al., 1970). The largest N exports after logging were reported in northern hardwoods ecosystems, while coniferous forests are generally thought to be less susceptible to N losses (Binkley and Brown, 1993; Lamontagne et al., 2000; Martin et al., 1984).

⁎ Corresponding author at: Department of Forestry, Michigan State University, Natural Resources Building, 480 Wilson Road, East Lansing, MI 48824, USA. Tel.: +1 517 488 4145. E-mail address: [email protected] (S. Grand).

http://dx.doi.org/10.1016/j.catena.2014.06.004 0341-8162/© 2014 Elsevier B.V. All rights reserved.

S. Grand et al. / Catena 122 (2014) 18–26

Vitousek et al. (1982) proposed that N availability prior to disturbance was one of the main factors determining N leaching losses after disturbance, with N-poor sites having better N retention after disturbance as a result of lower N mineralization rates and higher N immobilization capacity. Several N-poor sites also exhibited lags in nitrification, probably due to small initial nitrifier populations. Conifers in particular are known to produce allelophatic chemicals that inhibit nitrifiers (Paavolainen et al., 1998; White, 1994). This nitrification lag may prevent N export to stream altogether if vegetation re-establishment is rapid enough (Vitousek et al., 1982).

19

we expect the increase in inorganic N concentration on cut sites to be limited due to the low initial soil N availability. We however expect that the loss of biotic regulation (Bormann and Likens, 1979) induced by forest harvest will have an impact on the variability of soil chemical attributes. Finally, we hypothesize that Roberts Creek's low pH soils may be at risk of acidification and decreasing base saturation following logging, which could negatively impact soil quality. 2. Methods 2.1. Field sites

1.2. Acidification and base cations Classic studies of forest harvest impacts conducted in the northern hardwood forest biome showed that in many watersheds, clearcutting results in soil acidification, increase in soluble Al, and a shortlived increase in NO− 3 concentration in soil and drainage water (Dahlgren and Driscoll, 1994; Hendrickson et al., 1989; Johnson et al., 1991; Likens et al., 1970, 1978; McHale et al., 2007). Likens et al. (1970) proposed that the increases in NO− 3 and acidity were linked since increases in nitrification rates result in the production of H+ ions. More generally, decarboxylation of organic anions and ammonification of organic N tend to increase soil pH (Mengel, 1994; Xu et al., 2006) while nitrification of mineralized N (Conyers et al., 1995; Dahlgren and Driscoll, 1994), production of organic acids and oxidation of reduced S (Devries and Breeuwsma, 1987) all tend to decrease it. Actual soil acidification or alkalization following disturbance will depend on the relative importance of these processes. In watersheds experiencing strong acidification after harvest, the base saturation of soil exchange complexes tends to be reduced as base cations are displaced by acid forming ones and eventually flushed out of the system (e.g. Lawrence et al., 1987; Neal et al., 1992). On the other hand, when acidification is mild, base cation prevalence on exchange complexes either remains constant or increases as decomposing organic matter releases significant amounts of Ca and K (Hendrickson et al., 1989; Johnson et al., 1997; Snyder and Harter, 1985). Of the base cations, K shows the most frequent change after logging (see for instance Johnson et al., 1991; Likens et al., 1970; Mann et al., 1988; Mroz et al., 1985). This is not surprising since the control on K concentration is mainly organic while Ca, Mg and Na are thought to be predominantly influenced by mineral weathering (Vitousek, 1977). Although forest harvesting is thought to increase weathering rates (Johnson, 1989; Likens et al., 1970), the resulting release pulse is generally not as dramatic as changes in the concentration of elements under strong biotic control. 1.3. Objectives The objectives of this study are to determine the effects of forest harvest on soil acidity, nutrients and labile ions in an economically important, managed coastal forest of southwestern Canada. We complement prior chronosequence studies focused on long-term (N200 years) soil ecological dynamics conducted in the region (e.g. Fons and Klinka, 1998; Preston et al., 2002) by exploring nutrient dynamics for the first 15 years post-logging, which is the timeframe in which acute negative environmental effects are most likely to occur. We focus on the detection of treatment (harvest) mean effects as well as dispersion effects (difference in variance between treatments), which are not commonly emphasized in the literature but can have important ecological impacts on regenerating vegetation and nutrient leaching potential. We relate our findings to stream chemistry studies previously conducted in our study forest. We hypothesize that changes in nutrients and labile ions concentrations in harvested plot will be relatively small, because (1) soils are not nutrient-rich and (2) harvested plots showed a good retention of their soil organic matter stock (Grand and Lavkulich, 2012). In particular,

This study was conducted at the Roberts Creek study forest (49° 27′ N, 123°41′ W) on the Sunshine Coast of southwestern British Columbia (BC). The study forest lies within the Coastal Western Hemlock, Drier Maritime variant (CWHdm) biogeoclimatic zone (Pojar et al., 1991) and experiences a mean annual temperature of 10.2 °C and mean annual precipitation of 1369 mm (Environment Canada, 2013). Elevation ranges from 350 to 590 m above sea level and the forest is situated on a gentle (~15%) southerly slope (Hudson, 2001). The current forest originated following wildfires approximately 140 years (D'Anjou, 2002). The dominant canopy species consisted of Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) interspersed with smaller diameter western redcedar (Thuja plicata (Donn. Ex E. Don) Spach) and western hemlock (Tsuga heterophylla (Raf.) Sarg.). The cedar and hemlock components gained dominance later in succession. The shrub and herb layers mostly comprised salal (Gaultheria shallon Pursh) and brackenfern (Pteridium aquilinum (L.) Kuhn) characteristic of the Salal site series of coastal BC (Green and Klinka, 1994). Vegetation in harvested stands was dominated by the herb layer (particularly fireweed, Epilobium angustifolium L.) for the first 3 to 5 years post logging, then by Douglas-fir saplings. Soils of Roberts Creek were Albic Gleyic podzol (IUSS Working Group WRB, 2006) of sandy loam to loamy sand texture that developed following glacial retreat from a basal till deposited on granodioritic bedrock. The sola ranged from 40 to over 120 cm in thickness. The following sequence of horizon was observed: Oi, Oe + Oa, E, Bs1, Bs2, BCg and Cr. Description of a typical soil profile and average horizon properties are reported in Grand and Lavkulich (2011). 2.2. Experimental design We sampled a disturbance chronosequence of forest plots that had been clear-cut 2 to 15 years prior to sampling to determine short to medium-term differences in nutrient and labile ions concentrations. Nine soil pits were located on undisturbed forested plots (control), 11 were located on cleared stands (logged 2–5 years before sampling) and 7 in regenerating stands (logged 8–15 years before to sampling). The harvest method was a clearcut with bole-only removal by cable yarding and slash left untreated on site. Variable retention occurred in some of the harvested plots; in this case, we sampled clear-cut portions of the plot, maintaining a minimum distance of at least 12 m to the nearest retained tree. Samples from harvested stands spanned seven harvest clusters distributed throughout the experimental forest. Control locations were interspersed in the undisturbed forest between and around logged plots and at a distance of at least 30 m from the edge of the disturbance. The distribution of sampling sites is recorded in Grand and Lavkulich (2012). Sampling was carried out during the dry season (August 2005). When sampling harvested plots, our objective was to gain insight about the in-situ effects of vegetation removal over time, rather than the extent of mechanical disturbance caused by logging equipment. We sampled morphologically undisturbed soil profiles with no signs of mechanical disruption or water erosion. We avoided old logging roads, equipment tracks and preferential flow channels. The use of a cable yarding system for timber extraction limited soil disturbance;

20

S. Grand et al. / Catena 122 (2014) 18–26

we visually estimated the area showing signs of soil impacts to approximately 10% of stand area. 2.3. Soil sampling and analyses 2.3.1. Sampling Soil pits of approximately 100 cm in diameter and 120 cm in depth were manually excavated and sampled by morphological horizon. The forest floor was separated into two components: (1) fresh litter (Oi) and (2) humic and fibric layer (Oe + Oa). The litter layer was thin and patchy, and often dominated by coarse woody debris. Only the Oe + Oa layer was sampled. 2.3.2. Analyses pH was measured potentiometrically on field moist samples within 48 h of sample collection in deionized water and 0.01 M CaCl2. The soil/solution ratio was 1:2 for mineral horizons and 1:4 for organic horizons (Schofield and Taylor, 1955; Van Lierop, 1990). Salt-extractable NO3-N and NH4-N were measured by extracting 20 g of field moist, b2 mm soil with 60 mL of 2 M KCl followed by analysis on a Lachat QuikChem 8000 Flow Injection Analysis System (Hach Company, 2009). Nitrate was determined using the cadmium reduction method followed by spectrophotometric detection of the nitrite formed (ISO standards, 1996). Ammonia was reacted with sodium salicylate method and the resulting indophenol was determined spectrophotometrically (Verdouw et al., 1978). Exchangeable acidity (Al3 + and H+) was determined on the abovementioned KCl extracts according to the titration procedure of Thomas (1982). We partitioned total exchangeable acidity into H+ and Al3+ ions using the fluoride complexation method (Yuan, 1959). Other exchangeable and salt-extractable ions were extracted by leaching 10 g of soil with 160 mL of 1 N ammonium sulfate solution using a mechanical vacuum pump. Concentrations of Ca2 +, Mg2 +, Na+, K+ and PO34 − in the extract were determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES). The concentration of Cl− was measured on the Lachat Flow Injection Analysis System using the thiocyanate method (Hach Company, 2007). Ammonium sulfate was chosen as a displacing salt over the more common NaCl, KCl or NH4Cl salts because given the potential for maritime influence, Cl− was an ion of interest in the profile. In addition, SO2− 4 is believed to be a more effective anion displacer than Cl− (Katou et al., 1996). A 1 N solution of (NH4)2SO4 is weakly buffered at a pH of 5.2, which is close to the pH of the study soils. The effective cation exchange capacity (CECe, cmolc/kg) was determined by the sum of charges pertaining to exchangeable cations (Ca, Mg, Na and K) and exchangeable acidity. Total C and N were measured on air-dried forest floor material ground to pass through a 0.5 mm sieve by dry combustion using an induction furnace (LECO model CN-2000). Total P and base cations were extracted by the Parkinson and Allen (1975) digestion procedure followed by analysis on ICP-AES. This method was determined to be suitable for Podzols of coastal BC (Cade-Menun and Lavkulich, 1997). 2.4. Data analyses 2.4.1. Statistical model Statistical analyses were conducted using SAS version 9.3 software (SAS Institute Inc, 2010). The effects of harvest treatment were analyzed using a linear mixed model. Treatment effects (control/cleared/ regenerating), horizon effects and the treatment * horizon interaction were included as fixed effects. Observations were blocked by harvest plot using a random group effect (G-side). To avoid pseudo-replication with respect to horizon effect, the horizon effect was set as a repeated measure (R-side random effect). This sets a common correlation among all observations of the same soil profile. We used the first order autoregressive (AR1) covariance structure to model the correlation between horizons because the Bayesian information criteria (Schwarz, 1978)

compared favorably to other covariance structures, and there was no evidence of lack of fit. In order to allow for the possibility that treatment had an effect on the variance of the dependent variable (dispersion effect), covariance parameters were allowed to vary according to treatment groups. Variance component estimation was based on the restricted maximum likelihood procedure. Degrees of freedom were calculated using the procedure developed by Kenward and Roger (1997). Model diagnostics (normality, homoskedasticity, goodness of fit) were assessed on the conditional residuals (Haslett and Haslett, 2007). 2.4.2. Statistical tests We investigated treatment effects on the mean as well as the variance of the response variables. If the treatment * horizon interaction term was significant, treatment means and variances were computed and compared separately for each horizon. The significance level of all statistical tests was set to α = 0.05. Treatment means were compared using a t-test with no provision for multiple inferences (Webster, 2007). We tested for treatment heteroskedasticity by (1) performing Levene's one-way test for unequal variances (Levene, 1960) on the absolute treatment or treatment * horizon residuals and (2) by assessing the difference in model fit between models with and without different variance parameters for each treatment. The improvement in model fit was quantified by comparing the difference in −2RLL (restricted log likelihood) to a chisquare distribution with degrees of freedom equal to the difference in the number of parameters in the covariance structure (Littell et al., 2006). We took the conservative approach to consider that treatment heteroskedasticity was present when both Levene's and the chi-square tests returned significant values. 3. Results Significant treatment mean effects were detected for total P, the N/P ratio, salt-extractable P and K, and pH measured in salt (pHCaCl2 ). Dispersion effects were observed for the C/P and N/P ratios, total Al and Fe, exchangeable Na, Cl, Mg and K, and inorganic N. 3.1. Forest floor composition Forest floors of harvested (cleared and regenerating) plots had P concentrations 40% lower than control. The C/P and N/P ratios were correspondingly wider in harvested plots, and were also found to be more variable than in control. The average Al and Fe concentration in the O layer of cleared plot was approximately five times higher in cleared plots than in control, but this difference was not statistically significant. There was however very strong statistical evidence for an increase in variance in cleared plots. Treatment effects were not detected for other elements. 3.2. Acidity and cation exchange capacity The treatment * horizon interaction was significant throughout the profile for pH, salt-extractable Al (Alexch ) and CEC e. For these variables, treatment means were thus computed and compared separately for each horizon (Table 2). In Bf horizons, pH was slightly but consistently lower in harvested plots compared to control. The difference was statistically significant for pHCaCl2 but was small overall (0.2 to 0.3 pH units). Treatment effects were not detected for other variables. 3.3. Base cations and chloride Significant mean effects were observed for exchangeable K (Kexch). In the forest floor, Kexch concentration was lower in cleared and regenerating plots than in control (Table 3). The proportion of K on exchange complexes was also lower in logged plots (p = 0.04, data not

S. Grand et al. / Catena 122 (2014) 18–26

shown). In illuvial horizons, Kexch was higher in cleared plots than in control and regenerating plots. This trend was also reflected in the proportion of K on exchange complexes (p = 0.01, data now shown). Finally, the variance of Kexch was higher in illuvial horizons of cleared plots compared to control and regenerating plots (Table 3). Other base cations did not show any significant difference in mean values between treatments, but both Naexch and Mgexch showed significantly higher variances in the Ae horizon of regenerating plots. The variance of Caexch in the Ae horizon of regenerating plot was also higher than in control and cleared plots and just missed our criteria for statistical significance. Finally, Cl showed a higher variance in the forest floor of cleared plots compared to control or regenerating plots (Table 3). 3.4. Salt-extractable P and N In the forest floor, salt-extractable P (Pexch) was lower in harvested than in control plots (Table 3). Contrary to Pexch and all other saltextractable ions, and even though the criterion for statistical signifi+ cance was not met, inorganic N (NO− 3 and NH4 ) was higher in cleared plots than in control plots, which were similar to regenerating plots (Table 3). The difference was approximately 10-fold in the O and E layers, and 3 to 4-fold in the illuvial horizons. There was extremely strong evidence of heteroskedasticity, with a dramatic variance increase in cleared plots (Table 3). 4. Discussion 4.1. Experimental design 4.1.1. The chronosequence approach This study investigated forest harvest effects using the wellestablished disturbance chronosequence or space-for-time substitution approach. The first important assumption of chronosequence studies is that biotic and biotic conditions have remained temporally constant over the span of the study (Johnson and Miyanishi, 2008). Global environmental change may be a concern if it interacts with the effects of disturbance, but this interaction effect is likely to be small in short to medium-term studies. This may be particularly true for soil biogeochemical evolution, which is more likely to follow a predictable temporal trend than more idiosyncratic ecosystem characteristics such as species composition and abundance (Walker et al., 2010). Another necessary condition for chronosequence validity is that the only difference between sites should be their disturbance regime, and that all other site properties should be spatially homogeneous (Dyck and Cole, 1994). In Roberts Creek the maximum distance between sampling sites was 4 km. Vegetation, slope and aspect were similar throughout the study forest and we observed no altitudinal, longitudinal or latitudinal gradients in any of the soil properties measured (test results not shown). Since all sampling sites exhibited reasonable similarity in properties that are likely to affect the response variables, the chronosequence is expected to allow medium-term changes in soil fertility to be examined in a short time frame (Pennock and van Kessel, 1997). Finally, the statistical power of chronosequence designs is limited by the error term introduced in the experiment by spatial variability (Yanai et al., 2003). In Roberts Creek, within-plot variability was particularly high in the forest floor. The degree of replication was also low due to limitations in sampling and analytical resources as well as in the number of suitable study sites. This resulted in a high probability for type II error (false negative) (Eberhardt and Thomas, 1991). This means that only large treatment effects could be detected, commonly in the range of 35%–150% change. 4.1.2. Detecting dispersion effects In addition to changes in nutrient mean values, this study also reports dispersion effects (differences in variance between treatments). This is a less common focus, but we argue that including dispersion

21

effects in analyses of forest harvest impacts is essential to understand ecosystem disturbance and regeneration dynamics (Grossmann and Mladenoff, 2008). Increases in variance after disturbance can point to an increase in small to medium scale heterogeneity in the landscape, which has implications for nutrient leaching risk and regenerating forest nutrition. Statistical tools appropriate for the detection of dispersion effects are not presently as developed as tests for mean treatment effects. We used two different approaches to test for dispersion effects. The first method was based on Levene's test for homogeneity of variances (Levene, 1960). This is equivalent to a standard one-way ANOVA F-test performed on a dispersion variable (here the absolute difference between the observation and the treatment mean). The second method was based on a chi-square test for improvement in fit between models with and without different variance parameters for each treatment. Both measures of dispersion effects have significant shortcomings. Levene's test for unequal variances is only available for one-way models; random effects cannot be factored in the test. The chi-square test is on the other hand well-suited to evaluate residuals of complex mixed models. However, the chi-square test returns the probability of improved model fit by using different treatment variances, which is different from assessing whether treatment variances are significantly different. These two assessments are likely to be strongly related, but should not be confused. In the absence of algorithms allowing for Levene-type tests (F-tests) based on residuals of complex models, we are limited in our assessments of dispersion effects for datasets with low withinblock replication, such as those resulting from many ecological and forestry experiments. Using Levene's test on a simplified model in conjunction with a chi-square test for goodness of fit on the full model is however likely to be a robust approach for detecting unequal variances, particularly when adopting the conservative approach to consider that treatment heteroskedasticity is present when both Levene's and the chi-square tests returned significant values. 4.2. Impacts of forest harvest 4.2.1. Forest floor dynamics In temperate forests, the forest floor acts as a major reservoir of both rapidly and slowly available nutrients (Yanai, 1998). Monitoring changes in forest floor composition is thus one of the keys to understanding the evolution of other nutrient pools in the soil (Johnson et al., 1985). In Roberts Creek, the total P concentration was 40% lower in the forest floor of harvested plots than in control plots (Table 1). Salt-extractable P was also lower in harvested than in control plots (Table 3). These lower P concentrations could not be attributed to the concomitant decrease in C concentration (Table 1), itself likely linked to forest floor disturbance and admixing of mineral soil (Grand and Lavkulich, 2012), since the C/P and N/P ratios also significantly widened. The data instead suggest that the forest floor of harvested plots was depleted in P relative to C and N. Possible causes include leaching losses of labile P species, P translocation into decomposing coarse woody debris and stumps by fungi (Palviainen et al., 2010) and/or a dilution effect due to inputs of low-P residues. The thickness of logging debris commonly reached 20 to 40 cm above the forest floor and the Oe + Oa layer thickness increased from 6 to 11 cm between control and harvested plots, most likely because of the gradual incorporation of slash into the forest floor (Grand and Lavkulich, 2012). This supports the idea of P pool dilution and P exports to decomposing woody debris, which can lead to an effective decrease in P availability at least in the medium term. The other statistically significant treatment effect observed in the forest floor was a higher variability in the Al and Fe concentration of harvested plots. Al and Fe are not generally thought to be controlled by biocycling; hence the most likely explanation for changes in Al and Fe distribution is forest floor disturbance and uneven mineral soil admixing during harvest operations.

22

S. Grand et al. / Catena 122 (2014) 18–26

Table 1 Effects of forest harvest on forest floor composition measured by a total elemental analyzer (C and N) and Parkinson and Allen digestion (all other elements). Variable

Control

Cleared

Regenerating

(Mean ± SEM) C (%) N (%) P (%) C/P N/P Ca (g/kg) Mg (g/kg) K (g/kg) Na (g/kg) Al (g/kg) Fe (g/kg)

38.1 1.06 0.099 a 391 10.9 a 7.79 0.30 1.41 0.75 0.36 0.52

± ± ± ± ± ± ± ± ± ± ±

2.13 0.07 0.005 18 d 0.68 d 0.59 0.07 0.13 0.15 0.08 d 0.16 d

30.7 ± 0.82 ± 0.059 b ± 659 ± 16.9 b ± 6.65 ± 0.37 ± 1.06 ± 1.29 ± 2.08 ± 2.74 ±

4.72 0.12 0.011 173 e 3.07 e 0.59 0.07 0.13 0.41 0.97 e 1.19 e

29.2 1.03 0.060 b 497 17.8 b 7.36 0.28 0.96 1.21 0.55 0.91

± ± ± ± ± ± ± ± ± ± ±

3.65 0.14 0.006 63 de 2.69 e 0.63 0.08 0.14 0.21 0.11 d 0.17 d

Mean effects,

Dispersion effects

F-test

χ2 Test

Levene's test

0.12 0.13 0.003 0.16 0.05 0.42 0.68 0.10 0.27 0.18 0.11

0.14 0.21 0.86 b0.0001 0.001 0.78 0.13 0.82 0.86 b0.0001 b0.0001

0.03 0.44 0.34 0.04 0.04 0.88 0.34 0.40 0.11 0.04 0.02

Means are given ± the standard error of the mean (SEM). The F-test documents treatment effects on variable means; means followed by different letters are significantly different at the α = 0.05 level (bold p-values). The χ2 and Levene's tests document treatment effects on variances; standard errors followed by different letters are significantly different at the α = 0.05 level (bold p-values).

4.2.2. Soil pH SoilpHCaCl2 was slightly lower in the first podzolic horizon of harvested plots compared to control (Table 2), but there was no corresponding pH change in overlying horizons. This pattern is consistent with the hypothesis of a disruption of the N cycle. Nitrification may be more active in harvested plots due to increased soil temperature and moisture, reduced plant uptake and increase in nitrifiers population (Smith et al., 1968). In the O layer, hydrolysis of organic N consumes protons, which can counter the acidifying effects of nitrification (Qing-ru et al., 2006) and result in an approximately constant pH. Some of the ammonium ions released from the top part of the profile may be leached to deeper horizons and undergo nitrification, which results in a net release of protons. Overall, the pH decrease was small and was not accompanied by an increase in Alexch or a decrease in base saturation. This could be due to the low rate of acid deposition in the area. Most studies that have reported strong acidification following clear felling were conducted on soils that were had been strongly affected by acid precipitation (Dahlgren and Driscoll, 1994; Dise and Gundersen, 2004; Johnson et al., 1991; Likens et al., 1970; McHale et al., 2007; Snyder and Harter, 1985). Acid deposition leads to the accumulation of N and S in organic matter, which causes a large release of acidity upon mineralization (Devries and Breeuwsma, 1987). Roberts Creek soils are likely to have

a larger buffering capacity than these acidified soils, thereby potentially reducing the observed effects of forest harvest. 4.2.3. Exchangeable cations and chloride The CECe of harvested plots was either similar to, or higher than the CECe of control plots. This was most likely due to the good retention of organic matter in the profile after harvest (Grand and Lavkulich, 2012), since most of the cation exchange sites are provided by organic matter in these acid, coarse-textured soils (Grand and Lavkulich, 2013; Johnson, 2002; Ross et al., 2008). More intensive forestry practices that do not retain as much organic matter on site, such as whole-tree harvest for the purpose of bioenergy generation, would most likely have very different consequences on nutrient cation retention in this ecosystem. In the forest floor, Kexch was lower in cleared plots than in control, whereas the reverse was observed in illuvial horizons (Table 3). A previous study of Roberts Creek stream chemistry (Hiebler-Chariarse, 2003) provided data supporting a large K+ increase in harvested reaches of Roberts Creek's streams. This suggests that clear felling resulted in K mobilization from Roberts Creek soils, with some K exported to drainage waters and some retained on B-horizon exchange complexes. Potential K sources include forest floor leachates, since K readily leaches out of fresh organic material (Carlisle et al., 1967), and

Table 2 Effects of forest harvest on soil pH, Alexch and CECe. Variable

Horizon

Control

Cleared

Regenerating

Mean ± SEM pHH2 O

pHCaCl2

Alexch (mg/kg)

CECe (cmolc/kg)

Oe + E Bs1 Bs2 BCg Oe + E Bs1 Bs2 BCg Oe + E Bs1 Bs2 BCg Oe + E Bs1 Bs2 BCg

Oa

Oa

Oa

Oa

4.57 4.57 5.37 5.60 5.51 3.56 3.59 4.54 a 4.75 4.71 313 272 87.4 53.5 51.3 37.6 4.54 2.06 1.34 0.99

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

0.15 0.13 0.06 0.08 0.07 0.13 0.12 0.09 0.07 0.07 143 72.2 21.11 13.8 21.9 4.18 0.31 0.34 0.25 0.18

4.71 ± 4.64 ± 5.31 ± 5.46 ± 5.58 ± 3.72 ± 3.66 ± 4.34 b ± 4.60 ± 4.65 ± 286 ± 285 ± 125 ± 48.9 ± 46.6 ± 30.8 ± 4.98 ± 2.59 ± 1.11 ± 0.98 ±

0.16 0.13 0.09 0.05 0.06 0.14 0.10 0.06 0.06 0.04 84.3 61.1 19.1 12.1 7.73 4.00 0.86 0.29 0.21 0.15

4.31 ± 4.60 ± 5.16 ± 5.41 ± 5.48 ± 3.48 ± 3.65 ± 4.27 b ± 4.48 ± 4.61 ± 221 ± 163 ± 167 ± 59.0 ± 35.6 ± 34.9 ± 3.77 ± 3.06 ± 1.27 ± 0.85 ±

0.18 0.14 0.09 0.04 0.08 0.15 0.12 0.04 0.08 0.04 46.6 71.4 23.9 14.5 8.10 4.56 0.47 0.34 0.25 0.18

Mean effects,

Dispersion effects

F-test

χ2 Test

Levene's test

0.27 0.92 0.08 0.12 0.58 0.49 0.89 0.04 0.10 0.55 0.72 0.43 0.06 0.87 0.64 0.54 0.43 0.15 0.78 0.83

0.95 0.70 0.90 0.17 0.61 0.58 0.55 0.09 0.25 0.23 0.03 0.21 0.08 0.45 0.61 0.95 0.12 0.21 0.20 0.86

0.80 0.62 0.93 0.27 0.56 0.31 0.33 0.38 0.37 0.21 0.40 0.06 0.40 0.67 0.60 0.98 0.04 0.23 0.43 0.74

The treatment * horizon interaction was significant throughout the profile and treatment means were computed and compared separately for each horizon. Means are given ± the standard error of the mean (SEM). The F-test documents treatment effects on variable means; means followed by different letters are significantly different at the α = 0.05 level (bold p-values). The χ2 and Levene's tests document treatment effects on variances; standard errors followed by different letters are significantly different at the α = 0.05 level (bold p-values).

S. Grand et al. / Catena 122 (2014) 18–26

23

Table 3 Effects of forest harvest on salt-extractable cations and anions (Mexch). Variable

Horizon

Control

Cleared

Regenerating

Mean ± SEM Caexch (mg/kg)

Mgexch (mg/kg)

Kexch (mg/kg)

Naexch (mg/kg)

NH4exch (mg NH3-N/kg)

NO3exch (mg NO3-N/kg)

Pexch (mg tot P/kg)

Clexch (mg/kg)

Oe + Oa E illuvial Oe + Oa E illuvial Oe + Oa E illuvial Oe + Oa E illuvial Oe + Oa E illuvial Oe + Oa E illuvial Oe + Oa E illuvial Oe + Oa E illuvial

5590 ± 165 ± 104 ± 343 ± 15.1 ± 7.04 ± 761 a ± 26.3 ± 19.1 a ± 105 ± 11.4 ± 9.75 ± 6.87 ± 0.50 ± 0.58 ± 1.93 ± 0.20 ± 0.48 ± 174 a ± 3.53 ± 0.82 ± 251 ± 24.7 ± 21.0 ±

773 26.7 28.0 42.1 2.09 d 1.43 69.9 8.61 2.24 d 12.7 1.16 d 1.65 0.75 0.30 d 0.33 d 1.20 d 0.13 d 0.13 d 26.3 1.18 0.07 29.3 d 3.17 2.51

4594 ± 209 ± 113 ± 298 ± 15.8 ± 6.53 ± 384 b ± 22.9 ± 28.7 b ± 72.5 ± 9.95 ± 7.20 ± 66.9 ± 6.11 ± 2.08 ± 16.9 ± 1.73 ± 1.66 ± 74.2 b ± 2.77 ± 0.80 ± 219 ± 16.9 ± 21.8 ±

718 27.4 18.2 42.1 2.20 d 0.55 69.9 3.20 3.26 e 12.3 0.76 d 0.45 37.4 2.67 e 0.83 e 14.1 e 0.71 e 0.53 e 26.3 0.38 0.06 53.0 e 3.55 2.78

5172 ± 1063 246 ± 62.4 109 ± 50.7 334 ± 45.5 20.8 ± 5.06 e 7.25 ± 1.72 393 b ± 75.5 20.4 ± 2.91 15.7 a ± 2.10 d 122 ± 14.1 16.0 ± 3.22 e 9.27 ± 0.60 13.2 ± 1.69 0.51 ± 0.30 d 0.74 ± 0.11 d 2.82 ± 0.96 d 0.35 ± 0.19 d 0.48 ± 0.13 d 98.4 b ± 28.4 2.47 ± 0.46 0.68 ± 0.07 159 ± 15.9 d 20.6 ± 2.94 18.2 ± 0.94

Mean effects,

Dispersion effects

F-test

χ2 Test

Levene's test

0.66 0.39 0.97 0.72 0.63 0.89 0.002 0.78 0.02 0.09 0.20 0.14 0.19 0.17 0.12 0.58 0.15 0.13 0.04 0.73 0.31 0.26 0.32 0.37

0.25 0.07 0.27 0.74 0.01 0.11 0.39 0.27 0.03 0.86 0.001 0.15 0.007 0.05 0.01 b0.0001 b0.0001 b0.0001 0.39 0.55 0.49 0.02 0.32 0.002

0.77 0.04 0.22 0.76 0.01 0.11 0.24 0.09 0.04 0.64 0.02 0.14 0.07 0.04 0.003 0.04 0.02 0.001 0.39 0.28 0.68 0.0008 0.19 0.14

There was no treatment * horizon interaction in the Bs–BC horizons, which were grouped together under the name ‘illuvial'. Means are given ± the standard error of the mean (SEM). The F-test documents treatment effects on variable means; means followed by different letters are significantly different at the α = 0.05 level (bold p-values). The χ2 and Levene's tests document treatment effects on variances; standard errors followed by different letters are significantly different at the α = 0.05 level (bold p-values).

accelerated mineral weathering after clear-cutting (Bormann and Likens, 1979). In regenerating plots, the resumption of vegetation uptake may contribute to lower K concentration. Other studies have similarly reported strong K retention in the solum (Dahlgren and Driscoll, 1994; Snyder and Harter, 1985). It has been shown that some subsoil horizons can preferentially retain K (Ludwig et al., 2001), likely due to selective sorption onto amorphous and free Fe and Al phases (Chu and Johnson, 1979; Snyder and Harter, 1985; Van Reeuwijk and de Villiers, 1968). Reactive Al and Fe phases are abundant in Roberts Creek's subsoil (Grand and Lavkulich, 2008) and were positively correlated with Kexch concentration (r = 0.47, p = 0.02 for short-range order Al phases, represented by oxalateextractable Al, and r = 0.88, p b 0.01 for free oxides, represented by dithionite-extractable Al and Fe). This suggests that poorly crystalline and free Al and Fe phases likely contributed to K retention in the profile. 4.2.4. Inorganic nitrogen Nitrogen availability in Roberts Creek forested plots was moderate to low, as is the case in most western Canadian forest soils (Gessel et al., 1973). In the mineral soil of control plots, inorganic N concentrations were extremely low, around 0.5 mg/kg for both NO3-N and NH4-N. Inorganic N concentrations were non-negligible in the forest floor (Table 3), indicating that the organic layer likely played a large role in plant N nutrition. Ammonium-N was more abundant than nitrate-N, suggesting that NO− 3 is subject to preferential biological uptake and/or leaching, or that nitrifiers' activity is limited. We previously calculated that the mineral soil contained on average 0.55 kg/m2 of total N (Grand and Lavkulich, 2011), while the lower limit proposed by Gessel et al. (1973) for adequate Douglas fir forest N supply is 0.50 kg N/m2. Therefore even small N losses after harvest could negatively impact forest productivity. Hudson and Tolland (2002) presented evidence that Roberts Creek drainage waters could register a large increase in NO3 after harvest. In one of the studied streams, the authors measured a 4, 30 and 60-fold increase in nitrate level 1, 2 and 3 years after logging, exceeding responses measured in other watersheds in the region. In the second surveyed stream, changes in nitrate concentration were more moderate. The

NO3 levels were unchanged in year 1, and then increased 12- and 18-folds in years 2 and 3 after logging. These water chemistry results are consistent with our soil data, in which cleared plots had extreme NO3 values and a significant increase in variability. This suggests that N leaching risk is likely to be inconsistent from plot to plot. Initial N availability can be an important factor determining NO3 production and losses after disturbance (Vitousek et al., 1982). Fig. 1 shows NO3 concentration in the mineral soil as a function of the C/N ratio, chosen as an indicator of N availability. Increases in NO3exch above 2 mg N/kg all occurred in plots with a C/N ratio of 25 or lower. This is consistent with the idea that net nitrification is generally inversely related to the C/N ratio, with a C/N ratio of 25–30 considered to be a threshold below which net nitrification and nitrate leaching may take place (Gundersen and Rasmussen, 1990; Gundersen et al., 1998). These data suggest that increased nitrification at favorable sites played a role in the occurrence of observed high NO3exch values. Soil moisture was higher in harvested plots compared to control (Grand and Lavkulich, 2012) and may have contributed to the presence of elevated NO3exch values at cleared sites through the alleviation of diffusional and physiologic limitations to nitrifiers (Stark and Firestone, 1995). This effect is likely to be most evident during the dry summer period; the observed pattern in inorganic N distribution is thus likely to vary seasonally. Elevated values of NO3exch over 2 mg N/kg all occurred in plots logged 1 to 3 years prior to sampling. The timing of the observed increase in N availability (1–3 years) is in good agreement with previous studies (Dahlgren and Driscoll, 1994; e.g. Hendrickson et al., 1989; Hudson and Tolland, 2002; Kranabetter et al., 2006). In 5 years old plots and regenerating plots 8 to 15 years old, NO3exch was similar to control levels. This is likely due to the establishment of tree seedlings and competing vegetation. Early seral vegetation probably played a prevalent role in 5 year old plots, where the establishment of tree seedlings was not pervasive. Early seral vegetation dynamics has previously been invoked to explain differences in soil inorganic N levels after clearcutting (Grenon et al., 2004; Strahm et al., 2005) and is likely to play a major role in reducing N availability and N leaching losses after disturbance (Slesak et al., 2010). From a long-term stand nutrition and

24

S. Grand et al. / Catena 122 (2014) 18–26

Fig. 1. Concentration of salt-extractable nitrate (NO3exch) as a function of the carbon to nitrogen mass ratio (C/N) in the mineral soil of control, cleared and regenerating forest plots.

water quality protection standpoint, suppression of competing vegetation in regenerating plots may thus be an objectionable practice. 4.2.5. Implications Overall, this study showed that the effects of forest harvest on Roberts Creek's soil chemistry were not severe, but the elements showing the strongest treatment effect were also the ones likely to limit ecosystem productivity (N and P) (Gessel et al., 1973; Weetman et al., 1997). The lower P concentration observed in the forest floor of harvested plots may adversely affect regenerating vegetation nutrition, as the organic layer is known to supply a large portion of P requirements to humid conifer forests (Ballard, 1980). There was a dramatic increase in inorganic N variability after harvest. This indicates that average soil concentrations cannot be assumed to be representative of micro to mesoscale field conditions such as N availability and leaching potential. This also interferes with the power of common statistical tests that assume homogeneous variances across treatments. Average NO3exch concentrations remained modest (b1.8 mg N/kg in the mineral soil, b 18 mg N/kg in the forest floor), but some of the mineral samples from harvested plots exceeded 10 mg N/kg and some of the organic samples exceeded 110 mg N/kg. Taken together with studies of drainage water chemistry showing that stream NO3 concentration increased after logging in Roberts Creek (Hiebler-Chariarse, 2003; Hudson and Tolland, 2002), these data suggest that some areas of cleared plots are actively producing and exporting NO3. 5. Conclusion This study measured differences in the distribution and concentration of labile and exchangeable ions in soils of undisturbed and clear-cut forest plots. Differences in soil chemistry overall were not dramatic. Acidification was detected in the subsoil of cleared plots but was mild, and in most cases was not accompanied by an increase in exchangeable Al or a decrease in base saturation. This was likely due to the absence of prior soil acidification, suggesting that ecosystem response to disturbance is likely to be strongly influenced by anthropogenic impact history. Compared to control, harvested plots recorded differences in macronutrients concentration and distribution including K, P and N. Potassium was leached from the forest floor but retained in illuvial horizons, likely

due to interactions with soil's reactive Al and Fe phases. Phosphorus concentration was lower in the forest floor of harvested plots, raising concerns for regenerating vegetation nutrition, especially given the low P availability in the mineral soil. Differences in average inorganic N concentrations were relatively moderate but there was a dramatic increase in variability in cut plots, with some of the samples from cleared sites recording extreme inorganic N values. This suggests that mineralization and nitrification processes are locally very active. Taken together with the results of local drainage water composition studies, these results indicate that these coniferous forests produce and export a significant amount of NO3 following logging. Any N losses are a concern for long-term forest productivity, given the N-poor status of the soils. The variance of several elements increased in harvested plots, including N, K and P stochiometry. This suggests that the logging disturbance resulted in a disruption of steady state nutrient cycling (Marks and Bormann, 1972). This post-harvest increase in small to medium scale heterogeneity has implications for the nutrition of regenerating vegetation, nutrient leaching potential as well as our ability to detect harvest-induced changes and predict them. We recommend that future studies of the impacts related to forest management activities include an analysis of treatment dispersion as well as mean effects. Acknowledgments This study was funded through a Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant. We thank Dr. Hans Schreier and an anonymous reviewer for their comments which led to the improvement of this manuscript. We also thank field assistants Bryan Forrest, Marina Romeo and Peter Shanahan, laboratory manager Carol Dyck and laboratory technician Keren Ferguson for their assistance with field sampling and analyses. References Ballard, R., 1980. Phosphorus nutrition and fertilization of forest trees. In: Khasawnek, F.E. (Ed.), The Role of Phosphorus in Agriculture. ASA, CSSA and SSSA, Madison, WI, pp. 763–804. Binkley, D., Brown, T.C., 1993. Forest practices as nonpoint sources of pollution in North America. J. Am. Water Resour. Assoc. 29, 729–740. Bormann, F.H., Likens, G.E., 1979. Reorganization: loss of biotic regulation. Anonymous Pattern and Process in a Forested Ecosystem: Disturbance, Development and the Steady State Based on the Hubbard Brook Ecosystem Study. Springer-Verlag, New York, pp. 81–102.

S. Grand et al. / Catena 122 (2014) 18–26 Bradley, R.L., Kimmins, J.P., Martin, W.L., 2001. Post-clearcutting chronosequence in the B.C. Coastal Western Hemlock Zone—II. Tracking the assart flush. J. Sustain. For. 14, 23–43. Burns, D., Murdoch, P., 2005. Effects of a clearcut on the net rates of nitrification and N mineralization in a northern hardwood forest, Catskill Mountains, New York, USA. Biogeochemistry 72, 123–146. Cade-Menun, B.J., Lavkulich, L.M., 1997. A comparison of methods to determine total, organic, and available phosphorus in forest soils. Commun. Soil Sci. Plant Anal. 28, 651–663. Carlisle, A., Brown, A.H.F., White, E., 1967. The nutrient content of tree stem flow and ground flora litter and leachates in a sessile oak (Quercus petraea) woodland. J. Ecol. 55, 615–627. Chu, C.H., Johnson, L.G., 1979. Cation-exchange behavior of clays and synthetic aluminosilica gels. Clays Clay Minerals 27, 87–90. Conyers, M.K., Uren, N.C., Helyar, K.R., 1995. Causes of changes in pH in acidic mineral soils. Soil Biol. Biochem. 27, 1383–1392. Dahlgren, R.A., Driscoll, C.T., 1994. The effects of whole-tree clear-cutting on soil processes at the Hubbard Brook Experimental Forest, New Hampshire, USA. Plant Soil 158, 239–262. D'Anjou, B., 2002. Roberts Creek Study Forest: harvesting, windthrow and conifer regeneration within alternative silvicultural systems in Douglas-fir dominated forests on the Sunshine Coast. Forest Research Technical ReportResearch Section, Vancouver Forest Region, B.C. Ministry of Forestry, Nanaimo, pp. 1–17 (TR-018). Devries, W., Breeuwsma, A., 1987. The relation between soil acidification and element cycling. Water Air Soil Pollut. 35, 293–310. Dise, N.B., Gundersen, P., 2004. Forest ecosystem responses to atmospheric pollution: linking comparative with experimental studies. Water Air Soil Pollut. 4, 207–220. Dyck, W.J., Cole, D.W., 1994. Strategies for determining consequences of harvesting and associated practices on long-term productivity. In: Dyck, W.J., Cole, D.W., Comerford, N.B. (Eds.), Impacts of Forest Harvesting on Long-Term Site Productivity. Chapman & Hall, London, UK, pp. 13–40. Eberhardt, L.L., Thomas, J.M., 1991. Designing environmental field studies. Ecol. Monogr. 61, 53–73. Environment Canada, 2013. Canadian climate normals 1981–2010. http://climate. weatheroffice.gc.ca/climate_normals/index_e.html (Accessed August 9, 2013). Fons, J., Klinka, K., 1998. Temporal variations of forest floor properties in the Coastal Western Hemlock zone of southern British Columbia. Can. J. For. Res. 28, 582–590. Gessel, S.P., Cole, D.W., Steinbrenner, E.C., 1973. Nitrogen balances in forest ecosystems of the Pacific Northwest. Soil Biol. Biochem. 5, 19–34. Grand, S., Lavkulich, L.M., 2008. Reactive soil components and pedogenesis of highly productive coastal Podzols. Geochim. Cosmochim. Acta (Suppl. 72), A323 (Goldschmidt Abstracts). Grand, S., Lavkulich, L.M., 2011. Depth distribution and predictors of soil organic carbon in Podzols of a forested watershed in southwestern Canada. Soil Sci. 176, 164–174. Grand, S., Lavkulich, L.M., 2012. Effects of forest harvest on soil carbon and related variables in Canadian Spodosols. Soil Sci. Soc. Am. J. 76, 1816–1827. Grand, S., Lavkulich, L.M., 2013. Potential influence of poorly crystalline minerals on soil chemistry in Podzols of southwestern Canada. Eur. J. Soil Sci. 64, 651–660. Green, R.N., Klinka, K., 1994. A field guide to site identification and interpretation for the Vancouver Forest Region. B.C. Ministry of Forests, Victoria, B.C. Land Management Handbook No. 28. Grenon, F., Bradley, R.L., Joanisse, G., Titus, B.D., Prescott, C.E., 2004. Mineral N availability for conifer growth following clearcutting: responsive versus non-responsive ecosystems. For. Ecol. Manag. 188, 305–316. Grossmann, E.B., Mladenoff, D.J., 2008. Farms, fires, and forestry: disturbance legacies in the soils of the Northwest Wisconsin (USA) Sand Plain. For. Ecol. Manag. 256, 827–836. Gundersen, P., Rasmussen, L., 1990. Nitrification in forest soils: effects from nitrogen deposition on soil acidification and aluminum release. Rev. Environ. Contam. Toxicol. 113, 1–45. Gundersen, P., Callesen, I., de Vries, W., 1998. Nitrate leaching in forest ecosystems is related to forest floor C/N ratios. Environ. Pollut. 102, 403–407. Hach Company, 2007. Chloride. Method 8113: Mercuric thiocyanate method. Hach Company world headquarters, USA. DOC316.53.01017, pp. 1–6. Hach Company, 2009. Lachat intruments. www.lachatinstruments.com (Accessed December 21, 2013). Haslett, J., Haslett, S.J., 2007. The three basic types of residuals for a linear model. Int. Stat. Rev. 75, 1–24. Hendrickson, O.Q., Chatarpaul, L., Burgess, D., 1989. Nutrient cycling following whole-tree and conventional harvest in northern mixed forest. Can. J. For. Res. 19, 725–735. Hiebler-Chariarse, J.A., 2003. Effect of Forest Harvesting on Biological Processes in Headwater Streams Within the Flume Creek Experimental Watershed, British Columbia. M.Sc. thesis University of British Columbia, Vancouver. Hudson, R., 2001. Roberts Creek Study Forest: preliminary effects of partial harvesting on peak streamflow in two S6 creeks. Forest Research Technical ReportResearch Research Section, Vancouver Forest Region, B.C. Ministry of Forestry, Nanaimo, pp. 1–9 (EN-007). Hudson, R., Tolland, L., 2002. Roberts Creek Study Forest: effects of partial retention harvesting on nitrate concentrations in two S6 creeks three years after harvesting. Forest Research Technical ReportResearch Section, Vancouver Forest Region, B.C. Ministry of Forestry, Nanaimo, pp. 1–20 (TR-019). ISO standards, 1996. Determination of nitrite nitrogen and nitrate nitrogen and the sum of both by flow analysis and spectrophotometric detection. Water QualityISO central office, Geneva, Switzerland, pp. 1–18 (ISO 13395). Johnson, C.E., 1989. Effects of whole-tree harvesting on chemical weathering rates in a small forested catchment. Trans. Am. Geophys. Union 70, 1115.

25

Johnson, C.E., 2002. Cation exchange properties of acid forest soils of the northeastern USA. Eur. J. Soil Sci. 53, 271–282. Johnson, E.A., Miyanishi, K., 2008. Testing the assumptions of chronosequences in succession. Ecol. Lett. 11, 419–431. Johnson, J.E., Smith, D.W., Burger, J.A., 1985. Effects on the forest floor of whole-tree harvesting in an Appalachian oak forest. Am. Midl. Nat. 114, 51–61. Johnson, C.E., Johnson, A.H., Siccama, T.G., 1991. Whole-tree clear-cutting effects on exchangeable cations and soil acidity. Soil Sci. Soc. Am. J. 55, 502–508. Johnson, C.E., Romanowicz, R.B., Siccama, T.G., 1997. Conservation of exchangeable cations after clear-cutting of a northern hardwood forest. Can. J. For. Res. 27, 859–868. Katou, H., Clothier, B.E., Green, S.R., 1996. Anion transport involving competitive adsorption during transient water flow in an andisol. Soil Sci. Soc. Am. J. 60, 1368–1375. Kenward, M.G., Roger, J.H., 1997. Small sample inference for fixed effects from restricted maximum likelihood. Biometrics 53, 983–997. Klopatek, J.M., Barry, M.J., Johnson, D.W., 2006. Potential canopy interception of nitrogen in the Pacific Northwest, USA. For. Ecol. Manag. 234, 344–354. Kranabetter, J.M., Sanborn, P., Chapman, B.K., Dube, S., 2006. The contrasting response to soil disturbance between Lodgepole Pine and Hybrid White Spruce in subboreal forests. Soil Sci. Soc. Am. J. 70, 1591–1599. Lamontagne, S., Carignan, R., D'Arcy, P., Prairie, Y.T., Pare, D., 2000. Element export in runoff from eastern Canadian Boreal Shield drainage basins following forest harvesting and wildfires. Can. J. Fish. Aquat. Sci. 57, 118–128. Lawrence, G.B., Fuller, R.D., Driscoll, C.T., 1987. Release of aluminum following whole-tree harvesting at the Hubbard-Brook-Experimental-Forest, New-Hampshire. J. Environ. Qual. 16, 383–390. Levene, H., 1960. Robust tests for the equality of variance. In: Olkin, I. (Ed.), Contributions to Probability and Statistics. Stanford University Press, Palo Alto, CA, pp. 278–292. Likens, G.E., Bormann, F.H., Johnson, N.M., Fisher, D.W., Pierce, R.S., 1970. Effects of forest cutting and herbicide treatment on nutrient budgets in the Hubbard Brook watershed-ecosystem. Ecol. Monogr. 40, 23–47. Likens, G.E., Bormann, F.H., Pierce, R.S., Reiners, W.A., 1978. Recovery of a deforested ecosystem. Science 199, 492–496. Littell, R.C., Milliken, G.A., Stroup, W.W., Wolfinger, R.D., Schabenberger, O., 2006. SAS for Mixed Models. SAS Institute Inc., Cary, NC. Ludwig, B., Khanna, P.K., Anurugsa, B., Folster, H., 2001. Assessment of cation and anion exchange and pH buffering in an Amazonian Ultisol. Geoderma 102, 27–40. Mann, L.K., Johnson, D.W., West, D.C., Cole, D.W., Hornbeck, J.W., Martin, C.W., Riekerk, H., Smith, C.T., Swank, W.T., Tritton, L.M., Van Lear, D.H., 1988. Effects of whole-tree and stem-only clearcutting on postharvest hydrologic losses, nutrient capital, and regrowth. For. Sci. 34, 412–428. Marks, P.L., Bormann, F.H., 1972. Revegetation following forest cutting: mechanisms for return to steady-state nutrient cycling. Science 176, 914–915. Martin, C.W., Noel, D.S., Federer, C.A., 1984. Effects of forest clearcutting in New England on stream chemistry. J. Environ. Qual. 13, 204–210. McHale, M., Burns, D., Lawrence, G., Murdoch, P., 2007. Factors controlling soil water and stream water aluminum concentrations after a clearcut in a forested watershed with calcium-poor soils. Biogeochemistry 84, 311–331. Mengel, K., 1994. Symbiotic dinitrogen fixation—its dependence on plant nutrition and its ecophysiological impact. J. Plant Nutr. Soil Sci. 157, 233–241. Mroz, G.D., Jurgensen, M.F., Frederick, D.J., 1985. Soil nutrient changes following whole tree harvesting on three Northern Hardwood sites. Soil Sci. Soc. Am. J. 49, 1552–1557. Neal, C., Reynolds, B., Smith, C.J., Hill, S., Neal, M., Conway, T., Ryland, G.P., Jeffrey, H., Robson, A.J., Fisher, R., 1992. The impact of conifer harvesting on stream water pH, alkalinity and aluminium concentrations for the British uplands: an example for an acidic and acid sensitive catchment in mid-Wales. Sci. Total Environ. 126, 75–87. Paavolainen, L., Kitunen, V., Smolander, A., 1998. Inhibition of nitrification in forest soil by monoterpenes. Plant Soil 205, 147–154. Palviainen, M., Finér, L., Laiho, R., Shorohova, E., Kapitsa, E., Vanha-Majamaa, I., 2010. Phosphorus and base cation accumulation and release patterns in decomposing Scots pine, Norway spruce and silver birch stumps. For. Ecol. Manag. 260, 1478–1489. Parkinson, J.A., Allen, S.E., 1975. A wet oxidation procedure suitable for the determination of nitrogen and mineral nutrients in biological material. Commun. Soil Sci. Plant Anal. 6, 1–11. Pennock, D.J., van Kessel, C., 1997. Clear-cut forest harvest impacts on soil quality indicators in the mixedwood forest of Saskatchewan, Canada. Geoderma 75, 13–32. Pojar, J., Klinka, K., Demarchi, D.A., 1991. Coastal western hemlock zone. In: Meidinger, D., Pojar, J. (Eds.), Ecosystems of British Columbia. B.C. Ministry of Forests, Victoria. Preston, C.M., Trofymow, J.A., Niu, J., Fyfe, C.A., 2002. Harvesting and climate effects on organic matter characteristics in British Columbia coastal forests. J. Environ. Qual. 31, 402–413. Qing-ru, Z., Bo-han, L., Li-tian, Z., Xi-hong, Z., Hong-xiao, T., 2006. Short-term alleviation of aluminum phytotoxicity by urea application in acid soils from south China. Chemosphere 63, 860–868. Ross, D.S., Matschonat, G., Skyllberg, U., 2008. Cation exchange in forest soils: the need for a new perspective. Eur. J. Soil Sci. 59, 1141–1159. SAS Institute Inc, 2010. SAS. Version 9.3. Cary, NC. Schimel, J.P., Bennett, J., 2004. Nitrogen mineralization: challenges of a changing paradigm. Ecology 85, 591–602. Schofield, R.K., Taylor, A.W., 1955. The measurement of soil pH. Soil Sci. Soc. Am. Proc. 19, 164–167. Schwarz, G., 1978. Estimating the dimension of a model. Ann. Stat. 6, 461–464. Slesak, R.A., Harrington, T.B., Schoenholtz, S.H., 2010. Soil and Douglas-fir (Pseudotsuga menziesii) foliar nitrogen responses to variable logging-debris retention and competing vegetation control in the Pacific Northwest. Can. J. For. Res. 40, 254–264. Smith, W., Bormann, F.H., Likens, G.E., 1968. Response of chemoautotrophic nitrifiers to forest cutting. Soil Sci. 106, 471–473.

26

S. Grand et al. / Catena 122 (2014) 18–26

Snyder, K.E., Harter, R.D., 1985. Changes in solum chemistry following clearcutting of northern hardwood stands. Soil Sci. Soc. Am. J. 49, 223–228. Spielvogel, S., Prietzel, J., Kögel-Knabner, I., 2006. Soil organic matter changes in a spruce ecosystem 25 years after disturbance. Soil Sci. Soc. Am. J. 70, 2130–2145. Stark, J.M., Firestone, M.K., 1995. Mechanisms for soil moisture effects on activity of nitrifying bacteria. Appl. Environ. Microbiol. 61, 218–221. Strahm, B.D., Harrison, R.B., Terry, T.A., Flaming, B.L., Licata, C.W., Petersen, K.S., 2005. Soil solution nitrogen concentrations and leaching rates as influenced by organic matter retention on a highly productive Douglas-fir site. For. Ecol. Manag. 218, 74–88. Sudduth, E.B., Perakis, S.S., Bernhardt, E.S., 2013. Nitrate in watersheds: straight from soils to streams? J. Geophys. Res. 118, 291–302. Thomas, G.W., 1982. Exchangeable cations. In: Page, A.L. (Ed.), Methods of Soil Analysis. American Society of Agronomy, Madison, WI, pp. 159–165. Van Lierop, W., 1990. Soil pH and lime requirement determination. In: Westerman, R.L. (Ed.), Soil Testing and Plant Analysis. Soil Science Society of America, Madison, Wisconsin, USA, pp. 73–92. Van Reeuwijk, L.P., de Villiers, J.M., 1968. Potassium fixation by amorphous aluminosilica gels. Soil Sci. Soc. Am. Proc. 32, 238–240. Verdouw, H., Van Echteld, C.J.A., Dekkers, E.M.J., 1978. Ammonia determination based on indophenol formation with sodium salicylate. Water Res. 12, 399–402. Vitousek, P.M., 1977. The regulation of element concentrations in mountain streams in the northeastern United States. Ecol. Monogr. 47, 65–87.

Vitousek, P.M., Gosz, J.R., Grier, C.C., Melillo, J.M., Reiners, W.A., 1982. A comparative analysis of potential nitrification and nitrate mobility in forest ecosystems. Ecol. Monogr. 52, 155–177. Walker, L.R., Wardle, D.A., Bardgett, R.D., Clarkson, B.D., 2010. The use of chronosequences in studies of ecological succession and soil development. J. Ecol. 98, 725–736. Webster, R., 2007. Analysis of variance, inference, multiple comparisons and sampling effects in soil research. Eur. J. Soil Sci. 58, 74–82. Weetman, G.F., Prescott, C.E., Kohlberger, F.L., Fournier, R.M., 1997. Ten-year growth response of coastal Douglas-fir on Vancouver Island to N and S fertilization in an optimum nutrition trial. Can. J. For. Res. 27, 1478–1482. White, C.S., 1994. Monoterpenes: their effects on ecosystem nutrient cycling. J. Chem. Ecol. 20, 1381–1406. WRB IUSS Working Group, 2006. World reference base for soil resources. World Soil Resources Reports No. 103FAO, Rome. Xu, J., Tang, C., Chen, Z., 2006. The role of plant residues in pH change of acid soils differing in initial pH. Soil Biol. Biochem. 38, 709–719. Yanai, R.D., 1998. The effect of whole-tree harvest on phosphorus cycling in a northern hardwood forest. For. Ecol. Manag. 104, 281–295. Yanai, R.D., Currie, W.S., Goodale, C.L., 2003. Soil carbon dynamics after forest harvest: an ecosystem paradigm reconsidered. Ecosystems 6, 197–212. Yuan, T.L., 1959. Determination of exchangeable hydrogen in soils by a titration method. Soil Sci. 88, 164–167.