Long term accumulation of nitrogen in soils of dry mixed eucalypt forest in the absence of fire

Long term accumulation of nitrogen in soils of dry mixed eucalypt forest in the absence of fire

Forest Ecology and Management 256 (2008) 1133–1142 Contents lists available at ScienceDirect Forest Ecology and Management journal homepage: www.els...

719KB Sizes 0 Downloads 37 Views

Forest Ecology and Management 256 (2008) 1133–1142

Contents lists available at ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Long term accumulation of nitrogen in soils of dry mixed eucalypt forest in the absence of fire John Turner a,*, Marcia Lambert a, Vic Jurskis b, Huiquan Bi c,d a

Forsci Pty Ltd., Unit 10/124, Rowe Street, Eastwood, NSW 2122, Australia Forests NSW, P.O. Box 100, Beecroft NSW 2119, Australia c Forest Resources Research, Science and Research Division, NSW Department of Primary Industries, P.O. Box 100, Beecroft, NSW 2119, Australia d School of Forest and Ecosystem Science, University of Melbourne, Victoria, Australia b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 9 May 2008 Received in revised form 6 June 2008 Accepted 9 June 2008

In the Eden area in NSW, Australia, low fertility granitic surface soils were sampled from 156 sites and analysed for pH, organic C, total N, total P, available P, exchangeable bases and exchangeable Al. Fifty eight of these sites were also sampled to a depth of 40 cm. Time since fire ranged from 1 to 39 years and was used in the analysis as a surrogate for fire frequency. No information was available on fire intensity. No significant relationships were found between time since fire and P or base cations. However, the quantities of organic matter and total N (kg ha1), and the C/N ratio were significantly related to both time since fire alone and to the combination of time since fire and soil total P. Based on these relationships, it was estimated that there were average net increases of between 11 and 21 kg N ha1 year1 in surface soil, the actual quantity depending on the level of soil total P. There was little change in N in the initial 10 years after fire and there was a peak in N accumulation about 24 years after fire. The C/N ratio and surface soil pH decreased with time since fire. Accumulation of N and reductions in pH and C/N ratio were studied further in a small scale paired plot analysis. The repeatedly burnt plots had lower levels of both litter and understorey and the overstorey trees generally had healthier crowns than in the unburnt plots. The differences between the repeatedly burnt and the unburnt plots matched the models developed from the general survey. There were no significant changes in the C/N ratio, but the unburnt sites had higher levels of extractable mineral N and the relationships between the mineral N and the C/N ratio for burnt and unburnt sites were statistically significant. The quantities of extractable mineral N in the unburnt soils (2.3 kg N ha1) were about twice the levels in the burnt soils (1.2 kg N ha1). The pH of the surface soil (4.4 in 1:1 water) in the regularly burnt area was higher than in the unburnt area (pH 4.1) and the exchangeable aluminium also differed (0.62 c mol1 in the burnt area and 1.3 c mol1 in the unburnt). The combined data indicate that changes occur in forest soils when there is a long period of exclusion of fire. It is suggested that these changes generally lead to secondary changes, such as in pH and availability of other elements such as aluminium. The study highlights a number of issues including the rates of inputs of N to the system and the question of N saturation and its long term interaction with plant species. It is hypothesised that reduced burning leads to increased N availability and other soil changes which negatively impact on tree health. ß 2008 Elsevier B.V. All rights reserved.

Keywords: Nitrogen accumulation Nutrient cycling Fire Tree health

1. Introduction Crown dieback and stand deterioration of extensive and increasing areas of Eucalyptus forests in Australia have been variously reported together with a number of hypotheses for possible causes (Jurskis and Turner, 2002; Jurskis, 2005a,b; Turner and Lambert, 2005). A fundamental hypothesis is that in some

* Corresponding author. Tel.: +61 2 9804 8292; fax: +61 2 9804 8302. E-mail address: forsci@fluoroseal.com.au (J. Turner). 0378-1127/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2008.06.021

forest types, pre-European fire regimes and some post-European fire and/or grazing regimes maintained relatively stable N cycles, whereas recent reduced disturbance or ‘‘non-intervention’’ in many forests has initiated progressive changes in nutrient cycling, primarily N cycling, which modify some nutrient-related processes leading to chronic decline in trees. That is, extended periods of minimal disturbance are associated with N accumulation which can lead to high mineral N production and this, either directly or indirectly, causes stress within trees. The stresses include direct affects due to tree roots with high levels of elements such as aluminium and manganese or the effects of nutritional imbalances

1134

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

on tree biochemistry, such as increased levels of some amino acids (for example, arginine). The stresses cause trees to be more predisposed to attack by insects, disease and folivores leading to crown dieback and mortality (Jurskis and Turner, 2002; Jurskis, 2005a,b; Turner and Lambert, 2005). The converse is that disturbance leading either to reduction of N inputs (for example, by reduction of N-fixing plants) and/or direct loss of N due to fire, reduces levels of mineral N and other elements and maintains tree health. This recognises that in a number of forest ecosystems, such as wet sclerophyll forests with good soil structure and relatively high fertility, decline is less of an issue (for example, some forests of Eucalyptus regnans in Victoria or Eucalyptus delegatensis in southern NSW). This hypothesis is compatible with some of those developed for N saturation and decline in Northern Hemisphere forests; however, in those situations there is the added impact of high anthropogenic atmospheric N inputs (Aber, 1992; Fenn et al., 2006). Aber (1992) reviewed the effects of increasing N within an ecosystem identifying a series of key parameters including an increased nitrate component of N cycled by plants, a declining fraction of mycorrhizal fungi, increasing nitrate export into streams and increases in free amino acids in tree foliage. Tree decline and mortality as a result of adding N have been reported from several long term ecological research areas in North America (e.g. Magill et al., 2004; Wallace et al., 2007). The hypothesis outlined above proposes a number of interconnected stages or processes which may be evaluated individually. Basic to the hypothesis is the accumulation of N in the absence of disturbance and an increase in mineral N. Nitrogen from the atmosphere is fixed in soils by various organisms but the actual rates of fixation in Australian east coast forests have been quantified in only a few cases. Adams and Attiwill (1984a,b) provided tentative estimates of N fixation as 15– 20 kg N ha1 year1, using mass balance (budget approach) in E. regnans forests and 12–32 kg N ha1 year1 using acetylene reduction. A number of field experiments involving periodic low intensity burning (termed ‘‘burning studies’’) have been established and soils have been monitored but most of these studies have either been of short duration or limited in replication, and have been compromised by over-riding disturbance such as wildfire or have not been monitored and reported (Van Loon, 1969a,b; Tolhurst et al., 1992; Guinto et al., 1999a, 2001; Hopmans, 2003; Bridges, 2005). Additionally, it has often been difficult to accurately detect changes in N in fertile soils by mass balance due to large background levels and high spatial variability (e.g. Ryan et al., 2000). Generally, assessment of changes in N quantities at the ecosystem level has often shown estimated increases in N in the absence of disturbance but the reported quantities are higher than expected from known symbiotic fixation (Richards and Voight, 1963; Binkley et al., 2000; Bormann et al., 2002). That is, N is accumulating but at higher than predicted levels. A number of studies across different soils and/or forest types have shown that there is a significant increase in the production of mineral N (ammonium and/or nitrate) where soil C/N ratios are low. Low C/N ratios are usually associated with fertile soils supporting rainforests or wet sclerophyll forests dominated by species such as E. regnans or rainforests that have adaptations to high nitrate or ammonium levels (Spain et al., 1983; Ellis, 1985; Adams and Attiwill, 1988; Turner et al., 1989; Polglase et al., 1992; Attiwill and May, 2001). Forest N budgets are directly affected by fires, and the impacts vary with fire intensity (Raison et al., 1985; Johnson et al., 2007). Consumption of organic matter by fire leads to the loss to the atmosphere of the contained N. The organic matter affected by fire is primarily in the forest floor (fuel), with some in the soil A horizon

Fig. 1. Reported biomass (t ha1) and N content (kg ha1) of forest floor in Australian east coast Eucalyptus forests. (Sources: Park, 1975; Turner and Lambert, 1977, 2008; Feller, 1980; Charley and Richards, 1983; Baker and Attiwill, 1985; Turner and Kelly, 1985, Adams and Attiwill, 1986; Hamilton et al., 1991; Raison et al., 1993).

and in components of the vegetation (e.g. Hamilton et al., 1991; Bridges, 2005). Low intensity fires release the N contained within the litter and in some of the understorey. The organic matter and N will subsequently be replaced by litterfall after uptake of the N from the soil by vegetation and it is this uptake process which will lead to removal of N from the soil, particularly from the A horizon. Higher intensity fires will result in greater N losses and also changes in availability of some soil nutrients such as P (Humphreys and Craig, 1981; Johnson et al., 2007). With higher intensity fires, there is the potential for greater tree mortality and subsequent rapidly growing regeneration. Fire may also stimulate regeneration of N-fixing species, such as Acacia spp., potentially leading to increases in rates of N fixation. Based on N losses from soil due to low intensity fires and estimated or assumed inputs, the potential replacement period can be calculated for various forest types and situations. Overall, the average quantity of N reported in the fine fuel component of the litter and the low understorey of Australian forests is about 100 kg N ha1 (Fig. 1, originally developed by Baker and Attiwill, 1981). Assuming 1 kg N ha1 year1 precipitation input (Turner et al., 1996) and 4–8 kg N ha1 year1 fixation (Hopmans et al., 1983), the replacement period for N is between 5 and 20 years although much lower if the fixation estimates of Adams and Attiwill (1984b) are used. The present paper reports on an evaluation of chemical analyses of relatively infertile granitic soils in the Eden Region of NSW. The objective of this analysis was to determine whether there are relationships between soil properties and fire regimes. The focus of the analysis was on nutritionally poor soils derived from granites where there has been no disturbance due to timber harvesting. 2. Site and methods The study area is located on the coastal zone in the south-east corner of New South Wales, from the NSW–Victorian border to latitude 378150 S and west from the coast to 1498400 E. The climate of the region varies from the coast to the inland forests (Table 1). On the coast at Merimbula, the daily mean maximum temperature is 18–20 8C and the minimum is 9–12 8C while the average annual rainfall is about 760 mm. Moving from the coast, there is a general increase in rainfall and a decrease in maximum and minimum temperatures. Six main geological types are represented within the region, namely Mallacoota Beds of Ordovician marine sediments comprising 33% of the area (phyllite, slate, quartzite, sandstone, greywacke

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142 Table 1 Climatic data for the study area (Bureau of Meteorology, 1988) Location

Bega Green Cape Eden Bondi State Foresta

Latitude (South)

Elevation (m)

0

36840 37816 378060 378090

11 18 33 914

Rainfall (mm)

879 751 904 1050

Mean Temperature (8C) Maximum

Minimum

22.0 18.1 21.1 17.0

8.2 12.4 7.9 2.0

Rainfall is mean annual rainfall. a Adjacent to the sampling area.

and shale); Bega Batholith of the early Devonian Intrusives comprising 45% of the area; Middle Devonian Gabo Island granite comprising 3% of the area, Merimbula group of Late Devonian Alluvials comprising 8% of the area, Middle Devonian Extrusives (Eden rhyolite) comprising 6% of the area and the Late Tertiary alluvial sediments comprising 5% of the area. A study of soils of the Bega Batholith (Kelly and Turner, 1978) showed their generally low nutrient status (Table 2). For example, the average total P for the Bega Batholith soils is less than 80 mg kg1 in contrast to many coastal forest soils which range between 200 and 400 mg kg1; and areas where there have been basalt influences and there is more than 1000 mg kg1. The vegetation in the study area is variable ranging from rainforests, through moist and dry eucalypt communities to shrub communities and there are relationships between the soil types and vegetation communities (Turner et al., 1978). The study focused on the soils derived from granitic parent materials and on eucalypt forest which was predominantly dry Eucalyptus forest (Turner et al., 1978; Richards et al., 1990). Under the classification of Keith and Bedward (1999a,b), the study includes Hinterland Fern Forest (Type 13), Wallagaraugh Dry Grass Forest (Type 30), Hinterland Dry Grass Forest (Type 31), Foothills Dry shrub (Type 44) and Timbillica Dry Shrub Forest (Type 46A). The dominant tree species include Eucalyptus consideniana, Eucalyptus cypellocarpa, Eucalyptus globoidea, Eucalyptus mullerana, Eucalyptus obliqua and Eucalyptus sieberi. Soils from forests in the Eden Region were sampled and chemically analysed for a number of different projects, commencing in the late 1960s when intensive harvesting was being planned in the Eden area, through to the early 1990s. The data sets were pooled, however, initially there was no single design for

1135

sampling and analysis. All the soils were from native forest areas where no recent harvesting was noted (there may have been selective harvesting on some areas prior to 1969 but information from specific sample plots was not available and no tree stumps were noted). At each site, a plot was established for assessment, the size and type of plot varying with the study but it was typically circular with a 20–30-m radius. Surface soils were sampled by collecting 10 cores along a transect across the centre of the plot. The cores were pooled for analysis providing an averaged estimate. No estimation of within plot variation was made. A pit was excavated in the centre of the plot and used for soil description and deeper sampling. Sampling included surface soils in all cases with deeper samples in a smaller number of studies. The projects included fire effects studies, regional soil assessments, Eden Catchment studies (commenced by J. Turner and J. Kelly in the late 1970s and early 1980s, followed by P. Ryan through the 1980s) and sampling undertaken for the Joint Scientific Committee (Richards et al., 1990). Sampling protocols were the same in all cases. There have been publications based on these studies (Kelly and Turner, 1978; Turner et al., 1978; Ryan, 1993), the most comprehensive analysis being that by the Joint Scientific Committee (Richards et al., 1990) but much of the data are unpublished. Soils from a range of locations were sampled on a number of occasions; however, the specific individual sites have not been re-sampled. At each site, bulk density was estimated on each cores and the proportion of soil was estimated from the single pooled sample. Soils were analysed for pH, organic matter, total N, total and available P, exchangeable bases and exchangeable Al (Lambert, 1989). A summary of the results of the chemical analyses is provided in Table 3. All the plot data were initially screened and soils that were derived from non-granitic parent materials, where chemical analyses were incomplete or where information on the timing of the last fire was not available, were rejected from the data set. For each of the remaining plots, the time since fire was known but the actual fire intensity and related information were generally not known. Time since last fire ranged from 1 to 39 years. It was used as an independent variable in the statistical analysis, recognising that this parameter relates to both the time since fire was recorded and fire frequency. A dendrochronological study in the least fire prone (coolest and wettest) part of the study area showed that over more than 210 years the interval between fires averaged about 10 years and never exceeded 20 years on exposed sites (Banks in Richards

Table 2 Summary of chemical analyses of surface soils (0–10 cm) in the Eden region together with some comparisons from other studies on the NSW north coast indicating the generally low fertility of the soils of the study Soil parent material

pH

Organic matter (%)

Total N (%)

Total P (mg kg1)

Exchangeable (c mol(+) kg1) Ca

Mg

K

Na

Al

Eden region Eden Rhyolitea Tertiary sedimentsa Bega Batholitha Devonian sedimentsa Malacoota Bedsa Gabo Island Granitea

5.31 4.49 5.26 5.01 4.96 5.42

4.85 5.20 6.75 7.85 9.15 7.60

0.08 0.11 0.12 0.11 0.18 0.15

61 40 78 82 125 137

1.71 0.97 2.65 2.56 2.61 2.39

1.82 1.16 1.77 2.23 2.50 2.22

0.43 0.24 0.42 0.52 0.80 0.72

0.24 0.30 0.22 0.23 0.40 0.24

1.29 0.90 0.67 0.98 1.98 1.41

NSW north coast Sandstoneb Rhyolitec Rhyolitec Sandstoneb Tertiary Basaltc

5.28 4.95 4.45 5.77 4.81

6.92 15.9 23.9 10.03 18.1

0.17 0.37 0.57 0.20 0.81

158 195 435 438 1340

5.19 4.50 3.56 11.06 9.30

3.56 3.45 2.70 6.90 3.43

0.41 0.74 0.78 1.37 1.08

0.21 0.34 0.52 0.29 0.37

0.25 3.05 2.88 0.28 1.18

The soils are listed according to increasing total P. a Kelly and Turner (1978). b Humphreys and Craig (1981). c Turner and Kelly (1981).

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

1136

Table 3 Summary of properties of soils used in the present study Depth

pH

Organic matter (%)

Total N (%)

Total P (ppm)

Exchangeable (c mol(+) kg1)

Soil

Ca

Mg

K

Na

Al

(%)

0–10 cm

Mean Min Max S.D.

4.83 3.37 5.87 0.55

5.92 1.96 17.80 2.81

0.106 0.030 0.440 0.066

53 11 329 33

2.30 0.16 11.83 1.92

1.46 0.11 7.06 0.99

0.46 0.03 1.93 0.20

0.24 0.02 1.84 0.23

0.96 1.07 3.47 0.68

83 56 100 9.8

10–30 cm

Mean Min Max S.D.

5.36 4.42 5.9 0.42

4.15 1.24 8.94 2.20

0.051 0.021 0.139 0.034

55 11 444 101

1.68 0.25 5.73 1.31

1.29 0.43 2.95 0.76

0.36 0.11 0.78 0.15

0.17 0.08 0.36 0.08

0.66 0.06 2.34 0.58

82 53 100 11.7

30–40 cm

Mean Min Max S.D.

5.48 4.65 6.01 0.402

2.43 0.75 5.67 1.31

0.050 0.010 0.090 0.025

58 11 249 60

1.13 0.20 6.95 1.48

1.08 0.45 4.12 0.87

0.40 0.11 1.92 0.38

0.17 0.08 0.51 0.10

0.66 0.04 2.29 0.57

73 41 99 16

et al., 1990). Most of the study area is warmer and drier, and has been used more by people. As a consequence, it is most likely that the interval between fires has generally been shorter. 3. Data analysis Two data sets were identified; one for surface soils (0–10 cm depth) and another where information was also available to a greater depth (40 cm). In total, surface soil samples (0–10 cm) were from 290 sites but only 156 contained data suitable for the present study; 58 soil samples were from the greater depth. Soil nutrient quantities (kg ha1) were calculated as soil volume  bulk density  percentage soil  nutrient or elemental concentration: The data were graphically screened to detect possible anomalies and to ascertain the patterns of change in soil chemical properties with time since fire in an exploratory analysis using SAS/INSIGHT. Since there were two sampling depths, the 156 surface soil samples and the 58 samples at a depth of 40 cm were initially examined separately. Soil N content, organic carbon, the C/ N ratio, total P, and exchangeable aluminium, calcium, magnesium and potassium were plotted against time since fire in a simple linear least squares regression to detect changes in soil chemical properties over time. Among these variables, soil nitrogen content exhibited an increasing mean trend with a monotone spread of data, a sign of the presence of heteroskedasticity. There were also more data points with smaller values than those having larger ones, an indication of positive skewness in the data. The C/N ratio showed a decreasing mean trend with time since fire, and heteroskedasticity was also present. Soil pH also appeared to decrease with time since fire. Only a slight increase of total phosphorus over time was detected for the surface soil, but not for the deeper samples, and the relationship was weak with a R2 value of 0.06 only. No time related changes were detected for other variables assessed. To overcome heteroskedasticity and positive skewness in the data, soil N content and the C/N ratio were log transformed. Since soil P content is a reliable indicator of soil fertility and it changed little over time, log transformed total P was included as an independent variable in a further analysis of changes in soil N content, the C/N ratio and soil pH with time since fire across levels of soil fertility. The log transformation was taken after comparing fit statistics such as R2 and root mean squared error against the untransformed form when incorporating total P as an independent variable in the regression. The time-related changes in soil N

content and in the C/N ratio were compared between the two soil sampling depths. For both variables, the rate of change over time was similar for the two soil depths as indicated by the parameter estimates associated with time since fire, although the intercept was different. Parameter estimates associated with total P were also compared between the sampling depths, and the differences were only marginally significant at a = 0.05. The comparison indicated the value of pooling the data sets from the two sampling depths in a combined analysis by incorporating a dummy variable representing the depth of soil sampling. Prior to pooling the two data sets for a combined analysis, the distribution patterns and associated univariate statistics of time since fire and total P were examined for each data set. There was little difference between the two data sets in the range, mean and standard deviation of time since fire. The shape of the distribution pattern of total P for the deeper soil depth closely resembled that for the data from the surface horizon. Although they had different means, the two distributions had similar values of skewness and kurtosis, and the same coefficient of variation. For total P to be a consistent independent variable in the combined analysis, total P values of the 58 deeper samples were converted to equivalent values at a sampling depth of 10 cm. The conversion was achieved by dividing total P values of the 58 deeper samples by their maximum value and then multiplied by the maximum total P value of the 156 surface samples. Following the exploratory analysis and with a further consideration of the realistic patterns of change in soil properties, a system of non-linear equations with cross-equation parameter constraints and cross-equation error correlation was specified for total organic carbon, soil nitrogen content and the C/N ratio as follows ln C ¼ a10 þ a11 ln P þ a12 I þ e1

(1)

 a23 þ e2 a24 T   a33 ln F ¼ a30 þ a31 ln P þ a32 I  exp þ e2 T a34 ln N ¼ a20 þ a21 ln P þ a22 I þ exp



(2)

(3)

where C and N represent soil organic carbon and soil N content in kg ha1, F is the carbon nitrogen ratio, T represents time since fire in years, P stands for total phosphorus in the surface soil in kg ha1, I is a dummy variable representing the two soil depths, with 0 for surface and 1 for the 40 cm depth, e1, e2 and e3 are the error terms of the corresponding equations. Because ln F = ln C  ln N, the crossequation parameter constraints are a30 = a10  a20, a31 = a11  a21, a32 = a12  a22, a33 = a23, and a34 = a24. Since the three error terms

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

are correlated, this system of equations falls into the framework of seemingly unrelated regressions (SUR) of Zellner (1962). Although Zellner’s SUR estimator was first derived for a system of linear equations, it can be extended to a system of non-linear equations such as the system specified above (Gallant, 1975; Srivastava and Giles, 1987; Judge et al., 1988). The parameters were estimated through the PROC MODEL procedure of SAS using the SUR method (SAS Institute Inc., 1988). Because the equations are estimated in log transformed form, the predicted values of C, N and F need to be corrected for log transformation bias that is inherent in multiplicative regression models with log normal errors (Flewelling and Pienaar, 1981). Although bias correction factor can be obtained through several estimators (Flewelling and Pienaar, 1981; Snowdon, 1991), the bias correction factor of Snowdon (1991) was calculated for both equations to obtain unbiased predictions of C, N and F. For soil pH, a non-linear model was specified after detailed exploratory analysis: pH ¼ a0 þ

a1 þ a2 P þ a3 N T þ 10

(4)

The parameters were also estimated through the PROC MODEL procedure of SAS (SAS Institute Inc., 1988). For each of the four equations, a generalised value of R2 was calculated to indicate the goodness of fit: PT 2 ðyi  yˆ i Þ R2 ¼ 1  Pi¼1 T 2 ¯ i¼1 ðyi  yÞ

(5)

where yi and y¯ are the observed values and their mean, and yˆ i is the predicted values. To illustrate the pattern of change in soil pH with time since fire, predicted values of N from Eq. (2) after correction for log transformation bias were substituted into Eq. (4). 3.1. Paired plot study An additional smaller study was undertaken in December 2006 using paired plots to assess the differences in soils between a repeatedly burnt area and an area not burnt for 34 years. The location was adjacent to the Eden Burn Study area (Bridges, 2005) along a fire break where one side was repeatedly burnt for protection purposes and the other had not been burnt since 1972. The understorey on the unburnt side of the track had developed strongly after the track was constructed in 1988. The overstorey species composition varied along the slope. Plots were paired according to position on slope, overstorey species composition, aspect and the shape of the ground surface. Seven pairs of plots were established from upper slope to lower slope positions. Two more plots were established in severely declining stands on the lower slope and creek flat below the paired plots but there were

1137

no similar stands available for comparison that had been repeatedly burnt. An additional plot nearby on the same geological substrate was established in a long unburnt, declining dry forest stand with a dense mid storey of Allocasuarina littoralis (a N-fixing species). Surface soils were sampled on a transect along the contour in each plot and 10 sub-samples were taken and pooled. Single deeper samples were collected from the centre of each transect. Soils were chemically analysed the same as for the major study and including extractable mineral N. Differences in soil properties were assessed using a paired t-test. Litter depth and percentage cover of understorey was estimated in an area 10 m wide along each soil sampling transect. Canopy condition was assessed for the three overstorey trees closest to each transect. These data were insufficient for statistical analysis and were considered qualitatively. 4. Results When estimated in the log transformed form, the equations explained 72%, 79% and 69% of the variation in soil organic carbon, soil N content and the C/N ratio in the data (Table 4). Observed values of log transformed C, N and C/N ratio plotted against the predicted values showed that model specification was adequate for the data (Fig. 2). Soil N content was relatively stable for the first 10 years after fire and then increased with time (Fig. 3). As time goes to infinity, the asymptotic maximum of soil N content ranges between 1299 and 2395 kg ha1 for the surface soil and between 3043 and 5610 kg ha1 for the deeper soil across the five levels of soil fertility in Fig. 3. The rate of N accumulation appeared to peak at 24 years after fire then decreased afterwards (Fig. 3). Correspondingly, C/N ratio changed little for the first 10 years after fire and then decreased with time. As time goes to infinity, the asymptotic minimum of C/N ratio ranges from 15.83 to 17.25 for the surface soil and from 14.81 to 16.14 for the deeper soil across the five levels of soil fertility in Fig. 3. The estimated equation of soil pH in relation to time since fire, total P and soil N content is: 6:7497 pH ¼ 4:6210 þ þ 0:006481P  0:00034N T þ 10 R2 ¼ 0:37; RMSE ¼ 0:3269 where RMSE stands for root mean squared error (Fig. 4). To illustrate the pattern of change in soil pH with time since fire, predicted values of N from Eq. (2) were substituted into this equation (Fig. 3). The asymptotic minimum of soil pH ranges from 4.32 to 4.47 across the five levels of soil fertility as time goes to infinity (Fig. 4). In the case of both the surface and deeper soils, no significant relationships were identified between time since the last fire and C,

Table 4 Parameter estimates and fit statistics for the system of non-linear equations ln C

ln F

ln N

Parameter/fit statistics

Estimated value

Parameter/fit statistics

Estimated value

Parameter/fit statistics

Estimated value

a10 a11 a12

9.033 0.318 0.785

a20 a21 a22 a23 a24

5.015 0.380 0.851 514.902 1.935

a30 a31 a32 a33 a34

4.018 0.062 0.066 514.902 1.935

R2 RMSE

0.72 0.254 1.036

R2 RMSE

u1

u2

RMSE stands for root mean squared error. u1, u2 and u3 are bias correction factors.

0.79 0.269 1.019

R2 RMSE

u3

0.69 0.170 1.013

1138

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

Fig. 2. Observed values of log transformed C, N and C/N ratio plotted against their predicted values with a diagonal line of unity. Open circles represent surface soil samples and closed ones indicate sampling depth of 40 cm.

exchangeable Al, Ca, Mg or K and only a very weak relationship (R2 = 0.06) was found for total P. Based on these analyses, no fertility bias was found according to a particular project or over time.

Fig. 3. Soil N content, rate of N accumulation and C/N ratio in relation to time since fire across five levels of total P of surface soil at 20, 40, 60, 80, 100 kg ha1. The five curves in each panel represent increasing levels of soil fertility from bottom up for soil N content and rate of N accumulation, and conversely for C/N ratio.

The results showed that the quantity of soil N increased with the length of time since the last fire (Fig. 2) although this change was non-linear with essentially no change in the initial 10 years after fire. There was a significant decline in soil C/N with increasing time since the last fire, and as C did not change significantly, this was mainly a result of change in soil N. The actual rate of change and specific levels were related to the soil fertility as indexed by soil P. The rate of increase in soil N, that is, the calculated overall annual changes in N (in kg ha1 year1) varied for the surface soil according to the fertility level and gave a different result when the depth to 40 cm was taken into account. The results showed little change for a decade followed by a sharp rise in accumulation up to about 24 years followed by a steady decline. At the peak of accumulation, the surface soil increased by between 18 and 35 kg N ha1 year1 while the average accumulation rate over 40 years is 11 kg N ha1 year1 for the low fertility soils and 21 kg N ha1 year1 for the highest fertility soils studied. A similar pattern was apparent when the profile to 40 cm was included with very little change in the initial stages and then a peak of accumulation at about 24 years followed by a decline. The low fertility soils peaked at about 40 kg N ha1 year1 while the higher fertility soils accumulated 85 kg N ha1 year1 and considered to be a very high estimate. Over 40 years, the low fertility soils accumulated approximately 27 kg N ha1 year1 and the higher fertility soils about 49 kg N ha1 year1. The C/N ratio declined from about 45 to 22 over the length of the period studied (Fig. 3). The changes in pH indicated a declining trend representing approximately one pH unit over the period of the study (Fig. 4). In the paired plot study, there were changes in the properties of the soils on both sides of the track down the slope, and the properties on either side of the track increasingly diverged down the slope. The condition of the canopy of trees in unburnt plots was assessed as being poorer than in burnt plots. The average total N in the surface soils was 749 and 945 kg N ha1 in the burnt and unburnt areas, respectively, representing an accumulation rate of about 5.7 kg N ha1 year1 (assuming a difference of 34 years since fire). At the soil depth of 40 cm, there was 1747 and 2205 kg N ha1 in the burnt and unburnt areas, respectively, an accumulation rate of about 13.5 kg N ha1 year1. These estimates matched the models developed from the general survey. There were no significant changes in the C/N ratio but the unburnt sites had higher levels of extractable mineral N and the relationships between the mineral N and the C/N ratio for burnt and unburnt sites were statistically significant. The quantities of extractable mineral N in the unburnt soils (2.3 kg N ha1) were about twice the levels in the burnt soils (1.2 kg N ha1). It was noted that the mineral N levels were low probably due to the very dry prevailing conditions. The pH (4.4 in 1:1 water) of the surface soil in the

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

1139

Fig. 4. Observed and predicted values of soil pH with a diagonal line of unity (left), and predicted changes of soil pH with time since fire (right) across five levels of total P of surface soil at 20, 40, 60, 80, 100 kg ha1 from bottom up.

regularly burnt area was higher than in the unburnt area (pH 4.1) and the exchangeable Al also differed (0.62 c mol(+) kg1 in the burnt area and 1.3 c mol(+) kg1 in the unburnt). Soils under the severely declining stands had higher levels of N and exchangeable Al, lower C/N ratio and pH than the paired plots. The site with a dense sub-canopy of Allocasuarina had even higher levels of N and mineral N and lower pH indicating that understorey development is a significant component of the N accumulation and subsequent problem of tree dieback. 5. Discussion Fire is generally recognised as a regular occurrence in native forest ecosystems in the Eden area of NSW. Little information is available on fire frequency or intensity. In the present analysis, soils were sampled for a number of studies and where fire had occurred in the sites, it was noted. The data for P, exchangeable cations, Al and C in the soil were analysed in relation to time since fire to determine if there was any obvious bias in the sampling system. No significant relationships were found for all except the very weak correlation for P and it was concluded that there was no bias or inherent nutritional gradient; hence changes detected in N were a result of accumulation in the absence of fire. That is, there was no detectable nutritional gradient such as reported by Turvey and Smethurst (1988) who concluded that the gradient led to erroneous estimates of N accumulation in their chronosequence analysis of plantations. Generally, lower slopes had higher N and P and a lower C/N ratio and also a lower frequency of fire, that is, the gullies were less frequently burned and it was considered that this factor was accounted for by the inclusion of P in addition to time since fire in the analyses. Sampling was confined to mature undisturbed forests. Thus none of the sampling sites had been affected by high intensity fires. Long unburnt sites were in sheltered positions where fires will not propagate under moderate conditions. Moderately long time (one to two decades) since fire is indicative of moderately sheltered sites or cool and moist climate, whilst sites that had been burnt within the previous decade were most likely to be in more exposed positions or warmer and drier areas where fires propagate readily under moderate conditions. With a long history of frequent ignitions by lightning and people in the study area, it is most unlikely that recently burnt sites had remained long unburnt prior to the most recent fire. Thus time since fire is considered to provide a broad indication of the historical frequency of fires at the study sites. Based on the N relationships in Table 4, the implied average N accumulation is generally in the range of 11–21 kg N ha1 year1 for surface soils and 27–49 kg N ha1 year1 when deeper horizons

were included. Even though there is often development of some leguminous species such as Acacia after fire, there appears to be little change of N in this period but it starts to increase after about a decade and this is generally related to the significant development of symbiotically N-fixing species such as Casuarina. It is possible that in the initial decade there is active N fixation, however, it is being accumulated in biomass and forest floor, two components not included in this present study. The average annual N input in rainwater was less than 1 kg N ha1 year1 (Turner et al., 1996) so additional increases in the soil are assumed to be from N fixation due to symbiotic and/or free living organisms. This is similar to that shown by Baker and Attiwill (1981) who estimated asymbiotic N fixation of approximately 1.5 kg N ha1 year1 in E. obliqua forests and they considered that symbiotic fixation rates might have been as high as 20–50 kg N ha1 year1, especially immediately after a fire in areas where there was development of N-fixing species. As well as an influx of N-fixing species (such as Acacia) immediately after fire, there can also be ongoing recruitment of other N-fixers, such as Allocasuarina, leading to continued inputs. Increased N has been noted beneath Allocasuarina in comparison with adjacent areas, and higher levels occurred under dense Allocasuarina than in any of the other long unburnt paired plots in the additional study. Litter mass was not estimated on the sample plots, however, indicative estimates can be obtained from a study of litter mass by Bridges (2004) over 8 years at Eden. In multi-aged stands that had not been disturbed by fire or harvesting for at least 5 years, the mean litter weight over all years was 10.4 t ha1 while between years it ranged from 8.9 to 14.4 t ha1. Taking the overall average weight and applying average N concentrations to each component, the potential quantities of contained N can be estimated and hence the reduction in N by burning can be estimated (Table 5). The estimated N content of about 62 kg N ha1 was below the average reported in Fig. 1. Bridges (2005) reported that after burning, there was a reduction of 31% in fine fuel weight from a level of 16 t ha1 in unburnt stands adjacent to our paired plot study. This is Table 5 Components of forest floor mass (Bridges, 2004) and N concentrations in E. sieberi in the Eden area Component

N (%)

Leaf Twig Bark Understorey Fine material

0.69 0.43 0.38 0.80 0.73

Total

Biomass (t ha1)

N (kg ha1)

2.188 3.751 0.729 1.563 2.189

15.1 16.1 2.8 12.5 16.0

10.420

62.5

1140

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

equivalent to a removal of about 30 kg N ha1 by low intensity burning. At another dry forest site in the region, Bridges (2004) reported a 54% reduction in fine fuel from a level of 15 t ha1 representing a removal of 48 kg N ha1 in a prescribed burn. In a longer term study in dry sclerophyll E. pilularis forest, there was an average 59% reduction by prescribed fires (Birk and Bridges, 1989). In E. pilularis forest in Queensland, Guinto et al. (2001) found that biennial burning reduced total N and organic C levels in the topsoil by 40% compared with long time unburnt areas, whilst quadrennial burning reduced them by 12 and 10%, respectively. In our paired plots study, total N in the repeatedly burnt plots was 21% lower than in the long unburnt plots, suggesting that relative rates of N accumulation in the absence of fire are similar between the different forest types. The soil C/N ratio was found to be negatively related to time since fire and was also related to soil fertility. The annual change in C/N ratio over a 30-year period averaged about 0.5 units each year but was difficult to detect in the short term, especially in the period immediately after burning. Other studies indicate that when the C/ N ratio declines below about 25, potential increases in total mineral N (NH4+ and NO3) are expected with large and continued increases with a C/N ratio less than 20 (Attiwill et al., 1996). Our additional study indicated there was a relationship between the C/ N ratio and extractable mineral N and that this relationship differed between plots periodically burnt or with low fire frequency. The results of this study were consistent with estimates made in the general survey study. The increased N and resultant nitrification is one possible cause of decline in soil pH. A number of studies within undisturbed forest ecosystems have found greater accumulation of N than would be expected based on estimates of direct fixation of N by both symbiotic and free living organisms (Son, 2001). The identification and quantification of processes to account for N accumulation have been recognised as a problem for some time (for example, see Richards and Voight, 1963). Bormann et al. (2002) concluded that unexplained N accumulation (that is, not accounted for by identified fixing organisms) in field experimental studies was 40– 150 kg N ha1 year1 over a 5-year period. While this has been explained in some instances due to issues associated with methodology, substantial evidence remains for consistent and long term increases in total ecosystem N in the absence of disturbance. The questions relate more to the actual rate of increase, the processes involved in the increase and the effect of such long term accumulation. N fixation or accumulation rates are difficult to estimate but in temperate forests, they have been reported to be in the range from low to more than 100 kg N ha1 year1. Attiwill and May (2001) working with E. regnans on fertile soils, reported that the rate of symbiotic N fixation was dependent on the number of silver wattle (Acacia dealbata) present and at 3 years of age, burnt and unburnt coupes were accumulating 187 and 74 kg N ha1 year1 respectively, or more than an order of magnitude over expected estimates. Post-fire accumulation of N by N-fixing shrubs in dry area forests in the USA were reported by Johnson et al. (2005) as 36–48 kg N ha1 year1 and this was low compared with other work in moister forests of 70–100 kg N ha1 year1 (Youngberg and Wollum, 1970; Binkley et al., 1982; McNabb and Cromack, 1983). In Northern Hemisphere coniferous forests, N fixation rates of 0.3–38 kg N ha1 year1 have been reported (Granhall and Lindberg, 1978; Todd et al., 1978). The estimates of N accumulation in the present study fall within the general levels reported in the literature. While analyses of N mineralisation were not undertaken in this study, the relationship between soil C/N and mineralisation in eucalypt forest soils has been reported (Attiwill et al., 1996). In the

Fig. 5. Summary of C/N ratio and extractable mineral N (mg g1) in surface soils in Australian forests.

reported function (Y = 1340  100.267*X, where Y is the rate of nitrification in mg g1 day1 and X is the C/N ratio), mineral N production becomes significant when the C/N ratio is below 20 and then rises rapidly as the C/N ration declines below that amount. This relationship has been developed for E. regnans and evidence from other forest types indicates a similar pattern although quantitatively different, such as E. obliqua stands in Tasmania showing greater mineralisation at higher C/N ratios (Adams and Attiwill, 1988). This concept was extended (Fig. 5) using other published data from a range of native Eucalyptus studies where the soil C/N ratio and extractable mineral N were available (Charley and Richards, 1977; Turner and Kelly, 1981; Baker and Attiwill, 1981; Ellis, 1986; Bale, 1980; Turner and Lambert, 1983; Adams and Attiwill, 1986, 1988; Grove et al., 1986; Polglase et al., 1992; Birk, 1993; Granger et al., 1994; Attiwill et al., 1996; Guinto et al., 1999b; Cannell and Thornley, 2000). Soil types strongly affect the species composition and there are strong links between species type, C/N and mineral N. Basically though, when the C/N ratio is below 25, mineral N increases. The pattern indicates that some forest types/soils have generally low C/N and generally high mineral N while the converse is the case with some other forest types. The present study demonstrated that N accumulates with time since fire in dry eucalypt forests at Eden, and that relatively high N, low C/N, low pH and high Al occurred in three severely declining stands. The accumulation of N in eucalypt forests has been suggested as a basic component of the process of decline in forest health (Jurskis and Turner, 2002; Jurskis, 2005a,b; Turner and Lambert, 2005). Ongoing studies in Western Australia and Tasmania have confirmed accumulation of N in long unburnt declining eucalypt stands (Dugald Close, pers. comm.). The processes are comparable with those reported in the Northern hemisphere forests where correlations between declining forest health and elevated nitrogen inputs (natural and anthropogenic) have been demonstrated (Aber, 1992; Magill et al., 2004; Fenn et al., 2006; Wallace et al., 2007). Less rigorous studies of nitrogen and forest decline have been reported in Australian forests. Granger et al. (1994) reported tree decline in E. ovata and E. camphora communities where the soils had lowered C/N ratio, increased mineral N and increased foliar N concentrations. In NSW, Stone (2005) reported a relationship between soil N content and poor health of eucalypts although a relationship was reported with inherent soil fertility rather than as a result of reducing the frequency of fire. In a small study near Sydney, Stone and Simpson (2006) reported a weak correlation between poor eucalypt health and increasing soil N although there were confounding factors of geology, topography and land management. Based on observations

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

and preliminary analyses, it appears that in susceptible stands which are dependent on frequent fire, tree health is generally maintained with a C/N ratio above 25–30. To maintain a stable C/N ratio with low mineral N production, a fire periodicity is required where N losses are about equivalent to N inputs. If inputs of N were about 12 kg N ha1 year1 and losses in low intensity fire were 65 kg N ha1, a period between fires of about 5 years would maintain stability. This would vary according to the fertility of the soil and the fire intensity. In addition to the changes in soil N, it was found that pH was higher and exchangeable Al was lower in periodically burnt plots than in long unburnt plots. As there are no significant changes in soil organic matter levels, the change in pH is most probably as a result of changes in base cations as a result of tree uptake and increased levels of nitrification. The soils in the current studies have low fertility with a low buffering capacity. Over a period of 39 years without fire, there was a decrease of about 0.6 pH unit. Similar changes occurred in a dry sclerophyll forest in Queensland (Guinto et al., 1999b) where the pH was 5.4 in the surface soil in a long unburnt area, compared with 6.1 in an annually burnt area, but the differences were smaller on a wet sclerophyll site. Humphreys and Craig (1981) studied different burning regimes commencing in 1967 in E. pilularis at Manning River, NSW. Two of the replicates were on high fertility soils and two on low fertility. After 12 years, the unburnt lower fertility soils had a pH of about 5.1 whereas the regularly burnt soils were about 5.7. There had also been a gradual but significant increase in exchangeable aluminium in the unburnt plots. There was little change on the more fertile, more highly buffered soils. The present study on low fertility granitic soils in the Eden area shows there were increases in soil N, acidity and exchangeable Al in the absence of burning in the forests which are adapted to low intensity fires. Changes due to the absence of fire can impact on tree health (Aitken, 1992; Wallace et al., 2007). The soils have a low buffering capacity and impacts can be quite rapid. Comparisons with other studies show that similar patterns occur on poorer fertility soils but as fertility increases, so does the buffering capacity and effects are less noticeable. A decline in pH will produce a less suitable environment for roots due to higher levels of Al and Mn which will have detrimental effects on tree health. The increases in mineral N will also impact on the nutritional balance within the trees and it is proposed that this imbalance will make the trees more susceptible to folivores, pests and pathogens. Burning trials in forests are often reported as if fire is a treatment and unburnt areas are controls. However, in fireadapted forests, the imposition of fire restriction is the treatment. The ‘‘control’’ or baseline should be the periodically burnt plots. When such trials are studied in a time sequence, the properties of soils in burnt plots are generally found to be stable. Dry eucalypt forests are adapted to frequent fire, and stable soil conditions are maintained, particularly in relation to N. Elimination of frequent fire leads to either high intensity fires and tree mortality or long term modification to the soil process leading to reduced tree health and mortality. 6. Conclusions Nitrogen appears to increase in quantity in the soil with time since fire, and the rate of increase is related in part to the basic soil fertility as indicated by soil phosphorus levels. The apparent rate of increase in the surface soils is approximately 11– 21 kg N ha1 year1 with the potential for higher levels on more fertile soils. The source of the N is assumed to primarily be N fixation, as the measured atmospheric inputs were of the order of 1 kg ha1 year1. The increases in N lead to a reduced soil C/N

1141

ratio, higher N mineralisation and reduced pH. It is proposed that the reduced pH is a result of a combination of nitrification, related to the increased soil N, and a reduction in base cations through uptake by vegetation. It is hypothesised that these changes create a poorer root environment and nutritional status for eucalypts, and these changed conditions can impact directly on tree health and increasing susceptibility to pests and pathogens. In the subsidiary study of paired plots found there was declining tree health in the long unburnt areas related to changes in soil characteristics, compared with the adjacent regularly burnt areas. Acknowledgements We wish to thank Forests NSW for assistance with early soil sampling and chemical analyses. Financial support from the Bushfire Cooperative Research Centre, Hobart, Tasmania is gratefully acknowledged. References Aber, J.D., 1992. N cycling and N saturation in temperate forest ecosystems. Trees 7, 220–223. Adams, M.A., Attiwill, P.M., 1984a. Role of Acacia spp. in nutrient balance and cycling in regenerating Eucalyptus regnans F.Muell. forests. I. Temporal changes in biomass and nutrient content. Aust. J. Bot. 32, 205–215. Adams, M.A., Attiwill, P.M., 1984b. Role of Acacia spp. in nutrient balance and cycling in regenerating Eucalyptus regnans F.Muell. Forests. II. Field studies of acetylene reduction. Aust. J. Bot. 32, 217–223. Adams, M.A., Attiwill, P.M., 1986. Nutrient cycling and N mineralisation in eucalypt forests of south-eastern Australia. I. Nutrient cycling and N turnover. Plant and Soil 92, 319–339. Adams, M.A., Attiwill, P.M., 1988. Nutrient cycling in forests of north-east Tasmania. Research Report No. 1, Tasmanian Forest Research Council Inc., 214 pp. Aitken, R.L., 1992. Relationships between extractable Al, selected soil properties, pH buffer capacity and lime requirement in some acidic Queensland soils. Aust. J. Soil. Res. 30, 119–130. Attiwill, P.M., Polglase, P.J., Weston, C.J., Adams, M.A., 1996. Nutrient cycling in forests of south-eastern Australia. In: Attiwill, P.M., Adams, M.A. (Eds.), Nutrition of Eucalypts. CSIRO, Melbourne, pp. 191–227. Attiwill, P.M., May, B.M., 2001. Does N limit the growth of native eucalypt forests: some observations for mountain ash Eucalyptus regnans. Mar. Freshwater Res. 52, 111–117. Baker, T.G., Attiwill, P.M., 1981. N in Australian eucalypt forests. Proceedings on Australian Tree Nutrition Workshop: Productivity in Perpetuity, CSIRO, pp. 159–172. Baker, T.G., Attiwill, P.M., 1985. Above-ground nutrient distribution and cycling in Pinus radiata D. Don and Eucalyptus obliqua L’Herit. forests in southeastern Australia. For. Ecol. Manage. 13, 41–52. Bale, C.L., 1980. The impact of topography on patterns and processes in forested ecosystems on the mid North Coast, New South Wales. Ph.D. Thesis. UNE, Armidale, 365 pp. Binkley, D., Cromack, K., Fredrikson, R.L., 1982. N accretion and availability in snowbrush ecosystems. For. Sci. 28, 720–724. Binkley, D., Son, Y., Valentine, D.W., 2000. Do forests receive occult inputs of N? Ecosystems 3, 321–331. Birk, E.M., 1993. Effects of an exotic forest species on soil N mineralization and nitrification potentials. State Forests of NSW. Unpublished Report. Birk, E.M., Bridges, R.G., 1989. Recurrent fires and fuel accumulation in even-aged blackbutt (Eucalyptus pilularis) forests. For. Ecol. Manage. 29, 59–79. Bormann, B.T., Keller, C.K., Wang, D., Bormann, F.H., 2002. Lessons from the Sandbox Is unexplained N real? Ecosystems 5, 727–733. Bridges, R.G., 2004. Fine fuel in the dry sclerophyll forests of south-eastern New South Wales. Aust. For. 67, 88–100. Bridges, R.G., 2005. Effects of logging and burning regimes on forest fuel in dry sclerophyll forests in south-eastern New South Wales. Initial results (1984– 1993) from the Eden Burning Study Area. State Forests of NSW Research Paper No. 40, 79 pp. Bureau of Meteorology, 1988. Climatic Averages Australia. Meteorological Summary April 1988. Commonwealth Information Services, Australian Government Printer, Canberra. 532 pp. Cannell, M.G.R., Thornley, J.H.M., 2000. Nitrogen states in plant ecosystems: a viewpoint. Ann. Bot. 86, 1161–1167. Charley, J.L., Richards, B.N., 1977. Mineral cycling in rain forests. In: Golstein, W. (Ed.), Rain Forests. National Parks and Wildlife Service, Sydney, p. 107. Charley, J.L., Richards, B.N., 1983. Nutrient allocation in plant communities: mineral cycling in terrestrial ecosystems. In: Lange, O.L., Nobel, P.S., Osmond, C.B., Ziegler, H. (Eds.), Physiological Plant Ecology IV. Responses to the Chemical and Biological Environment. Springer-Verlag, Berlin, pp. 5–45.

1142

J. Turner et al. / Forest Ecology and Management 256 (2008) 1133–1142

Ellis, R.C., 1985. The relationship among eucalypt forest, grassland and rainforest in a highland area in north-eastern Tasmania. Aust. J. Ecol. 10, 297–314. Ellis, R.C., 1986. Mineralization of nitrogen in soils of clear felled and burnt coupes in Southern Tasmania. In: Rummery, R.A., Hingston, F.J. (Eds.), Managing N economies of natural and man made forest ecosystem. CSIRO Div. Land Use Management, Perth, Australia, pp. 293–302. Feller, M.C., 1980. Biomass and nutrient distribution in two eucalypt forest ecosystems. Aust. J. Ecol. 5, 309–333. Fenn, M.E., Poth, M.A., Aber, J.D., Baron, J.S., Bormann, B.T., Johnson, D.W., Lemly, A.D., McNulty, S.G., Ryan, D.F., Stottlemyer, R., 2006. N excess in North American ecosystems: predisposing factors, ecosystem responses, and management strategies. Ecol. Appl. 8, 706–733. Flewelling, J.W., Pienaar, L.V., 1981. Multiplicative regression with lognormal errors. For. Sci. 27, 81–189. Gallant, A.R., 1975. Seemingly unrelated non-linear regressions. J. Econ. 3, 35–50. Granger, L., Kasel, S., Adams, M.A., 1994. Tree decline in southeastern Australia: nitrate reductase activity and indications of unbalanced nutrition in Eucalyptus ovata (labill.) and E. camphora (R.T. Baker) communities at Yellingbo, Victoria. Oecologia 98, 178–192. Granhall, U., Lindberg, T., 1978. N fixation in some coniferous forest ecosystems. In: Granhall, U. (Ed.) Environmental Role of N-fixing Blue-green Algae and Asymbiotic Bacteria, Ecol. Bull. (Stockholm) 26, pp. 172–177. Grove, T.S., O’Connell, A.M., Dimmock, G.M., 1986. Nutrient changes in surface soils after an intense fire in jarrah (Eucalyptus marginata Donn ex Sm.) forest. Aust. J. Ecol. 11, 303–317. Guinto, D.F., Zu, Z.H., House, A.P.N., Saffigna, P.G., 1999a. Impacts of repeated fuel reduction burning on tree growth, mortality and recruitment in mixed species eucalypt forests in southeast Queensland. Aust. For. Ecol. Manage. 115, 13–27. Guinto, D.F., Saffina, P.G., Xu, Z.H., House, A.P.N., Perera, M.C.S., 1999b. Soil N mineralisation and organic matter composition revealed by 13C NMR spectroscopy under repeated prescribed burning in eucalypt forests of south-east Queensland. Aust. J. Soil Res. 37, 123–135. Guinto, D.F., Zu, Z.H., House, A.P.N., Saffigna, P.G., 2001. Soil chemical properties and forest floor nutrients under repeated prescribed-burning in eucalypt forests in south-east Queensland, Australia. N. Z. J. For. Sci. 31, 170–187. Hamilton, S.D., Lawrie, A.C., Hopmans, P., Leonard, B.V., 1991. Effects of fuelreduction burning on a Eucalyptus obliqua forest ecosystem in Victoria. Aust. J. Bot. 39, 203–217. Hopmans, P., 2003. Effects of repeated low-intensity fire on carbon, N and phosphorus in the soils of a mixed eucalypt foothill forest in south-eastern Australia. Department of Sustainability and Environment, Forest Science Centre, Research Report No 60, Melbourne, Australia, 34 pp. Hopmans, P., Douglas, L.A., Chalk, P.M., 1983. N fixation associated with Acacia dealbata as estimated by the acetylene reduction assay. Aust. J. Bot. 31, 331–339. Humphreys, F.R., Craig, F.G., 1981. Effects of fire on soil chemical, structural and hydrological properties. In: Gill, A.M., Groves, R.H., Noble, I.R. (Eds.), Fire and the Australian Biota. Australian Academy of Science, Canberra, pp. 177–214. Johnson, D.W., Murphy, J.F., Susfalk, R.B., Caldwell, T.G., Miller, W.W., Walker, R.F., Powers, R.F., 2005. The effects of wildfire, salvage logging, and post-fire Nfixation on the nutrient budgets of a Sierran forest. For. Ecol. Manage. 220, 155– 165. Johnson, S.W., Murphy, J.D., Walker, R.F., Glass, D.W., Miller, W.W., 2007. Wildfire effects on forest carbon and nutrient budgets. Ecol. Eng. 31, 183–192. Judge, G.G., Hill, R.C., Griffiths, W.E., Lutkepohl, H., Lee, T.C., 1988. Introduction to the theory and practice of econometrics, 2nd ed. Wiley, New York, 1024. Jurskis, V., 2005a. Decline of eucalypt forests as consequence of unnatural fire regimes. Aust. For. 68, 257–262. Jurskis, V., 2005b. Eucalypt decline in Australia, and a general concept of tree decline and dieback. For. Ecol. Manage. 215, 1–20. Jurskis, V., Turner, J., 2002. Eucalypt dieback in Eastern Australia: a simple model. Aust. For. 65, 87–98. Keith, D.A., Bedward, M., 1999a. Native vegetation of the South East Forests region, Eden, New South Wales. Cunninghamia 6, 1–60. Keith, D.A., Bedward, M., 1999b. Native vegetation of the South East Forests region, Eden, New South Wales. Appendix: descriptive profiles of floristic assemblages. Cunninghamia 6, 61–218. Kelly, J., Turner, J., 1978. Soil nutrient–vegetation relationships in the Eden area, N.S.W. I. Soil nutrient survey. Aust. For. 39, 127–134. Lambert, M.J., 1989. Methods for chemical analysis of forest soils, plant material and waters. Forestry Commission of NSW Technical Paper No. 25, third ed., 187 pp. Magill, A.H., Aber, J.D., Currie, W.S., Nadelhoffer, K.J., Martin, M.E., McDowell, W.H., Melillo, J.M., Steudler, P., 2004. Ecosystem response to 15 years of chronic N additions at the Harvard Forest LTER, Massachusetts, USA. For. Ecol. Manage. 196, 7–28. McNabb, D.H., Cromack, K., 1983. Di-N N-fixation by a mature Ceanothus velutinus (Dougl) stand in the western Oregon Cascades. Can. J. Microbiol. 29, 1014–1021. Park, G.N., 1975. Variation in nutrient dynamics and secondary ecosystem development in subalpine eucalypt forests and woodlands. Ph.D. Thesis. ANU, Canberra.

Polglase, P.J., Attiwill, P.M., Adams, M.A., 1992. N and phosphorus cycling in relation to stand age in Eucalyptus regnans F.Muell. II. N mineralization and nitrification. Plant Soil 142, 167–176. Raison, R.J., Khanna, P.K., Woods, P.V., 1985. Mechanisms of element transfer to the atmosphere during vegetation fires. Can. J. For. Res. 15, 132–140. Raison, R.J., O’Connell, A.M., Khanna, P.K., Keith, H., 1993. Effects of repeated fires on N and phosphorus budgets and cycling processes in forest ecosystems. In: Trabaud, L., Prodon, R. (Eds.), Fire in Mediterranean Ecosystems. Publication EUR 15089 EN of the Commission of European Communities, Brussels, pp. 347– 363. Richards, B.N., Voight, G.K., 1963. N accretions in coniferous forest ecosystems. In: Youngberg, C.T. (Ed.), Forest–Soil Relationships in North America. Proceedings of the Second North American Forest Soils Conference. OSU Press, Corvalis, pp. 105–116. Richards, B.N., Bridges, R.G., Curtin, R.A., Nix, H.A., Shepherd, K.R., Turner, J., 1990. Biological conservation of the south-east forests. Report of the Joint Scientific Committee, Department of Primary Industries and Energy, Canberra, 407 pp. Ryan, P.J., 1993. Soil formation on the Wallagaraugh Adamellite, southeastern N.S.W., Australia. Catena 20, 543–561. Ryan, P.J., McKenzie, N.J., Connell, D.O., Loughhead, A.N., Leppert, P.M., Jacquier, D., Ashton, L., 2000. Integrating forest soils information across scales: spatial prediction of soil properties under Australian forests. For. Ecol. Manage. 138, 139–157. SAS Institute Inc., 1988. SAS/ETS User’s Guide, Version 6, 2nd ed. SAS Institute Inc., Cary, N.C., 1022 pp. Snowdon, P., 1991. A ratio estimator for bias correction in logarithmic regressions. Can. J. For. Res. 21, 720–724. Son, Y., 2001. Non-symbiotic N fixation in forest ecosystems. Ecol. Res. 16, 183–196. Spain, A.V., Isbell, R.F., Probert, M.E., 1983. Soil organic matter. In: Soils and Australian Viewpoint, CSIRO Division of Soils, pp. 551–563. Srivastava, V.K., Giles, D.A., 1987. Seemingly Unrelated Regression Equations Models: Estimation and Inference. Marcel Dekker, Inc., New York, 374 pp. Stone, C., 2005. Bell-miner-associated dieback at the tree crown scale: a multitrophic process. Aust. For. 68, 237–241. Stone, C., Simpson, J., 2006. Leaf, tree and soil properties in a Eucalyptus saligna forest exhibiting canopy decline. Cunninghamia 9, 507–520. Todd, R.L., Meyer, R.D., Waide, J.B., 1978. N fixation in a deciduous forest in the south-eastern United States, Granhall, U., Environmental Role of N-fixing Bluegreen Algae and Asymbiotic Bacteria, Ecol. Bull. (Stockholm) 26, 172–177. Tolhurst, K.G., Flinn, D.W., Loyn, R.H., Wilson, A.A.G., Foletta, I., 1992. Ecological Effects of Fuel Reduction Burning in Dry Sclerophyll Forest. A Summary of Principal Research Findings and their Management Implications. Department of Conservation and Environment, Forests Research Centre, Melbourne, Australia, 52 pp. Turner, J., Lambert, M.J., 1977. The Response of Forest Ecosystems to Disturbance: Regeneration of Forest Stands Through Vegetation Successions Following Disturbance. Proceedings of the CSIRO Symposium on Nutrient Cycling in Indigenous Forest Ecosystems, Perth, Australia, pp. 125–135. Turner, J., Kelly, J., Newman, L.A., 1978. Soil nutrient–vegetation relationships in the Eden area, N.S.W. II. Vegetation-soil associations. Aust. For. 41, 223–231. Turner, J., Kelly, J., 1981. Relationships between soil nutrients and vegetation in a North Coast Forest, New South Wales. Aust. For. Res. 11, 201–208. Turner, J., Lambert, M.J., 1983. Nutrient cycling within a 27-year-old Eucalyptus grandis plantation in New South Wales. For. Ecol. Manage. 6, 156–168. Turner, J., Kelly, J., 1985. Effect of radiata pine on soil chemical characteristics. For. Ecol. Manage. 11, 257–270. Turner, J., Lambert, M.J., Kelly, J., 1989. Nutrient cycling in a New South Wales subtropical rainforest. Ann. Bot. 63, 635–642. Turner, J., Lambert, M.J., Knott, J., 1996. Nutrient inputs from rainfall in New South Wales State Forests. State Forests of NSW Research Paper No 17, 49 pp. Turner, J., Lambert, M.J., 2005. Soil and nutrient processes related to eucalypt forest dieback. Aust. For. 68, 251–256. Turner, J., Lambert, M.J., 2008. Nutrient cycling in age sequences of two Eucalyptus plantation species. For. Ecol. Manage. 255, 1701–1712. Turvey, N.D., Smethurst, P.J., 1988. Apparent accumulation of N in soil under radiata pine: misleading results from a chronosequence. In: Dyck, W.J., Mees, C.A. (Eds.), Research strategies for long-term site productivity. IEA/BE Report No 8. New Zealand Forest Research Institute, Rotorua, New Zealand, pp. 39–43. Van Loon, A.P., 1969a. Investigations into the effects of prescribed burning of young, even aged blackbutt. Forestry Commission NSW, Research Note No. 23, 17 pp. Van Loon, A.P., 1969b. Bushland fuel quantities in the Blue Mountains, litter and understorey. Forestry Commission of NSW, Research Note No 33, 22 pp. Wallace, Z.P., Lovett, G.M., Hart, J.E., Machona, B., 2007. Effects of N saturation on tree growth and death in a mixed-oak forest. For. Ecol. Manage. 243, 210–218. Youngberg, C.T., Wollum, 1970. Nonleguminous symbiotic nitrogen fixation. In: Youngberg, C.Y., Davey, C.B. (Eds.), Tree Growth and Forest Soils. Oregon State University Press, Corvallis, pp. 383–395. Zellner, A., 1962. An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. J. Am. Stat. Assoc. 57, 348–368.