Net nitrogen mineralization and nitrification of different landscape positions in a lowland subtropical rainforest in Taiwan

Abstract

Measurements of soil nitrogen (N) transformation rate are needed to be able to assess the plant availability and forest ecosystem losses of N, but little is known about lowland subtropical rainforests. The objectives of the present study were to determine the rates of N mineralization and nitrification in different seasons (January and August 2006) and in different landscape positions (footslope and summit) in the Nanjenshan forest of southern Taiwan, where vegetation types and soil properties vary among different landscape positions. Net N mineralization and nitrification were measured using 28-day in-situ open core and capped core incubations. The results from January 2006 showed that the concentrations of ammonium-N (NH₄ ⁺-N), mineral N (NH₄ ⁺-N plus NO₃ ⁻-N) and nitrate-N (NO₃ ⁻-N) were not significantly different between open and capped cores, or between summit and footslope. However, the results of August 2006 showed that the NH₄ ⁺-N concentrations of the summit soils were significantly higher than those of the footslope soils (P < 0.05), and that NO₃ ⁻-N concentrations in the open cores were significantly lower than those in the capped cores because of higher rainfall in the summer. In general, concentrations of mineral N and NO₃ ⁻-N in August were higher than those in January (both P = 0.0003), and NH₄ ⁺-N concentrations were significantly different between the different landscape positions (P < 0.05). There were larger soil C and N and microbial N pools at the summit position; however, footslope soils showed higher net N mineralization and nitrification rate as expressed on the basis of per unit C or N. Our results suggested that the substrate properties of the footslope position contributed to the higher net N mineralization and nitrification rate, and that the differences in N transformation rates between the landscape positions appeared to be related to vegetation type.

Key words:

• landscape position
• lowland subtropical rainforest
• net nitrogen mineralization and nitrification
• soil microbial biomass
• vegetation type

Introduction

Soil scientists and ecologists have long focused on the study of soil nitrogen (N) mineralization and nitrification rates when they seek to understand controls on terrestrial productivity, nutrient cycling and microbial function (Schimel and Bennett 2004). Rates of soil N mineralization and nitrification often differ between forest types (Finzi et al. 1998; Knoepp and Swank 1998; Reich et al. 1997; Venterea et al. 2003), elevations (Bohlen et al. 2001; Venterea et al. 2003) and landscape positions (Ohrui et al. 1999; Venterea et al. 2003; Zak et al. 1991) owing to site variations with different soil organic matter content (Burke et al. 1999; Hobbie et al. 2007), temperature and soil moisture (Knoepp and Swank 2002). Although a great deal of attention has focused on how little is known about tropical rainforests, even less is known about subtropical rainforests. This ignorance results primarily from the fact that, although they were once widespread, most subtropical rainforests were converted long ago to agriculture before they could be studied (Sun et al. 1998).

The Kenting National Park on the Hengchun Peninsula of Taiwan, located south of the Tropic of Cancer, is geographically divided into a tropical climate region. Forest ecosystems of the Kenting National Park have high species richness and plant diversity because of its special geographic location and, characteristically, the strong effect of the northeastern monsoon winds during the winter (Chen et al. 1997; Hsieh et al. 2000; Su and Su 1988; Sun et al. 1998). The Nanjenshan Nature Reserve of Kenting National Park conserves the last native lowland subtropical rainforest in Taiwan. Moreover, it is noteworthy that the Nanjenshan forests conserve most lowland plant species of Taiwan, and there are also numerous rare and locally endemic species in the Hengchun flora (Sun et al. 1998). Despite the low elevation, it is surprising that temperate-zone tree taxa are also present in the subtropics (Liao 1995; Sun et al. 1998).

Compared with vegetation research little is known about the soil nutrient cycles and dynamics in the Nanjenshan forests. To maintain and assess the ecological functions and biodiversity of the Nanjenshan forests under the impacts of global climate change, it is important to determine soil N transformation rates as basic information for further research. The objective of the present study was to examine the rates of soil N mineralization and nitrification in different seasons (summer and winter) and in different landscape positions (footslope and summit) at Mt Nanjenshan in southern Taiwan. Although net mineralization assays suffer from technical limitations when used as measures of N cycling dynamics, net N mineralization is discussed as an index of plant-available N (Schimel and Bennett 2004), and it is still a useful measure for comparing the differences among various landscape positions at the study site.

Materials and methods

Site description

The study site (22°37′N, 120°10′E) is located within the Nanjenshan Nature Reserve of Kenting National Park on the Hengchun Peninsula at the southern end of Taiwan (Fig. 1). Annual precipitation is approximately 3,000 mm and mean monthly air temperatures are highest in July (28°C) and lowest in January (18°C) (Fig. 2). Typhoons are very common during the summer and the northeastern monsoon winds usually start in late October and continue until late March of the following year (Fig. 2).

A transect experimental site 450 m long and 10–40 m wide was established on the northwestern ridge of Mt Nanjenshan in 1994. A vegetation census and analysis indicated that there was a phenomenon of vegetation compression with great plant species richness over a short range of 200–400 m in elevation (Liao 1995). The vegetation along this 1-ha transect site consists of 139 free-standing woody species in 91 genera and 49 families, which can be divided into three distinct vegetation types (Liao 1995; Wu et al. 2007). Vegetation families found mainly in the tropics dominate the footslope forest. Conversely, families showing definite evidence of northern and eastern Asia affinities are better represented in the summit forest. Compared with the vegetation species distributed on the footslope and summit positions, those on the backslope position show a transition status in forest structure, floristic composition and habitat.

Figure 1 Geographical location of the long-term ecological research site in Nanjenshan Mountain in southern Taiwan.

The underlying bedrock within the study area consists of sandstone and shale of Miocene age. Soils located on the summit position are classified as Typic Paleudults based on Soil Taxonomy and have an argillic horizon (Bt) resulting from strong leaching and illuvial processes of ions and clay particles. Soils located on the unstable backslope position associated with steep slopes and soils located on the footslope position are classified as Typic Dystrudepts and have a cambic horizon (Bw) resulting from weak leaching processes (Hseu et al. 2001). However, the clay contents of the surface soils (<15 cm) range from 27% at the summit to approximately 30% at the footslope; this difference is not significant (Hseu et al. 2001). Tsui et al. (2004) indicated that some soil chemical properties were significantly different between the summit and footslope positions of the transect plot, and Wu et al. (2007) further indicated that the soil nutrient status appeared to be more related to the mineral nutrient concentrations of the dominant tree species than to other environmental factors.

Figure 2 Monthly mean rainfall, temperature and wind velocity at the Nanjenshan Nature Reserve. Windward data were obtained from the meteorological observatory at nearby Nanjen Lake. Leeward data were obtained from the meteorological observatory close to the footslope position of the transect site.

Field and laboratory methods

To determine net rates in situ of N mineralization and nitrification in different vegetation types and landscape positions at the study site, two sets of three 2 m × 2 m plots were established on the summit and footslope positions, respectively. All six plots were chosen under well-covered canopy with dominant tree species in December 2005. In January and August 2006, in situ soil net N mineralization and nitrification were measured using both open and capped core incubations (Raison et al. 1987).

Within each plot, 20 polyvinyl chloride (PVC) cores (5 cm diameter × 24 cm depth) were driven 20 cm into the soil surface. The thin forest floor layer above the mineral soil was included in the PVC cores (Ollinger et al. 2002; Stump and Binkley 1993). Four PVC cores were removed immediately, and from each core the 0–15 cm soil layer was collected and mixed in a polyethylene bag. These bulked soil samples were kept in bags and stored in a cooling system until the soil samples were carried to the laboratory for the time zero determination of ammonium-N and nitrate-N concentrations. The focus was on soil from the surface to a depth of 15 cm because these soils usually have the highest N transformation rates. For the other 16 PVC cores, eight cores were capped with aluminum lids to prevent leaching loss by rainfall. There were two bores near the top side in each core for gas exchange. The other eight open cores were covered with 1 mm nylon nets to prevent new litterfall dropping into the core. After incubation in the field for 7, 14, 21 and 28 days, two open cores and two capped cores were removed from each plot at each sampling date. After taking out the soil, the cores were inserted into the original holes and capped to avoid water accumulating in the empty core and disturbing the soils. The soil temperature at a depth of 15 cm and the air temperature were measured, and soil volumetric moisture was also determined by a TDR (time domain reflectometry) measurement (TRIME-FM Version 3, IMKO GmbH, Ettlingen, Germany) during soil sampling. When new field incubations began, we also collected forest floor samples from within a 50 cm × 50 cm area near each plot to determine the concentrations of C, N, P, K, Ca and Mg in litter at footslope and summit landscape positions.

All collected soil samples were kept in a cooling system until we returned to the laboratory as previously described and stored at 4°C until analysis (within 48 h). Field-moist soils were sieved to <5.66 mm and visible litter and stones were removed. Duplicate 20 g fresh soil samples were shaken with 100 mL of 2 mol L⁻¹ KCl for 1 h and then filtered. The concentrations of ammonium-N and mineral N (ammonium plus nitrate) in the filtered extracts were determined using the steam distillation method (Keeney and Nelson 1982). At the beginning (0th day) and end (28th day) of the field experiment, soil microbial biomass carbon and nitrogen were measured using the chloroform fumigation incubation method (Jenkinson and Powlson 1976), and the procedures proposed by Rice et al. (1996) were followed. Two samples of field-moist soil were weighed (25 g) and adjusted to 55% of water-holding capacity (WHC) and pre-incubated for 7 days. After the pre-incubation period, one set of the duplicate soil samples was placed in a vacuum desiccator and fumigated with ethanol-free chloroform for 24 h; the other set of duplicate soil samples served as the control (unfumigated sample). After the fumigation period, the chloroform was removed and each sample was incubated in a canning jar for 10 days in the dark at room temperature (25°C). A vial containing 2 mL of 2 mol L⁻¹ NaOH was placed in the jar to trap CO₂. After the 10-day incubation the excess NaOH was titrated with a standard concentration of HCl to a phenolphthalein endpoint. To determine the amount of microbial biomass N, the soil was extracted with 2 mol L⁻¹ KCl at a solution : soil ratio of 4:1. The differences between CO₂-C evolution from the fumigated and unfumigated samples and the microbial N flush were converted to microbial biomass C and microbial biomass N using a k factor of 0.5 (Adams and Laughlin 1981; Jenkinson and Powlson 1976).

Spare soil was air-dried, ground and sieved to <2 mm. The gravimetric soil moisture contents of the fresh and air-dried soils were determined by the mass difference before and after drying at 105°C for 48 h. Soil N was assessed using the total Kjeldahl nitrogen procedure (Bremner and Mulvaney 1982) and determined by the steam distillation method. Soil organic C content was determined using the modified Walkley–Black method (Nelson and Sommers 1982). Litter samples were oven-dried at 65°C for 48 h and milled into powder. The total N and organic C contents of the litter were determined using the same methods as those used for the soil samples. The contents of litter P, K, Ca and Mg were determined by nitric-perchloric acid digestion (Miller 1998).

Net N mineralization and net nitrification were determined as the increase in ammonium plus nitrate (mineralization) or nitrate-N (nitrification) relative to time zero for each field incubation. All data were presented on an oven-dry weight basis, and soil N transformation rates and concentrations were calculated on an area basis using previously determined soil bulk density at the study site.

Statistical analyses

Statistical analyses were conducted using the SAS system (SAS Institute 2003). Soil and microbial biomass concentrations of C and N were averaged within each 2 m × 2 m plot for each incubation time interval. A repeated measures multivariate analysis of variance (manova) was used to test the changes in inorganic N concentrations with different landscape positions, core methods, seasons (months) and incubation time. The landscape positions, core methods and seasons (months) served as between-subject factors, and incubation time served as the within-subject factor. The repeated measures manova was carried out using the general linear model (GLM) procedure. Comparisons of soil and litter CN and microbial properties among the different landscape positions and core incubation methods were made using Student’s t-tests. Correlations were tested with the Pearson correlation coefficient using the CORR procedure in SAS.

Results and discussion

Temperature and moisture regime during the field incubation

There was little rainfall during the field incubation in January and higher rainfall in August, particularly in the last two intervals (Table 1). The gravimetric soil moisture contents did not differ between the open and capped cores (data not shown), thus were averaged in each plot. The gravimetric soil moisture contents of the summit soils were significantly higher than those of the footslope soils (P < 0.05) owing to higher soil organic matter (SOM) content on the summit (Tsui et al. 2004) that increasing soil water holding capacity, however, volumetric soil moisture contents were not different. The gravimetric soil moisture content of each plot in August was slightly less than in January, probably due to the higher evapotranspiration during summer. Both soil and air temperatures on the summit position were lower than those on the footslope position, although the elevation difference was <200 m (Fig. 2).

Carbon and N concentration of the soil and litter before incubation

The soil organic carbon (OC) and total nitrogen (TN) concentrations of the summit position were significantly higher than those of the footslope position before incubation (Table 2), and the soil OC contents were in the range reported by Tsui et al. (2004). The soil C:N ratio of the summit position was also greater than the ratio of the footslope position in January, but not in August 2006. On the contrary, litter C and N concentrations did not show any differences between the summit and footslope positions in our collected samples. Only the C:N ratio of the summit litter was significantly greater than the ratios of the footslope litter in January. The N concentration of the litter was lower than the concentrations reported by Wu et al. (2007) who measured the fresh leaves of dominant tree species at the study site. These researchers also observed that foliar N concentrations of the tree species that dominated the footslope position tended to be higher. Wu et al. (2007) further indicated that the dominant tree species of the footslope position tended to accumulate some nutrient elements, including N.

Table 1 Gravimetric soil moisture (moisture; w/w%), volumetric soil moisture (TDR; v/v%), air temperature and soil temperature at a depth of 15 cm for each sampling date at the Nanjenshan transect site

	January 2006				August 2006
	Moisture (%)	TDR (%)	Air temperature (°C)	Soil temperature (°C)	Moisture (%)	TDR (%)	Air temperature (°C)	Soil temperature (°C)
	19 January 2006				22 August 2006
F	31.1 ± 2.2^b	32.2 ± 2.2^a	22.1 ± 0.5^a	19.1 ± 0.1^a	29.9 ± 2.7^b	31.7 ± 0.4^a	26.7 ± 0.2^a	25.6 ± 0.1^a
S	42.2 ± 3.7^a	32.8 ± 3.8^a	20.6 ± 0.1^b	17.9 ± 0.4^b	42.0 ± 3.2^a	35.7 ± 1.4^a	26.7 ± 0.4^a	24.5 ± 0.1^b
	26 January 2006 (0 mm)^†				29 August 2006 (22.5 mm)
F	31.2 ± 1.6^b	30.9 ± 1.1^a	19.2 ± 1.2^a	17.4 ± 0.3^a	30.1 ± 3.2^b	–	29.2 ± 0.3^a	25.3 ± 0.2^a
S	42.7 ± 2.4^a	31.4 ± 1.5^a	17.3 ± 1.0^a	16.8 ± 0.2^b	41.2 ± 2.2^a	36.2 ± 2.7	26.5 ± 0.4^b	24.4 ± 0.3^b
	3 February 2006 (1.5 mm)				4 September 2006 (10.5 mm)
F	32.2 ± 1.8^b	37.2 ± 0.9^a	19.0 ± 0.2^a	17.7 ± 0.2^a	27.8 ± 3.3^b	–	27.1 ± 0.2^a	25.2 ± 0.1^a
S	43.7 ± 2.7^a	34.8 ± 1.9^a	17.1 ± 0.2^a	16.9 ± 0.2^b	39.2 ± 0.9^a	–	27.1 ± 0.2^a	24.3 ± 0.1^b
	9 February 2006 (1.5 mm)				12 September 2006 (87.5 mm)
F	32.4 ± 1.7^b	38.4 ± 1.6^a	15.5 ± 0.4^a	16.8 ± 0.6^a	30.9 ± 2.4^b	–	26.3 ± 0.3^a	24.1 ± 0.0^a
S	44.5 ± 0.9^a	37.9 ± 1.2^a	13.3 ± 0.1^b	16.2 ± 0.3^a	40.0 ± 2.1^a	–	25.3 ± 0.6^a	23.8 ± 1.0^a
	16 February 2006 (0.2 mm)				19 September 2006 (43.0 mm)
F	33.4 ± 2.8^b	36.9 ± 1.0^a	22.7 ± 1.5^a	18.0 ± 0.3^a	30.2 ± 2.5^b	35.1 ± 1.5^a	24.7 ± 0.5^a	22.8 ± 0.1^a
S	45.7 ± 2.6^a	35.3 ± 1.5^a	21.5 ± 0.5^a	16.6 ± 0.3^b	42.9 ± 3.7^a	34.5 ± 2.5^a	22.7 ± 0.3^b	22.0 ± 0.1^b
^†Data in parentheses are cumulative rainfall between two sampling dates. Values are mean ± standard deviation (n = 3). Data with different letters between landscape positions are statistically significant at P < 0.05 (Student’s t-test). F, footslope; S, summit; –, data not available.

Table 2 Organic carbon and total nitrogen of the soil and litter properties for the different landscape positions before incubation

	January 2006		August 2006
	Footslope	Summit	Footslope	Summit
Soil
OC (g kg^−1)	12.3 ± 1.71^b	23.4 ± 3.88^a	13.2 ± 0.78^b	27.1 ± 2.46^a
TN (g kg^−1)	1.11 ± 0.15^b	1.61 ± 0.34^a	0.95 ± 0.09^b	1.84 ± 0.20^a
C:N ratio	10.5 ± 1.7^b	14.7 ± 1.3^a	13.9 ± 1.7^a	14.8 ± 0.4^a
Litter
C (g kg^−1)	376 ± 99^a	417 ± 16^a	402 ± 31^a	418 ± 63^a
N (g kg^−1)	12.8 ± 2.4^a	10.7 ± 0.5^a	10.2 ± 1.9^a	11.0 ± 0.8^a
C:N ratio	29.3 ± 4.5^b	39.2 ± 0.7^a	40.1 ± 5.2^a	38.3 ± 8.0^a
P (g kg^−1)	0.62 ± 0.06^a	0.48 ± 0.02^b	0.47 ± 0.08^a	0.43 ± 0.01^a
K (g kg^−1)	5.15 ± 1.05^a	4.93 ± 0.72^a	5.90 ± 0.99^a	4.58 ± 0.86^a
Ca (g kg^−1)	4.56 ± 2.50^a	3.67 ± 0.74^a	6.17 ± 1.44^a	4.57 ± 1.19^a
Mg (g kg^−1)	3.13 ± 0.25^a	2.90 ± 0.08^a	3.08 ± 0.52^a	2.67 ± 0.31^a
Values are mean ± standard deviation (n = 3). Data with different letters between footslope and summit are statistically significant at P < 0.05 (Student’s t test). OC, organic carbon; TN, total nitrogen.

Vegetation type and tree species have a great effect on surface soil C and N dynamics (Finzi et al. 1998; Hobbie et al. 2007; Lovett et al. 2004; Ollinger et al. 2002). At the transect site, the species composition of the summit forest is discriminated from the footslope forest, although the elevation difference between the summit and footslope is small (Wu et al. 2007; Table 1). Characteristically temperate zone tree taxa (e.g. Fagaceae) occur at the summit position, as well as on the windward slope of the other sites in the Nanjenshan forest (Chen et al. 1997; Sun et al. 1998). Kuo and Lee (2003) indicated that windward species had sclerophyllous leaves with significantly higher leaf mass per area (LMA) than leeward species, and the decomposition rate was found to be negatively correlated with LMA (Kurokawa and Nakashizuka 2008). Most of the windward species studied by Kuo and Lee (2003) were dominant tree species at the summit position at the transect site, whereas the leeward species dominated the footslope position. In view of the leaf morphology, we suggest that the leaf litter of the summit tree species is comparably less decomposable than that of the footslope tree species. In addition, litter chemical composition is related to decomposition. For example, N, P and S concentrations often limit the decomposition rate in the early phase; in contrast, concentrations of Ca and Mn, which are associated with the growth of white-rod fungal species or the activity of lignin degrading enzyme, have positive relationships to the limit value of decomposition in the very late stage (Berg 2000). The mean concentrations of these elements in our litter were not significantly different in the different landscape (Table 2); however, higher concentrations of P, Ca and Mn in soils (Tsui et al. 2004) and P and Ca in the dominant tree species (Wu et al. 2007) may contribute to greater litter decomposition at the footslope position. Although the C and N concentrations of the litter between summit and footslope were similar, a difference in litter decomposability probably resulted in the higher OC accumulated in the summit soils.

Soil microbial biomass

Comparisons of soil microbial properties at the beginning and at the end of the in situ incubations of soils sampled in January and August 2006 are shown in Table 3. After a 28-day incubation in the field, soil microbial C (C_mic), soil microbial N (N_mic) and the C_mic : N_mic ratio between the open and capped core incubations did not differ significantly. The results indicated that the soil microbial properties were not influenced by the different field incubation methods used. In our study, mean microbial biomass C and N concentrations in soils in the footslope position were lower than those in the summit soils. In contrast, the C_mic : N_mic and C_mic : OC ratios in the footslope soils were significantly higher than those in the summit soils (P < 0.05) (Table 3).

Table 3 Comparisons of soil microbial properties at the beginning and end of the in-situ core incubation

	January 2006			August 2006
	Beginning	End		Beginning	End
	0 day	28 day (open)	28 day (capped)	0 day	28 day (open)	28 day (capped)
Cmic (mg CO2-C kg^−1)
Footslope	863 ± 130^b	906 ± 174^a	861 ± 210^a	962 ± 67^b	925 ± 36^b	1012 ± 219^a
Summit	1158 ± 46^a	1106 ± 213^a	1061 ± 189^a	1259 ± 80^a	1349 ± 127^a	1321 ± 70^a
Nmic (mg N kg^−1)
Footslope	32 ± 14^b	43 ± 17^b	44 ± 18^b	71 ± 9^a	32 ± 7^b	50 ± 21^b
Summit	102 ± 14^a	123 ± 22^a	119 ± 25^a	140 ± 28^a	90 ± 20^a	95 ± 16^a
Cmic : Nmic ratio
Footslope	29.3 ± 8.8^a	22.6 ± 4.5^a	20.8 ± 4.6^a	13.7 ± 1.2^a	29.6 ± 4.6^a	22.0 ± 5.7^a
Summit	11.6 ± 1.4^a	9.0 ± 1.1^b	9.0 ± 1.8^b	9.2 ± 1.4^b	15.4 ± 2.2^a	14.2 ± 1.6^a
Cmic : OC (%)
Footslope	7.0 ± 1.1^a	7.4 ± 1.4^a	7.0 ± 1.7^a	7.3 ± 0.5^a	7.0 ± 0.3^a	7.7 ± 1.7^a
Summit	4.9 ± 0.2^a	4.7 ± 0.9^a	4.5 ± 0.8^a	4.6 ± 0.3^b	5.0 ± 0.5^b	4.9 ± 0.3^a
Nmic : TN (%)
Footslope	2.9 ± 1.3^b	3.8 ± 1.6^b	4.0 ± 1.6^a	7.5 ± 0.9^a	3.4 ± 0.7^a	5.2 ± 2.2^a
Summit	6.3 ± 0.9^a	7.7 ± 1.4^a	7.4 ± 1.6^a	7.6 ± 1.5^a	4.9 ± 1.1^a	5.1 ± 0.9^a
Values are mean ± standard deviation (n = 3). Data with different letters between landscape positions are statistically significant at P < 0.05 (Student’s t test). OC, organic carbon; TN, total nitrogen.

The ratio of microbial C to soil organic C has been used as an indicator of C availability or decomposability of the substrate (Bauhus and Khanna 1999; Sparling 1992); it also can reflect the climatic, vegetation, cropping and management history (Sparling 1992). Ratios of C_mic : OC at the Nanjenshan transect site (Table 3) were greater than those recorded from temperate forests (0.5–2.3%, Joergensen et al. 1995; 0.3–9.9%, with a mean value of 1.6%, Bauhus and Khanna 1999), New Zealand pasture (0.99–4.30%; Sparling 1992), Alaskan boreal forest floors (2.4–3.3%; Vance and Chapin 2001) and tropical forest (2.8%; Salamanca et al. 2002). However, there was an obvious pattern in that the C_mic : OC ratios in the summit soils (4.5–5.0%) were comparably lower than those in the footslope soils (7.0–7.7%) at our study site. Vance and Chapin (2001) showed that a lower C_mic : OC ratio implies a lower substrate availability in spruce forest soils corresponding with a higher C:N ratio. Although summit soils had higher OC and C_mic levels, the results suggest that the C bioavailability of footslope soils is higher than that of summit soils, and a relatively larger portion of soil OM is resistant to decomposition on the summit position, as previously suggested.

The incorporation of N into the microbial biomass is affected by the availability of C and N and can be expressed by the proportion of microbial N to total soil N (Joergensen et al. 1995). The N_mic : N_t ratios, which were more variable than the C_mic : OC ratios (Table 3), were within the ranges reported by Bauhus and Khanna (1999) for a large dataset (0.5–10.5%, with a mean value of 3.4%), but our present study did not show any relationship with landscape position or sampling time. The C_mic : N_mic ratios of the footslope (13.7–29.6) were significantly greater than those of the summit (9.0–15.4) because of the lower concentration of N_mic in the footslope soils (Table 3). Anderson and Domsch (1980) reported that the C_mic : N_mic ratios of in vitro bacteria and fungi were 5.6 and 7.1–8.3, respectively, and a recent review by Cleveland and Liptzin (2007) indicated that the ratios ranged from 2.6 to 20.6. Many other researchers have obtained similar values; however, direct estimates of the C:N ratios of fungi growing in situ in forests range from 14.1 to 29.0, with a mean value of 20.2 (Wallander et al. 2003). Our results were higher than those recorded by Anderson and Domsch (1980), Cleveland and Liptzin (2007) and most other reports (particularly for the C_mic : N_mic ratio at the footslope), but still reasonable. The higher ratio caused by higher C_mic values possibly resulted from limitations in the chloroform fumigation incubation method, liberated degradable C from the non-living part of the OM or the higher mineralization of fungi (Martens 1995). As fungi have a higher C:N ratio than bacteria (Wallander et al. 2003), and large microbial C:N ratios are caused by increased fungal-to-microbial biomass ratios (Joergensen et al. 1995), we suggest that there is a larger proportion of fungi distributed in the footslope soil microbial communities than in the summit communities (i.e. the soil microbial populations differed with landscape position). Further research to clarify soil microbial composition in relation to landscape position and vegetation type within the transect site is necessary.

Net nitrogen mineralization and nitrification of the in situ incubations

During the field incubation in January 2006, the soil mean ammonium-N concentrations of the study site ranged from <1 to 4.51 mg N kg⁻¹, the soil mean nitrate-N concentrations ranged from <1 to 7.86 mg N kg⁻¹, and the mean concentrations of mineral N ranged from 1.73 to 11.5 mg N kg⁻¹ (Fig. 3). Both concentrations of ammonium-N and mineral N were significantly different (P < 0.05) between the open core soils and the capped core soils in the first week of sampling, but did not different on the other sampling dates (Table 4). Soil nitrate-N concentrations did not differ significantly between different landscape positions or PVC core methods. Throughout the entire in situ incubation period in January, the soil N concentrations did not differ significantly between the landscape positions or core methods (Table 5).

During the field incubation in August 2006, soil mean ammonium-N concentrations varied from approximately 2 to 7.68 mg N kg⁻¹, soil mean nitrate-N concentrations ranged from 3.81 to 17.4 mg N kg⁻¹, and soil mean mineral N ranged from 5.77 to 23.8 mg N kg⁻¹ (Fig. 4). Throughout the entire incubation period in the field, soil ammonium-N concentrations at the summit were significantly higher than those at the footslope (Fig. 4; Table 5; P = 0.003). A sharp decrease in soil N concentrations in the open core soils on day 21 was observed and was caused by rainfall leaching starting 2 days before the sampling date and continuing for a couple of days at the site (Table 1), resulting in significantly reduced concentrations of mineral N and nitrate-N in the open cores compared with the capped cores (Fig. 4; Table 4). Thus, the soil nitrate-N concentrations differed significantly between the core methods (Table 5; P = 0.04). However, soil mineral N (ammonium-N plus nitrate-N) was not significantly different in the different landscape positions or core methods sampled in August 2006.

We estimated net N mineralization and nitrification rates using a 28-day incubation. However, some of the highest N concentrations were observed before the 28th day during the incubation period (Figs 3,4). For example, the mineral N and nitrate-N concentrations at the footslope (in both open and capped cores) in January were highest on day 21 (Fig. 3), indicating that net mineralization and nitrification rates based on 28-day calculation data could be underestimated. These results also imply that soil N immobilization was greater within the PVC cores over a longer incubation period at the footslope position in January 2006. In addition, differences in the soil N concentration between the open and capped core methods show that there was N leaching in this subtropical rainforest (Fig. 4), as shown in other humid ecosystems. Raison et al. (1987) indicated that N leaching could be assessed by comparing the N quantity presented in these two core methods; however, better field methods have been suggested, for example, the resin core method, to estimate N mineralization and N leaching simultaneously during incubation (Binkley and Hart 1989). In the present study, we focused only on net N mineralization and net nitrification, and other incubation techniques are recommended to precisely assess the N leaching in the future.

As we put the data from two field incubations together and took month as one source of variation, the results of the repeated measures manova indicated that soil ammonium-N concentrations were significantly different between landscape positions (P = 0.01; data not shown), whereas soil mineral N and nitrate-N concentrations were significantly different between months (both P = 0.0003; data not shown). As there were higher concentrations of ammonium-N (particularly in August) (Fig. 4), OC, TN and C:N ratio (Table 2) in the summit soils, our results are consistent with Booth et al. (2005) who found that soil ammonium-N concentration was positively correlated to soil C and N concentration and the soil C:N ratio.

Figure 3 Concentrations of soil ammonium-N (NH4+-N), mineral N and nitrate-N (NO3−-N) during in situ incubation in January 2006 at the Nanjenshan transect site. Error bars are the standard deviation (n = 3); where no bars are shown only one data point (core) is available.

Net rates and amounts of mineralization and nitrification in the two field incubation experiments are shown in Table 6. In January 2006, the amount (mg N m⁻² 28 day⁻¹) of mineralized and nitrified N and the N transformation rate (mg N m⁻² day⁻¹) at the footslope were similar to those at the summit. Comparing these two in situ incubations, the net rates or amounts of mineralization and nitrification in August were 1.3-fold and 2.1-fold higher than those in January at the footslope and 2.29-fold and 2.46-fold higher at the summit, respectively, both as a result of significantly higher concentrations in August. Our results indicated that net N mineralization and nitrification differed greatly between January and August, which represented the coldest/driest month and the hottest/wettest month of the Nanjenshan forest (Fig. 2). At the footslope position, the concentrations of soil mineral N and soil nitrate-N were significantly correlated with soil temperature (r = 0.43, P < 0.01 and r = 0.38, P < 0.05, respectively), but not with soil moisture. The correlations between N concentrations and soil temperatures were stronger at the summit position (r = 0.58 for ammonium-N, P < 0.01; r = 0.60 for mineral N, P < 0.001; r = 0.47 for nitrate-N, P < 0.05), and the N concentrations were not significantly correlated with soil moisture content. At our study site, air and soil temperatures in August were higher than those in January, but soil water regimes at one landscape position were not significantly different between January and August (Table 1). It appeared that at one landscape position, seasonal variation in N transformation was mainly affected by the temperature regime and less affected by the soil moisture regime at the Nanjenshan transect site. This result confirmed the results recorded from laboratory incubations showing that N mineralization was more responsive to temperature change than to moisture change (Knoepp and Swank 2002; Wang et al. 2003).

Table 4 Statistical analysis using the general linear model procedure for soil nitrogen concentrations during in situ incubation periods at the transect site in January and August 2006

Source of variation	January 2006						August 2006
	NH4 ^+-N		(NH4 ^++NO3 ^−)		NO3 ^−-N		NH4 ^+-N		(NH4 ^++NO3 ^−)		NO3 ^−-N
	F	P > F	F	P > F	F	P > F	F	P > F	F	P > F	F	P > F
Week 0
Land	0.04	0.84	0.16	0.71	0.58	0.49	5.78	0.047	0.16	0.70	1.15	0.32
Core	0.40	0.56	0.23	0.66	0.52	0.51	0.09	0.77	0.03	0.87	0.07	0.81
Land × Core	0.53	0.51	1.33	0.31	0.07	0.80	0.09	0.77	0.03	0.87	0.07	0.81
Week 1
Land	3.47	0.14	6.91	0.06	4.99	0.09	7.93	0.026	3.29	0.11	0.04	0.85
Core	22.50	0.009	11.57	0.03	2.65	0.18	0.96	0.36	0.79	0.40	0.09	0.77
Land × Core	8.80	0.04	3.35	0.14	0.43	0.55	1.19	0.31	0.11	0.75	0.39	0.55
Week 2
Land	1.64	0.27	0.81	0.42	0.06	0.81	14.17	0.007	2.90	0.13	0.86	0.39
Core	0.11	0.75	0.64	0.47	0.88	0.40	0.62	0.46	0.04	0.84	0.25	0.63
Land × Core	1.04	0.37	0.00	0.97	0.22	0.67	0.57	0.47	0.28	0.62	0.17	0.69
Week 3
Land	2.53	0.19	2.49	0.19	0.60	0.48	7.28	0.03	0.60	0.46	0.56	0.48
Core	0.24	0.65	0.00	0.99	0.03	0.87	1.50	0.26	18.55	0.003	10.97	0.01
Land × Core	3.87	0.12	0.08	0.79	0.20	0.68	0.65	0.45	2.76	0.14	3.88	0.09
Week 4
Land	0.00	0.98	0.10	0.76	0.22	0.67	2.84	0.14	0.92	0.37	0.01	0.93
Core	0.03	0.88	0.43	0.55	1.23	0.33	0.01	0.91	6.04	0.04	14.96	0.006
Land × Core	1.20	0.34	0.00	0.96	1.12	0.35	0.13	0.73	0.14	0.72	0.06	0.81
Values in bold are significant.

Table 5 Repeated-measures multivariate ANOVA of landscape position (landscape), core method (core) and in-situ incubation period (time) on soil nitrogen concentrations at the transect site in January and August 2006

Source of variation	df		P values
			NH4 ^+-N	(NH4 ^++ NO3 ^−)-N	NO3 ^−-N
			January 2006
Between-subject effect
Landscape	1		0.75	0.54	0.64
Core	1		0.24	0.42	0.81
Landscape × core	1		0.06	0.60	0.58
Error	4
Within-subject effect	Num. df	Den. df
Time	4	1	0.47	0.14	0.29
Time × landscape	4	1	0.43	0.43	0.77
Time × core	4	1	0.65	0.38	0.42
Time × landscape × core	4	1	0.89	0.49	0.88
			August 2006
Between-subject effect
Landscape	1		0.003**	0.14	0.94
Core	1		0.65	0.06	0.04*
Landscape × core	1		0.48	0.76	0.49
Error	7
Within-subject effect	Num. df	Den. df
Time	4	4	0.56	0.12	0.03*
Time × landscape	4	4	0.19	0.47	0.69
Time × core	4	4	0.50	0.03*	0.04*
Time × landscape × core	4	4	0.40	0.42	0.47
Values in bold are significant. *P < 0.05; **P < 0.01 ; df, degrees of freedom.

Figure 4 Concentrations of soil ammonium-N (NH4+-N), mineral N and nitrate-N (NO3−-N) during in situ incubation in August 2006 at the Nanjenshan transect site. Error bars are the standard deviation (n = 3).

Table 6 Net rates and amounts of mineralization and nitrification at the Nanjenshan transect site

	January 2006		August 2006
	Footslope	Summit	Footslope	Summit
Net amount of N mineralization
(mg N m^−2 28 day^−1)	844	868	1130	1990
Net rate of N mineralization
(mg N m^−2 day^−1)	30.1	31.0	40.2	71.3
(mg N g C^−1 day^−1)	0.018	0.010	0.023	0.019
(mg N g N^−1 day^−1)	0.201	0.143	0.313	0.287
Net amount of nitrification
(mg N m^−2 28 day^−1)	691	737	1440	1810
Net rate of nitrification
(mg N m^−2 day^−1)	24.7	26.3	51.5	64.6
(mg N g C^−1 day^−1)	0.015	0.008	0.029	0.018
(mg N g N^−1 day^−1)	0.165	0.121	0.402	0.260
Net rates and amounts of mineralization and nitrification were calculated from differences in the mean concentrations between the beginning and end of the in situ incubation.

In addition to seasonal variation at the same landscape position, the differences in N transformation rates (mg N m⁻² day⁻¹) between the footslope and summit positions were greater in August, and net N mineralization rates and net nitrification rates at the summit were 1.8-fold and 1.3-fold higher than those at the footslope, respectively. In addition to substrate concentration (soil C and N), SOM composition also plays a role in regulating N mineralization, and differences in N mineralization rates among ecosystem types reflect the importance of SOM composition (Booth et al. 2005). Therefore, we also expressed N mineralization and nitrification on the basis of per gram OC and per gram N (mg N g OC⁻¹ day⁻¹ and mg N g N⁻¹ day⁻¹, respectively), and net N mineralization and net nitrification rates at the footslope were both higher than the rates recorded at the summit in January and August (Table 6). This is in contrast to our previous discussion on a per square meter basis and reveals that differences in N transformation rates between landscape positions are affected by substrate quality, which is directly linked to vegetation type (Knoepp and Swank 1998) and/or tree species (Hobbie et al. 2007). Although there were larger soil C and N pools at the summit (Table 2), the higher foliar N concentrations (Wu et al. 2007) and substrate quality (Table 2) of the footslope soils are more favorable to decomposition, suggesting that N in the footslope soil is more labile than N in the summit soil. Our results are consistent with most observations recorded in temperate forests, which indicate that foliar N concentrations are positively correlated with mineralization and/or nitrification and that C:N ratios are negatively correlated with mineralization and/or nitrification (Finzi et al. 1998; Lovett et al. 2004; Ollinger et al. 2002; Stump and Binkley 1993; Venterea et al. 2003).

In addition to substrate quantity and quality, soil microbes play a key role in regulating N transformations. Plant detritus is an important source of labile C that drives soil microbial growth and regulates the balance of soil N mineralization and immobilization (Hooker and Stark 2008). Root litter and foliage litter are important input sources to soil organic C. Stump and Binkley (1993) found that fine-root litter was generally higher in N than foliage within tree species, and root lignin : N and C:N ratios were correspondingly lower. A recent study by Russell et al. (2007) found that the quantity of fine-root inputs influenced SOC stocks more than the quantity of above-ground inputs in a lowland tropical moist forest in Costa Rica. Moreover, Hobbie et al. (2007) indicated that net N mineralization and net nitrification were positively related to root N concentration in both laboratory and in situ assays. Although the fine-root properties have not been determined, we suggest that the quantity and quality of root litter among vegetation types may also affect the microbial biomass and N dynamics at the Nanjenshan transect site.

Conclusions

In general, net N mineralization and nitrification at the Nanjenshan transect site differed with landscape position. Despite the short elevation difference, soil properties and vegetation compositions differed significantly between the summit and footslope positions owing to the special geographic location of the Hengchun Peninsula. The results of our 28-day field incubations indicated that the concentrations of soil ammonium-N varied significantly between the different landscape positions, whereas the concentrations of soil mineral N and nitrate-N determined in August were significantly higher than those determined in January. There were larger soil C and N and microbial N pools at the summit position; however, footslope soils showed higher net N mineralization and nitrification rates (expressed on the basis of per unit C or N). Our results suggest that the substrate properties of the footslope position contribute to the higher net N mineralization and nitrification rate, and that differences in N transformation rates between the landscape positions appear to be related to vegetation type.

Acknowledgments

We are grateful to the staff of the Soil Survey and Remediation Laboratory, Department of Agricultural Chemistry, National Taiwan University, for their assistance during soil sampling at the study site. Financial support for this work was provided by the National Science Council (NSC), Executive Yuan, Government of Taiwan, through grant number NSC 91-2621-B-002-003 and NSC 92-2621-B-002-013.