Introduction

Fine root production and turnover are widely recognized as critical parameters in estimating carbon (C) allocation and nutrient cycling in forest ecosystems (Norby and Jackson 2000). With increases in atmospheric nitrogen (N) deposition, root production and turnover may be profoundly altered (Nadelhoffer 2000). However, studies on the influence of N on root systems have often yielded conflicting results, with some studies showing the increase in fine root production and turnover following N fertilization (e.g., Majdi 2001; Son and Hwang 2003), and others reporting opposite patterns (e.g., Albaugh et al. 1998; Burton et al. 2000). These uncertainties may be related to variations in climate (Vogt et al. 1996) and tree species (Withington et al. 2006), so the investigations on the responses of fine root production and turnover to soil N addition in multiple tree species should help to resolve these contradictary results (Hendricks et al. 2006; Guo and Fan 2007).

The responses of fine root production and turnover to changes in soil N availability may also be related to methodological differences (Hendricks et al. 1993; Gill and Jackson 2000; Nadelhoffer 2000; Guo and Fan 2007). For example, Albaugh et al. (1998) reported that, in a Pinus teada forest, N additions decreased fine root production and turnover estimated by sequential soil cores; this contrasts with the increased results estimated by minirhizotron in the same species (King et al. 2002). Several techniques have been developed to quantify fine root production and turnover, but each method has both advantages and disadvantages (Vogt et al. 1998). The sequential soil core together with decision matrix calculations may either underestimate or overestimate the production and turnover in some forests (Hertel and Leuschner 2002). The ingrowth core can be suitable for the comparisons among different sites or treatments (Hertel and Leuschner 2002), even though it greatly modifies the soil environment (Vogt et al. 1998). The minirhizotron technique has advantages in monitoring root dynamics in situ (Vogt et al. 1998), but is more likely to overestimate fine root turnover because this method focuses mainly on the root tips, which are the most dynamic fraction of the root system (Wells et al. 2002). To date, there is no consensus for researchers to determine which approach is the best for their systems (Vogt et al. 1998; Hendricks et al. 2006; Noguchi et al. 2007). It has been suggested that several approaches should be employed simultaneously to estimate fine root production and turnover at the same site (Hertel and Leuschner 2002; Trumbore et al. 2006; Hendricks et al. 2006). Yet such studies are still limited, particularly when fine root production and turnover are considered along with fine root morphology and biomass to examine their responses to N fertilization.

In the temperate region, Larix gmelinii Rupr. (larch) and Fraxinus mandshurica Rupr. (ash) are dominant tree species with extensive plantations in Northeastern China. These two species differ in taxonomy rank (i.e., gymnosperm vs angiosperm) and in the type of mycorrhizal infection (i.e., ectomycorrhizal vs endomycorrhizal). Previous studies in this region on the soil of Hap-Boric Luvisols found that annual soil N mineralization rate was higher in ash plantations than that in larch (Chen et al. 1999). Fine root standing biomass of ash plantation was also larger than that of larch (Cheng et al. 2005; Mei et al. 2009). These suggested that soil N availability may influence fine root production and turnover in both species, and the magnitude and direction to which the effect of N on fine roots may differ with species and methods examined, but all of these still remained unclear. In this study, sequential soil core and ingrowth core methods were employed to evaluate the responses of fine root (≤1 mm) standing biomass (living and dead), root length density (RLD), specific root length (SRL), production and turnover to N fertilization in larch and ash plantations. Our hypotheses were: (1) fine root biomass, RLD, and production of both species would be reduced following N fertilization, but turnover would be increased, because maintaining a small quantity of roots in fertile soil may satisfy the resource acquisition, and shorter lifespan would maximize the efficiency of nutrient uptake according to the cost–benefit theory (Eissenstat and Yanai 1997); (2) given fine root biomass and RLD decrease after N fertilization, we predict that SRL may be unchanged, due to the efficiency of resource acquisition of individual roots which should be maintained to compensate for the decline in total length of explorative roots (Eissenstat and Yanai 1997); and (3) fine root production and turnover, estimated by different methods, may display different responses to N fertilization, due to these two variables being largely influenced by the specific approach used (Vogt et al. 1998; Hendricks et al. 2006).

Materials and methods

Study site and N fertilization

The study site was located at Maoershan Experimental Station of Northeast Forestry University, in Heilongjiang, China (45°21′–45°25′N, 127°30′–127°34′E). The site has a continental temperate monsoon climate, with mean January, July and annual temperatures at −19.6, 20.9 and 2.8°C, respectively, and a growing season ranging from 120 to 140 days. The mean annual precipitation is 723 mm with 477 mm distributed in June, July and August (Zhou 1994). Mean air temperature, precipitation and evaporation during the 10 years prior to this study (from 1993 to 2002) are shown in Fig. 1. Soils are Hap-Boric Luvisols developing from granite bedrock (Gong et al. 1999), and exceed 50 cm in depth with high organic matter content and are well drained (Wang et al. 2006). The site has an elevation ranging from 450 to 500 m above sea level.

Fig. 1
figure 1

Mean monthly temperature (°C), rainfall (mm) and evaporation (mm) from 1993 to 2002 in the study area

Full size image

Plantations of larch and ash were established in 1986 by planting 2-year-old seedlings. The stand properties and soil characters of the two plantations are shown in Table 1. In May 2002, six plots (20 × 30 m) were randomly chosen for each species. Shrubs in the study plots were removed in September 2002. From May 2003 to September 2005, ammonium nitrate (NH4 +–NO3 −) was applied at a rate of 10 g N m−2 year−1 in three plots for each species, and the other three plots was left as control. The proportion of the 10 g N m−2 year−1 was applied each month during the 5-month-long growing season: 15.25% in May and September, 21% in June and August, 27.5% in July, respectively. This application pattern was designed to track the natural temporal patterns of N mineralization occurring during that particular month (see also Chen et al. 1999).

Table 1 Stand age, diameter (DBH), density and soil properties in Larix gmelinii and Fraxinus mandshurica plantations
Full size table

Sequential soil coring and measurements of root morphological indices

Eight soil cores (60.4 mm internal diameter) were sampled in 0–30 cm of the soil at 10-cm intervals (i.e., 0–10, 10–20, and 20–30 cm) from each plot every month (or sometimes every 2 weeks) from April to October in 2003 and 2004. Soil samples were placed in the plastic bags, sealed and transported to the laboratory and stored at 2–4°C until later processing (within 1 month of the sampling). In the laboratory, roots were sorted carefully out of the soil, and root samples were washed free of soil particles by deionized water (1–2°C) through a 0.42-mm mesh. Roots were then sorted into living and dead based on color and resilience (Bengough et al. 2000). Both fractions were then separated into fine roots (diameter ≤1 mm) and coarse roots (1–2 mm).

Root samples were scanned with an Epson digital scanner (Expression 10000XL 1.0) and analyzed using the software WinRhizo Pro (S) v. 2004b (Regent Instruments, Canada) to obtain the root length. After length measurements, roots were dried at 65°C to a constant weight for the determination of dry mass. The SRL (m g−1) was calculated from total fine root length in each soil core divided by their dry weight (Eissenstat and Yanai 1997), and the RLD (m m−2) were calculated by dividing total fine root length by soil core area. The average living biomass, necromass, and RLD for each sampling time were all expressed as per unit area for 0–30 cm depth.

Ingrowth core

Ingrowth cores were installed according to Neill (1992). Sixteen soil cores with an inner diameter of 100 mm were sampled in September 2003 in each plot. Each core was then divided by depth at 10-cm intervals. All visible living and dead root materials were picked out (Hertel and Leuschner 2002), and the remaining soil material was put back into the same hole with the center marked with a flag. These ingrowth cores were harvested with a soil core of 60.4 mm internal diameter 12 months later. In the laboratory, living and dead roots of ingrowth cores with different diameter classes were sorted and weighed following the same procedure described for sequential soil cores. Fine root biomass in the cores was calculated as fine root production (Vogt et al. 1998) and was expressed as g m−2 year−1.

Fine root production and turnover rate

The fine root production were estimated following the procedures below: (1) the sequential core with “max–min” calculation (SC-MM) (Son and Hwang 2003); (2) the sequential core with the “positive increments” calculation (SC-PI) (Lauenroth 2000); (3) the sequential core with “decision matrix” calculation (SC-DM) (Hertel and Leuschner 2002); and (4) the ingrowth core method (IC) (Hertel and Leuschner 2002). Specifically, for the SC-MM method, biomass production was calculated from the significant difference between minimum and maximum of fine root standing biomass within each year (Hendricks et al. 2006; Jourdan et al. 2008). Mean values for fine root biomass per plot were calculated on each sampling date. In the SC-PI method, the production was considered as the sum of all positive increments in both living and dead root biomass between sampling intervals (Lauenroth 2000). The SC-DM method was estimated by the relative changes of living and dead root biomass according to the decision matrix (Fairley and Alexander 1985). For the IC method, we calculated fine root production in the cores as the increase in root biomass since root buried to the harvesting (i.e., from September 2003 to October 2004). Fine root turnover rates were estimated by dividing productions from sequential soil cores and ingrowth cores by standing biomass, respectively (Vogt et al. 1998; Gill and Jackson 2000).

Statistical analysis

For each species at a particular sampling date, root standing biomass, necromass, RLD and SRL were calculated for the two treatments (fertilization vs control). All root variables were calculated using a diameter of ≤1 mm. The differences in fine root living biomass, necromass, RLD and SRL between fertilization and control and between the two species were tested by a mixed-level (2 × 2 × 17) three-way (species, N-treatment and sampling date) factorial ANOVA. Differences in production and turnover rates among the species, year, treatment and method were tested by a mixed-level (2 × 2 × 2 × 4) four-way (species, year, treatment and method) factorial ANOVA. All calculations were performed by the SAS software V. 6.12 (PROC GLM procedures; SAS Institute).

Results

Living root biomass, necromass, RLD and SRL

Ash had greater fine root biomass, necromass, RLD and SRL than larch across treatments and sampling dates (all P values <0.01; Table 2; Figs. 2 and 3). Living fine root biomass (mean ± SE) for ash and larch in control plots, averaged across sampling dates, were 130.7 ± 11.8 and 81.5 ± 8.0 g m−2, respectively. Both mean RLD and mean SRL for ash were roughly twice as large as those for larch (Fig. 3a, b). Seasonal patterns of fine root variables (living biomass, necromass, RLD and SRL) also differed between species (Table 2; Figs. 2 and 3). For ash in the control plots, fine root biomass was higher in spring (April) in 2003 and early summer (June) in 2004 than other seasons (Fig. 2b), RLD and SRL were higher in summer (July–August) than other seasons in both years (Fig. 3a, b). But for larch, higher fine root biomass, RLD (only in 2003) and SRL occurred in summer (July–August) than other seasons (Figs. 2a and 3a, b).

Table 2 ANOVA of species, year, treatment, sampling date and method on fine root (≤1 mm) living biomass, necromass, root length density (RLD), specific root length (SRL), production and turnover in Larix gmelinii and Fraxinus mandshurica plantations
Full size table
Fig. 2
figure 2

Seasonal patterns of fine root (≤1 mm) biomass (g m−2) and necromass (g m−2) for Larix gmelinii (a) and Fraxinus mandshurica (b) plantations at 0–30 cm depth in control and N-fertilized plots in 2003 and 2004. Bars Standard errors (n = 3)

Full size image
Fig. 3
figure 3

Seasonal patterns of fine root (≤1 mm) length density (RLD, m m−2) (a) and specific root length (SRL, m g−1) (b) for Larix gmelinii and Fraxinus mandshurica plantations at 0–30 cm depth in control and N-fertilized plots in 2003 and 2004. Bars Standard errors (n = 3). Lg, Larix gmelinii; Fm, Fraxinus mandshurica

Full size image

In both species, N fertilization significantly reduced living root biomass and RLD (two P values <0.01; Table 2), but not necromass and SRL (P > 0.05; Table 2). Compared with the control plots, average living root biomass for ash decreased by 26% in the N-fertilized plots in both years, but that for larch did not change significantly (Fig. 2a, b). There were also significant interactions between species and N treatment for fine root biomass, RLD and SRL (all P values <0.05; Table 2), but not for necromass. Moreover, N fertilization did not change the seasonal patterns of these fine root variables (Figs. 2 and 3), confirmed by the non-significant interactions between N fertilization and sampling date (Table 2).

Fine root production

Fine root production differed significantly between species, years and methods (all P values <0.01; Table 2). Ash had greater production than larch (Table 3). Fine root production estimated by sequential soil cores differed with the calculation method (Table 3). Among the three calculation methods, SC-PI and SC-DM yielded higher production estimates (165.18 and 157.80 g m−2 year−1 for larch, 240.84 and 206.34 g m−2 year−1 for ash) than SC-MM (52.12 g m−2 year−1 for larch and 79.92 g m−2 year−1 for ash) (Table 3). The ingrowth core method yielded lower production estimates than the soil core methods, particularly in larch (Table 3). N fertilization did not significantly change production estimates yielded by sequential soil cores with all three calculation methods in both species (Table 3). However, with the IC method, N fertilization reduced fine root production dramatically (by 70% for larch and 54% for ash comparing with the control plots) (P < 0.05; Table 3).

Table 3 Fine root (≤1 mm) production (g m−2 year−1) and turnover rate (year−1) in Larix gmelinii and Fraxinus mandshurica plantations
Full size table

Turnover rate

Fine root turnover rates were similar in the two species, but varied significantly with year, treatment and method (all P values <0.01; Table 2). The interactions among year, treatment, and method were also significant (all P values <0.05; Table 2). Turnover rate estimated by sequential soil cores varied with calculation methods (Table 3). In the control plots, SC-PI and SC-DM yielded higher turnover rates than SC-MM and IC (Table 3) in both species. N fertilization significantly reduced turnover rates in both species in the IC method (P < 0.05; Table 3), but had no significant influence in all three soil core methods.

Discussion

N fertilization effects on fine root biomass, RLD and morphology

Our results indicated that N fertilization significantly reduced fine root biomass and RLD in ash (Table 2; Figs. 2b and 3a), but not in larch, partly supporting our first hypothesis. The response of fine root biomass to N fertilization in ash roots was expected and is consistent with the previous reports of decreased standing root biomass (<1 mm) associated with increased soil N availability, such as in a Pinus resinosa plantation in Wisconsin, USA (Haynes and Gower 1995). Lower standing root biomass under higher soil resource availability is also consistent with the predictions based on cost-benefit analysis, i.e., under higher soil resource availability, more resources are devoted to aboveground (Eissenstat and Yanai 1997). Significantly lower RLD in the fertilized plots in ash was essentially the same response as the lower root biomass (Fig. 3a).

In contrast, N fertilization did not change standing fine root biomass and RLD in larch (Figs. 2a and 3a). There are various reasons for this response. First, larch is an ectomycorrhizal species, and its nutrient uptake may be highly dependent on mycorrhizal hypha (Lilleskov et al. 2002), and it is likely that it was these hypha, but not roots themselves that were sensitive to changes in soil N availability (Treseder and Allen 2000). Although we did not directly assess the production of mycorrhizal hypha in response to N fertilization, our previous study did show that N fertilization significantly reduced the mycorrhizal colonization rate in larch (Sun et al. 2007). Second, larch generally has a lower ratio of total root biomass to aboveground biomass (0.18) than ash (0.27) in the upper 30 cm (Mei et al. 2009), so maintaining a lower fine root biomass is perhaps a common pattern in conifer species compared with hardwoods (Jackson et al. 1997). Under a low root biomass, further reductions in biomass are less likely. Third, the response of root standing biomass to soil N availability may be species-specific (Burton et al. 2000). Inagaki et al. (2009) suggested that conifers tend to have a lower demand of N, but higher N use efficiency than hardwood species.

The finding that SRL was not altered by N fertilization in both species was expected, confirming our second hypothesis. To acquire plenty of nutrients or water, trees generally employ either an extensive strategy, in which root biomass and total length increase, or N intensive strategy via root morphological change (cf. Ostonen et al. 2007). Under high nutrient status, trees may decrease their C investment in explorative fine root length growth, shown by the lower root biomass and RLD, but the SRL remains unchanged (Fig 2). On the other hand, root morphology is generally considered a conservative trait (Pregitzer 2008) and may be insensitive to changes in soil environment. Other studies also found similar results. For example, root morphology was often not affected by N additions in nine North American tree species including both conifers and hardwood species (Pregitzer et al. 2002). In contrast with the lack of response of root morphology to N fertilization, SRL differed with sampling date (Table 2; Fig. 3b), suggesting that root morphology may differ with season (Wang et al. 2006).

N fertilization effects on fine root production and turnover

Fine root production varies widely between methods, and across species and ecosystems (Vogt et al. 1996). The effects of soil N availability on fine root production have been studied during the past decades, and these studies reported an increase, decrease, or lack of change in fine root production (see reviews in Nadelhoffer 2000; Guo and Fan 2007). Our study showed that fine root production estimated by sequential soil cores (e.g., SC-DM) was not altered significantly by N fertilization in both species, but that estimated using the IC method, it was decreased significantly by N fertilization (Table 3). Hendricks et al. (2006) suggested that sequential soil cores may not be a reliable method for estimating fine root production because it lacks the ability to capture simultaneous root birth and death, among other shortcomings. Even through the SC-DM approach considered the differences among root live biomass, necromass and decomposition in sampling interval, its estimates are still full of uncertainty (Hendricks et al. 2006). By contrast, IC may be more reliable in comparing different treatments, and previous studies have also suggested that this method may be suitable for comparisons among different sites (Tateno et al. 2004) or species (Hertel and Leuschner 2002). However, the production estimates yielded by the IC method seem low, and Hendricks et al. (2006) also reported low production estimates yielded by IC methods as compared to minirhizotron methods. The low estimates in the IC method may be due to the lag of root growth, because it often takes some time for roots to reach the soil within the cores (Hendricks et al. 2006). Moreover, fine roots growing into the ingrowth cores were mainly lower order roots (1st–3rd orders; Liu et al. 2009), whose biomass only makes up a fraction of total fine root biomass <1 mm (which may include 1st–4th order roots and some 1st–5th order roots; see Wang et al. 2006), so the lower production estimates by IC seem reasonable as compared with soil core estimates. Nonetheless, the reduction in root production by N fertilization is probably closer to reality as it not only agrees with the standing biomass estimates (lower root biomass in fertilized plots) but also confirms the results by our minirhizotron measurements during 2004 and 2005 in the same study site, which showed that N fertilization significantly reduced root production (Yu et al. 2009).

Similarly, responses of fine root turnover rate to N fertilization also differed between methods (Table 3). Sequential soil cores showed that fine root turnover rates did not significantly change under N fertilization in both species though there was a decreasing trend in larch and an increasing trend in ash (Table 3). In comparison, the IC approach showed that turnover rates decreased substantially in both species (from 0.23 to 0.08 year−1 for larch and 1.08 to 0.71 year−1 for ash; Table 3). These results support our third hypothesis that the responses of fine root production and turnover to soil N fertilization yielded different results in different methods, and also rejected the proposition in the first hypothesis that turnover rate would increase. Although soil core and IC methods yielded different results (Table 3), the effects of N on root turnover rate suggested that the IC method was probably more reliable. Therefore, our results seem to support the view that fine root turnover rate decreased after N fertilization (Nadelhoffer 2000; Guo and Fan 2007). This finding is similar to a number of previous reports (e.g., Burton et al. 2000), and suggests that roots may be maintained as long as the benefit (e.g., nutrients) they provide outweighs the C cost of keeping them alive (Eissenstat and Yanai 1997; Burton et al. 2000). Further evidence supporting this finding is the significant increase in fine root diameter in both species by N fertilization (average diameter increased from 0.46 to 0.55 mm for larch, and from 0.42 to 0.54 mm for ash) as reported by Yu et al. (2009). Increasing root diameter may increase root lifespan and thus reduce fine root turnover rate (Kern et al. 2004). Clearly, how root turnover rate responds to N fertilization deserves more study, and adopting the new branch order approach (see Pregitzer et al. 2002; Guo et al. 2008) may increase the precision in detecting the responses of root turnover rate to changes in soil N availability, as different branch orders within the fine root group may show markedly different turnover rate.

In conclusion, our study showed that N fertilization significantly changed fine root standing biomass, RLD, production and turnover rate in larch and ash, but the direction and magnitude of change depended critically on the species and methods. N fertilization significantly reduced living root standing biomass and RLD in ash, but not in larch. Necromass, SRL and seasonal patterns of all root variables were not altered by N fertilization in both species. None of the methods associated with the sequential soil cores showed any N fertilization effects on fine root production and turnover rate, but the IC method yielded significant lower estimates for root production and turnover rate under N fertilization in both species. Our results suggested that the responses of fine root production and turnover to N addition are species-specific. Future studies including more species and methods may further improve our understanding of the relationship between fine root dynamics and soil N availability.