Trajectories in nitrogen availability during forest secondary succession: illustrated by foliar δ¹⁵N

Abstract

Background

Forest succession is an important ecological process and has been studied for more than a century. However, changes in nitrogen (N) availability during succession remain unclear as they may lead to either N saturation or N limitation. Here, we propose a conceptual model to illustrate changes in N availability during four stages of secondary succession using the natural abundance of ¹⁵N in plant leaves (foliar δ¹⁵N). We predicted that N availability would decline in the early stages of succession and then increase in late stages, coinciding with the changes in foliar δ¹⁵N, with the inflection point varying in different climate zones. Data on foliar δ¹⁵N from 16 succession sequences were synthesized to explore changes in N availability during forest succession.

Results

The compiled data were consistent with the proposed conceptual model. Foliar δ¹⁵N in boreal and temperate forests decreased significantly in the first two stages of succession (estimated to last at least 66 years in temperate forests), at a rate of 0.18‰ and 0.38‰ per decade, respectively, and decreased slightly in tropical forests in the first 23 years. Foliar δ¹⁵N is projected to increase in later stages in all forests, which is supported by observations in both temperate and tropical forests. The inflection points of N availability when N limitation peaked during succession were different in different climate zones, implying different ecosystem N turnovers.

Conclusions

Our study reconciles the controversies regarding changes in N availability during forest secondary succession. Our findings are also useful for predicting the recovery of N and carbon accumulation during succession. Nonetheless, studies on forest secondary succession using foliar δ¹⁵N have thus far been limited, and more research should be conducted to further verify the conceptual model proposed here.

Introduction

Forests are important terrestrial ecosystems that cover 20% of land area and acts as a pivotal C sink in alleviating global warming (Pan et al. 2011). From 2011 to 2020, the global anthropogenic carbon dioxide (CO₂) emissions were 10.6 Pg C yr⁻¹, of which 24% of the emitted CO₂ has been sequestered by forests (Friedlingstein et al. 2021). To date, disturbance and land-use change have increased the proportion of secondary forests worldwide. Although secondary forest contributes higher rates of net primary production (NPP) and, therefore, a larger C sink (He et al. 2017; Yang et al. 2011; Zhao et al. 2021), the process of forest recovery after disturbance, i.e., secondary succession, is likely to be limited by nutrient availability.

Nitrogen is a major limiting nutrient that affects primary production (LeBauer and Treseder 2008) and C accumulation (Luo et al. 2004) during secondary succession. Nitrogen availability, defined here as the supply of N relative to the demands of plants and microbes, changes during forest succession and is likely to regulate the progress of other ecosystem processes (McLauchlan et al. 2007; Odum 1969; Vitousek and Reiners 1975). During the early and middle stages of secondary succession, forests recover biomass quickly and C accumulates rapidly, resulting in low soil N supply relative to the plant N demand. In the later stages, forest ecosystems reach a steady state, with net primary production approaching respiration losses. In these stages, net C accumulation rate would be close to zero (Odum 1969; Vitousek and Reiners 1975). Forest ecosystems reach N saturation with higher N leaching and gaseous losses. Vitousek and Reiners (1975) suggested that forest ecosystems would eventually reach an equilibrium stage with N inputs balancing N outputs (Fig. 1). However, changes in N availability during forest development have not been well examined and these conclusions are still tentative. For example, McLauchlan et al. (2007) suggested that N availability gradually decreased in temperate secondary forests of the United States using tree δ¹⁵N data. In contrast, Sullivan et al. (2019) noted that N availability in tropical forests showed an upward trend by synthesizing soil and foliar N metrics from 10 tropical forest studies. Furthermore, the inflection point of time in N changes during secondary succession remains unknown.

The lack of consensus on the changes in N during forest succession can be attributed to the following reasons. First, the process of forest succession is slow, especially in colder climate. Forest successions lasting several hundred years are difficult to study (Buma et al. 2017, 2019). Second, parameters for quantifying the changes in N varied in different studies, so direct comparisons are difficult. General parameters include ecosystem N input and output (Oulehle et al. 2021), soil available N (Xiao et al. 2018; Yan et al. 2009), soil mineralization and nitrification rates (Dai et al. 2020; Figueiredo et al. 2019; Fisk et al. 2002), and plant and soil N contents or C/N ratios (Johnson et al. 2001; Yanai et al. 2013). However, ecosystem N budgets require long-term monitoring and repeated sampling, resulting in few data sets thus far. Soil available N and soil mineralization and nitrification rates also have large spatial and temporal heterogeneities. In addition, although plant and soil N contents and C/N ratios can reflect changes in ecosystem N status, they cannot be considered reliable sensitive indicators due to the internal stability strategies of plant nutrients (i.e., stoichiometric homoeostasis) (Wang et al. 2021).

Stable nitrogen isotope ratios (¹⁵N:¹⁴N, expressed as δ¹⁵N values) in foliage or tree rings can provide an integrated index to reveal changes in N status during forest succession (Compton et al. 2007; McLauchlan et al. 2007; McLauchlan and Craine 2012; Poulson et al. 1995; Sullivan et al. 2019). Foliar or tree ring δ¹⁵N has been demonstrated to decrease with the decline in N availability (Craine et al. 2009; McLauchlan et al. 2017). McLauchlan et al. (2007) proposed four stages of changes in N availability during forest secondary succession: stage (I) recovery of uptake potential, stage (II) recovery of immobilization, stage (III) equilibration of litter pools, and stage (IV) progressive increase in base cation limitation. Their conceptual model was confirmed by tree ring δ¹⁵N data from Mirror Lake in the northeastern United States (McLauchlan et al. 2007). However, this concept has not been well verified by other studies. On the basis of this model and the general patterns of forest development (Odum 1969; Oliver 1980; Vitousek and Reiners 1975), we further propose that foliar δ¹⁵N can be used to indicate the change in N availability during forest secondary succession (Fig. 1). In the first stage of forest succession, the soil N supply exceeds plant N demand, i.e., high N availability, with reduced soil N immobilization after forest disturbance and a large fraction of ¹⁵N-depleted N lost through nitrification and denitrification. Therefore, foliar δ¹⁵N and soil δ¹⁵N are relatively high in this stage. Forest recovers biomass quickly and plant N demand increases. Nitrogen availability, N loss and foliar δ¹⁵N begin to decrease with forest succession. In the second stage, tree biomass accumulation continues to increase. Meanwhile, soil organic matter accumulates gradually and a portion of soil N is immobilized in organic matter. The soil N supply may be insufficient to meet plant N demands, resulting in a progressive decline in N availability, N loss and foliar δ¹⁵N. The third stage is the mature stage with the recovery of soil N immobilization and equilibrium of the soil N pool. The soil N supply exceeds N demand for plant growth, indicating that N availability would increase gradually. During the last successional stage, i.e., the fourth stage, N is no longer the limiting nutrient for plant growth, while base cation limitation increases.

Fig. 1

Conceptual model of changes in the rate of biomass accumulation, N leaching, N availability and foliar δ¹⁵N during forest secondary succession. Stage (I): The early stage. In the beginning of secondary succession after disturbances by forest fire or logging, there was little aboveground biomass. The soil surface temperature increased as did the rate of soil mineralization and nitrification. The plant N demand was much lower relative to the soil N supply, i.e., high N availability. With the rapid recovery of tree biomass, N leaching, N availability and δ¹⁵N decreased rapidly from the initial maximum due to the great N demand for plant growth. Stage (II): Plant biomass accumulation (but the accumulation rate declined), soil organic matter accumulation, and soil C/N ratio increased in this stage. Although soil mineralization and nitrification were still higher than those before disturbances, a portion of soil N was immobilized in organic matter and the N demand for plant growth increased gradually. Therefore, soil N supply could not meet the N demand for plants. Soil N leaching progressively declined to extremely low levels, and N availability and foliar δ¹⁵N also decreased. Stage (III): The mature stage. Forest biomass accumulation still increased, although the accumulation rate continued to decrease. At this point, the N demand for plants began to decline with the re-accumulation of soil N. The soil N supply exceeded the plant N demand. Nitrogen availability, foliar δ¹⁵N and N leaching began to increase. Stage (IV): The late stage of succession. The ecosystem gradually reached a new steady state as plant biomass accumulation ceased. Ecosystem N input approximately equaled the output. Nitrogen was no longer the limiting nutrient for plant growth while base cation limitation increased. The horizontal dashed lines in b, c and d represent the N input equal output, pre-disturbance N availability and pre-disturbance δ¹⁵N, respectively

In the present study, we synthesized data on foliar δ¹⁵N from succession sequences in different climate zones using “space-for-time” substitutions and examined whether the changes in N availability during forest secondary succession were consistent with the conceptual model proposed above. If the changes in N availability were consistent with the conceptual model, we hypothesized that the inflection point of changes in N availability from the second stage to the third stage would vary in different climate zones. The inflection point of changes in N availability during secondary succession in boreal and temperate forests would take a longer time than that in tropical forests, depending on the ecosystem N turnover.

Methods

We collected foliar N content and δ¹⁵N data from peer-reviewed literature. From the Web of Science, we used the key words “leaf”, “leaves”, “foliar”, “forest”, “nitrogen stable isotope”, “delta N-15”, “¹⁵N natural abundance”, “succession”, “secondary succession”, “development”, and “chronosequence” in different combinations to search targeted publications. Studies related to forest retrogression and primary succession were not considered, and studies that did not collect samples based on forest development time were also excluded from this study. Finally, 14 forest succession sequences from the literature were selected, including 6 tropical forest successions from Sullivan et al. (2019) (detailed information can be found in Additional file 1: Table S1). All studies were of secondary succession, which regenerated after previous disturbances. The “space-for-time” substitutions combined with stable isotope techniques of ¹⁵N natural abundance in leaves were used to investigate the changes in N status during secondary succession. We extracted the data from the figures in references using the software “GetData”.

Data were obtained from 2 boreal forests, 4 temperate forests, and 8 tropical forests. The study sites were located from 65.6°N to 56.6°N, 44.6°N to 39.3°N, and 23.2°N to 22.6°S in boreal, temperate, and tropical forests, respectively. The mean annual temperature was − 2 °C to − 1.7 °C, 4.7 °C to 15 °C, and 21.0 °C to 25.8 °C in boreal, temperate, and tropical forests, respectively. Successional durations differed in the studied biomes, with the longest succession history of 350–400 years in boreal and tropical forests versus approximately 150 years in temperate forests. Among all studies, only the tropical Dinghushan (Fang et al. 2008, 2010) in China had slopes of 15° to 35°, and the others were gently sloping. The dominant species varied in different regions and succession stages. The foliar N contents in seven succession sequences were also compiled.

Furthermore, foliar δ¹⁵N data from two long-term N addition experiments in Swedish boreal forests were collected. This result could indicate the changes in N availability with forest succession. The Norrliden site was in a Pinus sylvestris forest in northern Sweden (Hasselquist and Hogberg 2014; Högberg et al. 2011). Data from Norrliden are available between 1970 and 2009. The first sampling was when this plantation was 18 years. The Stråsan site (Blaško et al. 2013; Högberg et al. 1992) was located on a steep slope in central Sweden with the dominant species Picea abies. The experiment at Stråsan started in 1967 and lasted for 43 years after sampling in 2009. The locations of the study sites are shown in Fig. 2.

Fig. 2

Distribution of 16 forest secondary successions in this study, including 4 boreal forests (2 sites by continuous observation and 2 sites by “space-for-time” substitutions), 4 temperate forests (by “space-for-time” substitutions), and 8 tropical forests (by “space-for-time” substitutions). Detailed information on these sites is available in Additional file 1: Table S1.

Statistical analysis

To find the changes in N availability during forest succession, we used an “lm” fitting regression model (package “lme4” in R) to evaluate the variation in foliar δ¹⁵N with stand age at each site. For all forest sites, linear regression analysis was first used to analyze the overall trends of foliar δ¹⁵N overtime. Then, quadratic regressions were used to explore the temporal pattern with multiple sampling times and two continuous sampling observations of Swedish boreal forests. However, the succession sequences with only two or three available time points were not used for quadratic equations due to insufficient sampling.

For all “space-for-time” substitution sequences, linear regression analysis was performed in N content with succession. For two continuous sampling observation of Swedish boreal forests, quadratic regression fitting was used to assess the temporal pattern in N content with succession.

The changes in N availability with forest successions in different climate regions were also analyzed. To minimize differences in isotope ratios induced by variations in regions, sites, climate and species, foliar δ¹⁵N data were standardized relative to the average value of the first 50 years since succession. Then quadratic regression was used to assess the temporal pattern of normalized foliar δ¹⁵N with forest succession.

The piecewise linear regression function in the “segmented” package was also used for each individual succession sequence and the three climate regions to find the time inflection point of changes in foliar δ¹⁵N. Data were analyzed statistically in R (4.0.3).

Results

Foliar δ¹⁵N and N contents with forest succession by continuous observation at the same site

To date, foliar δ¹⁵N has been monitored in only two Swedish forests for a long time period. At Norrliden site (Hasselquist and Hogberg 2014; Högberg et al. 2011), the N content in leaves ranged from 0.9 to 1.5% (with an average of 1.2%), and showed a pattern of first decreasing, and then increasing (Additional file 1: Figure S1). The foliar N content varied significantly with succession time (P < 0.05). Foliar δ¹⁵N decreased significantly with forest succession. The values decreased from 0.4‰ to −3.1‰, with an average rate of decline of 0.93‰ per decade (Fig. 3).

At Stråsan site (Blaško et al. 2013; Högberg et al. 1992), the foliar δ¹⁵N was generally lower than that at Norrliden, ranging from −0.9‰ to −4.7‰. As at Norrliden, the foliar δ¹⁵N at Stråsan declined over the measurement period. The foliar δ¹⁵N in Stråsan forest decreased rapidly in the first 35 years, by an average of 1.5‰ per decade. However, the foliar δ¹⁵N increased significantly in the following 20 years from −4.7‰ to −3.6 ‰ (Fig. 3). The leaf N content at this site did not change significantly (Additional file 1: Figure S1).

Fig. 3

Changes in foliar δ¹⁵N with succession in two Swedish forests by continuous observations. The red solid circle represents the mean value of foliar δ¹⁵N in the same sampling time. The gray solid circle in b represents the foliar δ¹⁵N in the same sampling plot. The quadratic regression fitting shows that the changes in foliar δ¹⁵N are significant at these two sites. a Norrliden site (Hasselquist and Hogberg 2014; Högberg et al. 2011): y = 0.003x² − 0.2778x + 4.886, R² = 0.85, P < 0.001. b Stråsan site (Blaško et al. 2013; Högberg et al. 1992): y = 0.006x² − 0.385x + 1.051, R² = 0.64, P < 0.001.

Foliar δ¹⁵N and N contents with forest succession by “space-for-time” substitutions

The data sets of foliar δ¹⁵N using the “space-for-time” substitutions were synthesized from two boreal, four temperate, and eight tropical forest sequences. The foliar δ¹⁵N in three climate regions differed significantly from each other (P < 0.001), with tropical forests having the highest value (2.1 ± 2.3‰), temperate forests having an intermediate value (1.5 ± 2.1‰), and boreal forests having the lowest value (− 4.0 ± 1.4‰) (Fig. 4b). In the same climate region, the foliar δ¹⁵N differed significantly among different succession sequences within temperate and tropical forest biomes (Fig. 4a, P < 0.001). However, the foliar δ¹⁵N in boreal forest biome was not significantly different (Fig. 4a).

Fig. 4

Foliar δ¹⁵N of different forest successions in three climate regions. a Foliar δ¹⁵N at each site; b Foliar δ¹⁵N in each climate region. The box mid line, diamond, lower and upper edges, and error bar outside the boxes represent the median and mean values, the upper and lower quartiles, and the < 25th and > 75th percentiles of all data, respectively. The gray blank circle represents the distribution of foliar δ¹⁵N. The asterisks in b represent significant differences in foliar δ¹⁵N among three climate regions. Details of the site information are listed in Additional file 1: Table S1

The foliar δ¹⁵N in two boreal forest sequences (Hu et al. 2014; Hyodo et al. 2013) decreased at a rate of 0.11‰ and 0.44‰ per decade, respectively, which was a lower rate than that in the two Swedish forests using continuous observations. Further quadratic regression indicated that the foliar δ¹⁵N had a slightly shifting trend with succession in boreal forests (Additional file 1: Table S2). The time inflection points of the Alberta and northern Sweden sites were 52 and 123 years after the beginning of succession, respectively, using piecewise linear regressions (Fig. 5).

Fig. 5

Changes in foliar δ¹⁵N with succession by “space-for-time” substitutions. Gray solid circles and black solid circles represent the foliar δ¹⁵N at each site and the mean value of foliar δ¹⁵N at the same sampling time, respectively. The blue solid, black solid and red dotted lines represent the simple linear regression, quadratic regression and piecewise linear regression lines, respectively. The inflection points in Boreal Alberta, Boreal northern Sweden (SE), Temperate Michigan, Tropical Molienda and Tropical Santa Rosa were at the 52nd, 123rd, 36th, 22nd, and 40th years, respectively. No inflection points were found in other sites. Details of the site information are listed in Additional file 1: Table S1

The foliar δ¹⁵N in the four temperate forest sequences (Compton et al. 2007; LeDuc et al. 2013; Li et al. 2013; Wang et al. 2007) also showed a decreasing trend with succession. However, compared with boreal forests, the rate of decline was faster, with an average rate of 0.2‰ to 1.31‰ per decade (on average 0.73‰). Further quadratic regression for the Michigan forest with multiple sampling times and data indicated that the foliar δ¹⁵N first decreased and then increased with succession (Additional file 1: Table S2). Through piecewise linear regression analysis, we found that the inflection point appeared in the 36th year (Fig. 5).

Only five succession sequences (Broadbent et al. 2014; Buzzard et al. 2015; Davidson et al. 2007; Gehring et al. 2005; Piccolo et al. 1994) were sampled more than three times in tropical forests, and their temporal patterns were inconsistent. Among these five sites, a similar trend with boreal and temperate forests was found in Molienda, showing a marked decrease with succession. The average rate of decline was 0.32‰ per decade, which was faster than the observed rates in temperate and boreal forests. Piecewise linear regression analysis indicated that the inflection point was in the 22nd year. The foliar δ¹⁵N in Santa Rosa showed an increasing trend and subsequently decreased with succession, with the inflection point at the 40th year. The foliar δ¹⁵N in other three forests increased significantly with succession (Fig. 5).

The pooled data from each of three climatic regions showed that the foliar δ¹⁵N decreased with succession in boreal and temperate forests. The inflection point of foliar δ¹⁵N in boreal forest was not found by analyzing data from all of “space-for-time” substitutions and continuous observations. We further assessed the asymptotic point of inflection using a single-exponential model and found that the inflection point from the second to third stage occurred over 350 years (Additional file 1: Figure S2). The inflection point in temperate forests was found in the 66th year, and the foliar δ¹⁵N increased with succession. The overall trend in tropical forest succession was indistinct, although it exhibited a slight increase (Fig. 6). The inflection point was found in the 23rd year based on analyses of data less than 100 years in the tropical forest.

Fig. 6

Changes in foliar δ¹⁵N with forest succession in different climatic regions. Foliar δ¹⁵N was standardized to the mean value of foliar δ¹⁵N in the first 50 years of succession at the regional level. Blank black circles represent the mean value of foliar δ¹⁵N in the same succession time. Blank circles in other colors represent foliar δ¹⁵N in different succession sequences. The black solid circle in c represents the outlier. The red line represents the trajectory of foliar δ¹⁵N with forest succession. The blue line represents the piecewise linear regression line. The inflection points in temperate and tropical forests were at the 66th and 23rd years, respectively. No inflection point was found in the boreal forest

Only seven forest successions by the “space-for-time” substitutions in this study reported foliar N content. The foliar N content of the tropical Para and Dinghushan sites increased significantly with succession (P < 0.001). For the other forests, the foliar N content did not change significantly with succession (Additional file 1: Figure S3).

Discussion

Across all data from forest secondary succession studies, we found no distinct pattern in the foliar N content (Additional file 1: Figure S2). Furthermore, the N content among the different climatic zones did not differ significantly. These results indicated that foliar N content could not be used as a sensitive indicator to reflect N availability. In contrast, temporal patterns of foliar δ¹⁵N were clear in 12 of 14 forest succession sequences with relatively complete data. The foliar δ¹⁵N in three climate biomes was significantly different, with the maximum value in tropical forest, followed by that in temperate and boreal forests (Fig. 4b), suggesting the differences in N availability. Our results are consistent with those of studies of N availability in boreal (Blaško et al. 2013; Högberg et al. 1992, 2011), temperate (McLauchlan et al. 2007, 2017), and tropical forests (Hietz et al. 2011), and suggest that foliar δ¹⁵N can be used as a sensitive indicator to evaluate N availability.

Either by continuous observation at the same site (Fig. 3) or by “space-for-time” substitutions (Figs. 4 and 5), the foliar δ¹⁵N in boreal and temperate forests supports our proposed conceptual model, which was consistent with the patterns of tree ring δ¹⁵N. For instance, McLauchlan et al. (2007) analyzed the tree ring δ¹⁵N of temperate forests at Mirror Lake in the northeastern United States and indicated that δ¹⁵N decreased significantly with succession. They then measured δ¹⁵N for more tree rings from 49 undisturbed forest sites in the United States and found that tree ring δ¹⁵N declined since 1850 (McLauchlan et al. 2017). In addition, the trajectory of tree ring δ¹⁵N in the United States since 1970 suggested that N availability gradually decreased, i.e., forests became more N limited. The authors, however, suggested that the potential reason for the decline in N availability was the “fertilization effect” of increasing atmospheric CO₂ concentration, enhancing plant growth and thus progressive N limitation (McLauchlan et al. 2017).

Previous studies demonstrated that nitrate (NO₃⁻) concentrations in stream water changed with forest succession by the long-term (continuous decades) monitoring of N loss within the same forest catchments (Bosch and Hewlett 1982; Vitousek et al. 1979), which can be used to reflect the changes in N availability. For example, Bernal et al. (2012) analyzed the related factors affecting the N cycle of the Hubbard Brook Forest in North America, and found that the NO₃⁻ concentration of stream water could be influenced by forest logging. They suggested that the NO₃⁻ concentration changed with the succession process using the 18-year NO₃⁻ record combined model. When tree biomass continues to increase, more N will be needed for tree growth, resulting in a reduction in the NO₃⁻ concentration in streams. Tokuchi and Fukushima (2009) carried out a more comprehensive study on the NO₃⁻ concentration from 0- to 87-year catchments after afforestation in Japan. They demonstrated that the NO₃⁻ concentration of stream water was related to the stand age, namely, the stages of forest succession. In their study, the NO₃⁻ concentration increased sharply after cutting, but the NO₃⁻ concentration gradually decreased with forest age. When forest was in the mature stage (approximately 25 years), the NO₃⁻ concentration was lowest and remained stable for the long term. These results indicated the inflection point of N availability in the Japanese temperate forest, and the value was similar to our results. Furthermore, Lucas et al. (2016) found that the inorganic N concentration in nine different forest watersheds in northern Sweden decreased from 1985 to 2011, as forest biomass accumulation increased over time. Their results suggested that these forests were still under fast growth and needed more N, which was consistent with our results here. In summary, these studies on the NO₃⁻ concentration of stream water in forest watersheds agreed with our proposed model, indicating that the N supply is insufficient to meet the N demand for tree growth in the early and middle stages of forest succession, therefore, leading to a declining trend of ecosystem N availability.

Unlike boreal and temperate forests, the foliar δ¹⁵N in tropical forest succession exhibited diverse patterns. Among these forests, the foliar δ¹⁵N increased with succession in three of five sequences. The reasons for the different patterns remain unclear. One potential reason might be that trees in colder climate zones need a longer time to reach peak growth and maturity than those in the tropics. Conversely, tropical climates are suitable for tree growth, so trees grow fast and reach their prime quickly. Sierra et al. (2021) indicated that it took approximately 30–50 years to mature according to their study on C turnover time in tropical forests. Poorter et al. (2016, 2021) pointed out that the biomass recovery rate reached 85% after 20 years of succession. They suggested that tropical secondary forests probably reached maturity in approximately just over 20 years, which was consistent with our results that the transition of forest status occurred in the 23rd year. However, the rate of soil N turnover in tropical forests differed from that in boreal and temperate forests (Berg et al. 1993). The process of soil N cycling is influenced by climate conditions, vegetation, soil microbial community, soil properties and their interactions. Previous studies have demonstrated that soil organic layers are thicker in boreal and temperate forests than in tropical forests, and there are slower decomposition rates in boreal and temperate forests than in tropical forests (Melillo et al. 1982; Prescott 2010; Vivanco and Austin 2006). This may lead to a large amount of N being immobilized in the organic layer, resulting in the lack of N available for tree growth in boreal and temperate forests (Attiwill and Adams 1993). For example, the results of ¹⁵N-tracer experiments in temperate forests have suggested that most added inorganic N was immobilized in the organic layer (Li et al. 2019). In contrast, the fast turnover within the organic layer in tropical forests might release the N retained and facilitate plant N uptake (Giweta 2020). A ¹⁵N-tracer study in tropical forest indicated that the organic layer was a short-term sink for N, with most of the N lost after 1 year (Wang et al. 2018).

Conclusions

Our study proposed a conceptual model of N availability during forest succession and explored trajectories of N availability based on foliar δ¹⁵N from different forest successions. Our results suggest that although the general patterns of N availability were consistent among different climate regions, the trajectories of N availability were nonlinear. The inflection points of N availability in succession sequences varied among different climate regions. The shifting time from the second stage to the third stage during secondary succession in boreal forests was much longer than that in temperate forests, while the inflection point in tropical forests appeared the earliest. Our findings are useful for predicting the recovery of N and C accumulation during forest succession in the context of future climate change. However, it is worth noting that studies on foliar δ¹⁵N during succession are still limited. More studies should be conducted to further confirm the conceptual model we proposed. Nonetheless, our results have reconciled some controversial issues about N availability in different studies.