Introduction

Giant miscanthus (Miscanthus × giganteus) is a perennial warm-season C4 grass, capable of reaching heights of 3.5 m and yields of 30 Mg ha−1 at winter harvest [1]. A native of Eastern Asia, giant miscanthus is an infertile hybrid of Miscanthus sinensis and Miscanthus sacchariflorus [2]. Previous research has indicated that giant miscanthus requires few fertilizer inputs [3], which, along with high annual biomass yield, make giant miscanthus a promising low-input bioenergy crop.

Giant miscanthus stands have been known to survive for at least 10 to 15 years [4]. Since incorporation of P after the planting year is not feasible, and P is primarily immobile in soil, appropriate P fertilizer application and incorporation or banding prior to crop establishment may be important, particularly on very low P-testing soils. A recent study reported that annual P removal by giant miscanthus ranged from 2 to 5 kg P ha−1 year−1, which was similar to other perennial grasses including switchgrass and big bluestem (Andropogon gerardii Vitman) but less than the removal from annual crops including corn and sorghum, which ranged from 20 to 54 kg P ha−1 [5]. Studies monitoring giant miscanthus nutrient cycling have found that nutrient concentrations reach their peak at the start of the growing season, become diluted as above-ground biomass increases, and decline further as the plant senesces [3, 6]. According to Himken et al. [3], 36–50 % of P in leaves and stems can be remobilized to the rhizomes. Through this remobilization, giant miscanthus is able to store nutrients during dormancy and recycle stored nutrients to the growing plant tissue in the next growing season. Himken et al. [3] reported that giant miscanthus remobilized 21 % to 46 % of N from leaves and stems to rhizomes and roots. By harvesting after senescence, the plants will not only have nutrient reserve for the next growing season, but the biomass will contain a smaller amount of nutrients that may become atmospheric pollutants during combustion, such as N and S [7].

Researchers who have studied giant miscanthus throughout the world have come to different conclusions about the nutrient requirements by this plant [8]. Yield responses to N have been recorded in some studies [9–12], while others have reported no response [3, 13–22]. Miguez et al. [23] created a model that simulates the effects of N fertilization has on giant miscanthus yield, by compiling recent European research (31 studies). Based on their model, they stated that there is no evidence of response to N fertilizer within the first 3 years of growth. Furthermore, they concluded that compared with typical row crops, considerable yields could be obtained with very little N fertilizer. Most recently, Maughan et al. [24] reported no response to N fertilizer rates (0, 60, 120 kg N ha−1) during establishment of giant miscanthus (first 3 years after planting) at four locations in the US: Illinois, Nebraska, New Jersey, and Kentucky.

The goal of this study was to evaluate the effects of N and P fertilization on the establishment of giant miscanthus in western and central North Carolina. Specific objectives were to (1) determine the influence of increasing rates of P and N fertilizer on biomass yield, (2) identify interactions between P uptake and N uptake, and (3) measure change in extractable P and inorganic N in the soil during the establishment period. This research will provide valuable information to future researchers and growers of giant miscanthus in North Carolina and the US Mid-Atlantic/Southeast Regions.

Materials and Methods

This study was conducted at two sites in North Carolina: the Mountain Horticultural Crops Research Station in Mills River (N 35°25′0.5196″, W 82°33′32.241″) and the Tobacco Research Station in Oxford (N 36°19′39.5256″, W 78°39′20.5986″). Soil types were Bradson gravelly loam (clayey, parasesquic, mesic, Typic Hapludult) at Mills River and Cecil sandy loam (fine, kaolinitic, thermic, Typic Kanhapludult) at Oxford. Plot areas at both sites contained unfertilized fescue (Festuca arundinacea L.) hay for >3 years, prior to the experiment. Sites were moldboard plowed in late winter of each planting year, then disked and tilled with a rotovator two times prior to planting. Pre-plant soil characteristics to a depth of 91 cm for each site are given in Table 1. Temperature and precipitation data were obtained from the North Carolina CRONOS database and are given in Figs. 1 and 2, respectively.

Table 1 Preplant soil characteristics by depth in 2008 at Mills River and 2009 at Oxford, NC
Full size table
Fig. 1
figure 1

Average monthly temperatures and 30-year average (1971–2000) from May 2008 to March 2011 at Mills River and April 2009 to March 2011 at Oxford, NC

Full size image
Fig. 2
figure 2

Monthly sum of daily precipitation and 30-year averages (1971–2000) from June 2007 to March 2011 at Mills River and Oxford, NC. Source: NC CRONOS Database, State Climate Office of North Carolina, 2011

Full size image

Soil samples were collected before planting and before emergence each year. Each subplot was sampled to a depth of 20 cm using a 1.7-cm-diameter hand probe, mixing 8–10 cores to form a composite sample from each plot. Additionally, each of the plots with 135 kg N ha−1 were sampled to a depth of 91 cm at increments of 0–20, 20–41, 41–61 and 61–91 cm using a 5-cm diameter hydraulic probe, mixing four cores from each plot. Samples were dried at 38 °C and ground to pass through a 2-mm sieve.

Soil samples were sent to the NC Department of Agriculture and Consumer Services (NCDA&CS) Agronomic Division soil testing laboratory for routine analysis [25]. Nutrient analysis was performed by Mehlich-3 extraction using inductively coupled plasma (ICP) [26]; cation exchange capacity (CEC) was determined by summation of basic cations (excluding Na) and buffer acidity [27]; soil pH was determined on a 1:1 soil/water volume ratio; and humic matter was determined using NaOH digestion with colorimetric determination [26]. Inorganic N (NO3 and NH4) was measured by KCl extraction using 2.5 g soil with 25 ml of 1 M KCl. Samples were shaken for 1 h on a reciprocal shaker, filtered with Fisherbrand P4 filter paper, and refrigerated at 2–4 °C. Extracts were analyzed colorimetrically using Lachat methods 12-107-04-1-I and 12-107-06-2-A for NO3 and NH4, respectively [28].

Treatments were arranged in a split-plot randomized complete block with four replications; main plots included 0, 147, 294, and 441 kg P ha−1with subplots of 0, 45, 90, and 135 kg N ha−1. Individual subplots included 16 plants on 1-m grid spacing. Phosphorus was applied as triple superphosphate (0–45–0) and was pre-plant incorporated to about 15 cm soil depth. Nitrogen was surface broadcast as ammonium nitrate (34–0–0) pre-plant as well as pre-emergence in subsequent years. Lime was pre-plant applied and incorporated (to 15 cm) into the entire plot area at Mills River at 3.36 Mg ha−1 on 11 June 2008. Herbicide was applied pre-emergence and post-emergence each year at both the Mills River and Oxford sites: pre-emergence herbicide used was S-metolachlor plus atrazine (Bicep II Magnum; Syngenta, Greensboro, NC, USA), post-emergence herbicide was paraquat (Firestorm; Chemtura, Middlebury, CT, USA) and was spot-sprayed where necessary.

Giant miscanthus plants were obtained from Hoffman Nursery, Inc. (Rougemont, NC, USA). Plants were then propagated by divisions and grown in greenhouse 38-cell flats at the Mountain Horticultural Crops Research Station in Mills River, North Carolina. When transplants were 30–45 cm tall they were planted by hand in the Mills River research plots on 17 June 2008 using a 1 × 1 m grid. Planting success 4 weeks after planting was approximately 95 %; plants that had died at that point were replaced with transplants remaining from the initial planting. Rhizome pieces were hand harvested from border plants of the Mills River site in late March 2009 (after one full growing season), rinsed, separated, and cut into pieces approximately 7.5–10 cm in length and were not used unless at least three buds were visible. The rhizome pieces were hand-planted on 22 April 2009 to establish the Oxford site. Neither site was irrigated in the planting year or thereafter. A weed infestation the first year at Oxford resulted in loss of 314 of the 1,024 originally planted trasnsplants. As such, the dead plants were replaced on 8 March 2010 with giant miscanthus plants dug from a nearby research field where rhizomes leftover from 22 April 2009 had been planted on the same day. However, plants missing from the center 4 of each plot were only replaced with border plants from within the same plot. An area of approximately 45-cm diameter by 20-cm depth was excavated for each individual plant that was replanted into the plots. This method allowed all of the plants within the plots at Oxford to be the same age and approximate root mass, and although the replants had to reestablish, created the most uniform plot stand possible within the site.

Plants were measured and harvested after senescence in winter of each year. Plant heights were taken by measuring the tallest part of the plant (the tallest stem in each cluster) of the center 4 plants in each plot. From the 2008 through the 2010 growing seasons, biomass dry yieldwas determined by hand harvesting and weighing the four center plants of each subplot, with the exception of the 2009 growing season at Oxford. Harvests at Mills River took place on 6 February 2009, 17 December 2009, and 4 January 2011. Oxford harvest dates were 5 January 2010 and 21 December 2010. The Oxford harvest on 5 January 2010 included harvest of all plants within each plot (mean number of plants harvested was 10.8 ± 2.4). For other harvests where plants were missing, the number of plants harvested was noted, but if more than two of the four center plants were missing, the border plants were harvested. Beginning with the 2010 growing season harvests, plots were harvested with a mechanical forage plot harvester (Wintersteiger Inc., Salt Lake City, UT, USA) at a height of ~10 cm and included the center 4 plants of each plot. After harvest of record plants in each plot, all other plants in the plot areas were harvested and removed from the field. A biomass subsample from each subplot was ground through a chipper/shredder on site and saved for further preparation and analysis. Biomass samples were immediately weighed after initial field grinding to determine wet weight, dried at 80 °C for 3 days, and re-weighed to determine moisture content. Tissue samples were later ground to pass through a 1-mm mesh screen and sent to the NCDA&CS tissue testing laboratory to determine nutrient concentration and thereby calculate plant nutrient uptake and removal.

Statistical analyses of yield, tissue, plant height, and nutrient removal responses to N and P fertilizer treatments were performed separately for each site and year using SAS PROC MIXED [29]. Type 3 tests of fixed effects were considered significant at the P < 0.05 level. Repeated measures analyses were used to determine effects of time and depth on soil N and P concentrations. To compare surface (0–20 cm) N and P concentrations with the initial years at each site (when samples were only taken for each P-treatment block), N and P concentrations were averaged for each block at each site in subsequent years.

Results and Discussion

Plant Height

Mean plant height in 2008 was greater for plots receiving P fertilizer compared with the control (data not shown). Unfertilized plots averaged 1.93 m while P fertilized plots averaged 2.10 m, with no height difference among the different N or P application rates. In 2009, average height was greater for plants receiving 0 N (3.42 m) than those receiving N fertilizer (average of 3.30 m). Similar results occurred in 2010 when plants receiving 0 and 45 kg N ha−1 had a higher average height (3.72 m) compared with the average height of plants receiving 135 kg N ha−1 (3.57 m). Plants receiving 90 kg N ha−1 averaged 3.67 m, which was not significantly different from the other N rates.

Plant height data were not recorded at Oxford in 2009 due to the poor plant stand. The mean plant height across treatments at Oxford in 2010 was 2.65 m. In 2010, there were no differences in plant height among treatments.

The negative response in plant height to N application rate in 2009 and 2010 at Mills River was likely caused by increased lodging, particularly since the yield in these same plots did not decrease with N fertilization. Increased lodging of giant miscanthus may be related to N application rates. Kaack and Schwarz [30] found that giant miscanthus plants fertilized with 0 or 150 kg N ha−1 had a significantly higher percentage of lodging than those with 75 kg N ha−1.

Studies of other bioenergy crops such as Arundo donax have shown a correlation between plant height and yield [31]. Height data could potentially be a predictor of yield, however studies that have measured height and yield of giant miscanthus have not attempted to determine a relationship between the two factors [17, 32, 33]. Clifton-Brown et al. [34] combined data for 15 Miscanthus genotypes and found that yield and height were not significantly correlated. Across all N rates, our data showed a positive correlation between plant height and yield when data from all three harvests at Mills River and both harvests at Oxford were combined (Fig. 3). However, on an individual site and year basis, the correlation between plant height and biomass yield at harvest was not consistent.

Fig. 3
figure 3

Plant height versus dry biomass yield in 2008, 2009, and 2010 at Mills River and 2010 at Oxford, NC. Data points represent individual subplot

Full size image

Dry Matter Yield

A positive yield response to P fertilizer (P = 0.0019) was observed at Mills River in 2008, where control plots receiving 0 P produced less biomass (2.40 Mg ha−1) than P treatment plots (average of 3.16 Mg ha−1 across all P rates) (Table 2). However, there was no difference in yield among any P treatments indicating that the yield response observed occurred at or below the lowest P rate (147 kg P ha−1). Biomass harvests from the second (2009) and third (2010) growing seasons at Mills River showed no response to P fertilizer. During the establishment year (2009) at Oxford, severe weed pressure negatively impacted biomass yield. Weed control was improved in 2010, resulting in increased plant survival. Yield results from 2010 show a significant response (P = 0.0047) to N fertilizer rate with plots receiving 0 N producing less biomass (average of 6.35 Mg ha−1) than plots receiving N fertilizer (average of 9.29 Mg ha−1 across N-fertilized plots).

Table 2 Mean biomass dry yield at Mills River in 2008, 2009, and 2010 and Oxford, NC, in 2009 and 2010 in response to N and P fertilizer rate
Full size table

Results from this study can be compared to productivity trials across Europe (15 locations) which were initiated in 1993 as part of the European Miscanthus Network (EMN) [34]. First year winter harvest dry matter yields (out of those which survived the winter) ranged from 1.0 to 8.9 Mg ha−1 (in Ireland and England, respectively); second year yields ranged from 2.0 to 32.4 Mg ha−1 (in Portugal and Sicily, respectively; both sites had received regular irrigation); and third year yields ranged from 3.1 to 26.8 Mg ha−1 (at the same sites given in the second year range) [34]. Average yields across all EMN sites were 4.0 Mg ha−1 in year 1, 11.1 Mg dm ha−1 in year 2, and 15.8 Mg ha−1 in year 3 [34]. Compared with the EMN studies, our first year yields were below average, Oxford was below average for the second year, and Mills River was above average for the second and third years. All of the third year yields in the EMN as well as our study were exceeded by third year yields of 37.3 Mg ha−1 in southern Illinois [1].

Results at Mills River generally support previous research where giant miscanthus requires few fertilizer inputs during establishment years [3, 35]. Prior to our study, there had been no published literature reporting giant miscanthus yield response to P fertilizer rates. The P response we observed during the first year at Mills River is likely the result of the relatively low P availability in the soil. As giant miscanthus becomes established, the root system becomes more extensive (with roots able to reach a depth of 250 cm), allowing for better nutrient uptake and survival in soils with low nutrient availability [36]. This could help explain the lack of P response in subsequent years. Also, P is remobilized in giant miscanthus more than N or K [8]. The percentage of P in above-ground biomass recycled into rhizomes has been measured at 33 % [6] and 50 % [3].

Yield results from the first year at Oxford (2009) are not a typical representation of potential first year growth in similar soils or climatic conditions due to the high plant mortality resulting from weed pressure. The positive response to N fertilizer found at Oxford in 2010, is contrary to most previously published studies, which have commonly reported no or very limited N response especially during the first few years of growth [23]. Soil inorganic N (NO3 − and NH4 +) analysis results from samples collected in February 2010 indicate relatively low available N at the beginning of the growing season at Oxford: 6.3 mg kg−1 at 0–20 cm depth (Table 1). Also, little is known about the soil and climate preferences of the N-fixing bacteria Azospirillum doebereinerae associated with giant miscanthus, which has been used to explain a lack of N response in other studies [37].

Although studies reporting no yield response to N fertilizer [3, 13–22] outnumber the relatively few studies that have found an N response [9–12], the latter reported research provides evidence that under some conditions, giant miscanthus will indeed respond to N fertilizer. Heaton et al. [18] reported a compilation of the results of 21 studies measuring giant miscanthus yield response to N application rate in different parts of the world. The test of fixed effects of N on giant miscanthus yield was significant at P < 0.1, indicating a positive response to N. A study in south Germany compared available N (mineral N plus fertilizer) to yield and using the “boundary line approach” concluded that giant miscanthus responds strongly to low N supply by producing lower yields when available N is less than 50 kg N ha−1 year−1 [38]. In contrast, Himken et al. [3] found no yield response to N even with 25 kg N ha−1 in the upper 30 cm of soil. However, Himken et al. [3] only tested N treatments on an already established, 4-year-old crop. The former studies describing no yield response to N either had: relatively higher soil N concentrations [21] than at Mills River or Oxford, only studied crops after they had: (1) at least 3 years of establishment [13, 14, 19, 20, 22], (2) did not use a 0 N control [16, 17], or (3) did not provide pre-plant soil N data [15].

Nutrient Concentration and Removal

Decreasing plant tissue N concentration with time is likely due to the increase in biomass yield, which resulted in dilution of plant tissue N. Lower N concentration in older plants may also be related to an improved ability to recycle nutrients into the rhizomes as giant miscanthus becomes more established. For example, Heaton et al. [7] reported that during the third growing season, tissue N concentration in giant miscanthus decreased from a peak concentration of 1.17 % in the spring to 0.33 % at senescence in late fall, but in the following year, N concentration decreased by a greater amount (from 1.63 % to 0.29 % in the same time frame).

Nutrient dilution may also account for the higher N concentration in plants given 0 P (9,365 mg kg−1) than those with P fertilizer (average of 8,219 mg kg−1) in 2008 at Mills River. A corresponding response was seen in dry matter yield, where plants fertilized with P produced higher yield than the 0 P control. In 2010 at Mills River, plants receiving 135 kg N ha−1 had a greater N concentration (1,951 mg kg−1) than those with less or no N applied (average of 1,707 mg kg−1). Because there was no corresponding yield response in 2010, the increased concentration is likely the result of luxury uptake. The differences in N concentration between P treatments in 2010, however, were likely a result of sample variability as there was no P response to yield and concentration does not increase or decrease consistently with change in P fertilizer rate.

In 2010 at Oxford, the plants receiving 135 kg N ha−1 had a greater N concentration (2,297 mg kg−1) than those with less or no fertilizer (average of 1,912 mg kg−1). This is also probably the result of luxury uptake, similar to the response observed at Mills River in 2010. Beale and Long [6] reported a close similarity of N, P and K concentrations between higher- and lower-yielding stands of giant miscanthus, suggesting that there was no luxury uptake of these nutrients. However, Beale and Long [6] only studied the results of plants receiving 120 kg N ha−1 applied N and did not compare with lower N rates or a 0 N control. There is some precedent for luxury uptake with giant miscanthus. For example, Cadoux et al. [8] concluded that soils high in K may lead to luxury uptake of K by giant miscanthus.

Luxury uptake is also apparent in the analysis of plant tissue P. Every year except for 2008 at Mills River, average P concentration tended to increase with increasing P application (the differences in Mills River 2009 and Oxford 2009 were not statistically significant). Biomass yield in 2008 at Mills River responded to P fertilizer and, correspondingly, the plants with higher biomass yield had lower P concentrations: an average of 310 mg P kg−1 in the plots with P fertilizer (which produced more yield) versus 347 mg P kg−1 in the plots with no P fertilizer and a lower yield. Therefore, the difference in P concentration was likely a result of nutrient dilution. In 2010 at Mills River, 0 P control plots resulted in an average P concentration of 107 mg P kg−1 which was less than the 230 mg P kg−1 found in the 294 kg P ha−1 and 441 kg P ha−1 treatments. Plants given 147 kg P ha−1 contained 165 mg P kg−1, which was not significantly different from those given 0, 294, or 441 kg P ha−1. In 2010 at Mills River, the plants given 0 N had greater tissue P concentrations (275 mg P kg−1) than those receiving N fertilizer (regardless of rate, average of 153 mg P kg−1).

Phosphorus concentration averaged 176 mg kg−1 at Oxford (2010) in plots receiving 0 and 147 kg P ha−1 which was less than the 323 mg P kg−1 in plots receiving 441 kg P ha−1 (data not shown). The 294 kg P ha−1 treatment resulted in an average plant P concentration of 231 mg P kg−1, which was not significantly different from those in the 0, 147, or 441 kg P ha−1 plots. Plants given 0 N had greater tissue P concentrations than those receiving N fertilizer (regardless of rate), which was similar to results at Mills River the same year (2010). At Oxford, the 0 N plants contained 410 mg P kg−1 while the N fertilized plants contained an average of 165 mg P kg−1. The P concentration change at Oxford could be accounted for by the difference in biomass yield—plants that produced more biomass had more diluted concentrations of P. However, yield differences do not account for the P concentration differences in Mills River.

Average N removal across treatments at Mills River ranged from 24.7 to39.5 kg N ha−1and from 2.9 to 17.0 kg N ha−1 at Oxford (Table 3). Average P removal across treatments at Mills River ranged from 0.95 to 3.64 kg P ha−1 and from 0.09 to1.7 kg P ha−1 at Oxford (Table 4). Differences in N and P removal among fertilizer treatments generally correspond to the treatment responses observed for yield and nutrient concentration.

Table 3 Mean N removal at Mills River in 2008, 2009, and 2010 and at Oxford, NC, in 2009 and 2010 in response to N and P fertilizer rate
Full size table
Table 4 Mean P removal at Mills River in 2008, 2009, and 2010 and at Oxford, NC, in 2009 and 2010 in response to N and P fertilizer rate
Full size table

Yield, N concentration, P concentration, and N removal each responded to P rate in 2008 at Mills River. However, P removal among P fertilizer rates was not significant. Nitrogen removal was greatest in the plots receiving P fertilizer (average of 25.76 kg N ha−1), while N removal in the 0 P plots was 21.6 kg N ha−1. While N removal was not affected by any treatments in 2009 at Mills River, P removal was affected by P rate. Just as P concentration increased with increasing applied P, P removal also increased. Removal was greatest (2.04 kg P ha−1) for the highest P rate, and other plots averaged removal of 1.55 kg P ha−1 (147 kg P ha−1 treatment resulted in removal of 166 kg P ha−1, which was not significantly different from the 0, 294, and 441 kg P ha−1 treatments). Nitrogen removal in 2010 at Mills River varied among P treatments, though observed differences did not follow a consistent trend. Phosphorus rate also affected P removal in that plants receiving P fertilizer removed more P (average of 4.14 kg P ha−1) than plots with 0 applied P (2.12 kg P ha−1). Nitrogen fertilizer rate also impacted plant P removal, though an inverse relationship was found and P removal decreased with increasing N rate. Mean P removal decreased with increasing N rates at Mills River (Table 4). The 0 N control plots had significantly greater P removal than the other N treatments.

Neither P nor N removal was affected by fertilizer treatments in 2009 at Oxford, whereas in 2010, N removal increased with increasing N fertilizer rate (Table 3). Nitrogen removal rates for 0 N control plots were significantly less than plots receiving N. Nitrogen removal was less in plots given 45 kg N ha−1 than those given 135 kg N ha−1 and N removal in the plots given 90 kg N ha−1 was not different from plots given 45 or 135 kg N ha−1. Trends in P removal in 2010 at Oxford were similar to P removal in 2010 at Mills River (Table 4). At Oxford in 2010, increasing P fertilization increased P removal. Phosphorus removal at Oxford in 2010 also responded to N fertilizer, where P removal was significantly greater in the 0 N treatment (2.37 kg P ha−1) than the 45, 90, and 135 kg N ha−1 treatments (average of 1.47 kg P ha−1).

Previous studies on crops similar to giant miscanthus have shown a relationship between fertilizer treatment and nutrient uptake. Newman et al. [39] found that perennial warm season forages stargrass (Cynodon nlemfuensis Vanderyst), bahiagrass (Paspalum notatum Flugge), and limpograss (Hemarthria altissima [Poir] Stapf & Hubb) each increased P removal with increasing N fertilizer application. However, we measured a decrease in P removal with increasing N application in 2010 at both sites and no response the previous years. Hussaini et al. [40] reported that N fertilizer up to 60 kg N ha−1 increased the concentrations of N, P, K, Ca, and Mg in dry season maize grain (Zea mays L.), however in corn stover, N fertilizer only increased tissue concentrations of N and Ca. The positive correlation between N applied and tissue N concentration is similar to our results in 2010 at both sites. We also found a positive correlation between N applied and tissue Mg concentration in 2010 at both sites. Hussaini et al. [40] also reported that P fertilization of 40 kg P ha−1 increased N and Ca concentration in maize stover, while reducing K concentration. Also, according to Khan et al. [41], greater applications of P fertilizer increased N uptake in wheat. Our results did not show an increased N or Ca concentration with P fertilization. However, Khan et al. [41] found that increasing P fertilizer rates of 34, 67, 101, and 135 kg ha−1 resulted in decreasing N concentrations of corn leaves, which is similar to the treatment response we found in tissue N concentrations at Mills River in 2008. Our study results provide evidence that giant miscanthus is capable of luxury consumption when supplied with more N and P than needed.

Soil N and P

Results of soil analyses indicated no significant difference in inorganic N concentration of the surface 20 cm among plots for any year at either location. By the time of sampling, inorganic N from fertilizer treatments had likely either been absorbed by plants or lost via denitrification and/or leaching. Soil inorganic N decreased each year of the study at both Mills River and Oxford except between 2008 and 2009 at Mills River when surface inorganic N did not change significantly. Christian et al. [42] reported that more inorganic N fertilizer was recovered by 3-year-old giant miscanthus than by 2- or 1-year-old plants. Therefore, more N was probably absorbed by roots and stored in the rhizomes each year as the plants became established, leaving less inorganic N in the soil. Inorganic N moves with rainfall and normally does not accumulate in North Carolina surface soils that are sufficiently drained for crop production [43]. Therefore, rainfall patterns may help explain changes in soil inorganic N with depth (Fig. 4). Less precipitation and less time between fertilizer application and soil sampling (9 months as opposed to 11 and 10 months in 2010 and 2011, respectively) may account for the higher inorganic N at 41–61 cm in 2009 than in 2008, 2010, and 2011 at Mills River.

Fig. 4
figure 4

Inorganic N (NO3 and NH4) concentration with depth for plots with 135 kg ha−1 applied N from 2008 to 2011 at Mills River and 2009 to 2011 at Oxford, NC. Means with the same letter at individual sites and depths do not differ by year at P < 0.05. Soil inorganic N changed with depth in years 2008, 2009 and 2010 at Mills River and in 2009, 2010 and 2011 at Oxford

Full size image

Soil P concentration in the upper 20 cm changed with time and P fertilizer treatment (Fig. 5). Pre-plant soil test Pin the upper 20 cm of soil was 29.4 mg P kg−1 across all treatments at Mills River. After P fertilizer was incorporated and giant miscanthus plantlets grew in the soil for a year, average soil P concentrations in the upper 20 cmwere66, 118, and 143 mg kg−1, in the 147, 294, and 441 kg P ha−1 treatments, respectively (there was no change in soil P in the 0 P control plots). After the third year of growth, soil P concentrations in the upper 20 cm decreased in the 294 and 441 kg P ha−1 treatments, which was probably due to a combination of P fixation and uptake and storage in the giant miscanthus rhizomes. Time and P treatment did not significantly affect P concentration at depths below 20 cm except for a few instances, which were likely caused by sample variability.

Fig. 5
figure 5

Mehlich-3 P concentration in the 0–20 cm depth of soil from 2008 to 2011 at Mills River and from 2009 to 2011 at Oxford, NC by amount of P applied. Means with the same letter are not different by year at P < 0.05 for a given site and P treatment

Full size image

Soil Mehlich-3 P concentration in the upper 20 cm decreased between the second and fourth growing seasons by 37 and 29 mg P kg−1 in the 294 and 441 kg P ha−1 treatments, respectively, at Mills River and decreased between the second and third growing seasons by 34, 68, and 139 mg P kg−1 in the 147, 294, and 441 kg P ha−1 treatments, respectively, at Oxford. This drop in soil P concentration is not due to leaching deeper into the profile, nor can it be accounted for by P removal alone since P removal rates (approximately 2–4 kg P ha−1) were much lower than fertilization rates (147–441 kg P ha−1).

After incorporating lime into the acidic (pH = 4.9) soils at Mills River, pH remained constant at 5.2 each year, which is still low enough to cause P fixation by Al. Soil nutrient concentrations will need to continue to be measured for the next several years to determine the long-term effects of growing and harvesting giant miscanthus on soil test levels in these soils.

Conclusions

Giant miscanthus produced yields of up to 24 Mg ha−1 during the third growing season. Biomass yield and plant height increased each year and productivity was generally better at Mills River than at Oxford, NC, probably due to poor establishment at Oxford. Phosphorus fertilization only increased biomass yield the first year at Mills River, and yield did not increase beyond the lowest P treatment (147 kg P ha−1). Nitrogen fertilization only increased yield the second year at Oxford, which did not increase beyond the lowest N treatment (45 kg N ha−1). No interactions occurred between N and P treatments.

Plant tissue N, Ca, and Mg concentrations decreased each year, as yields increased. Plant tissue P concentrations decreased from year 1 to year 2 at both sites, but increased from year 2 to 3 at Mills River. Apparent luxury consumption for N and P occurred in 2010 at both sites. An increase in soil Mehlich-3 P was measured after P treatments were applied and the soil test P decreased with time in plots with the greatest P fertilizer rates (294 and 441 kg P ha−1). Soil Mehlich-3 P did not change with time in the 0 P control plots.

Giant miscanthus successfully established at Mills River and Oxford, NC with minimal fertilizer inputs on soils with low initial N and P concentrations. Weed control was crucial during the first year of growth. Generally, the nutrient removed in greatest amounts was K, therefore additional research focusing on K requirements would be beneficial. Further research should also be done to determine long-term effects of growing and harvesting giant miscanthus on soil quality, particularly to provide appropriate P fertilizer recommendations. Studies specifically determining the role of N-fixing bacteria associated with giant miscanthus would improve understanding of N fertilization requirements.