About Us | Help Videos | Contact Us | Subscriptions
 

Soil Science Society of America Journal - Soil & Water Management & Conservation

Soil-Test Biological Activity with the Flush of CO2: II. Greenhouse Growth Bioassay from Soils in Corn Production

 

This article in SSSAJ

  1. Vol. 82 No. 3, p. 696-707
    unlockOPEN ACCESS
     
    Received: Jan 10, 2018
    Accepted: Mar 15, 2018
    Published: May 10, 2018


    * Corresponding author(s): alan.franzluebbers@ars.usda.gov
 View
 Download
 Alerts
 Permissions
Request Permissions
 Share
  1. Alan J. Franzluebbers *a and
  2. Mary R. Pershingb
  1. a USDA-ARS 3218 Williams Hall NCSU Campus Box 7620 Raleigh, NC 27695
    b North Carolina State Univ. Dep. of Crop and Soil Sciences Raleigh, NC 27695
doi:10.2136/sssaj2018.01.0024
Core Ideas:
  • Grass growth in the greenhouse was dependent on soil nitrogen mineralization.
  • Soil-test biological activity was a valuable indicator of nitrogen mineralization.
  • Biological activity, residual inorganic nitrogen, and total nitrogen were most important.

Abstract

Soil nitrogen (N) mineralization is variably affected by management and edaphic conditions. A routine soil test that reflects both soil biological activity and N mineralization could improve predictions for N fertilizer recommendations to cereal grains on different soil types and landscape settings. We collected soils from 47 corn production fields in North Carolina and Virginia at depths of 0 to 10, 10 to 20, and 20 to 30 cm and evaluated soil C and N characteristics in association with sorghum-sudangrass [Sorghum bicolor (L.) Moench ssp. Drummondii] dry matter production and N uptake during 6 to 8 wk of growth in the greenhouse. Plant dry matter and N uptake were strongly associated, as expected. Plant available N (sum of net N mineralization during 24 d of aerobic incubation + residual inorganic N) had the strongest association with plant dry matter production (r2 = 0.76) and N uptake (r2 = 0.85). However, the flush of CO2 during a 0- to 3-d period following rewetting of dried soil was nearly equally effective at r2 = 0.74 and r2 = 0.76, respectively. Multiple regression models with 4 ± 2 additional variables led to r2 = 0.88 ± 0.10 among different separations of data based on depth, region, and soil textural class. We suggest the optimum combination of variables to predict soil N availability would be the flush of CO2, residual inorganic N, and total soil N concentration, as they balance relevant scientific information with limited soil-testing resources (time and labor). We demonstrated that the flush of CO2 was a rapid and reliable indicator of soil N availability.


Mineralization of N from soil has a firm scientific foundation from various studies conducted under standardized laboratory conditions (Stanford and Smith, 1972; Gianello and Bremner, 1986; Cabrera and Kissel, 1988; Curtin et al., 1998). A wide variety of aerobic and anaerobic incubations and chemical extractions have been evaluated to estimate soil N availability (Griffin, 2008; Schomberg et al., 2009). Variations in laboratory methods for estimating N mineralization cause some discrepancies in how to relate soil N mineralization to potential plant N uptake (Griffin, 2008).

Laboratory-determined soil N availability indices should associate with plant N uptake if there are no confounding factors on plant growth. Association of N availability indices with plant N uptake has largely been successful if the index of N availability is robust (Haney et al., 2001; Nyiraneza et al., 2009). Therefore, controlled greenhouse studies can provide a reasonable validation of N availability from N indices. From seven New Jersey soils varying in soil organic C concentration from 9 to 41 g kg−1, wheat (Triticum aestivum L.) dry matter production and N uptake during a 2-mo growth period in the greenhouse was strongly related (r2 = 0.95) to plant available N, as determined by acid hydrolysis (Purvis and Leo, 1961). However, growing wheat on these same soils again 1 yr later in the greenhouse resulted in weaker association (r2 = 0.47). From 39 soils collected across the United States, Stanford et al. (1973) showed strong association (r2 = 0.85) of sudangrass [Sorghum bicolor (L.) Moench spp. Drummondii] N uptake during 6 wk of growth in a greenhouse with N mineralization potential based on successive leaching and incubation of soil. They found that 75% of N that should have been available (based on temperature adjustment of the nonlinear rate constant) was actually recovered by plants in the greenhouse.

In an evaluation of several short-term chemical indices of N availability from a diversity of soils from the southern United States, Schomberg et al. (2009) found that net N mineralization during 24 d of aerobic incubation related very closely with potentially mineralizable N via successive leaching and incubation. Additionally, the simple metric of the flush of CO2 combined with total soil N determination was the best predictive combination of N mineralization potential. Several other studies have shown the strong association of the flush of CO2 with net N mineralization during 24 d of aerobic incubation (Franzluebbers and Brock, 2007; Franzluebbers and Stuedemann, 2008; Franzluebbers and Haney, 2017). Evaluations extending beyond laboratory determined N mineralization are needed to test if the flush of CO2 has practical value for N prediction in the field. In Texas, bermudagrass (Cynodon dactylon L.) dry matter production was associated with the flush of CO2 in 1 d, as modulated by dairy manure application (Haney et al., 2001). In Michigan, corn (Zea mays L.) production as a result of long-term management differences was associated with the flush of CO2 in 1 d and anaerobic N mineralization (Culman et al., 2013).

The aim of this study was to test under semi-controlled greenhouse conditions whether the flush of CO2 would relate to unamended plant growth as an index of N availability across a diversity of soils in the Mid-Atlantic region of the United States. We also evaluated a variety of soil C and N fractions, along with routine soil-testing procedures for other nutrients from soils at depths of 0 to 10, 10 to 20, and 20 to 30 cm to see if there were other important contributors to N availability than just soil biological activity alone. The optimum sampling depth for evaluation of soil biological activity has not been clearly described, although we do know that the top few cm is often the most biologically active in conservation agricultural systems (Franzluebbers and Stuedemann, 2008, 2015). We hypothesized that plant available N (i.e., inorganic + mineralizable N) would be the most reliable predictor of unamended plant growth production. However, determining net N mineralization during 24 d of incubation requires significant time and analytical resources, so we explored use of the flush of CO2 as a surrogate. Specifically, we wanted to select the best combination (i.e., most influential, but least number of different methods) of soil C and N properties associated with soil N availability through a greenhouse growth bioassay as a function of differences in management from a variety of soil textures and physiographic regions. Our broader goal is to characterize the flush of CO2 as a potential soil-testing tool to make predictions of soil N availability, which could subsequently be used to inform crop N management decisions.


MATERIALS AND METHODS

Soil was collected from 47 fields throughout North Carolina and Virginia (Table 1). Sites were sampled in spring of 2014, 2015, or 2016 and that would be in corn production. This soil evaluation preceded an anticipated evaluation of corn production response to N fertilizer (Franzluebbers, 2018). Sites were located in four distinct physiographic regions, including the Coastal Plain (9 sites total in Camden, Lenoir, Halifax, and Washington counties in North Carolina), Piedmont (15 sites total in Montgomery, Rowan, and Stanly counties in North Carolina and Culpeper and Fauquier counties in Virginia), Great Valley (20 sites total in Augusta and Rockingham counties in Virginia), and Blue Ridge (3 sites total in Henderson County in North Carolina). Soils were predominately Ultisols (27 sites as Endoaquults, Hapludults, Kanhapludults, Paleaquults, Paleudults, and Umbraquults) and Alfisols (11 sites as Hapludalfs), followed by Inceptisols (7 sites as Dystrudepts and Eutrudepts) and Entisols (2 sites as Udifluvents). Mean annual temperature ranged from 11.2 to 16.0 and precipitation ranged from 859 to 1300 mm.


View Full Table | Close Full ViewTable 1.

Location, management history, soil taxonomy, and mean climatic conditions of field sites across several counties in North Carolina and Virginia.

 
Location Previous crop Soil taxonomy
Coastal Plain
Lenoir Co NC Soybean Lynchburg sandy loam (fine-loamy, siliceous, semiactive, thermic Aeric Paleaquults); Portsmouth loam (fine-loamy over sandy or sandy-skeletal, mixed, semiactive, thermic Typic Umbraquults)
Camden Co NC Corn Perquimans silt loam (fine-silty, mixed, semiactive, thermic Typic Endoaquults)
Halifax Co NC Soybean + multi-species cover; Corn + multi-species cover Grantham loam (fine-silty, siliceous, semiactive, thermic Typic Paleaquults); Nahunta silt loam (fine-silty, siliceous, subactive, thermic Aeric Paleaquults)
Washington Co NC Soybean Cape Fear loam (fine, mixed, semiactive, thermic Typic Umbraquults); Portsmouth fine sandy loam (fine-loamy over sandy or sandy-skeletal, mixed, semiactive, thermic Typic Umbraquults)
Piedmont
Stanly Co NC Corn Chewacla loam (fine-loamy mixed, mesic Typic Dystrudepts)
Montgomery Co NC Soybean Creedmoor-Brickhaven complex (fine, mixed, semiactive, thermic Aquic Hapludults); Tillery silt loam (fine-silty, siliceous, subactive, thermic Aquic Hapludults)
Stanly Co NC Cotton + multi-species cover; Cotton + rye winter cover Badin channery silt loam (fine, mixed, semiactive, thermic Typic Hapludults); Tarrus channery silty clay loam and Georgeville silty clay loam (fine, kaolinitic, thermic Typic Kanhapludults)
Rowan Co NC Corn + barley cover with dairy slurry; Corn with heavy poultry litter; Corn + rye cover; poultry litter; Corn + multi-species cover; Soybean Dorian fine sandy loam (fine, mixed, semiactive, thermic Aquic Hapludults); Lloyd clay loam (fine, kaolinitic, thermic Rhodic Kanhapludults); Mecklenburg clay loam (fine, mixed, active, thermic Ultic Hapludalfs)
Culpeper Co VA Soybean Rapidan-Penn complex (fine, mixed, semiactive, mesic Typic Hapludults)
Fauquier Co VA Corn + rye cover; Corn silage with dairy manure; Soybean + multi-species cover Albano silt loam (fine, mixed, active, mesic Typic Endoaqualfs); Ashburn silt loam (fine-silty, mixed, active, mesic Oxyaquic Hapludalfs); Dulles silt loam (fine, vermiculitic, mesic Aquultic Hapludalfs)
Great Valley
Rockingham Co VA Barley silage + sorghum-sudangrass hay; Corn and barley silage with dairy manure; Corn + multi-species cover; Corn silage + multi-species cover Craigsville cobbly fine sandy loam (loamy-skeletal, mixed, superactive, mesic Fluventic Dystrudepts); Frederick silt loam (fine, mixed, semiactive, mesic Typic Paleudults); Lodi silt loam (fine, mixed, mesic Typic Hapludults)
Augusta Co VA Corn; Corn with dairy manure; Corn + barley silage with dairy slurry; Corn + multi-species cover; Corn + rye cover with poultry litter; Corn silage + multi-species cover; Corn silage + rye cover with poultry litter; Soybean with poultry litter; Soybean with poultry litter and biological amendments; Soybean + rye cover with poultry litter + beef bedpack Chagrin loam (fine-loamy, mixed, active, mesic Dystric Fluventic Eutrudepts); Chavies fine sandy loam (coarse-loamy, mixed, active, mesic Ultic Hapludalfs); Edom silt loam (fine, illitic, mesic Typic Hapludalfs); Endcav silt loam (very-fine, mixed, mesic Typic Hapludalfs); Frederick silt loam (fine, mixed, semiactive, mesic Typic Paleudults); Frederick-Christian silt loam (fine, mixed, semiactive, mesic Typic Paleudults/Hapludults); Lodi silt loam (fine, mixed, mesic Typic Hapludults); Sequoia-Berks silt loams (fine, mixed, semiactive, mesic Typic Hapludults/loamy-skeletal, mixed, active, mesic Typic Dystrudepts)
Blue Ridge
Henderson Co NC Corn; Wheat Codorus loam (fine-loamy, mixed, active, mesic Fluvaquentic Dystrudepts); Comus fine sandy loam (coarse-loamy, mixed, active, nonacid, mesic Typic Udifluvents)

Soil was typically collected from four replicate blocks within both research station trials (4 locations and total of 11 site-years) and private farms (19 locations and total of 36 site-years). Soil was sampled at depths of 0 to 10, 10 to 20, and 20 to 30 cm with a hydraulic probe (4 cm i.d.) or same-sized hand probe (a few sites could not be sampled at 20- to 30-cm depth due to very dry conditions). Typically, eight cores within a block were composited in a paper bag, transported to the laboratory, and dried by placing in an oven at 55°C for ≥3 d until constant weight (sometimes initially by blowing room-temperature air over the sample on a paper plate for ≤1 d followed by oven drying). Soil was then gently crushed with a pestle over a screen with 4.75-mm openings. Stones and residues not passing the screen were weighed (if greater than ∼5% of total weight) and removed from the sample during further processing.

Soil physical, chemical, and biological properties were described more fully in Franzluebbers et al. (2018). Briefly, laboratory methods included:

  • Dispersion of soil with density of solution from a hydrometer to calculate clay concentration;

  • Collecting sand-sized material following dispersion and drying to determine sand concentration, which was also considered the particulate organic fraction;

  • Dry combustion using a Leco TruMac CN analyzer for whole-soil and particulate organic C and N fractions;

  • Log [H+] using glass electrode from 1:2 (v/v) soil/water solution for pH;

  • Mehlich-III extraction followed by determination by inductively coupled spectroscopy for concentration of Ca, Cu, K, Mg, Mn, Na, P, S, and Zn;

  • Chloroform fumigation-incubation without subtraction of a control using an efficiency factor of 0.41;

  • Aerobic incubation of rewetted soil at 50% water-filled pore space and 25°C for the flush of CO2 during 3 d, basal soil respiration from 10 to 24 d, and cumulative C and net N mineralization during 24 d (CO2 detected by acid titration of alkali trap);

  • Extraction of inorganic N with 2 mol L–1 KCl and subsequent determination of NH4–N and NO3–N + NO2–N using salicylate-nitroprusside and hydrazine auto analyzer techniques, respectively, for calculation of residual inorganic N species and mineralizable N during 24 d;

  • Calculation of plant available N from the sum of residual inorganic N and mineralizable N in 24 d.

Routine soil nutrient analyses were conducted by Soil Testing Services of the North Carolina Department of Agriculture and Consumer Services in Raleigh, NC. These analyses included sieved density (w/v), humic matter (%), cation exchange capacity (meq dm−3), base saturation (%), soil pH, acidity (meq dm−3), and extractable Ca, Cu, K, Mg, Mn, Na, P, S, Zn (mg dm−3). Soil organic C and N fractions were determined according to methods of Franzluebbers and Stuedemann (2008). Soil properties were sorted into physical (clay, sand, sieved density), chemical (total organic C and N, particulate organic C and N, residual inorganic N species and sum, pH, acidity, cation exchange capacity, base saturation, and extractable Ca, Cu, K, Mg, Mn, Na, P, S, Zn), and biological components (flush of CO2, cumulative C mineralization, basal soil respiration, microbial biomass C, net N mineralization, and plant-available N).

Dried and sieved soil from each of the 541 samples (not all sites could be sampled at 20- to 30-cm depth due to dry conditions) was thoroughly mixed and scooped into triplicate greenhouse growth tubes (164 mL volume, 21 cm tall, 3.8 cm i.d. at top; Ray Leach SC10 cone-tainers). Prior to filling, a cotton ball was placed at the bottom of the tube to prevent soil from leaking out, but to allow water to wick upward. Dry weight of a fixed volume in the tube was recorded to adjust results for differences in density of soils. Soil weight in this standard volume was ∼140 to 170 g. Soil was wetted with 10 to 20 mL of water prior to placing five pre-germinated sorghum-sudangrass seeds in small indentations at the soil surface. Sorghum-sudangrass was selected as a standard grass test crop that could take up large quantities of N if available and persist reasonably well during low N conditions. Approximately 20 g of washed, coarse sand was placed on top of seeds and the tubes were watered again with ∼10 mL. Water was added from the top this way on a daily basis until seedlings emerged (typically 3 to 4 d from planting). Once the majority of seeds had sprouted, trays of samples were transferred from the laboratory (∼23°C) to a greenhouse (target daytime maximum of 30°C and nighttime minimum of 20°C) for growth for up to 8 wk. Supplemental light was offered to achieve 12 h of total daylight (7:00 AM to 7:00 PM).

Samples were randomly assigned a position within a set of racks. Replicate subsamples were assigned to different blocks within the greenhouse. Samples were watered by capillarity from the bottom every 1 to 3 d (40 to 100% of water-holding capacity). Racks of samples were placed into trays filled with ∼5 cm of water and soil samples absorbed water for ∼20 min, at which time racks were returned to position in the greenhouse. No nutrients were added to the soil or solution. About midway through the 8-wk growth period, plants typically showed signs of N deficiency and so growth until 6 to 8 wk was more than sufficient to exhaust the readily mineralizable sources of N.

A total of three growth trials was conducted—one for each set of soils collected in 2014 (5 sites), 2015 (25 sites), and 2016 (17 sites). Growth conditions varied slightly among sets, with temperature (mean ± standard deviation of hourly values) of 24.9 ± 4.1°C in Trial 1, 23.8 ± 4.3°C in Trial 2, and 24.2 ± 4.1°C in Trial 3. Greenhouse trials were conducted from 22 Oct. to 15 Dec. 2014, from 15 Sept. to 15 Nov. 2015, and from 4 Oct. to 14 Nov. 2016. Another difference among trials was that smaller containers were used in Trial 1 than the other two trials (1.5 cm diam., 16 cm tall) and a single harvest at the end of 6 wk occurred in Trial 3 compared with dual harvest at 4 and 8 wk in the other two trials. To understand the implications of changing growth conditions, a selection of the same soils was tested with small and large containers, which resulted in no appreciable difference in dry matter accumulation when expressed per unit of soil weight. The difference in growth period did not result in appreciable difference in dry matter accumulation based on exploratory tests (not reported here) that resulted in greatly subdued growth after 4 wk. Total aboveground dry matter production was 2.5 ± 0.7 g kg−1 in Trial 1, 2.9 ± 1.2 g kg−1 in Trial 2, and 3.6 ± 1.7 g kg−1 in Trial 3. Mean values were likely more affected by the type of soil present than growth conditions.

At termination, plants were cut at a consistent level near the soil surface and transferred to a paper envelope to be dried in the oven at 55°C for 3 d until constant weight. Total aboveground dry matter (DM) was determined from weight and adjusted per kg of soil originally placed in each tube (g DM kg−1). Plants were subsequently ground to pass a 1-mm screen in the Udy mill. All samples were scanned with a Model 5000 near-infrared spectrometer (NIRS) and processed with WinISI v.1.5 software (Foss North America, Inc., Eden Prairie, MN). Spectra were evaluated for outliers (H > 3.0) prior to sample selection for calibration with total N determination with dry combustion. Using H > 0.6 as criterion, samples were selected with discernable spectral differences (n = 210). Selected samples were analyzed for total N concentration with a Leco TruMac combustion analyzer (Leco Corp, St. Joseph, MI). A calibration was developed for N concentration using modified partial least squares regression with four cross validations. After calibration development, all samples were estimated for N concentration using NIRS spectra.

Plant N uptake (mg N kg−1) was calculated from N concentration (mg N g DM−1) × dry matter production (g DM kg−1) of sorghum-sudangrass. Dry matter and plant N uptake were averaged across the three subsample replicates for each of the 541 soil samples.

Simple- and multiple-linear regression of dry matter production and plant N uptake against soil variables were performed with SAS v. 9.4 (SAS Institute, Inc., Cary, NC). Data were grouped by soil depth (0–10, 10–20, and 20–30 cm), geographic region (Coastal Plain, Piedmont, Great Valley, and Blue Ridge), and soil textural class (sandy loam, sandy clay loam, loam, clay loam, silt loam, silty clay loam, silty clay, and clay) and separate multiple regression analyses were conducted for each of these levels within a group to see if consistencies remained or if particular variables might be more important under specific conditions. Multiple regressions were also conducted using the flush of CO2 as the only soil biological property available and with all soil biological properties available (i.e., the flush of CO2, basal soil respiration, cumulative C mineralization, soil microbial biomass C, net N mineralization, and plant-available N) to test if the flush of CO2 was indeed the most appropriate choice to reflect soil N availability. Data were plotted with SigmaPlot v. 13 (Systat Software, Inc., San Jose, CA). Significant associations were considered at p ≤ 0.01 (a strict threshold due to the large number of samples evaluated).


RESULTS AND DISCUSSION

Dry matter production during 6 to 8 wk of unamended growth varied from as little as 1.0 g DM kg−1 to as much as 7.5 g DM kg−1. Median DM production was 2.3 g DM kg−1 in 2014, 2.6 g DM kg−1 in 2015, and 3.4 g DM kg−1 in 2016. Differences in production among greenhouse trials was likely more due to soil type (and its management) than due to minor environmental and procedural conditions. The middle 50% of DM production values was 2.1 to 3.0 g DM kg−1 in 2014, 1.9 to 3.7 g DM kg−1 in 2015, and 2.2 to 4.9 g DM kg−1 in 2016. The middle 50% of plant N uptake was 22 to 34 mg N kg−1 in 2014, 19 to 46 mg N kg−1 in 2015, and 18 to 45 mg N kg−1 in 2016. Median plant N uptake was 26 mg N kg−1 in 2014, 29 mg N kg−1 in 2015, and 27 mg N kg−1 in 2016.

Plant N uptake was greater (p < 0.001) in soils from the Piedmont and Great Valley than from the Coastal Plain and Blue Ridge regions at both 0- to 10- and 10- to 20-cm depths, but was not different among regions at 20 to 30 cm. Dry matter production at 0- to 10-cm depth was also greater in soils from the Piedmont (4.7 ± 1.2 g DM kg−1) and Great Valley (5.1 ± 1.0 g DM kg−1) regions than from the Coastal Plain (3.7 ± 0.9 g DM kg−1) and Mountain (3.1 ± 0.8 g DM kg−1) regions.

Dry matter production had median values of 4.5, 2.6, and 1.9 g DM kg−1 at depths of 0 to 10, 10 to 20, and 20 to 30 cm, respectively. Plant N uptake had median values of 52, 26, and 17 mg N kg−1 at depths of 0 to 10, 10 to 20, and 20 to 30 cm, respectively. The stratified depth distribution of plant growth properties was expected due to similarly stratified depth distribution of many soil C and N fractions (Franzluebbers et al., 2018). Depth stratification of soil organic C and N fractions is typical in soils of the southeastern United States with a long history of no-tillage management (Franzluebbers, 2002), as was the case in many of these sites.

Plant N uptake during unamended sorghum-sudangrass growth in the greenhouse was strongly associated with dry matter production in each of the greenhouse trials (Fig. 1). These data excluded three samples in 2015 that had abnormally high N concentration and associated plant N uptake. All three of these samples were from a Piedmont location (PRSB5) at depth of 0 to 10 cm, a site that had recent heavy poultry litter application. It is possible that some inhibition of plant development may have occurred, as several micronutrients were at very high levels. These three data points were excluded from remaining analyses. Across all data remaining (n = 538), dry matter explained the majority of variation in plant N uptake (PNU = −3.7 + 12.0 DM; r2 = 0.77). Individual greenhouse trials explained another 7% of variation, with slopes between plant N uptake and dry matter varying from 11.9 ± 0.6 mg N g DM−1 (mean ± standard error) in 2014, 15.1 ± 0.4 mg N g DM−1 in 2015, and 11.2 ± 0.4 mg N g DM−1 in 2016.

Fig. 1.
Fig. 1.

Association of sorghum-sudangrass nitrogen uptake with dry matter production in unamended soils in three different trials in the greenhouse. Data are across soil depths (0–10, 10–20, and 20–30 cm) and replications (typically 4) of sampling sites.

 

Dry matter production during 6 to 8 wk of greenhouse growth (n = 538) was highly related with net N mineralization during 24 d of aerobic incubation (DM = 1.6 + 0.032 NMIN; r2 = 0.74), with plant available N (residual inorganic N + net N mineralization during 24 d) (DM = 1.5 + 0.028 PAN; r2 = 0.76), and with the flush of CO2 in 3 d (DM = 1.7 + 0.009 FCO2; r2 = 0.74). By itself, residual inorganic N explained only 43% of total variation in dry matter production. Plant available N (i.e., inorganic + mineralizable N) was hypothesized to be the most reliable estimate of unamended plant growth production and it was. Interestingly though, the flush of CO2 in 3 d was nearly equally effective in predicting unamended plant growth in the greenhouse as was plant available N. Given the shorter time period and the simpler laboratory approach, the flush of CO2 was considered further as a valuable indicator of plant N availability, at least under greenhouse growth conditions. The strong association of the flush of CO2 with net N mineralization under standard laboratory conditions has been previously noted in these same soils (Franzluebbers et al., 2018), as well as in soils from Georgia (Franzluebbers and Stuedemann, 2003, 2008; Franzluebbers et al., 2007), Wyoming (Ingram et al., 2005), and Brazil (Green et al., 2007).

Plant N uptake was also highly related with several soil biological properties (Table 2). The best associations were with plant available N (r2 = 0.85), net N mineralization during 24 d (r2 = 0.81), the flush of CO2 in 3 d (r2 = 0.76), cumulative C mineralization during 24 d (r2 = 0.72), basal soil respiration (r2 = 0.68), and soil microbial biomass C (r2 = 0.67). Other highly significant correlations (p < 0.001) with plant N uptake occurred for total soil N (r2 = 0.63), particulate organic C (r2 = 0.61), particulate organic N (r2 = 0.59), residual inorganic N (r2 = 0.56), extractable Zn (r2 = 0.38), residual soil nitrate (r2 = 0.38), cation exchange capacity (r2 = 0.35), extractable Cu (r2 = 0.33), extractable Ca (r2 = 0.30), extractable P (r2 = 0.25), negative of sieved soil density (r2 = 0.25), extractable K (r2 = 0.23), total organic C (r2 = 0.22), base saturation (r2 = 0.06), extractable Mn (r2 = 0.06), extractable Mg (r2 = 0.05), and soil pH (r2 = 0.02). These associations make it clear that soil biological activity in the form of C and N transformation processes were the most responsible for plant N uptake in these greenhouse growth bioassays. As a simple and rapid proxy, the flush of CO2 had very strong association with plant N uptake, similar to many other more methodologically complex and labor costly biological indicators.


View Full Table | Close Full ViewTable 2.

Regression equations and strength of association between sorghum-sudangrass N uptake during 6 to 8 wk of greenhouse growth and various soil C and N fractions (n = 538).

 
Y intercept Slope Independent variable r2
mg N kg−1
10.6 0.405 Plant available N (mg N kg−1) 0.85
12.5 0.460 Net N mineralization (mg N kg−1)0–24 d 0.81
13.8 0.125 Flush of CO2 (mg CO2–C kg−1)0–3 d 0.76
13.1 0.051 Cumulative C mineralization (mg kg−1)0–24 d 0.72
16.7 1.80 Basal soil respiration (mg CO2–C kg−1 d−1) 0.68
9.8 0.040 Soil microbial biomass C (mg kg−1) 0.67
8.4 21.1 Total soil N (g N kg−1) 0.63
21.9 0.772 Total organic C (g C kg−1) 0.22

Multiple Regression

Although soil biological properties were able to explain a vast majority of variation in plant N uptake (67 to 85% of total variation), multiple regression analysis with other soil chemical properties further improved associations to 88 to 90% of total variation (Table 3). All models had 8 to 10 entries of significance, although the soil biological component explained 75 to 94% of the explained variance and the next couple of entries most of the remaining variance. Prediction of plant N uptake with plant available N (residual inorganic + net N mineralization in 24 d) was aided minimally with knowledge of total soil N and Zn concentration (partial r2 = 0.01 each). Prediction of plant N uptake with the flush of CO2 was aided with knowledge of residual inorganic N (partial r2 = 0.07) and total soil N (partial r2 = 0.03). Prediction of plant N uptake with basal soil respiration was aided with knowledge of residual inorganic N (partial r2 = 0.10), total soil N (partial r2 = 0.04), and the negative of total organic C (partial r2 = 0.02). Residual inorganic N was frequently an important secondary variable among the various models (partial r2 = 0.08 ± 0.04) with different soil biological properties as primary component and total soil N was another factor of minor importance across models (partial r2 = 0.03 ± 0.01) (Table 3). Residual inorganic N was more of a factor in models when net N mineralization was selected than it was in models when plant available N was selected, because residual inorganic N was already attributed in the calculation of plant available N. Individual components of residual soil nitrate and residual soil ammonium occasionally appeared in multiple regression models. Other soil properties provided only minor contributions to model strength. Extractable Cu (positive effect) and Mn (negative effect) were always present as a small contribution to explain plant N uptake in the greenhouse, which may have been a reflection of sites with history of broiler litter application that contributed to overall soil fertility.


View Full Table | Close Full ViewTable 3.

Stepwise regression analysis (forward selection; p = 0.01 for entry) to predict sorghum-sudangrass N uptake during 6 to 8 wk of greenhouse growth (mg N kg–1), as led by different individual soil biological components (plant available N [PAN], net N mineralization [NMIN], flush of CO2 [FCO2], basal soil respiration [BSR], cumulative C mineralization [CMIN], and soil microbial biomass [SMBC], which were all highly correlated).†

 
Variable Estimate SE F-value Partial r2 Variable Estimate SE F-value Partial r2
PAN NMIN
Intercept 27.3 5.3 26 Intercept 27.3 5.3 26
PAN 0.231 0.016 205 0.849 NMIN 0.231 0.016 205 0.805
TOC −0.361 0.055 43 0.005 TOC −0.361 0.055 42 0.005
TSN 13.4 1.4 94 0.011 TSN 13.4 1.4 94 0.013
Clay −17.8 2.9 37 0.004 Clay −17.8 2.9 37 0.004
RIN 0.295 0.046 41 0.008 RIN 0.526 0.039 178 0.050
Density −16.0 4.1 15 0.003 Density −16.0 4.1 15 0.003
Cu 1.07 0.12 74 0.007 Cu 1.07 0.12 74 0.007
Zn −0.529 0.063 72 0.014 Zn −0.529 0.063 72 0.014
FCO2 BSR
Intercept 22.2 5.8 15 Intercept 12.4 2.0 39
FCO2 0.0530 0.0046 134 0.716 BSR 0.567 0.066 73 0.683
TOC −0.419 0.059 50 0.008 TOC −0.529 0.060 77 0.024
TSN 17.1 1.4 148 0.028 TSN 21.1 1.3 250 0.038
Clay −12.9 3.2 16 0.002 Sand −8.6 2.8 10 0.002
RSN 0.47 0.12 15 0.004 Clay −20.1 3.9 27 0.004
RIN 0.21 0.10 4 0.069 RIN 0.633 0.043 216 0.102
Density −12.1 4.5 7 0.002 K −0.0081 0.0030 7 0.010
K −0.0046 0.0029 3 0.006 Cu 1.11 0.15 58 0.005
Cu 1.06 0.14 61 0.005 Zn −0.505 0.075 46 0.011
Zn −0.465 0.071 43 0.009
CMIN SMBC
Intercept 23.3 5.7 16 Intercept 11.8 1.9 39
CMIN 0.0198 0.0017 135 0.723 SMBC 0.0172 0.0014 157 0.673
TOC −0.497 0.056 79 0.013 TSN 12.2 0.9 178 0.031
TSN 18.6 1.3 194 0.032 Sand −9.9 2.5 15 0.004
Clay −13.9 3.2 19 0.003 Clay −23.7 3.5 45 0.006
RSA −0.42 0.12 12 0.002 RSA −0.38 0.12 10 0.002
RIN 0.722 0.050 209 0.098 RIN 0.727 0.049 222 0.132
Density −13.4 4.5 9 0.002 HM −2.27 0.21 113 0.027
K −0.0049 0.0029 3 0.007 Mn 0.0163 0.0038 19 0.003
Cu 0.97 0.14 50 0.003 Cu 1.11 0.13 72 0.009
Zn −0.451 0.071 41 0.009 Zn −0.449 0.063 51 0.009
Abbreviations and units: BSR, basal soil respiration (mg CO2–C kg–1 d–1); Clay (g kg–1); CMIN, cumulative C mineralization (mg CO2–C kg–1)0–24 d Cu, extractable copper (mg dm–3); Density, sieved density (g cm–3); FCO2, flush of CO2 (mg CO2–C kg–1)0–3 d HM, humic matter (%); K, potassium (mg dm–3); Mn, extractable manganese (mg dm–3); NMIN, net N mineralization (mg N kg–1)0–24 d PAN, plant-available N (mg N kg–1)0–24 d RIN, residual inorganic N (mg N kg–1); RSA, residual soil ammonium (mg NH4–N kg–1); RSN, residual soil nitrate (mg NO3–N kg–1); Sand (g kg–1); SE, standard error; SMBC, soil microbial biomass C (mg C kg–1); TOC, total organic C (g C kg–1); TSN, total soil N (mg N kg–1); Zn, extractable zinc (mg dm–3).

Soil Depth

Prediction of sorghum-sudangrass dry matter production and N uptake with various soil properties was also tested when data were sorted by depth, since there was a clear gradient of these responses by sampling depth. When any soil biological property were allowed to be selected, plant available N was the dominant component to predict plant dry matter production from soil at depth of 0 to 10 cm and the flush of CO2 was the dominant component at depths of 10 to 20 and 20 to 30 cm (Table 4). Other variables that explained minor variations (3–10%) in plant dry matter production were clay and sand concentration, extractable Mn and K, and total soil N. At depth of 0 to 10 cm, plant available N, extractable Mn, and total soil N had positive influence on dry matter production, while clay concentration had negative influence in the multiple regression model. When only the flush of CO2 was allowed to be selected as a soil biological property, several other variables were selected to complement the strong association with dry matter production (Table 4). Cumulative model strength was similar when any and multiple soil biological properties were allowed to be selected compared with isolation of only the flush of CO2 at depths of 10 to 20 and 20 to 30 cm. However, the coefficient of determination was 0.10 greater at depth of 0 to 10 cm with the flush of CO2 as biological property (largely a result of more variables that were significant in the multiple regression).


View Full Table | Close Full ViewTable 4.

Multiple regression models explaining sorghum-sudangrass dry matter production and N uptake when separated into soil depth increments (0–10, 10–20 and 20–30 cm). Positive versus negative effects are indicated with (+) and (−) respectively.†

 
0 to 10 cm
10 to 20 cm
20 to 30 cm
Variable Cumulative model r2 Effect Variable Cumulative model r2 Effect Variable Cumulative model r2 Effect
Dry matter production with any soil biological component
PAN 0.497 + FCO2 0.482 + FCO2 0.470 +
Clay 0.557 K 0.582 + PAN 0.535 +
Mn 0.591 + Clay 0.634 Sand 0.585 +
TSN 0.617 + Mn 0.663 + Mn 0.607 +
RSN 0.687 + TSN 0.630 +
POC 0.704 + Density 0.650 +
BS 0.666 +
Dry matter production with flush of CO2 as biological component
FCO2 0.433 + FCO2 0.482 + FCO2 0.470 +
RSN 0.496 + K 0.582 + TSN 0.518 +
P 0.551 + Clay 0.634 Mn 0.561 +
Mn 0.579 + Mn 0.663 + Density 0.612 +
Clay 0.599 RSN 0.687 + RSN 0.631 +
Sand 0.627 POC 0.704 + pH 0.652 +
TSN 0.649 +
PON 0.668
POC 0.691 +
TOC 0.715
Plant N uptake with any soil biological component
PAN 0.659 + PAN 0.587 + PAN 0.539 +
SMBC 0.713 + SMBC 0.633 + SMBC 0.595 +
NMIN 0.743 RSN 0.666 + Mg 0.622
K 0.760 FCO2 0.730 + HM 0.650
Cu 0.771 + S 0.744 + TSN 0.670 +
Zn 0.780 RSN 0.697 +
TSN 0.794 +
Clay 0.804
Sand 0.817
POC 0.827 +
Mg 0.835 +
CMIN 0.846 +
P 0.852 +
Mn 0.862 +
pH 0.873
Plant N uptake with flush of CO2 as biological component
FCO2 0.451 + FCO2 0.526 + FCO2 0.329 +
RIN 0.606 + RSN 0.669 + RIN 0.510 +
TSN 0.662 + TSN 0.716 + TSN 0.611 +
K 0.705 TOC 0.751
Cu 0.724 + S 0.767 +
Zn 0.734
TOC 0.745
Clay 0.756
Sand 0.774
Abbreviations and units: BS, base saturation (%); Clay (g kg–1); CMIN, cumulative C mineralization (mg CO2–C kg–1)0–24 d Cu, extractable copper (mg dm–3); Density, sieved density (g cm–3); FCO2; flush of CO2 (mg CO2–C kg–1)0–3 d HM; humic matter (%); K, potassium (mg dm–3); Mg, extractable magnesium (mg dm–3); Mn, extractable manganese (mg dm–3); NMIN, net N mineralization (mg N kg–1)0–24d P, extractable phosphorus (mg dm–3); PAN, plant-available N (mg N kg–1)0–24 d pH (−log[H+]); POC, particulate organic C (g kg–1); PON, particulate organic N (g kg–1); RIN, residual inorganic N (mg N kg–1); RSN, residual soil nitrate (mg NO3–N kg–1); S, extractable S (mg dm–3); Sand (g kg–1); SMBC, soil microbial biomass C (mg C kg–1); TOC, total organic C (g C kg–1); TSN, total soil N (mg N kg–1); Zn, extractable zinc (mg dm–3).

Prediction of plant N uptake was dominated by plant available N at all three depths (Table 4). Soil microbial biomass C was the second-most influential variable at all three depths (partial r2 = 0.05 ± 0.01). Additional variables that appeared in at least two of the depths were extractable magnesium, total soil N, and residual soil nitrate. At a depth of 0 to 10 cm, 4 of the 15 variables selected were of soil biological nature, with no other variable explaining >2% of variation. Plant N uptake was clearly driven by variations in soil biological properties associated with N transformations.

When the flush of CO2 was isolated as the only soil biological property to be selected, prediction of plant N uptake with multiple regression within individual depth increments was also good (Table 4). Cumulative model strength had partial r2 = 0.09 ± 0.01 lower than with all biological variables available. However, models with the flush of CO2 as dominant component had only 3 to 9 variables selected, as compared with 5 to 15 variables selected with all biological variables available. This might imply that the flush of CO2 encompassed more variation associated with plant N uptake by itself than a combination of other variables. Total soil N and either residual inorganic N or residual soil nitrate were key secondary variables in all depths.

Region

Prediction of sorghum-sudangrass dry matter production and N uptake with various soil properties was tested when data were sorted by major land resource area (i.e., 108 observations in the Coastal Plain of North Carolina, 179 observations in the Piedmont of North Carolina and Virginia, 215 observations in the Great Valley of Virginia, and 36 observations in the Blue Ridge of North Carolina). Generally, regions had differences in topography (broad, level landscape in the Coastal Plain and river bottoms of the Blue Ridge compared with undulating hills in the Piedmont and Great Valley), temperature (typically colder in Great Valley and Blue Ridge than in Piedmont and Coastal Plain), and soil texture (generally sandier soils in Coastal Plain and Blue Ridge than in Piedmont and Great Valley). With any soil biological property available for selection, dry matter production was predicted dominantly with plant available N in the Coastal Plain (partial r2 = 0.76) and Piedmont (partial r2 = 0.80) regions, with total organic C (partial r2 = 0.78) in the Great Valley, and with net N mineralization (partial r2 = 0.83) in the Blue Ridge (Table 5). All regions had models that explained 84 to 89% of total variation with any soil biological property available for selection. Additional variables that contributed ≥3% to the model were basal soil respiration, clay concentration, particulate organic N, and soil microbial biomass C. When the flush of CO2 was the only soil biological property allowed, cumulative model strength increased by r2 = 0.03 in the Coastal Plain and decreased by r2 = 0.06 ± 0.02 in the Piedmont, Great Valley, and Blue Ridge regions (Table 5). As the only soil biological property allowed, the flush of CO2 only appeared as the dominant variable in the Piedmont (partial r2 = 0.79), was a secondary variable in the Coastal Plain and Blue Ridge, and was not present as a significant variable in the Great Valley. The strong correlation between the flush of CO2 and total and particulate organic C and N fractions (Franzluebbers et al., 2018) likely contributed to its absence and minor significance in some regions.


View Full Table | Close Full ViewTable 5.

Multiple regression models explaining sorghum-sudangrass dry matter production and N uptake when separated into geographical regions. Positive versus negative effects are indicated with (+) and (−) respectively.†

 
Coastal Plain
Piedmont
Great Valley
Blue Ridge
Variable Model r2 Effect Variable Model r2 Effect Variable Model r2 Effect Variable Model r2 Effect
Dry matter production with any soil biological component
PAN 0.762 + PAN 0.801 + TOC 0.780 + NMIN 0.834 +
PON 0.818 + Clay 0.840 Clay 0.821 BSR 0.880 +
Sand 0.836 Mn 0.856 + SMBC 0.847
FCO2 0.869 + PAN 0.865 +
Density 0.875 + CEC 0.876 +
RSA 0.881 +
Acidity 0.885
Dry matter production with flush of CO2 as biological component
PON 0.693 + FCO2 0.787 + TOC 0.781 + PON 0.791 +
RSA 0.761 + Mn 0.824 + Clay 0.822 FCO2 0.846 +
Sand 0.812 HM 0.839 + Mn 0.838 +
Clay 0.841 K 0.860 + S 0.847 +
HM 0.868 + RSN 0.867 + CEC 0.853 +
POC 0.879 RIN 0.858 +
FCO2 0.887 +
Plant N uptake with any soil biological component
PAN 0.763 + PAN 0.878 + PAN 0.825 + PON 0.800 +
pH 0.828 FCO2 0.907 + TSN 0.857 + P 0.872
RSA 0.862 + Clay 0.914 NMIN 0.877 SMBC 0.901 +
S 0.880 + SMBC 0.920 + POC 0.891
TSN 0.890 + RIN 0.929 + CMIN 0.905 +
Clay 0.906 BSR 0.934 Sand 0.915 +
BS 0.914 + CMIN 0.937 + SMBC 0.921 +
FCO2 0.924
Plant N uptake with flush of CO2 as biological component
RSA 0.698 + FCO2 0.874 + TSN 0.814 + PON 0.800 +
Density 0.754 RSN 0.898 + RIN 0.867 + P 0.872 +
Clay 0.790 TSN 0.910 + Cu 0.878 + FCO2 0.898 +
S 0.835 + Ca 0.917 PON 0.889 Density 0.921
Sand 0.857 Zn 0.921 Clay 0.897
Cu 0.929 +
Clay 0.933
TOC 0.937 +
HM 0.941
Abbreviations and units: Acidity (meq dm–3); BS, base saturation (%); BSR, basal soil respiration (mg CO2–C kg–1 d–1); Ca, extractable calcium (mg dm–3); Clay (g kg–1); CEC, cation exchange capacity (meq dm–3); CMIN, cumulative C mineralization (mg CO2–C kg–1)0–24 d Cu, extractable copper (mg dm–3); Density, sieved density (g cm–3); FCO2, flush of CO2 (mg CO2–C kg–1)0–3 d HM, humic matter (%); K, potassium (mg dm–3); Mg, extractable magnesium (mg dm–3); Mn, extractable manganese (mg dm–3); NMIN, net N mineralization (mg N kg–1)0–24 d P, extractable phosphorus (mg dm–3); PAN, plant-available N (mg N kg–1)0–24 d pH (−log[H+]); POC, particulate organic C (g kg–1); PON, particulate organic N (g kg–1); RIN, residual inorganic N (mg N kg–1); RSA, residual soil ammonium (mg NH4–N kg–1); S, extractable S (mg dm–3); Sand (g kg–1); SMBC, soil microbial biomass C (mg C kg–1); TOC, total organic C (g C kg–1); TSN, total soil N (mg N kg–1); Zn, extractable zinc (mg dm–3).

Plant N uptake was predicted with dominance using plant available N in the Coastal Plain, Piedmont, and Great Valley and using particulate organic N in the Blue Ridge (Table 5). In the Coastal Plain, six additional variables added partial r2 = 0.05 to the best model. In the Piedmont, six additional variables added partial r2 = 0.06 to the best model. In the Great Valley, seven additional variables added partial r2 = 0.10 to the best model. In the Blue Ridge, only two variables added partial r2 = 0.10 to the best model. The greatly diminishing returns with additional variables was evident throughout these analyses. Allowing the flush of CO2 as the only soil biological property to be selected resulted in dominant influence of residual soil ammonium in the Coastal Plain (partial r2 = 0.70), the flush of CO2 in the Piedmont (partial r2 = 0.87), total soil N in the Great Valley (partial r2 = 0.81), and particulate organic N in the Blue Ridge (partial r2 = 0.80) (Table 5). Alone in the model, the flush of CO2 in association with plant N uptake had coefficient of determination of 0.41, 0.87, 0.69, and 0.70 in the Coastal Plain, Piedmont, Great Valley, and Blue Ridge regions, respectively. Few of the sites in the Coastal Plain had long-term use of no-tillage cropping, so the lack of soil organic matter accumulation at the soil surface may have limited the influence of soil biological activity to affect N transformation processes. The strong influence of residual soil ammonium in the Coastal Plain appeared to have expressed labile N transformations better, and residual soil ammonium can be elevated with soil organic matter accumulation in many soils (e.g., residual soil ammonium was associated with total soil N [r2 = 0.27] and cumulative C mineralization [r2 = 0.40], among other soil biological properties).

Soil Textural Class

Prediction of sorghum-sudangrass dry matter production and N uptake with various soil properties was tested against individual soil textural classes so that soil-specific relationships could be identified. Although this was not an a priori decision to balance the number of observations within a soil textural class, there were at least 24 observations in silt loam and silty clay textural classes and up to 160 observations in the clay loam textural class (Table 6). Our goal was to look for consistencies in associations among soil textural classes, and if not present, then explore options of how to develop optimal solutions for developing soil N availability indices.


View Full Table | Close Full ViewTable 6.

Multiple regression models (variables [Var.] and model r2) explaining sorghum-sudangrass dry matter production and N uptake when separated by soil textural classes.†

 
Sandy loam (n = 65)
Sandy clay loam (n = 36)
Loam (n = 118)
Clay loam (n = 160)
Silt loam (n = 24)
Silty clay loam (n = 75)
Silty clay (n = 24)
Clay (n = 36)
Var. r2 Var. r2 Var. r2 Var. r2 Var. r2 Var. r2 Var. r2 Var. r2
Dry matter production with any soil biological component
PAN 0.89 NMIN 0.83 PAN 0.83 PAN 0.70 TOC 0.96 PAN 0.85 Zn 0.89 POC 0.75
HM 0.92 BSR 0.88 PON 0.86 K 0.74 S 0.98 RSA 0.89 SMBC 0.92 Mn 0.82
RSA 0.94 Clay 0.87 Clay 0.77 K 0.99 HM 0.90 BSR 0.95 PAN 0.87
Zn 0.95 NMIN 0.88 BSR 0.78 P 0.90
Dry matter production with flush of CO2 as biological component
FCO2 0.85 PON 0.79 PON 0.80 FCO2 0.69 TOC 0.96 TSN 0.83 Zn 0.89 POC 0.75
RIN 0.92 FCO2 0.85 RIN 0.84 TSN 0.75 S 0.98 RSA 0.86 Mn 0.82
pH 0.94 FCO2 0.86 K 0.77 K 0.99 Mn 0.88 RSN 0.87
RSA 0.95 Clay 0.87 PON 0.79
Zn 0.95 Clay 0.80
RSN 0.81
HM 0.83
Plant N uptake with any soil biological component
RIN 0.88 PON 0.80 PAN 0.81 PAN 0.83 TSN 0.97 PAN 0.90 PAN 0.94 PAN 0.86
CMIN 0.93 P 0.87 SMBC 0.89 SMBC 0.86 PON 0.98 SMBC 0.92 Sand 0.94
S 0.95 SMBC 0.90 Mg 0.90 Cu 0.87 CEC 0.99 RSN 0.94 POC 0.95
BSR 0.96 Mn 0.90 TSN 0.89 FCO2 0.95
Mn 0.97 CEC 0.91 Zn 0.90
S 0.92 CMIN 0.92
NMIN 0.93 TOC 0.92
PON 0.93
RSN 0.94
P 0.95
Plant N uptake with flush of CO2 as biological component
RIN 0.88 PON 0.80 FCO2 0.67 FCO2 0.76 TSN 0.97 TSN 0.87 TOC 0.83 TOC 0.85
Ca 0.93 P 0.87 RIN 0.79 TSN 0.82 PON 0.98 RSN 0.94 RSN 0.89 Sand 0.88
S 0.95 FCO2 0.90 Density 0.85 Cu 0.87 CEC 0.99 P 0.95 FCO2 0.93
Density 0.92 Mg 0.86 RSN 0.89 Cu 0.95 RIN 0.94
S 0.88 TOC 0.91
Clay 0.88 Zn 0.92
PON 0.93
P 0.94
Abbreviations and units: BS, base saturation (%); BSR, basal soil respiration (mg CO2–C kg–1 d–1); Ca, extractable calcium (mg dm–3); CEC, cation exchange capacity (meq dm–3); Clay (g kg–1); CMIN, cumulative C mineralization (mg CO2–C kg–1)0–24 d Cu, extractable copper (mg dm–3); Density, sieved density (g cm–3); FCO2, flush of CO2 (mg CO2–C kg–1)0–3 d HM, humic matter (%); K, potassium (mg dm–3); Mg, extractable magnesium (mg dm–3); Mn, extractable manganese (mg dm–3); NMIN, net N mineralization (mg N kg–1)0–24 d P, extractable phosphorus (mg dm–3); PAN, plant-available N (mg N kg–1)0–24 d pH (−log[H+]); POC, particulate organic C (g kg–1); PON, particulate organic N (g kg–1); RIN, residual inorganic N (mg N kg–1); RSA, residual soil ammonium (mg NH4–N kg–1); RSN, residual soil nitrate (mg NO3–N kg–1); S, extractable S (mg dm–3); Sand (g kg–1); SMBC, soil microbial biomass C (mg C kg–1); TOC, total organic C (g C kg–1); TSN, total soil N (mg N kg–1); Zn, extractable zinc (mg dm–3).

Prediction of sorghum-sudangrass dry matter production among soil textural classes was most often dominated by plant available N (four of eight soil textural classes) and by total organic C, particulate organic C, net N mineralization, and extractable Zn in the remaining classes (Table 6). When allowing the flush of CO2 as the only soil biological property in association with dry matter production, it was the dominant influence in sandy loam and clay loam classes and appeared in two other models. Total and particulate organic C and N fractions were dominant in five of the textural classes, and these properties were highly related with the flush of CO2. Total organic C was the dominant variable in the silt loam class (the flush of CO2 was highly related to total organic C in this class, r2 = 0.96). As well, total soil N dominated in the silty clay loam class (association with the flush of CO2 was r2 = 0.92), particulate organic C dominated in the clay class (association with the flush of CO2 was r2 = 0.74), and particulate organic N dominated in the sandy clay loam and loam classes (association with the flush of CO2 was r2 = 0.62 and 0.86, respectively).

Prediction of plant N uptake among soil textural classes was also most often dominated by plant available N (five of eight classes), and the dominating influence for other soil textural classes included total soil N, particulate organic N, and residual inorganic N (Table 6). Strength of multiple regression models was similar among soil textural classes for both dry matter production and plant N uptake (r2 = 0.94 ± 0.03). Secondary variables that explained at least 3% of variation included cumulative C mineralization, extractable P, soil microbial biomass C, and sand concentration. When the flush of CO2 was allowed as the only soil biological property, it was the dominant influence in loam and clay loam classes and appeared in models of two other classes. Like that for plant dry matter production, other dominant variables in association with plant N uptake were total and particulate organic C and N.

When exploring the actual associations of plant N uptake as a function of the flush of CO2 among soil textural classes, very little difference occurred in intercept and slope estimates (Fig. 2). Intercepts were 14 ± 2 mg N kg−1 and slope estimates were 0.13 ± 0.01 mg N(0–24 d) mg−1 CO2–C(0–3 d) (mean ± standard deviation among eight soil textural classes). With the high coefficient of determination among individual soil textural classes (r2 = 0.77 ± 0.10) and the overall high coefficient of determination (r2 = 0.76), along with intercept (13.8 ± 0.6 mg N kg−1) and slope (0.125 ± 0.003 mg N mg−1 CO2–C(0–3 d)) (mean ± standard error among 537 observations) estimates, we found little evidence to support selection of other soil biological variables to replace the simple and rapid approach of measuring the flush of CO2.

Fig. 2.
Fig. 2.

Association of sorghum-sudangrass N uptake during 6 to 8 wk of greenhouse growth with the flush of CO2 as an indicator of soil-test biological activity as affected by soil textural class.

 

Interpretations

There were broad and consistent indications that some measure of soil biological activity or biomass was essential in predicting soil N availability, as assessed in this bioassay with unamended soil using sorghum-sudangrass as test crop in the greenhouse for 6 to 8 wk. Although there were differences in variables chosen when data were separated by soil depth, geographic region, and soil textural class, the set of most influential variables affecting dry matter production and plant N uptake was biologically based. Variables most often chosen as dominant factor included plant available N, the flush of CO2, net N mineralization, particulate organic N, total organic C, and total soil N. Plant available N was the most oft selected top variable and we assumed it should have been the most indicative of unamended plant growth. We chose to focus on the flush of CO2 as a surrogate for plant available N, since it was the most correlated variable (r2 = 0.85), except for net N mineralization (r2 = 0.98), which was the dominating component of the plant available N variable (residual inorganic N + net N mineralization). The flush of CO2 is also much more rapid, inexpensive, and precise than determination of net N mineralization.

Although selecting more variables to measure will lead to greater predictability of plant dry matter and N uptake, the financial and human resources needed to get a more precise estimate has to be considered. Therefore, a balance between greatest precision and least amount of resource allocation should be pursued. From the overall data and separations based on soil depth, geographic region, and soil textural class, best-fit models had coefficients of determination of 0.88 ± 0.10 with 5 ± 3 variables for predicting dry matter production and plant N uptake. Using the flush of CO2 as the only predictive variable resulted in coefficient of determination with dry matter production of 0.74 and with plant N uptake of 0.76. Adding residual inorganic N to a multiple regression model to account for both organic and inorganic sources of N resulted in coefficient of determination with dry matter production of 0.76 and with plant N uptake of 0.83. Residual inorganic N was more important in prediction of plant N uptake than for dry matter production, although it could theoretically affect both. Adding further total soil N to the multiple regression model to account for a potentially intermediately available organic N pool resulted in coefficient of determination with dry matter production of 0.78 and with plant N uptake of 0.86. Since total organic C is often very highly related with total soil N, substitution of total soil N with total organic C in the multiple regression model led to similar coefficient of determination with dry matter production of 0.77 and with plant N uptake of 0.84.

Therefore, based on all analyses presented here and using scientific discretion to balance precision with resource allocation, it is suggested that a robust soil testing protocol to predict soil N availability would include the flush of CO2, residual inorganic N, and total soil N (or total organic C) in decreasing order of importance. This group of key secondary variables also complemented well multiple regression models when using net N mineralization, basal soil respiration, cumulative C mineralization, and soil microbial biomass as the primary soil biological property (Table 3).


CONCLUSIONS

We conclude that soil N availability should be assessed with some form of soil biological activity. The flush of CO2 has many of the qualities of an ideal soil-test protocol because it is highly related to soil N availability (as shown in this greenhouse bioassay and in previous correlations with net N mineralization under standard temperature and moisture conditions), it uses dried soil to normalize initial soil conditions and provide flexibility in processing, it is precise enough to relate to differences in plant growth, it relates strongly to other soil biological properties, and it is simple, rapid, and inexpensive to determine in the most basic laboratory conditions. Therefore, we conclude that the flush of CO2 is a robust indicator of soil-test biological activity, as evidenced here for predicting unamended plant growth under semi-controlled greenhouse conditions.

Acknowledgments

Ellen Leonard, Ashtyn Mizelle, and Erin Silva provided sound technical support in the laboratory. Wes Childres, Jeff Cline, Tim Cline, Carl Crozier, Brad Graham, Alec Lipscomb, Jay Marshall, Shelby Rajkovich, and Robert Shoemaker provided key logistical support at several field sites. Sincere appreciation is extended to all research station managers and collaborating farmers, including Jr. Beachy, Laverne Beachy, Anthony Beery, Harold Burton, Kevin Craun, Marty Diehl, Everette Gardner, Nathan Horst, Jeremy Kerns, Kyle Leonard, Nathan Lowder, Dale Reeves, John Pickler, Matt Rohrer, Dwight Swope, Ray Showalter, Kraig Smith, and Zeb Winslow. Financial support was provided by USDA-ARS and USDA-National Institute of Food and Agriculture, Competitive Grant Agreement No. 2013-67019-21369.

 

References

Footnotes


Comments
Be the first to comment.



Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.