Journal list menu
Statistics, Scoring Functions, and Regional Analysis of a Comprehensive Soil Health Database
Supplementary material is available online.
- Summary statistics were developed from a robust multiregional soil health (SH) dataset.
- Active carbon, organic matter, and penetration resistance were most useful soil health indicators.
- Midwestern soils had relatively lower mean values for measured biological properties than Northeast or Mid-Atlantic soils.
Soil health (SH) refers to the ability of a soil to function and provide ecosystem services. The Comprehensive Assessment of Soil Health (CASH) is an approach that measures 15 physical, biological, and chemical soil indicators, which are interpreted through scoring functions. This study reports on the SH status of 5767 samples from the Mid-Atlantic, Midwest, and Northeast regions of the USA as evaluated using CASH. Descriptive statistics and ANOVAs of subdatasets by region and soil textural group for SH indicators, in addition to correlation coefficients, principal component (PC) analysis, and best subsets regression (BSR) were performed. From these analyses, new CASH scoring functions were developed. Separate scoring functions by textural group (fine, medium, coarse) were necessary for Wet Aggregate Stability (WAS), Available Water Capacity (AWS), Organic Matter (OM), Active Carbon (AC), and Protein. Differences existed among regions, especially for WAS, OM, Protein, and Respiration (Resp), where the Midwest had relatively lower mean values compared to the Mid-Atlantic and Northeast. Biological properties and WAS showed moderately strong correlations (r = 0.58–0.78) and the highest loadings for the first two PCs. BSR results using the overall soil quality index as the response variable indicated that AC accounts for 45% of the variation, with additional predictability from Penetration Resistance, Resp, and WAS (68%). These four indicators are suggested for simplified SH tests. We conclude that the CASH approach can be successfully applied to evaluate the health status of soils with differing pedogenetic histories.
- Active Carbon
- Available Water Capacity
- Best Subsets Regression
- Comprehensive Assessment of Soil Health
- Cornell Soil Health Laboratory
- Cornell Soil Health Test, CND, Cumulative Normal Distribution
- Electrical Conductivity
- Loss on Ignition
- Principal Components Analysis
- Penetration Resistance
- Penetration Resistance 0 to 15 cm
- Penetration Resistance 15 to 45 cm
- Respiration over 4-d incubation
- Root Health Rating
- Soil Health, SMAF, Soil Management Assessment Framework
- Wet Aggregate Stability
Conceptually, soil health (SH) represents the emerging understanding of soil quality. Both terms refer to the ability of a soil to function and provide ecosystem services based on its inherent characteristics (e.g., texture, mineralogy) and environmental conditions (26; 4; 23). A soil's health status, within the context of land use and management goals, however, is consistent with the understanding of soils as a dynamic, complex, and living system (16). Intensive agriculture and poor land management practices have led to widespread soil degradation associated with increasing topsoil erosion, nutrient depletion, pollution, compaction, and loss of organic matter (31). A sustainable future with an ever-growing global population depends on healthy, well-functioning soils, which increase water and air quality, support human health and habitation, and sustain plant and animal productivity (25).
The assessment of soil health over time is recognized as a primary indicator of sustainable land management (26). It expands on traditional soil testing, which has largely focused on the measurement of chemical soil properties (i.e., soil pH and nutrient contents) to evaluate soil fertility (25; 34). This approach has proven useful for increasing agricultural production, but the narrow chemical focus has been a contributor to physical and biological soil degradation (45; 3; 26). The inadequacy of this conventional approach spurred the development of a more comprehensive assessment of soil health that evaluates multiple physical, biological, and chemical soil properties, or indicators, with an emphasis on those that are most sensitive to land management practices and correlated to ecosystem processes (25).
The Soil Management Assessment Framework (SMAF; 4) is a tool originally developed to evaluate how land management practices impact soil functioning. The SMAF proposes a three-step process, including (i) selection of a minimum data set of relevant soil health indicators, (ii) interpretation of measured indicator values using scoring functions, and (iii) integration of indicator scores into an overall soil health index. The projected range of values for each indicator is interpreted with consideration of local environmental conditions (e.g., precipitation, temperature, etc.), land use goals, and inherent soil properties (4). Using knowledge of the relationship between indicators and the relevant soil function(s), scoring functions are developed to translate measured indicator values into a unit-less score. The combination of all indicator scores into an additive index is meant to provide a single metric of overall soil health (4; 19).
The Comprehensive Assessment of Soil Health (CASH), initially referred to as the Cornell Soil Health Test (CSHT), was developed using the three-step process outlined for the SMAF. In the past decade, the CASH has been offered by the Cornell University Soil Health Lab (CSHL) as a commercial-scale, user-friendly soil health test designed to directly cater to the applied needs of agricultural land managers and owners. It places primary emphasis on the identification of specific soil constraints in agroecosystems, thereby aiding in the selection of land management solutions to increase land productivity and minimize environmental impact (23).
The CASH approach was based on data collected from 700 agricultural soils sampled from long-term controlled research sites and commercial growers' fields in the Northeast United States. The utility of 39 potential indicators of physical, biological, and chemical soil health was initially evaluated using a number of considerations, including cost of analysis, relevance to soil functions and processes, sensitivity to land management, reproducibility of measurements, sampling requirements, and potential to be estimated by statistical correlation or sensors (34). Reflecting only minor changes since its release in 2006, fifteen soil health indicators have been selected for inclusion in the 2016 CASH.
The assessment of dynamic soil properties is complicated by the intrinsic, cross-scale heterogeneity of soils that results from differing modes of formation and anthropogenic influences (25). Although the CASH was originally calibrated for use with soils of the northeastern United States, it is increasingly used for other geographic regions. Numerous soil health studies have been conducted at the plot (3; 5; 23; 15; 22) and landscape (27; 42; 51) scales, but few at the regional or national levels (8; 9; 10; 38). Using the CASH database of 5767 samples, we performed multiple statistical analyses, including an investigation into differences in soil health among three US regions that represent distinct soils, climates, topographies, and land uses.
The objectives of this study were to use a large soil health dataset to (i) describe basic statistics of soil health indicators, (ii) investigate the multivariate relations among multiple soil health indicators, (iii) analyze regional soil health differences, and (iv) develop updated scoring functions for CASH indicators.
MATERIALS AND METHODS
Data were compiled from all samples submitted for CASH analysis to the Cornell University Soil Health Laboratory (CSHL; Ithaca, NY) from 2007 to 2015. It was assumed that samples were collected from the 0- to 15-cm depth and composited by clients as per sampling guidelines (34). The CASH sample database is unique and large in its scope. However, it is comprised of soil samples that were passively received by the CSHL from many sources, and is not the product of a deliberate statistical sampling design. Therefore, it contains inherent biases, including geographical (i.e., greater representation of states in the Northeast United States).
Quantification of Soil Health Indicators
The current CASH offers measurement of soil texture, four physical indicators (Wet Aggregate Stability [WAS], Available Water Capacity [AWC], Penetration Resistance 0- to 15-cm [PR15], and Penetration Resistance 15- to 45-cm [PR45]), and five biological indicators (contents of Organic Matter [OM], Active Carbon [AC], and Autoclaved-Citrate Extractable Protein [Prot], Soil Respiration [Resp], and the Root Health Rating [Root]). Seven chemical properties (pH and Modified Morgan Extractable P, K, Mg, Fe, Mn, and Zn) are also assessed, representing a standard soil nutrient test.
Due to client package options and the delayed phase-in of some indicators (i.e., Prot and Resp), there is some variation in total sample size among measured indicators. Beginning in 2014, however, all soil samples submitted to the CSHL have been evaluated for texture, WAS, Resp, OM, and chemical properties. All laboratory measurements were performed on disturbed air-dried soil sieved past 2 mm unless otherwise noted. Appropriate corrections for sample water content after air-drying were determined after drying overnight in an oven at 105°C.
Soil texture was assessed using a rapid quantitative method developed by 29. A known weight of soil was dispersed with 3% hexametaphosphate [(NaPO3)6], and a combination of sieving and sedimentation steps was used to separate particle size fractions. For samples submitted for the most basic CASH package, texture was determined using the feel method as modified from 44. Soil textural grouping (coarse, medium, or fine) was determined from the texture analysis. Sand, loamy sand, and sandy loam classes were considered as “coarse”; sandy clay loam, loam, silt loam, and silt as “medium”, and sandy clay, clay loam, silty clay loam, silty clay, and clay as “fine” (39).
WAS was determined using a purpose-built rainfall simulator (0.59-m diam.) fitted with Teflon capillary microtubing and an adjustable Mariotte-type bubbling tube to control hydraulic pressure (36; 33). Approximately 30 g of dry soil aggregates (sieved past 0.25–2.00 mm) were evenly distributed on a 0.25-mm mesh sieve (200-mm diam.) and fitted above a filter funnel to collect slaked soil. From a distance of 0.5 m, samples received a total of 12.5 mm of reverse-osmosis water delivered as 4-mm drops over 300 s (1.9 J). The weight of slaked (‘unstable’) soil that fell through the sieve and into the filter was measured after oven drying at 105°C; this value was used to determine the proportion of water ‘stable’ aggregates remaining on the sieve by difference (34; 33).
For AWC, the difference between soil water content at field capacity (θfc) and permanent wilting point (θpwp) was assessed gravimetrically (g water g soil-1). Saturated soil subsamples were equilibrated to pressures of -10 kPa (θfc) and -1500 kPa (θpwp) for 7 d on ceramic high-pressure plates in air pressure chambers (Soil Moisture Equipment Corp., Goleta, CA; 46; 37).
Penetration resistance (PR) measurements were collected in the field by clients using a penetrometer or soil compaction tester (DICKEY-john Corp., Auburn, IL). Maximum PR was recorded for the depths of 0- to 15-cm (PR15) and 15- to 45-cm (PR45). PR values were not adjusted for water content, but guidelines recommend that readings are taken near field capacity conditions (17).
Active Carbon (AC) was assessed as permanganate-oxidizable carbon, measured in duplicate by reacting a 2.5 g soil sample with 20 mL 0.02 M potassium permanganate (KMnO4) solution (pH 7.2). Extracts were shaken (120 rpm, 2 min), then allowed to settle for exactly 8 min. An aliquot of solution was diluted 100 times before measurement for absorbance at 550 nm using a handheld spectrophotometer (Hach, Loveland, CO). Sample absorbance was calibrated with KMnO4 standard curves and converted to mg AC per kg soil using the equation of 50.
Soil heterotrophic respiration (Resp) was measured after 4-d incubation in duplicate using methods modified from 20 and 55. Soil sieved past 8 mm was weighed (20.0 g) into a perforated aluminum weighing boat that was put inside a glass jar sitting atop two staggered Whatman qualitative filter papers. A preassembled potassium hydroxide (KOH) trap (10 mL glass beaker adhered to a plastic stand) was placed onto the weighing boat and the beaker was filled with 9.0 mL 0.5 M KOH. Distilled water (7.5 mL) was pipetted into the jar along the side to facilitate rewetting of the sample via capillary rise. The amount of carbon dioxide (CO2) respired and absorbed by the KOH trap over the course of incubation was determined by measuring the change in electrical conductivity (EC) of the solution with an OrionTM DuraProbeTM 4-Electrode Conductivity Cell (ThermoFisher Scientific, Inc., Waltham, MA; 52; 53). The necessary background correction for atmospheric CO2 was quantified using blank (no soil) incubations.
Extractable protein (Prot) content of soil samples was determined by performing a neutral sodium citrate (pH 7) buffer extraction (28; 54). Extracts were shaken (5 min, 180 rpm) then subjected to high pressure and temperature in an autoclave (121°C, 20 min). After cooling to room temperature, an aliquot of each extract was clarified by centrifugation (10,000 x g, 3 min). The quantity of extracted protein in solution was measured using the colorimetric bicinchoninic acid (BCA; Thermo Pierce, Waltham, MA) assay with a 96-well spectrophotometric plate reader (BioTek Inc., Winooski, VT; 49). Sample absorbance readings were calibrated using standard concentration curves of Bovine Serum Albumin.
The root health bioassay was performed by planting replicate snap beans (Phaseolus vulgaris L., ‘Hystyle’) in nursery cones filled with 4-mm sieved soil. Plants were grown in a greenhouse for 5 to 6 wk, where they received regular watering and 12 to 14 h of light per day (1; 2). Once they had reached full flowering, plants were harvested and their roots were hand-washed using tap water. Cleaned roots were rated on a scale of 1 to 9, where lower ratings indicate a lack of and higher ratings indicate the strong presence of visible disease symptoms in hypocotyl and root (1).
Soil pH was measured from a suspension of 1:1 water to soil (v/v) ratio using a six-channel robotic pH tester equipped with refillable, double junction glass bulb pH electrodes (LIGNIN, LLC, Albuquerque, NM). Macronutrients and micronutrients in soil samples (P, K, and Mg, Fe, Mn, Zn, respectively) were determined from a Modified Morgan (MM; ammonium acetate plus acetic acid, pH 4.8) extraction using an inductively coupled plasma (ICP) emission spectrometer (SPECTRO Analytical Instruments Inc., Mahwah, NJ). The MM solution is the weakest of the commonly used soil extractants (e.g., Mehlich-3, Bray-I, and Olsen) and is thought to better target the bioavailable fractions that are most relevant to ecosystem processes on short time scales.
Initial database cleaning included removal of all samples having OM content greater than 10% (remaining n = 5767). Based on identification of the state from which they were submitted, samples were categorized into three U.S. regions for which there were sufficient sample sizes (Fig. 1 and Table S1). These regions included (i) Mid-Atlantic (DC, NJ, DE, MD, PA, VA, WV), (ii) Midwest (IA, IL, IN, MN, MO, WI), and (iii) Northeast (CT, MA, ME, NH, NY, RI, and VT). Major Land Resource Area (MLRA; 48) Regions K, L, M, N, O, P, R, S, and T cover states included in one of the three regions used to analyze soil health status for this study (N, P, R, S, and T for Mid-Atlantic; K, L, M, N, O for Midwest; and L, R, and S for Northeast). The Midwest region includes significant portions of the US primary grain belt, known for its deep soils, level or gently sloping topography, and high agricultural productivity. The Mid-Atlantic (nonglaciated) and Northeastern (glaciated) regions generally feature a higher diversity of landscapes and soils (48), with most CASH samples derived from the productive agricultural areas. MI and OH samples (n = 674) were excluded from the regional analysis because they are located in MLRA transition areas, and could not be assigned with confidence to a specific physiographic region.
Descriptive statistics and analysis of variance (ANOVA) were performed on the full dataset (n = 5767) and three regional subdatasets by textural grouping (coarse, medium, fine) for all soil health indicators. Mean separation was computed using the Tukey posthoc test after a significant F-ratio (p < 0.05) was observed. Pearson correlation coefficients for physical, biological, and chemical indicators were calculated using the full dataset (n = 5767, all textures). Multivariate statistics were evaluated using a subset of the database, including only samples that had measured values for all fifteen indicators (n = 930) regardless of textural group or region (Prot and Resp were added to the CASH for 2014 and 2015, which limited the sample size for these indicators). Principal component analysis (PCA) was conducted on the standardized measured values (subtraction of the mean and division by the standard deviation) of all CASH indicators. The first two PCs were visualized in two-dimensional space, and soil health indicator property weighting in the eigenvectors were used to determine whether the observed trends could be summarized via major components. A Best Subsets Regression (BSR) was performed where the overall soil health index (see Section 2.4) was predicted using subsets of individual soil health indicator scores. All statistics were computed using Minitab 17 software (32).
Development of Soil Health Indicator Scoring Functions
Although descriptive statistics were calculated for chemical indicators, scoring functions as in Eq.  were not used as interpretations for pH and nutrient concentrations as they are interpreted based on existing soil fertility recommendation systems. The scoring of chemical indicators is described below.
The CASH Summary Report includes an overall soil health index, which represents the mean of physical and biological indicators and pH, P, and K scores (secondary and trace nutrients are combined into a single metric, as detailed below).
RESULTS AND DISCUSSION
The CASH sample database included total sample numbers of 1889, 3605, and 273 for coarse, medium, and fine textural groups, respectively, and 420, 906, and 3762 samples (all textures) from the Mid-Atlantic, Midwest, and Northeast regions, respectively (Table S1). Statistical summaries of the CASH indicators by textural group are provided in Tables S2–4. CASH indicators exhibited a wide range of values due to the large variety of samples included in the database (e.g., location, topography, climate, time of sampling, sample handling, land use, and land management practices; Table 1). All indicators exhibited relatively high standard deviations compared with the means (Table 1), except for pH which is apparently effectively managed by farmers. The other chemical indicators (P, K, Mg, Fe, Mn, Zn) were the most variable and nonnormally distributed. Log10 transformation was successful in obtaining skew and kurtosis measures close to zero for chemical indicators (except pH, which is already defined using a logarithmic scale). Normality of pH improved with log transformation for fine textured soils, likely related to the lower relative sample size (Table S4). Compared to physical and other biological indicators, AC most closely followed a normal distribution and had the lowest variability within textural groups.
|Soil health indicator||p level for texture||Coarse||Medium||Fine|
|Sand (%)||0.000||65.6 (11.7), 1845||A||30.6 (13.7), 3514||B||16.5 (10.9), 234||C|
|Silt (%)||0.000||28.4 (11.1), 1845||C||55.8 (12.6), 3514||A||50.5 (13.6), 234||B|
|Clay (%)||0.000||6.0 (3.2), 1846||C||13.6 (6.1), 3515||B||33.0 (7.1), 234||A|
|WAS (%)||0.000||52.2 (23.8), 1889||A||42.2 (24.7), 3605||B||41.8 (20.0), 273||B|
|AWC (g g−1)||0.000||0.152 (0.068), 1846||C||0.208 (0.068), 3530||B||0.219 (0.060), 234||A|
|PR15 (kPa)||0.032||1158 (662), 1571||A||1110 (621), 2640||B||1110 (655), 128||AB|
|PR45 (kPa)||0.000||2199 (641), 1567||A||2041 (745), 2615||B||2048 (952), 128||B|
|OM (%)||0.000||3.26 (1.89), 1888||C||3.74 (1.51), 3604||B||4.42 (1.36), 273||A|
|AC (mg kg−1)||0.000||486.1 (241.1), 1844||C||531.2 (182.2), 3531||B||608.7 (168.4), 234||A|
|Prot (mg g−1)||0.000||10.2 (5.7), 512||A||7.0 (4.4), 1338||B||5.6 (2.4), 173||C|
|Resp (mg CO2 g−1)||0.519||0.64 (0.39), 555||A||0.62 (0.31), 1412||A||0.61 (0.27), 212||A|
|Root Health Rating (1–9)||0.373||4.5 (1.2), 1474||A||4.4 (1.2), 2489||A||4.2 (1.2), 142||A|
|pH||0.035||6.2 (0.7), 1889||AB||6.3 (0.8), 3605||A||6.1 (0.8), 273||B|
|P (ppm)||0.000||21.1 (47.3), 1889||A||12.9 (31.9), 3605||B||9.3 (16.4), 273||B|
|K (ppm)||0.000||122.8 (113.7), 1889||B||126.7 (90.8), 3591||B||207.6 (132.3), 272||A|
|Mg (ppm)||0.000||140.9 (127.8), 1889||C||242.2 (296.6), 3605||B||471.8 (428.1), 273||A|
|Fe (ppm)||0.002||8.6 (43.5), 1889||A||5.9 (14.0), 3605||B||5.6 (11.7), 273||AB|
|Mn (ppm)||0.000||10.5 (9.6), 1889||B||14.4 (10.6), 3605||A||10.2 (7.2), 273||B|
|Zn (ppm)||0.000||2.6 (5.8), 1889||A||1.4 (3.6), 3605||B||1.0 (1.2), 273||B|
- † Abbreviations: AC, Active Carbon; AWC, Available Water Capacity; OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm; Prot, Soil Protein; Resp, Soil Respiration; WAS, Wet Aggregate Stability.
Pearson product–moment correlation coefficients were computed for every pair of soil health indicators to create a correlation matrix (Table 2). Out of 171 pairs (including percent sand, silt, and clay), 129 were significantly correlated to each other (p ≤ 0.05). Strong positive correlations were observed, especially between biological indicators: OM with AC (r = 0.723), OM with Prot (r = 0.780), and OM with Resp (r = 0.671), as well as for WAS with Prot (r = 0.672), AC with Prot (r = 0.680), and Prot with Resp (r = 0.626). Moderately strong (r ≥ 0.5) positive correlations were observed for silt with AWC, WAS with OM, WAS with AC, and AC with Resp; moderate negative correlations were observed for sand with AWC. PR15, PR45, and Root Health Ratings exhibited little to no correlation with other indicators. In general, only very weak correlations with physical and biological indicators were identified for chemical indicators. The observed independence of measured chemical properties from physical and biological indicators is presumably because soil pH and nutrient status have historically been adjusted with fertilizer and lime additions as recommended by traditional soil tests.
- † Abbreviations: AC, Active Carbon; AWC, Available Water Capacity, OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm; Root, Root Health Rating; Prot, Soil Protein; Resp, Soil Respiration; WAS, Wet Aggregate Stability.
Principal Component Analysis
PCA was used to further examine the relationships between CASH indicators, with the goal of defining their linear alignment (n = 930). Six PCs accounted for 74% of the total variability in the raw data set and fulfilled criteria outlined by 24 of having eigenvalues greater than one and a cumulative fraction of total variance of at least 70% (Fig. 2; Table 3).
|Root Health Rating||0.102||−0.028||0.327||0.595||−0.043||−0.115|
- † Abbreviations: AC, Active Carbon; AWC, Available Water Capacity, OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm; Prot, Soil Protein; Resp, Soil Respiration; WAS, Wet Aggregate Stability.
PC1 accounted for 27% of the variance, with five variables having high positive loadings (OM > Prot > Resp > AC > WAS). PC2 had high positive loadings for pH, K, and Mg, and high negative loadings for Fe, representing 16% of the total variance. PC3 (9%) scores increased with increasing PR15 and PR45. For PC4 (7%), scores increased with increasing Root Health Rating (lower score) and decreasing AWC. PC5 (6%) had high negative loadings for P, Fe, and Zn and positive loadings for pH. Mn had high negative loadings for PC6 (5%). A two-dimensional visualization of the first two components (Fig. 2) showed little differentiation among textural groups. Samples were clustered slightly to the left side of the PC1 axis and equally spread out in the PC2 direction.
The relatively low explained variance for the primary PCs demonstrates the complexity of soil health assessment, but PCs did generally fall into categories that represent distinct soil processes: biological (PC1), chemical (PC2, PC5, PC6), and physical (PC3). The high dimensionality of the PC space (six with eigenvalues greater than one) and low dominance of the first PCs suggests that each indicator tends to represent a process that is differently expressed in each soil sample. These findings support the use of a multi-indicator assessment such as the CASH, rather than approaches targeting single or small sets of properties.
Best Subsets Regression
A Best Subsets Regression (BSR) was performed where the overall soil health index was predicted using subsets of individual soil health indicator scores, starting with subsets numbers of 1, 2, 3, … This approach evaluates which indicator(s) are most predictive of overall soil health, and are, therefore, most suitable for a simplified soil health assessment. It is noted that in the present analysis, the 15 indicator scores are used to calculate the overall soil health score through an unweighted mean, and then subsequently used as independent regression coefficients (i.e., it is recognized that they are not computationally independent). Therefore, the BSR analysis was limited to three variables in the prediction model.
When considering only a single variable, the three biological indicators AC, OM, and Resp were the most predictive of overall CASH score (R2–adj = 45.3%, 43.5%, and 35.8%, respectively; Table 4), corroborating the notion that soil organic matter (whether considering active, total, or mineralizable) is key to soil health (30). AC is suggested as the best single CASH indicator, as it has the advantages of being readily measured in the field and more responsive to land management (50; 12).
Active Carbon is always an included predictor in the best subsets model when considering multiple CASH indicators (Table 4). Penetration resistance (PR15 or PR45) combined with AC showed the highest predictability when considering two variables, indicating that the combination of a biological (AC) and physical (PR) indicator greatly improves predictability of overall CASH score (R2–adj = 57.2% for AC+PR15 and 59.5% for AC+PR45). When evaluating models with three CASH indicators, Resp increases predictability when supplemented with AC and PR measurements, reaching R2–adj values near 65%. It is concluded that (i) simplified soil health assessments can likely be performed at relatively low cost, possibly in the field, and (ii) biological and physical, rather than chemical, indicators incrementally contribute overall soil health information as measured by the CASH.
Comparisons of Soil Health Indicators Between Textural Groups
Descriptive statistics (mean, standard deviation, and sample size), p level from one-way ANOVA, and Tukey comparisons from the separation of the sample database by soil textural groups (no regions) are presented in Table 1. Effects of textural group at the 95% confidence level were observed for all indicators except Resp and Root Health Rating. For WAS, PR15, PR45, Prot, P, Fe, and Zn, coarse samples had the highest relative mean values (α = 0.05). Medium textured soils had the highest mean Mn, with means that fell in the middle between coarse and fine for AWC, OM, AC, Prot, and Mg (α = 0.05). Fine textured soils had the highest mean AWC, OM, AC, K, and Mg (α = 0.05). Tukey comparisons between coarse and medium groups were insignificant for Prot, pH, and K (α = 0.05); comparisons between medium and fine were statistically insignificant for WAS, PR15, PR45, P, Fe, and Zn.
Available Water Capacity, OM, AC, and K were all lower for coarse- than fine-textured soils. These findings are in line with conventional expectations, as soils with relatively high clay content improve water storage, retain more soil organic matter, and act as a slow-release source of K (7). In contrast, WAS, Prot, and P were all higher for coarse textured soils. These results were not expected, as the greater surface reactivity of clay-sized particles would be expected to promote soil aggregation and binding of OM and inorganic ions (11). Possible explanations include (i) higher Prot and P extraction efficiency for coarser soils, (ii) enhanced physical protection from raindrop impact provided by coarse sand-sized particles and related to the methodology of WAS determination, and/or (iii) sample bias, such as would result if submitted coarse samples have been previously managed to yield higher values (e.g., relatively more organic or livestock farms).
Comparisons of soil health indicators among regions
Results of one-way ANOVA by region for coarse, medium, and fine textural groupings (Tables 5, 6, and 7) were performed to address a possible need to develop regional interpretations and scoring functions for soil health indicators. Sample distributions by state for each region are provided in Table S1.
|Soil health indicator||p level for region||Regional means: coarse textural group|
|Sand (%)||0.000||68.6 (13.2), 101||A||65.9 (10.0), 63||AB||64.0 (11.4), 1374||B|
|Silt (%)||0.000||26.7 (11.7), 101||B||25.5 (8.1), 63||B||30.2 (10.9), 1374||A|
|Clay (%)||0.000||4.6 (4.3), 101||C||8.6 (4.0), 63||A||5.9 (3.0), 1374||B|
|WAS (%)||0.000||44.8 (20.0), 101||B||32.6 (16.8), 63||C||56.4 (23.9), 1417||A|
|AWC (g g−1)||0.000||0.11 (0.05), 101||C||0.14 (0.06), 63||B||0.17 (0.07), 1374||A|
|PR15 (kPa)||0.005||1252 (677), 83||AB||1537 (982), 46||A||1214 (650), 1251||B|
|PR45 (kPa)||0.425||2212 (579), 83||A||2339 (818), 45||A||2210 (650), 1248||A|
|OM (%)||0.000||2.20 (1.74), 101||B||2.53 (1.65), 63||B||3.70 (1.80), 1417||A|
|AC (mg kg−1)||0.000||335.4 (220.7), 101||B||477.5 (228.4), 63||A||537.3 (230.1), 1374||A|
|Prot (mg g−1)||0.000||6.4 (5.7), 23||B||5.5 (3.2), 52||B||11.4 (5.6), 383||A|
|Resp (mg CO2 g−1)||0.001||0.52 (0.43), 23||AB||0.47 (0.27), 52||B||0.67 (0.41), 426||A|
|Root Health Rating (1–9)||0.129||4.4 (1.3), 73||A||4.8 (1.0), 38||A||4.4 (1.2), 1079||A|
|pH||0.003||6.00 (0.53), 101||B||6.30 (0.47), 63||A||6.21 (0.69), 1417||A|
|P (ppm)||0.003||14.2 (40.5), 101||B||40.1 (57.4), 63||A||20.2 (48.9), 1417||B|
|K (ppm)||0.006||92.5 (85.7), 101||B||150.4 (95.6), 63||A||115.0 (114.3), 1417||B|
|Mg (ppm)||0.000||87.0 (71.5), 101||C||208.5 (148.4), 63||A||141.7 (135.6), 1417||B|
|Fe (ppm)||0.185||3.5 (5.0), 101||A||2.4 (4.0), 63||A||10.2 (50.1), 1417||A|
|Mn (ppm)||0.001||7.4 (7.4), 101||B||10.4 (9.4), 63||AB||11.1 (9.8), 1417||A|
|Zn (ppm)||0.105||1.9 (4.0), 101||A||4.0 (6.6), 63||A||2.7 (6.3), 1417||A|
- † Abbreviations: AC, Active Carbon; AWC, Available Water Capacity; OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm; Prot, Soil Protein; Resp, Soil Respiration; WAS, Wet Aggregate Stability.
|Soil Health Indicator||p level for region||Regional means: Medium textural group|
|Sand (%)||0.000||31.0 (11.6), 307||B||16.7 (12.5), 721||C||35.7 (10.1), 2139||A|
|Silt (%)||0.000||56.4 (12.7), 307||B||64.5 (13.3), 721||A||52.8 (9.9), 2139||C|
|Clay (%)||0.000||12.7 (5.8), 307||B||18.8 (5.1), 721||A||11.4 (5.2), 2139||C|
|WAS (%)||0.000||42.7 (24.4), 317||B||25.2 (14.2), 721||C||48.8 (24.7), 2219||A|
|AWC (g g−1)||0.000||0.22 (0.05), 317||B||0.23 (0.05), 721||A||0.19 (0.07), 2146||C|
|PR15 (kPa)||0.071||1346 (662), 258||A||1264 (753), 304||A||1050 (572), 1973||B|
|PR45 (kPa)||0.680||2005 (728), 258||A||2053 (718), 283||A||2037 (719), 1969||A|
|OM (%)||0.000||4.12 (1.76), 317||A||3.04 (1.02), 721||B||3.99 (1.51), 2219||A|
|AC (mg kg−1)||0.000||564.1 (231.7), 317||A||475.0 (154.2), 721||B||549.6 (174.5), 2146||A|
|Prot (mg g−1)||0.000||10.0 (6.1), 144||A||4.9 (1.8), 473||C||8.8 (4.7), 494||B|
|Resp (mg CO2 g−1)||0.000||0.86 (0.41), 145||A||0.47 (0.18), 473||C||0.70 (0.34), 567||B|
|Root Health Rating (1–9)||0.000||4.1 (1.2), 242||C||4.7 (1.3), 353||A||4.2 (1.1), 1715||B|
|pH||0.000||6.18 (0.61), 317||B||5.98 (0.54), 721||C||6.41 (0.84), 2219||A|
|P (ppm)||0.054||23.2 (73.7), 317||A||10.0 (19.4), 721||B||12.8 (26.1), 2219||B|
|K (ppm)||0.000||163.7 (102.5), 317||A||171.5 (89.4), 721||A||103.6 (80.4), 2205||B|
|Mg (ppm)||0.091||173.2 (99.4), 317||B||322.5 (174.1), 721||A||214.5 (344.3), 2219||B|
|Fe (ppm)||0.000||4.7 (9.7), 317||AB||3.1 (3.8), 721||B||6.7 (16.2), 2219||A|
|Mn (ppm)||0.000||17.2 (9.1), 317||A||12.6 (8.9), 721||C||14.0 (8.7), 2219||B|
|Zn (ppm)||0.001||2.0 (2.3), 317||A||1.2 (1.7), 721||B||1.3 (4.3), 2219||B|
- † Abbreviations: AC, Active Carbon; AWC, Available Water Capacity; OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm; Prot, Soil Protein; Resp, Soil Respiration; WAS, Wet Aggregate Stability.
|Soil Health Indicator||p level for region||Regional means: Fine textural group|
|Sand (%)||0.000||12.8 (9.8), 122||B||18.1 (8.6), 77||A|
|Silt (%)||0.000||56.1 (10.4), 122||A||46.8 (12.8), 77||B|
|Clay (%)||0.000||31.1 (5.4), 122||B||35.1 (8.8), 77||A|
|WAS (%)||0.000||37.7 (13.6), 122||B||48.8 (24.2), 116||A|
|AWC (g g−1)||0.000||0.23 (0.05), 122||A||0.19 (0.08), 77||B|
|PR15 (kPa)||0.026||924 (250), 36||B||1211 (746), 85||A|
|PR45 (kPa)||0.084||1883 (472), 36||A||2203 (1057), 85||A|
|OM (%)||0.001||4.27 (0.92), 122||B||4.84 (1.66), 116||A|
|AC (mg kg−1)||0.053||616.5 (143.0), 122||A||661.2 (177.9), 77||A|
|Prot (mg g−1)||0.000||5.6 (1.6), 113||B||7.3 (3.6), 35||A|
|Resp (mg CO2 g−1)||0.000||0.53 (0.21), 113||B||0.70 (0.28), 74||A|
|Root Health Rating (1–9)||0.000||4.6 (1.1), 51||A||3.9 (1.0), 65||B|
|pH||0.000||5.86 (0.55), 122||B||6.24 (0.82), 116||A|
|P (ppm)||0.020||10.9 (14.7), 122||A||6.3 (15.1), 116||B|
|K (ppm)||0.000||246.6 (80.9), 122||A||125.0 (110.6), 115||B|
|Mg (ppm)||0.035||521.8 (170.0), 122||A||399.3 (614.5), 116||B|
|Fe (ppm)||0.000||3.1 (4.1), 122||B||8.8 (16.6), 116||A|
|Mn (ppm)||0.000||7.2 (4.3), 122||B||13.0 (8.6), 116||A|
|Zn (ppm)||0.273||0.9 (0.7), 122||A||1.0 (1.6), 116||A|
- † Inadequate sample size for Mid-Atlantic Region (n = 2)
- ‡ Abbreviations: AC, Active Carbon; AWC, Available Water Capacity; OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm; Prot, Soil Protein; Resp, Soil Respiration; WAS, Wet Aggregate Stability.
Coarse textural group
The mean sand, silt, and clay contents for all regions in this group is represented by the sandy loam class (Table 4). The Mid-Atlantic region had the highest mean percent sand (69%), the Northeast the highest mean percent silt (30%), and the Midwest the highest percent clay (9%). No statistically significant differences (α = 0.05) among regions were observed for PR45, Root Health Bioassay, Fe, or Zn. The Northeast region samples had the highest mean WAS and AWC (56.4% and 0.17 g g-1, respectively; p < 0.001). For WAS, the lowest mean was observed for the Midwest (32.6%); for AWC, the Mid-Atlantic had the lowest mean (0.14 g g-1). The Northeast region had the highest mean values for all biological properties except Root Health Rating (for which low values are more favorable).
Medium textural group
Samples of the medium textural group made up the largest proportion of the database and all subdatasets (Tables 6 and S1). The representative texture class in this group for all regions was silt loam. Midwest soils had higher relative mean silt and clay, and lower relative mean sand, compared to the mean of the other two regions (p < 0.001). Highly significant differences among regions were observed for all indicators (p < 0.001) except for PR15, P, Mg, and Zn (p < 0.1) and PR45 (not significant). Mean WAS was lowest in the Midwest (25.2%), nearly half that of the Northeast (48.8%). Mean AWC followed the opposite trend (i.e., lowest in the Northeast at 0.19 g g-1 and highest in the Midwest at 0.23 g g-1), likely reflecting overall regional differences in soil texture (i.e., higher silt and clay contents in the Midwest). Mean PR15 was lowest in the Northeast (1050 kPa; p = 0.071) and mean Root Health Rating was highest in the Midwest. Mid-Atlantic medium textured soils had the highest mean for every biological indicator except Root Health Rating. Highest mean pH and Fe were observed in the Northeast (pH 6.41 and 6.7 ppm, respectively), highest mean P (23.2 ppm; p = 0.054), Mn (17.2 ppm; p < 0.001), and Zn (1.3 ppm; p < 0.001) in the Mid-Atlantic, and highest mean Mg (322.5 ppm; p = 0.091) in the Midwest. Mean K was higher in Midwest and Mid-Atlantic regions (172 and 164 ppm, respectively).
Fine textural group
The fine textural group had the lowest sample sizes. Regional analysis was limited to the Midwest and Northeast regions (only two samples from the Mid-Atlantic; Table 7). The representative texture class of this group for both regions was silty clay loam. Mean percent silt was higher and percent clay lower in the Midwest. No significant differences between regions were observed for Zn (p = 0.273). Regional comparisons between all other indicators were significant at p < 0.035. The Northeast had higher mean WAS, PR15, OM, Prot, Resp, pH, Fe, and Mn; the Midwest had higher AWC, Root Health Rating, P, K, and Mg.
These results offer insights into regional soil health differences that may be attributed to genetic and management factors. Somewhat unexpectedly, Midwest soils generally showed the least favorable mean values for the biological indicators (OM, AC, Prot, Resp, and Root Health Rating) and the physical indicator WAS. This was especially evident for the numerically dominant medium textured soils. These findings counter the common notion that Midwestern soils are of superior quality for agriculture than those in other regions, as well as critiques that soil health assessments are inherently biased to favor Mollisols or similar soil orders (41; 40). There are a few likely explanations for these findings. First, there is imperfection in the representation of samples due to a nondeliberate sampling design. Furthermore, Northeast and Mid-Atlantic samples generally receive greater organic inputs (especially manure), are more often managed to include rotations with perennial crops, and are more likely to be part of an organic farming system. In contrast, in the Midwest, intensive row cropping and fertilization with synthetic chemicals is relatively more common, and the soils appear to have slightly different textural components (i.e., more silt in the medium- and fine-textured groups, possibly related to more prevalent aeolian soil origins). In all, we infer from the large sample numbers, and the often highly significant regional differences observed, that the soil health dissimilarities are real, although confirmation through follow-up studies is warranted.
Informed by the results of this investigation into regional differences in soil health measures, CASH scoring functions were updated for use in reports of laboratory assessments provided to clients. For generic reporting without geographic considerations, the mean of the means and standard deviations of each region were determined for physical and biological indicators (i.e., accounting for regional differences in sample size). With acknowledgment of possible bias in the CSHL database and reflecting some necessary adjustment based on expert knowledge, client feedback, and experimental trials (unpublished data), the updated CASH scoring functions for physical and biological indicators are provided in Table 8. The overall effect of the 2016 changes is that measured values tend to be scored marginally higher and be more sensitive to change. All scores are interpreted using a five-color scale (red, orange, yellow, light green, dark green) to classify values as very low (0–20), low (20–40), medium (40–60), high (60–80), and very high (80–100), respectively (Fig. 3, 4).
|Soil health indicator||Associated soil processes||Type of scoring||2016 CASH scoring function(s)||Textural group|
|WAS (%)||Aeration, infiltration, shallow rooting, surface crusting, resistance to erosion, structural stability, surface runoff||More is better||Score = 100* CND(40, 23)||C|
|Score = 100* CND(31, 18)||M|
|Score = 100* CND(35, 19)||F|
|AWC (g g−1)||Water retention and availability, drought stress tolerance||More is better||Score = 100* CND(0.13, 0.07)||C|
|Score = 100* CND(0.16, 0.06)||M+F|
|PR15 (kPa)||Shallow rooting, aeration, water infiltration and transmission||Less is better||Score = 100* 1-CND(1131, 648)||C+M+F|
|PR45 (kPa)||Deep rooting, drainage, drought stress tolerance||Less is better||Score = 100* 1-CND(2068, 758)||C+M+F|
|OM (%)||Carbon storage, water and nutrient retention, biogeochemical cycling||More is better||Score = 100* CND(2.0, 0.8)||C|
|Score = 100* CND(3.1, 0.9)||M|
|Score = 100* CND(4.0, 0.9)||F|
|AC (mg kg−1)||Readily available energy for soil biological activity||More is better||Score = 100* CND(450, 200)||C|
|Score = 100* CND(500, 185)||M|
|Score = 100* CND(585, 200)||F|
|Prot (mg g−1)||Nitrogen availability, nutrient cycling||More is better||Score = 100* CND(7.4, 3.5)||C|
|Score = 100* CND(6.5, 3.3)||M|
|Score = 100* CND(5.9, 3.0)||F|
|Resp (mg CO2 g−1)||Microbial activity, decomposition||More is better||Score = 100* CND(0.6, 0.3)||C+M+F|
|Root Health Rating (1–9)||Soil-borne pathogen pressure||Less is better||Score = 100* 1-CND(4.5, 1.0)||C+M+F|
- † Abbreviations: AC, Active Carbon; AWC, Available Water Capacity; OM, Organic Matter; PR15, Penetration Resistance 0 to 15 cm; PR45, Penetration Resistance 15 to 45 cm.
Higher scores are associated with better soil health. Most physical and biological indicators are scored using a ‘more is better’ function, where scores are higher with increasing measured values. Scoring functions for penetration resistance (PR15 and PR45) and Root Health Rating use ‘less is better’ curves, where lower measured values or ratings are assigned a higher score. Scoring functions are textural group-dependent for all physical and biological indicators except PR15, PR45, Resp, and Root Health Rating.
Soil pH is scored using an optimum function, where values ranging from 6.3 to 7.2 are considered optimal (i.e., score = 100) for the availability of plant nutrient elements (47), agricultural crop growth (13), and soil microbial activity (43). A score of 0 is assigned when pH ≤ 5.4 or pH ≥ 7.7, with linear interpolation from the optimum values. A separate function is used for acid-loving crops, such as blueberries (Vaccinium corymbosum L.) or potatoes (Solanum tuberosum L.), set one pH unit lower (i.e., pH 5.3-to-6.2 is optimum, etc.). Extractable-P also uses an optimum scoring function, where concentrations ranging from 3.5-to-21.5 ppm are scored as 100. Negative impacts are expected when P is deficient (≤ 0.45 ppm) or excessive (≥ 100 ppm); in both cases, values are scored as 0 and interpolations are made from the optimum values. For Extractable-K, the scoring function is ‘more is better,’ where values greater than 74 ppm are scored as 100.
Secondary nutrients (Mg, Fe, Mn, Zn) are scored using a subscoring system (Fig. 4d). Each nutrient is assigned a sub-score of either 0 (suboptimum) or 100 (optimum) depending on measured values. The average of the four nutrient subscores is used to determine the secondary nutrient score. Suboptimal conditions for soil health are indicated when Mg and Zn are too low (< 33 ppm and < 0.25 ppm, respectively) and Fe and Mn are too high (> 25 ppm and > 50 ppm, respectively; 6; 18; 21; 14). The percentage of samples having suboptimal secondary element concentrations was generally low in our dataset (3.8% for Mg, 2.6% for Fe, 1.0% for Mn, and 7.2% for Zn).
The overall soil health index is calculated by the unweighted mean of individual indicator scores (secondary nutrients are combined into a single metric, as described above). Regional scoring functions can similarly be developed based on the regional soil health statistics (CND function, mean and standard deviation; Tables 5, 6, and 7).
The CASH approach evolved from extensive research (4; 33; 23) followed by a decade of commercial assessment of soil samples that yielded a large database (especially for the Northeast United States) with multiple measured soil health indicators. An investigation into regional differences (Mid-Atlantic, Midwest, and Northeast) by textural group found significant differences in mean measured values for most physical, biological, and chemical properties. Results were used to develop new CASH scoring functions, which are supported by the quantitative analysis of a geographically diverse dataset. Evidence suggests that the development of region-specific scoring functions may be appropriate after more complete regional soil health data analyses.
Multivariate analysis of the dataset revealed that some indicators (especially biological) are correlated, but still have sufficient levels of orthogonality that suggest independent effects. For example, most biological indicators show correlations with OM content, but still add additional information on the quality of the OM. Yet, the BSR analysis revealed that AC (and secondarily OM) was the best single predictor of overall CASH score, and that PR measurements were additionally informative, followed by Resp and WAS. These results suggest possibilities for simplified soil health assessments, although less informative than the comprehensive test. Chemical indicators exhibited little correlation with biological and physical indicators, supporting the notion that the nutrient status of agricultural soils has generally been independently managed through fertilizer and lime additions with little relation to other attributes of soil health. We conclude that the CASH approach can be successfully applied, with possible regional parameterization, to evaluate the soil health status of soils with differing pedogenetic histories.
Four supplementary tables are available online. Table S1 shows states included in the Mid-Atlantic, Midwest, and Northeast regions of the United States and the number of samples from each state analyzed, sorted by textural grouping. Table S2 shows summary statistics for all course textured soils without regional separation. Table S3 shows summary statistics for all medium textured soils without regional separation. Table S4 shows summary statistics for all fine textured soils without regional separation.
This research was funded by the USDA Natural Resources Conservation Service, grant number 77105.
|saj2sssaj2016090286-sup-0001.pdfPDF document, 289.3 KB||
Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
- 1, , and . 2004. Assessing root health by a soil bioassay with beans as an indicator of soil health. Phytopathology 94: S1.
- 2, and . 2000. Impact of soil health management practices on soil-borne pathogens, nematodes and root diseases of vegetable crops. Appl. Soil Ecol. 15: 37–47. https://doi.org/10.1016/S0929-1393(00)00070-6
- 3, and . 2001. Designing a soil quality assessment tool for sustainable agroecosystem management. Ecol. Appl. 11: 1573–1585. https://doi.org/10.1890/1051-0761(2001)011[1573:DASQAT]2.0.CO;2
- 4, , and . 2004. The soil management assessment framework: A quantitative soil quality evaluation method. Soil Sci. Soc. Am. J. 68: 1945–1962. https://doi.org/10.2136/sssaj2004.1945
- 5, , , , and . 2013. Developing a soil quality index to compare soil fitness for agricultural use under different managements in the Mediterranean environment. Soil Tillage Res. 130: 91–98. https://doi.org/10.1016/j.still.2013.02.013http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000318889100012&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 6, , , , , , and . 2011. Magnesium for field crops. Agronomy Fact Sheet Series, 59. International Plant Nutrition Institute and Nutrient Management Spear Program. Cornell University Cooperative Extension. Ithaca, NY.
- 7, and . 2008. The nature and properties of soils. 14th ed. Prentice Hall, Upper Saddle River, NJ.
- 8, , , , , and . 2000a. Distribution and variability of surface soil properties at a regional scale. Soil Sci. Soc. Am. J. 64: 974–982. https://doi.org/10.2136/sssaj2000.643974x
- 9, , , and . 2000b. Identification of regional soil quality factors and indicators: I. Central and Southern High Plains. Soil Sci. Soc. Am. J. 64: 2115–2124. https://doi.org/10.2136/sssaj2000.6462115x
- 10, , , and . 2000c. Identification of regional soil quality factors and indicators: II. Northern Mississippi Loess Hills and Palouse Prairie. Soil Sci. Soc. Am. J. 64: 2125–2135.
- 11, and . 2005. Soil structure and management: A review. Geoderma 124(1–2): 3–22. https://doi.org/10.1016/j.geoderma.2004.03.005
- 12, , , , , , , , , , , , , , , and . 2012. Permanganate oxidizable carbon reflects a processed soil fraction that is sensitive to management. Soil Sci. Soc. Am. J. 76: 494–504. https://doi.org/10.2136/sssaj2011.0286http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000300644400018&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 13 Cornell Cooperative Extension. 1987. Cornell field crops and soils handbook. 2nd ed. New York State College of Agriculture and Life Sciences. Ithaca, NY.
- 14 Cornell Cooperative Extension. 2002. Cornell guide for integrated field crop management. Media and Technology Services. Cornell University. Ithaca, NY.
- 15, and . 2014. Soil quality evaluation under different land management practices. Environ. Earth Sci. 72: 4531–4549. https://doi.org/10.1007/s12665-014-3353-z, http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000344737300030&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 16, and . 2000. Soil health and sustainability: Managing the biotic component of soil quality. Appl. Soil Ecol. 15: 3–11. https://doi.org/10.1016/S0929-1393(00)00067-6
- 17 2002. Diagnosing soil compaction using a penetrometer (Soil compaction tester). Agronomy Facts Sheet 63. Penn State Extension. The Pennsylvania State University. University Park, PA.
- 18, , and . 2007. Zinc. Agronomy Fact Sheet Series 32. Nutrient management spear program. Cornell University Cooperative Extension. Ithaca, NY.
- 19, and . 1992. The need for a soil quality index: Local and regional perspectives. Am. J. Alternative Agric. 7(1-2): 12–16. https://doi.org/10.1017/S0889189300004380
- 20, and . 2010. Simple and rapid laboratory method for rewetting dry soil for incubations. Commun. Soil Sci. Plant Anal. 41: 1493–1501. https://doi.org/10.1080/00103624.2010.482171, http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000279198600008&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 21, , and . 2010. Manganese. Agronomy Fact Sheet Series 49. Nutrient Management Spear Program. Cornell University Cooperative Extension. Ithaca, NY.
- 22, , , and . 1999. Adaptation of soil quality indices and application to three tillage systems in southern Illinois. Soil Tillage Res. 50(3-4): 237–249. https://doi.org/10.1016/S0167-1987(99)00012-4
- 23, , , , , , and . 2009. Use of an integrative soil health test for evaluation of soil management practices. Renew. Agric. Food Syst. 24: 214–224. https://doi.org/10.1017/S1742170509990068
- 24 2002. Principal component analysis. Springer, London, United Kingdom.
- 25, , and . 2003. Soil quality: Why and how? Geoderma 114(3-4): 145–156. https://doi.org/10.1016/S0016-7061(03)00039-9
- 26, , , , , and . 1997. Soil quality: A concept, definition, and framework for evaluation (A guest editorial). Soil Sci. Soc. Am. J. 61: 4–10. https://doi.org/10.2136/sssaj1997.03615995006100010001x
- 27, , , and . 2008. A preliminary watershed scale soil quality assessment in north central Iowa, USA. Soil Tillage Res. 99: 291–299. https://doi.org/10.1016/j.still.2008.03.002
- 28, and . 1980. Surface glycoproteins- evidence that they may function as the race specific phytoalexin elicitors of Phytophthora megasperma f. sp. glycinea. Physiol. Plant Pathol. 17: 175–192. https://doi.org/10.1016/0048-4059(80)90050-8
- 29, , and . 2001. Simplified method for soil particle-size determination to accompany soil-quality analyses. Soil Sci. Soc. Am. J. 65: 849–852. https://doi.org/10.2136/sssaj2001.653849x
- 30, and . 2009. Building soils for better crops: Sustainable soil management. Handbook Series Book 10. Sustainable Agric. Research and Extension, College Park, MD. p. 294
- 31, , , and . 1997. Agricultural intensification and ecosystem properties. Science 277: 504–509. https://doi.org/10.1126/science.277.5325.504
- 32 Minitab, Inc. 2016. Minitab statistical software. Release 17. Minitab, Inc., State College, PA.
- 33, , , , , and . 2007. Evaluation of laboratory-measured soil physical properties as indicators of soil quality. Soil Sci. 172: 895–912. https://doi.org/10.1097/ss.0b013e318154b520http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000251022000005&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 34, , , , , , , , , , , and . 2016. Comprehensive assessment of soil health– The Cornell framework manual 3rd ed. Cornell University, Geneva, NY.
- 35, and . 1996. Total carbon, organic carbon, and organic matter. In: D.L. Sparks, editor, Method of soil analysis. Part 3. SSSA Book Series 5. SSSA, Madison, WI. p. 961–1010.
- 36, , and . 1997. Miniature rain simulator for field measurement of soil infiltration. Soil Sci. Soc. Am. J. 61: 1041–1043. https://doi.org/10.2136/sssaj1997.03615995006100040008x
- 37, and . 2008. Soil water desorption and imbibition: Tension and pressure techniques. Soil sampling and methods of analysis. 2nd ed. CRC Press, Boca Raton, FL. p. 981–997.
- 38, and . 2000. Performance of soil condition indicators across taxonomic groups and land uses. Soil Sci. Soc. Am. J. 64: 300–311. https://doi.org/10.2136/sssaj2000.641300x
- 39 Soil Survey Division Staff. 1993. Examination and description of soils. Soil Survey Manual. Soil Conservation Service. United States Department of Agriculture Handbook 18. USDA-NRCS. Washington, D.C.
- 40, and . 1999. Reservations regarding the soil quality concept. Soil Sci. Soc. Am. J. 63: 1039–1054. https://doi.org/10.2136/sssaj1999.6351039x
- 41, , and . 2003. Quality soil management of soil quality management: performance versus semantics. Adv. Agron. 79: 1–68. https://doi.org/10.1016/S0065-2113(02)79001-9
- 42, , , , and . 2015. Mapping soil health over large agriculturally important areas. Soil Sci. Soc. Am. J. 79: 1420–1434. https://doi.org/10.2136/sssaj2014.09.0371, http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000365399100016&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 43, , , and . 2005. Principles and applications of soil microbiology 2nd ed. Pearson Education Inc., Upper Saddle River, NJ.
- 44 1979. A flow diagram for teaching texture by feel analysis. J. Agron. Educ. 8: 54–55.
- 45, , , , and . 2002. Agricultural sustainability and intensive production practices. Nature 418: 671–677. https://doi.org/10.1038/nature01014
- 46, , , and . 1993. Soil water desorption curves. In: M.R. Carter, editor, Soil sampling and methods of analysis. Canadian Society of Soil Science, Lewis Publishers, Boca Raton, FL.
- 47 1947. Soil reaction influence on availability of plant nutrients. Soil Sci. Soc. Am. Proc. 11: 305–308. https://doi.org/10.2136/sssaj1947.036159950011000C0057x
- 48 USDA-NRCS. 2006. Land resource regions and major land resource areas of the United States, the Caribbean, and the Pacific Basin. U.S. Department of Agriculture Handbook 296. USDA-NRCS. Washington, D.C.
- 49 2002. The bicinchonic acid (BCA) assay for protein quantitation. In: J.M. Walker, editor, The Protein Protocols Handbook. 2nd ed. Humana Press, Totowa, NJ. p. 11–14. https://doi.org/10.1385/1-59259-169-8:11
- 50, , , , and . 2003. Estimating active carbon for soil quality assessment: A simple method for laboratory and field use. Am. J. Alternative Agric. 18: 3–17. https://doi.org/10.1079/AJAA2003003
- 51, , , , , , and . 2016. Landscape-scale variability of soil health indicators: Effects of cultivation on soil organic carbon in the Usambara Mountains of Tanzania. Nutr. Cycling Agroecosyst. 105: 263–274. https://doi.org/10.1007/s10705-015-9750-1http://gateway.webofknowledge.com/gateway/Gateway.cgi?GWVersion=2&SrcApp=PARTNER_APP&SrcAuth=Agronomy_sub&KeyUT=WOS:000378548100008&DestLinkType=FullRecord&DestApp=WOS_CPL&UsrCustomerID=9992b2403adf8c36119d0b6fce39b97c
- 52, , and . 1952. An improved electrical conductivity method for accurately following changes in the respiratory quotient of a single biological sample. Plant Physiol. 27: 70–80. https://doi.org/10.1104/pp.27.1.70
- 53, and . 1970. A conductivity method for measuring microbially evolved carbon dioxide. Ecology 51: 155–156. https://doi.org/10.2307/1933610
- 54, and . 1996. Extraction of an abundant and unusual protein from soil and comparison with hyphal protein of arbuscular mycorrhizal fungi. Soil Sci. 161: 575–586. https://doi.org/10.1097/00010694-199609000-00003
- 55 1994. Carbon mineralization. In: P.S. Bottomley, J.S. Angle, and R.W. Weaver, editors, Methods of soil analysis: Part 2—Microbiological and biochemical properties. SSSA, Madison, WI, p. 835–863.