Soil Science Society of America Journal 67:1282-1286 (2003)
© 2003 Soil Science Society of America
DIVISION S-7—NOTES
Detecting nutrient pool changes in rocky forest soils
Andrew Kulmatiski*,a,
Daniel J. Vogtb,
Thomas G. Siccamac and
Karen H. Beardd
a Dep. of Biology and Ecology Center, Utah State Univ., Logan, UT 84322
b College of Forest Resources, Univ. of Washington, Box 352100, Seattle, WA 98195
c Yale School of Forestry and Environmental Studies, Greeley Lab., 310 Prospect Street, New Haven, CT 06511
d Dep. of Forest, Range, and Wildlife Sciences and Ecology Center, Utah State Univ., Logan, UT 84322
* Corresponding author (andrew{at}biology.usu.edu)
 |
ABSTRACT
|
|---|
Microsite heterogeneity often complicates the accurate measurement of soil properties. Many sampling techniques have been developed to overcome this difficulty, but use of these techniques requires the direct comparison of measurements from each technique. In this paper, we present estimates of C and N pool sizes determined from two commonly used, but previously not compared, techniques. Composite coring (core) and excavation mensuration (pit) techniques were performed in glacially stratified Inceptisols of southern New England. Estimates of total C pool size (forest floor [FF] -15 cm) from the pit and core techniques were significantly correlated (r2 = 0.61, P < 0.0001) and very similar (5.64 ± 0.32 and 5.63 ± 0.29, respectively). However, the core technique reduced variance in the sample population, allowing fewer samples to detect a 10% change in nutrient storage (21 core vs. 29 pit samples). In addition, sampling each plot with the core technique required less than one-half the sampling time of the pit technique. The pit technique, however, allowed quantitative sampling below 15 cm and direct measurement of large coarse fragments. Data from the pit technique revealed a strong exponential decline in nutrient storage with depth. This exponential decline allowed the extrapolation of C and N pools to greater depths. Carbon storage (kg m-2) was described by depth (cm) as: exp (5.53–0.04 depth); P < 0.0001. N storage (kg m-2) was described by depth (cm) and tree basal area (ba, m2 ha-1) as: exp (3.280–0.025 depth– 0.003 ba); P < 0.0001. Our data suggest that composite core sampling is more efficient than, but well supplemented by, pit sampling.
Abbreviations: ba, basal area • FF, forest floor • MDC, minimal detectable change
|
INTRODUCTION
|
|---|
FOREST MANAGERS, ecosystem ecologists, and climate modelers use estimates of soil nutrient pools as indicators of site productivity and biological activity. However, large microsite variability in forest soils often confounds sampling efforts (Huntington et al., 1988; Johnson et al., 1990; Palmer et al., 2002; Wilding et al., 2001). Wind-thrown trees, large roots, animal disturbances, and variable coarse fragment volumes all increase variability in forest soil nutrient pools. Due to the difficulty of accurately sampling extremely heterogeneous soils, soil nutrient models often lack effective ‘ground-truthing’ (Lexer et al., 1999; McGuire et al., 1995; Yost et al., 1993). Therefore, efficient means of collecting extensive baseline data are necessary (Palmer et al., 2002; Vincent and Chadwick, 1994).
The estimation of soil C or N content on an area basis requires information on the depth of sampling, rock volume content, soil bulk density, and C or N concentration, respectively (Boone et al., 1999). Soil sampling techniques commonly used to determine these parameters include the clod, core, irregular hole, pit, and sand cone techniques (Andraski, 1991; Lichter and Costello, 1994; Miller et al., 2001; Page-Dumroese et al., 1999). This wide variety of techniques reflects the variety of precision required, soil types to be sampled, and the resources available to investigators.
In general, sampling heterogeneous soils requires sampling at a scale that is large relative to the scale of variation in coarse fragments (Vincent and Chadwick, 1994; Wilding et al., 2001). Two general approaches can be used to capture soil heterogeneity within a sample: intensive and extensive. Intensive sampling is accomplished by excavating samples that are larger than the largest coarse fragments (Huntington et al., 1988). Extensive sampling is accomplished by extracting small samples from a wide area. For extensive sampling to be effective, the proportion of coarse fragments in a small (core) sample (e.g., 0.2–4 cm) must be similar to the proportion of coarse fragments in a large (pit) sample (e.g., 0.2–50 cm). These size ranges represent the coarse fragments sampled with a 4-cm core and 50-cm pit. However, this proportional assumption is not likely to be true when large coarse fragments are present and if so the pit technique would then be expected to provide better estimates of nutrient storage. Although, the accuracy gained with the pit technique may not outweigh the loss in sample sizes that result from an extensive sampling effort (Conkling et al., 2002).
The objective of this study was to compare the ability of two commonly used sampling techniques to detect a 10% change in total soil C and N pools in the soils of southern New England. To accomplish this, we compared estimates of nutrient pool size, the minimum detectable change (MDC) in these nutrient pools, the sampling effort required to detect that change, and the type (in regards to depth) of data produced from intensive and extensive soil sampling techniques. To our knowledge there has been no published comparison of soil C and N storage using these two techniques.
|
Materials and Methods
|
|---|
Site Description
The research was conducted in the Yale-Myers Forest (YMF) in northeastern Connecticut, USA (42° N lat., 72° W long.). This 3173-ha forest has been managed by the Yale School of Forestry and Environmental Studies since the 1930s. Parent material is mostly glaciofluvial and glacial till derived from schist and gneiss. The till common to the area is a poorly sorted mixture of silt, sand, and minor clay (<18%) with pebbles, cobbles, and boulders that underlay a thin layer of loess (Peper and Pease, 1975). Samples were taken from three soil catenas: Hollis/Charlton, Woodbridge/Paxton, and Hinckley. The Charlton, Paxton, and Woodbridge soils developed on glacial till. The Hinckley soils developed on glaciofluvial deposits. The Charlton soils are coarse-loamy, mixed, active, mesic Typic Dystrudepts. Paxton soils are coarse-loamy, mixed, active, mesic Oxyaquic Dystrudepts. Woodbridge soils are coarse-loamy, mixed, active, mesic Aquic Dystrudepts. The Hinckley soils are sandy-skeletal, mixed, mesic Typic Udorthents. The Hinckley soils are in the driest drainage class, excessively drained; followed by Charlton and Paxton, well drained; then by Woodbridge, moderately well drained. Soil survey maps indicate these to be the most common soils of upland forests in Connecticut, Rhode Island, and Massachusetts.
Field Sampling
Sampling was stratified by forest cover type and soil series and was designed to represent the natural distribution of forest cover types and soil series found on the landscape. The number of samples extracted from each treatment combination reflected the abundance of that treatment combination on the landscape as determined from forest inventories and soil survey maps. For example, 40% of the forest was covered in hemlock (Tsuga canadensis Carr.) and mixed hemlock stands so 40% of the soil samples were obtained from hemlock and mixed hemlock stands.
Forest cover type and soil series maps were overlayed to identify treatment combinations of sufficient size to sample (0.5 ha). Forest cover types were defined with the following percentages of ba by tree species: hardwoods: <25% hemlock, <50% oak (Quercus rubra L. and Q. alba L.), and <25% pine (Pinus strobus L. and P. resinosa Ait.); hemlock/hardwood: 25–60% hemlock, and <50% oak; oak: >50% oak; pine: ≥75% pine; pine/other: 25–75% pine; and hemlock: ≥60% hemlock. Sites with the largest area for a specific forest cover type/soil series combination were visited first. All subsequent plots were separated by at least 50 m, and were located throughout the study site. Forest cover type and soil series were verified in these plots using tree species ba and soil test pits, respectively. Sampling effort was divided evenly between the two soil sampling techniques. Soils were sampled during the summers of 1997 and 1998.
Core Technique
Fifty-six circular 0.1-ha plots were sampled at four depth increments (forest floor, 0–3, 3–6, and 6–15 cm) using the core technique (n = 56). Five subsamples were taken from the center and cardinal points of each plot with the four cardinal point subsamples being 17.8 m from the center subsampling point. Forest floor samples were removed after sawing around a 15 by 15 cm square template. The depth to mineral soil was measured on each of the four sides of the pedestal of forest floor material. The A horizon was distinguished from the Oa using a textural analysis done by hand to detect soil with >60% mineral content (Huntington et al., 1988). Woody material in forest floor samples, with a diameter >1 cm was discarded and not considered in analyses. After the forest floor was removed, a galvanized steel tube (4.1-cm i.d.) was plunged 6 cm below the mineral soil surface and soil was extracted. The 6-cm soil cores were measured from the bottom up, and cut at 3 cm, because compaction was assumed potentially large in the top 3 cm and relatively small in the bottom 3 cm. Soil samples of the 6- to 15-cm depth were taken directly below the location where the 0- to 6-cm samples were removed. Where coarse fragments prevented sampling directly below the 0- to 6-cm sample, samples were taken from the nearest undisturbed soils and within 0.5 m of the original sampling point. Subsamples from each depth strata in each plot were composited into a single bag and returned to the laboratory. Samples were refrigerated at 5 to 10°C before analysis.
Forest floor and mineral soil samples were dried to a constant weight at 70 and 105°C, respectively. Forest floor samples were ground in a Wiley or Cyclotech mill and analyzed for total C and N concentration via dry combustion in a CHN elemental analyzer (LECO CNH-600, Leco St. Joseph, MI). Carbon and N concentrations were determined from the average of three 200-mg subsamples. Inorganic carbonate-C was assumed absent because of the silaceous parent material and extreme acidity (pH 3–4). Coarse roots and rocks (>2 mm) were removed from mineral soil samples and weighed. Bulk density of the fine fraction was determined as the mass of the fine fraction (<2 mm) divided by the volume of fine fraction in the core. Fine fraction volume was calculated as sample volume minus coarse rock volume. Local rock density was determined to be 2.64 g cm-3.
Quantitative Pit Technique
The pit sampling technique was performed as described in Huntington et al. (1988) with some modifications. A 50 by 50 cm frame, marked with a 12.5 by 12.5 cm reference grid, was secured to the ground at a randomly selected point near the center of each 0.1-ha plot (n = 18). The pit was excavated in the following depth increments: forest floor, 0 to 3, 3 to 6, 6 to 15, 15 to 30, 30 to 45, and 45 to 60 cm. Soils were removed, sieved through a 2-mm sieve, and weighed. Coarse rocks and roots that did not pass through the 2-mm sieve were weighed separately. A subsample of 5 large (>2 cm) roots were returned to the lab, dried, ground, and analyzed for percentages of C and N. The average percentage of C (50%) and percentage of N (0.6%) from these roots were used to calculate root C and N storage for all pit samples. Bulk density of the fine fraction was calculated as the mass of the fine fraction (<2 mm) divided by the total volume of the fine fraction in the strata sampled (as in Huntington et al., 1988). Subsamples from each layer were removed from the sieved soils, returned to the lab, and prepared and analyzed in the same manner as the core samples.
Statistical Analyses
All statistical analyses were conducted using SAS for Windows v. 8 (SAS Institute, Cary, NC). With the exception of regression analyses for C and N storage by depth, data by soil depths were treated as independent within each plot because of the exponential decline in C and N storage with depth. Regression formulae reported represent the unbiased estimate of the median value of Y (soil C and N) given X (depth and ba). Assumptions of normality were met with log transformations when necessary. The prespecified level of significance for all analyses was set at 0.05.
The MDC and the sample size required to detect a 10% change in mean values (10% 
n) was calculated using the standard errors (SE) of the means with the following equations: MDC = (t)(SE), and (10% n) = {[(t)(CV)]/A}2, where t = the t-score of a two-tailed test with a sample size of n, CV = the coefficient of variation, and A = the allowable error, expressed as a percentage of the mean (Wilding et al., 2001). This assumes repeated sampling with the same technique and a normal distribution of real values (Avery and Burkhart, 1994; Huntington et al., 1988).
|
Results
|
|---|
Core vs. Pit Estimates of C, N, Bulk Density, and Percentage of Coarse Fragments
All ‘pit vs. core’ comparisons are made using data from 18 plots sampled using both techniques. Data from an additional 38 plots sampled with the core technique are reported to demonstrate sample size effects on the population estimates (n = 56). Estimates of C and N storage from the core samples were larger (7% for C and N) than estimates from the pit technique (Table 2), but not significantly (P = 0.2, 0.2, respectively). A regression between estimates of C and N from each depth (without coarse roots) from the two techniques revealed r2 values of 0.61 and 0.42, respectively (P < 0.001). Estimates of C storage from the two techniques were more similar when coarse root contributions were added to C stored in the mineral soil (hereafter referred to as total soil C or N). The pit technique estimated total C storage at 5.64 ± 0.32 kg m-2 and the core technique estimated total C storage at 5.63 ± 0.29 kg m-2. Coarse roots did not comprise a significant portion of the N pool as others have also shown (Ruark and Zarnoch, 1992). In addition to total C and N storage, the percentage of C and N were not significantly different between the two techniques (P = 0.2, 0.4, respectively) (Table 1). The MDC was also similar in both techniques (Table 2) and was largest in the forest floor pool for both C and N (Fig. 1). Despite these similarities the two techniques produced varied estimates of bulk density, percentage of rock volume, and root C and N storage (Table 1). In particular, the bulk density of the fine fraction (<2 mm) and the percentage of coarse fragment (in mineral soil) were 19 and 87% smaller with the core technique than with the pit technique.
View this table:
[in this window]
[in a new window]
|
Table 1. Estimates of the percentage of C, percentage of N, bulk density of the fine fraction (<2 mm), percentage of coarse rock volume in sample (>2 mm), and root C and N storage for both the pit and core technique (0–15 cm). Results shown for an equal number of samples for each technique (n = 18) as well as for similar sampling efforts (core, n = 56; pit, n = 18). Significance indicated for differences between paired samples (n = 18).
|
|
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 1. Carbon and N storage estimates, standardized to 1 cm, for samples taken in the forest floor, 0- to 3-, 3- to 6-, 6- to 15-, 15- to 30-, 30- to 45-, and 45- to 60-cm soil depth increments. Estimates do not include C and N stored in coarse (>2 mm) root material. Mean values are reported ±1 standard deviation for the 18 plots sampled. Note that storage and the variation in storage declines with depth. Equations describing the observed decline in storage with soil depth are reported in the results.
|
|
Pit Estimates
Carbon and N pool sizes (g cm-1 m-2) in the mineral soil decreased exponentially with depth (Fig. 1). Depth explained 73% of the variation in C pool sizes (exp[5.53–0.04 depth]; P < 0.0001). Depth and ba explained 54% of the variation in N pool sizes (exp[3.280–0.025 depth–0.003 ba]; P < 0.0001). The variation in C and N storage similarly decreased with depth: 58% of the variation in C and N storage occurred between the forest floor and 15 cm below the forest floor. Moving deeper into the soil profile, the C pool size (kg m-2), from the forest floor to 60 cm, averaged 7.91 ± 0.41 kg m-2 (range 4.0–11.3 kg m-2). Nitrogen storage from the forest floor to 60 cm averaged 0.56 ± 0.03 kg m-2 (range 0.31–0.82 kg m-2). Coarse rock fragment volumes for a whole pit (0–60 cm) averaged 32 ± 3.7% (range 10–54%).
Sampling Effort
Recording the amount of time spent at a plot standardized sampling efforts. Therefore, sampling times do not include site selection or travel time to and from plots. Sampling efforts were divided roughly evenly between the two techniques with 63 h spent sampling with the pit technique (n = 18) and 84 h spent sampling with the core technique (n = 56). Therefore, the core sampling procedure required 1.5 person-hours per plot to sample to 15 cm. The pit sampling procedure required 3.5 person-hours per plot to sample to 15 cm. Sample preparation and analysis for either technique required 2.5 h per plot. An additional 4.5 h in the field were required to sample to 60 cm using the pit technique.
|
Discussion
|
|---|
The relatively small amount of variation, and the consistent exponential decline in nutrient storage below 15 cm, indicates that this deeper nutrient storage pool is relatively insensitive to site conditions such as forest cover type, soil series, and topography (Fig. 1). Although these deeper nutrient storage pools did not vary widely across the landscape they did represent a large nutrient pool. Carbon and N estimates from the 15- to 60-cm strata, typically doubled storage estimates from 0 to 15 cm. Because nutrient storage is well described by an exponential decline, data from surficial (core) samples can be accurately extrapolated to depths of at least 60 cm. While the quantitative pit technique was used to sample only to 60 cm, one qualitative pit was dug to 1.7 m. Data from this pit provided anecdotal support for the use of regression equations of nutrient storage even below 60 cm (A. Kulmatiski, unpublished data, 2000).
Composite core and quantitative pit sampling techniques utilize extensive and intensive approaches, respectively, to capture variation in soil nutrient storage. The pit technique can be assumed to be more accurate because of a large (0.15 m3) sample volume and the direct measurement of large coarse fragments, but this accuracy comes at a cost (Vincent and Chadwick, 1994). For example, the sampling effort required for the detection of a 10% change in total C and N with the pit technique would be 102 and 172 h, respectively (Table 2). Whereas, the sampling effort required for the detection of a 10% change in total C and N with the core technique would only be 32 and 51 h, respectively. Therefore, core sampling to detect a 10% change in nutrient storage would require approximately one-third the field sampling effort required by the pit technique. Core sampling focused efforts on surficial depths (0–15 cm), where bulk density, nutrient storage, and nutrient concentrations varied most widely (Fig. 1, Table 1).
Carbon and N estimates are determined as the product of the concentration of the element, the mass of soil in the sample, and the volume of that sample. The core technique produced relatively low estimates of percentages of C and N, and bulk density, but relatively large estimates of soil volume (as a result of relatively small estimates of coarse rock volume), when compared with the pit technique. As a result, both techniques produced nearly identical estimates of total nutrient storage. While the core technique was anticipated to underestimate coarse rock volume, it was not anticipated to underestimate percentages of C and N or bulk density values. The core technique is restricted to sampling soils without large stones (<4cm). Results from the core technique indicate that these relatively stone-free microsites demonstrate lower bulk density and percentage of C and N, at least in the top 15 cm. Assuming equivalent organic input between rocky and less rocky soil microsites, it is reasonable to believe that rocky soils must concentrate organic materials in smaller soil volumes. This process could explain the observed differences in percentage of C and N between the core and pit technique estimates.
We believe that compositing core samples reduced sample variance allowing the detection of slightly smaller changes in soil status with less sampling effort, than the pit technique (Ruark and Zarnoch, 1992). Our data suggest that sampling efforts should emphasize the core technique and that data from these surficial samples can be extrapolated to greater depths using regression equations determined from a small number of local pit samples. Additional sampling in various regions will be necessary to determine the universality of the observed exponential decline in nutrient storage with depth.
|
ACKNOWLEDGMENTS
|
|---|
Funding was provided through the USDA NRCS, Grant No. 68-1106-6-02 and the Carpenter/Sperry Fund. J. Kimble (NSSC, USDA, NRCS) provided useful support and comments. A. Finkral and S. Childs provided guidance and information in the field. J. Barcroft, and I. Kelman provided field and laboratory assistance. E. Miller and two anonymous reviewers provided useful comments on earlier versions of the manuscript.
|
NOTES
|
|---|
Contribution of the Yale School of Forestry and Environmental Studies.
Received for publication March 12, 2003.
|
REFERENCES
|
|---|
- Andraski, B.J. 1991. Balloon and core sampling for determining bulk-density of alluvial desert soil. Soil Sci. Soc. Am. J. 55:1188–1190.[Abstract/Free Full Text]
- Avery, E.T., and H.E. Burkhart. 1994. Forest Measurements. 4th ed. McGraw-Hill Inc., New York.
- Boone, R.D., D.F. Grigal, P. Sollins, R.J. Ahrens, and D.E Armstrong. 1999. Soil sampling, preparation, archiving, and quality control. p. 3–28. In G. P. Robertson et al. (ed.) Standard soil methods for Long-term Ecological Research. Oxford University Press, New York.
- Conkling, B.L., C.M. Hoover, W.D. Smith, and C.J. Palmer. 2002. Using forest health monitoring data to integrate above and below ground carbon information. Environ. Pollut. 116:S221–S232.
- Huntington, T.G., D.F. Ryan, and S.P. Hamburg. 1988. Estimating soil-nitrogen and carbon pools in a northern hardwood forest ecosystem. Soil Sci. Soc. Am. J. 52:1162–1167.[Abstract/Free Full Text]
- Johnson, C.E., A.H. Johnson, and T.G. Huntington. 1990. Sample-size requirements for the determination of changes in soil nutrient pools. Soil Sci. 150:637–644.
- Lexer, M.J., K. Honninger, and M. Englisch. 1999. Estimating chemical soil parameters for sample plots of the Austrian Forest Inventory. Forstwiss. Centralbl. 118:212–227.
- Lichter, J.M., and L.R. Costello. 1994. An evaluation of volume excavation and core sampling techniques for measuring soil bulk density. J. Arboric. 20:160–164.
- McGuire, A.D., J.M. Melillo, D.W. Kicklighter, and L.A. Joyce. 1995. Equilibrium responses of soil carbon to climate change: Empirical and process-based estimates. J. Biogeography 22:785–796.
- Miller, R.E., J. Hazard, and S. Howes. 2001. Precision, accuracy, and efficiency of four tools for measuring soil bulk density strength—Introduction. p. 1–16. In USDA Forest Service Pacific Northwest Research Station Research Paper, USDA Forest Service Pacific Northwest Research Station, Portland, OR.
- Page-Dumroese, D.S., M.F. Jurgensen, R.E. Brown, and G.D. Mroz. 1999. Comparison of methods for determining bulk densities of rocky forest soils. Soil Sci. Soc. Am. J. 63:379–383.[Abstract/Free Full Text]
- Palmer, C.J., W.D. Smith, and B.L. Conkling. 2002. Development of a protocol for monitoring status and trends in forest soil carbon at a national level. Environ.Pollut. 116:S209–S219.
- Peper, J.D., and M.H. Pease. 1975. Geological map of the Westford quadrangle, GQ 1214. United States Geological Service, Reston, VA.
- Ruark, G.A., and S.J. Zarnoch. 1992. Soil carbon, nitrogen, and fine root biomass sampling in a pine stand. Soil Sci. Soc. Am. J. 56:1945–1950.[Abstract/Free Full Text]
- Vincent, K.R., and O.A. Chadwick. 1994. Synthesizing bulk-density for soils with abundant rock fragments. Soil Sci. Soc. Am. J. 58:455–464.[Abstract/Free Full Text]
- Wilding, L.P., L.R. Drees, and L.C. Nordt. 2001. Spatial variability: enhancing the mean estimate of organic and inorganic carbon in a sampling unit, p. 69–86. In R. Lal et al., (ed.) Assessment Methods for Soil Carbon. CRC Press, Boca Raton, FL.
- Yost, R., K. Loague, and R. Green. 1993. Reducing variance in soil organic-carbon estimates—soil classification and geostatistical approaches. Geoderma 57:247–262.
This article has been cited by other articles:
|
|
|
|
 
A. Kulmatiski, D. J. Vogt, T. G. Siccama, J. P. Tilley, K. Kolesinskas, T. W. Wickwire, and B. C. Larson
Landscape Determinants of Soil Carbon and Nitrogen Storage in Southern New England
Soil Sci. Soc. Am. J.,
November 1, 2004;
68(6):
2014 - 2022.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
