INFORMATION TO USERS The most advanced technology has been used to photo­ graph and reproduce this manuscript from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are re­ produced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. These are also available as one exposure on a standard 35mm slide or as a 17" x 23" black and white photographic print for an additional charge. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI U niversity Microfilms International A Bell & Howell Inform ation C o m p a n y 3 0 0 N orth Z e e b R o ad , A nn Arbor, Ml 4 8 1 0 6 -1 3 4 6 USA 3 1 3 /7 6 1 -4 7 0 0 8 0 0 /5 2 1 -0 6 0 0 O rder N u m b er 8912613 Soil nutrients in glaciated M ichigan landscapes: D istribution of nutrients and relationships w ith stand produ ctivity Merkel, Dennis Michael, Ph.D. Michigan State University, 1988 C opyright © 1988 by M erkel, D en n is M ichael. A ll rights reserved. UMI 300 N. Zeeb Rd. Ann Arbor, MI 48106 Soil Nutrients in Glaciated Michigan Landscapes: Distribution of Nutrients and Relationships With Productivity. BY Dennis Michael Merkel A DISSERTATION ^ 4 ^ ^ «■■■%*9 uuwmxwwQU a uv Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Forestry 1988 stand ABSTRACT SOIL NUTRIENTS DISTRIBUTION OF IN GLACIATED NUTRIENTS AND MICHIGAN LANDSCAPES: RELATIONSHIPS WITH STAND forest land PRODUCTIVITY by Dennis Michael Merkel Historically management the has evaluation not included beneath the surface 15-150 cm. soils, of soil soils for features or properties On sandy, glacially derived subsolum soil features have been found to influence site quality. The sandy soils of the Manistee National Forest are known to have subsolum textural strata with finer textures than the soil immediately above the strata. The purpose of this research was to examine the chemical properties of these soils and determine whether the presence and fertility of subsolum textural strata were associated with site quality of the stands above them. A stratified random sample of twenty nine stands was selected and soils were sampled to a depth of 450 centimeters. Total Kjeldahl nitrogen, phosphorus, and 1 N boiling nitric acid extractable potassium, calcium were determined. magnesium and Nutrient concentrations (mg kg'1) were converted into nutrient contents (kg ha'1) as a measure of site fertility. Differences in site fertility were determined with a one way analysis of variance (AOV). The AOV results were confirmed with a non-parametric Kruskal-Wallis analysis. Associations of site fertility with mean annual volume increment (MAI) were evaluated by multiple regression analysis and by principal components analysis. The evaluation of site fertility using subsolum textural strata (bands) found that soil series phases had nutrient contents and variabilities which differed noticeably between banded and unbanded phases of the same soil series. MAI was successfully estimated using site fertility variables incorporating either surface (forest floor to 45 cm) or subsolum data (forest floor to 450 cm). Adjusted R2 values were 0.86 for the subsolum data and 0.74 for the surface data, the standard errors of the estimate were 0.357 and 0.481 m3 ha'1 yr'1, respectively. Soil fertility variables from the forest floor to 200 cm were found to have significant associations with MAI. subsolum features, especially on soils with Examination of sandy surface textures, can lead to collection of information important in accurate determination of site quality and these should not be ignored in forest land evaluations. features Copyright by DENNIS MICHAEL MERKEL 1988 DEDICATION To my wife Cindy; and To the lads, Woody and Spinner v ACKNOWLEDGEMENTS The research reported in this dissertation was funded in part by the EPA, USDA Forest Agricultural Experiment Station. Service, and Michigan Without their support these investigations would not have been completed. Special thanks are extended to Mr. David T. Cleland, Soil Scientist for the Huron-Manistee N.F., whose tireless efforts on behalf of the ECS project allowed this research to be undertaken. I would like to recognize the work of the ECS crew who collected forest inventory, soils, and vegetative data during the summer of 1983. They were, Dr. George Host, Mr. Steve Westin, and Dr. Donald Zak. A great deal of credit for this work is due to the many individuals (too many to list separately) who assisted in sample preparation, laboratory analysis, and data entry. Sincere thanks also go to Drs. J.B. Hart Jr. and C.W. Ramm, whose thoughtful review comments were a welcome source of guidance during the preparation of this dissertation. Particular appreciation is offered to Dr. Phu V. Nguyen who often put aside his own work to predicaments in laboratory or data analysis. rescue me from His assistance was always rendered with a contagious enthusiasm which I will strive to maintain. vi TABLE OF CONTENTS List of T a b l e s .............................................. x List of F i g u r e s ............................................ xii Chapter 1 Introduction ............................................ 1 Chapter 2 Municipal Sludge Fertilization on Oak Forests In Michigan:Estimations of Long-Term Growth Responses ......................................... 7 Introduction ............................................ 8 O b j e c t i v e s .............................................. 10 Materials and Methods ................................... 10 Sludge Study A r e a ....................................... 12 Regional Study Areas 13 ................................... Sample Collection on the Sludge Study Area Sample Collection on Regional Stands ........... 15 ................. 15 Chemical Analysis ....................................... 16 Statistical Analysis ................................... 17 ................................. 19 Results and Discussion Acknowledgements ....................................... vii 27 Literature Cited ...................................... 28 Chapter 3 Distribution and Variability of Nutrients and Organic Carbon Across Glaciated Michigan L a n d s c a p e s ......................................... 30 Introduction ............................................ 31 O b j e c t i v e s .............................................. 35 Materials and Methods ................................... 36 Study A r e a .............................................. 36 Sample Collection ....................................... 39 Laboratory Analysis ..................................... 41 Statistical Analysis ................................... 43 ................................. 44 Characteristics of Sampled Stands ...................... 44 Nutrient and Organic Carbon Distributions ............. 46 Nutrient and Organic Carbon Variability ............... 50 Nutrient and Organic Carbon Distribution by Soil S e r i e s .............................................. 52 Nutrient and Organic Carbon Variability by Soil S e r i e s .............................................. 55 Nutrient and Organic Carbon Distributions by Phased Soil S e r i e s .............................................. 57 Nutrient and Organic Carbon Variability by Phased Soil S e r i e s .............................................. 61 C o n c l u s i o n s .............................................. 68 Acknowledgements ....................................... 72 Literature Cited ....................................... 73 Results and Discussion viii Chapter 4 Relationships of Soil Fertility and Stand Growth in Glaciated Michigan Landscapes ................. 77 Introduction ............................................ 78 O b j e c t i v e s .............................................. 82 Materials and Methods ................................... 83 Study A r e a .............................................. 83 Sample Collection ....................................... 85 Stand Growth Measurements ............................... 88 Laboratory Analysis ..................................... 88 Statistical Analysis ................................... 91 ................................. 94 Exploratory Data Analysis ............................... 94 Soil fertility relationships with stand growth: Soil sampling to 450 c m .......................... 94 Results and Discussion Soil fertility relationships with stand growth: Soil sampling to 45 c m ............................... 104 C o n c l u s i o n s .................................................113 Acknowledgements ......................................... 116 Literature Cited ......................................... 117 Chapter 5 C o n c l u s i o n s .................................................121 ix LIST OF TABLES Chapter 2. 2.1 2.2 Means of TKN and TKP contents for forest floor and soil components two years after treatment on the sludge fertilization study . . . . 20 Predicted mean annual increment over two years for sludge and control t r e a t m e n t s ..............21 Chapter 3. 3.1 3.2 3.3 3.4 3.5 3.6 Pre and post sampling stratification for the 30 sample s t a n d s .............................. 38 Stand volume grouped by series and series phase with one way AOV r e s u l t s .................... 45 Mean nutrient and organic carbon contents and coefficients of variability for all stands . . 47 Mean nutrient and organic carbon contents, coefficients of variability, and results of CHc VSy AOV iCi soil acixeS CjjTCupS • • • • • • • • 53 Mean nutrient and organic carbon contents, coefficients of variability, and results of one way AOV for soil series groups with p h a s e s .............................................. 58 Banding intensity designations ................... 65 Chapter 4. 4.1 Reduction of forest floor and pit fertility variable size for regression analysis ............. x 96 4.2 4.3 4.4 Regression summaries for forest floor and pit nutrient contents accumualted by 50 cm depths and descriptive statistics ................. 97 Correlations between MAI and forest floor total, or soil acid extractable K contents in soils grouped by presence or absence of subsolum textural strata .......................... 99 Summaries for forest floor and pit principal component regression analysis ...................... 101 4.5 Eigen values and vectors for PC's with significant regression coefficients: Forest floor and pit s a m p l e s ........................ 102 4.6 Regression summaries for forest floor and surface soil nutrient contents and descriptive statistics .............................. 107 4.7 Summaries for forest floor and surface soil principal component regression analysis ............ 109 4.8 Eigen values and vectors for PC's with significant regression coefficients: Forest floor and surface soil s a m p l e s ...................... Ill xi LIST OF FIGURES Chapter 2. Location of Sample stands ............... 14 Hypothetical MAI curve: Changes over time with three different nutrient retention rates for sludge nutrients > 2 yrs. old; a. 100%, b. initial rate, c. 75%. . . . 24 Chapter 3. Location of sample stands ........................ 37 Detail of forest floor and soil sampling; a. forest floor, b. soil pit, and c. bucket auger ................................... 40 TKN contents accumulated by 50 cm depths for all stands ..................................... 48 Forest floor total and soil acid extractable Mg contents accumulated by 50 cm depths for -.11 a n ..................... duanub 48 Percentages of K contents accumulated by forest floor and 50 cm depths to 450 cm ......... 49 Variability of TKP contents for all stands . . . . 50 Variability of forest floor total, and soil acid extractable Ca contents for all stands ......................................... 51 Forest floor total, and soil acid extractable K contents grouped by soil series ............... 54 Variability of TKP contents grouped by soil series ....................................... 56 xii 9a Forest floor total, and soil acid extractable K contents grouped by unbanded soil series phases ................................... 9b Forest floor total, and soil acid extractable K contents grouped by banded and unbanded phases of the Kalkaska soil series ........... 10a Variability of forest floor total and soil acid extractable K contents grouped by unbanded soil series phases ............................ 10b Variability of forest floor total and soil acid extractable K contents grouped by banded and unbanded phases of the Kalkaska soil series .......................... 11 Pit descriptions and acid extractable K contents for three stands across a range of banding intensities ............................ Chapter 4. 1 Location of sample stands ................. 2 Detail of forest floor and soil sampling; a. forest floor and surface soils, b. soil pit, and c. bucket auger ................. 3 Plot of forest floor and pit principal components having strongest association with M A I ................................... . \ Plot cf forest fleer and surface soil principal components having strongest association with M A I ...................... . xiii Chapter 1. Introduction My interest in forest soil fertility was renewed when the opportunity to participate in the Ecological Classification System Project on the Huron-Manistee National Forest presented itself. Upon a review of the literature it was apparent that almost all forest soil studies characterisatics of the upper soil. term inhabitant of a site, considerable soil volume, and were concerned with Since a tree is a long­ its roots can exploit a it did not seem unreasonable that features beneath the surface meter (subsolum features) had the potential to influence forest growth. The paucity of information on this topic left many questions to be answered and few hints on how to proceed. My own interests directed the course of this study towards examination of soil nutrients associated with subsolum features and the determination of their impact on forest productivity. These studies were undertaken to determine the extent that soil nutrients affect site quality and whether nutrients supplied by subsolum soil features (including strata of fine textured fluvial material) growth. thoroughly Samples were examine have any association with forest collected soil volumes to a depth exploited of by 450 tree cm to roots. Nutrients were expressed as contents (kg ha’1) rather than concentrations (mg Kg’1) to give a better perspective of the site nutrient resource. This dissertation consists of three main chapters which examined different aspects of site fertility. 2 The chapter 3 format was chosen publication. to facilitate their submission for The chapters are presented in the chronological order of their development. The conceptual whether regional basis site for Chapter fertility 2 studies was to could determine be used to detect changes in forest growth resulting from increases in nutrient contents due concurrent sludge opportunity to to municipal fertilization examine whether sludge project regressions additions. provided of A the growth and nutrients developed on a regional sample set would predict changes in forest long-term productivity due to fertilization by municipal sewage sludge additions. It was assumed that there was an association between nutrients and forest growth and that nutrient changes in the period since sludge application (2 years) would be confined to the forest floor and surface soils. Similar sampling and chemical analysis between the two studies allowed utilization of forest contents. relating floor These site and surface were used fertility and soil nitrogen to develop growth. an and phosphorus empirical Regressions of model mean annual volume increment and nitrogen and phosphorus contents in the forest floor and surface soils were used to predict changes in forest growth. Chapter 2 was presented at the Forest Land Applications Symposium held at the University of Washington in Seattle on June 25-28, 1985. It was co-authored by Drs. J.B. Hart Jr., 4 P.V. Nguyen, and C.W. Ramm and was published as Chapter 27 in the proceedings Washington of the Press. symposium It is by the reproduced University here with of their permission. Chapter 3 distribution is and a guantitative variability phosphorus, organic carbon, (potassium, magnesium and of and examination total acid calcium) of nitrogen, extractable contents. the total cations These accumulated by 50 cm increments to a depth of 450 cm. were Chapter 3 also investigates the influence of subsolum textural strata on soil nutrient contents. Soils are not uniform throughout their depth, especially soils of glacial and alluvial origin. exist which have distinctly surrounding soil matrix. soils can interface. cause Horizontal strata can finer textures than the The abrupt textural changes in such percolating water to accumulate at the It is not unreasonable to assume that nutrients being carried by the percolating water will also accumulate. Finer textured strata have increased water holding and cation exchange capacity, and should provide a greater nutrient and water resource for the stand growing above. The presence of textural strata on soils in the Huron-Manistee National Forest and a scarcity distributions of information on deep soil nutrient (on soils with and without subsolum textural strata) led to the investigation reported in Chapter 3. Use of the soil series (a taxonomic unit based on upper 5 soil features) has been forest applications. criticized The as a mapping sampling depth unit employed for in this study allows an examination of whether inclusion of a phase relating to the presence or absence of textural strata in the soil series taxonomic unit results in more homogenous mapping units with respect to nutrient differences between banded and contents. unbanded In phases addition, of a soil series will be tested for significance. Chapter 4 examines associations of site fertility with growth of upland oak and northern hardwood stands by multiple regression Conflicting methods, reports and of principal the utility components of soil analysis. nutrients as predictors of forest growth and yield motivated this study. It was hypothesized that the inclusion of subsolum nutrient data would lead to stronger relationships with stand growth and yield than examination of the surface soils alone. Stand growth was measured by mean annual volume increment (MAI). The associations of soil nutrient contents accumulated by forest floor and the surface 45 cm of the soil were compared to those obtained with the 50 cm accumulations. The chapters research is presented unique in collectively several aspects. in the First, following the soil sampling depth of 450 cm is much deeper than is routinely reported in the literature. This allowed a more comprehensive evaluation of site fertility for the entire volume of tree root exploitation and examination of subsolum nutrient distribution and variability. Second, these studies utilized a boiling nitric acid extraction to obtain a better representation of the site K, Mg, and Ca nutrient resources. soils has procedures. traditionally been Nutrient analysis of forest performed using agronomic Forest cover occupies a site for time spans much longer than agronomic crops and potentially interacts with a more diverse soil biota. These factors can lead to release and utilization of nutrients which are not extracted by the mild agronomic extractants. Third, this study used quantitative nutrient and organic carbon data of soils to evaluate whether the incorporation of landscape features as a phase of the soil series resulted in a more homogeneous mapping unit than the soil series alone. Finally, multivariate analyses of stand growth and yield and site nutrient variables were performed to determine which soil nutrients and what depths had significant associations with site productivity. Chapter 2. Municipal Sludge Fertilization on Oak Forests In Michigan: Estimations of Long-Term Growth Responses. INTRODUCTION Forest growth in regions of similar climate, soils, and stand histories supply are of nutrients. dependent on an adequate and balanced Nutrient contents (kg ha'1) to a given depth represent a site's resource based not only on nutrient concentrations, density. but Forest nutrients between nutrients also soil or horizon and bulk ecosystems are understood to actively cycle biotic and abiotic components. are taken up from the forest forest vegetation and utilized for growth. limiting growth, depths floor Available and soil by If a nutrient is then increases in forest floor and surface soils should relate to increases in overall stand growth. Correlations of forest growth with site nutrient contents have received little attention in the literature, yielded significant associations when applied. but have Site index of mature ponderosa pine trees in California showed a positive correlation with N content in the top 1.22 m (4 ft) of soil (Zinke 1960), and total height of 30 year old red pine trees in New York had a significant correlation with extractable potassium (K) contents in the top 152 cm (5 ft) of soil (White and Leaf 1964). 8 9 Trees growing on sandy outwash sands in northern lower Michigan rely on mineralization of soil organic matter for N requirements. In northern New increase soil soils York it nutrient of was similar found origin that reserves, and texture organic site amendments productivity, therefore site quality (Heiberg and Leaf 1961). in and The addition of organic matter in the form of municipal sewage sludge to sandy soils has been shown to significantly increase N and P levels in the forest floor of red and white pine stands in northern lower Michigan (Brockway 1983) and can be expected to increase water holding capacity of surface soil horizons. Few sludge studies have been concerned with long term growth changes expected that in mature forest stands. sludge additions will Intuitively it is result forest growth as do fertilizer applications. of basal analysis area at and 2 years radial DBH (Koterba growth et al. in increased However, studies from 1979) detailed and stem 14 months (Brockway 1983) after sludge applications found no significant increases in growth of mixed northern hardwoods or red pine, respectively. An inherent difficulty is the poor resolution obtained when measuring forest growth differences over a short time frame. until tree Growth lag periods, response, such from time of fertilization as the two to five year period reported for fertilized red pine in northern New York (Leaf et al. 1970) may further obscure short-term examinations. Growth effects from sludge applications have been 10 measured for time periods as short as two years, while effects of sludge application on forested systems may take longer to completely assess. One method of evaluation is with establishment of long-term sludge fertilization study sites. Such monitoring efforts have only been recently installed at the University of Washington in 1974, and at Michigan State University in 1981. Results from long-term measurements will not be available for some time. Until these studies reach fruition other methods of assessing long-term growth effects are needed. OBJECTIVES To nutrient exists, determine resources then to if and a correlation stand identify exists growth. nutrient If between a site relationship components with the strongest associations with stand growth. To use site nutrient resources to estimate changes in forest stand growth with sludge application. MATERIALS AND METHODS The forest stand was chosen as the unit for measurement of nutrient contents and stand growth. Mean annual volume 11 increment (MAI) was used as a measure of long-term growth and may be thought of as average growth over the age of a stand which integrates M A I 1s of stands site factors and climate. averaging 72 years In this old were used. study Stand growth is a complex response to many soil and environmental factors. No single site characteristic or measurement growth can completely quantify site quality. of We are using MAI as a growth measurement realizing the inherent limitations in its use arising from variations in stand age and species composition. Because of small nutrient fluctuations in soil at depths below 15 cm (6 in) northern Michigan (Hart et al. in sludged stands on similar (Brockway 1983), and on this 1984) , only forest floor contents were used. and soils in study area surface soil The Kjeldahl total contents for N and P are reported here. Acid extractable Ca, K, and Mg should also be related to MAI and are being evaluated. No heavy metals were examined due to the extremely low metal loadings with sludge application (Nguyen et al. 1985). Estimation of long-term growth changes encompassing a range of nutrient resources was accomplished by sampling a total of 29 stands within the region. across an upland productivity stands. scrub productivity oak to high The stands were located gradient ranging productivity from sugar low maple Regression equations were developed using data from the 29 stands for the nutrient contents of soil and forest 12 floor components exhibiting linear associations with stand growth. Nutrient contents for sludge and control plots on the sludge study area two years after treatment were inserted into the equation to applications. predict changes in growth from sludge The oak ecosystem in the sludge study has stand growth and composition which falls within the range sampled in the regional study; therefore, implications for changes in stand growth should be applicable. Sludge Study Area: Located in Montmorency county, Michigan, the sludge study was conducted on a 70 year old oak stand composed of red oak (Ouercus rubra L.) , white oak (Ouercus alba L.) , and red maple (Acer rubrum L.) aspen (Populus with scattered pines spp. L.). The (Pinus spp. experimental L.) design, and sludge loading rates, and further details of the site and methodology are presented 1985). The series, a elsewhere soils mixed, at (Hart this frigid et site Alfic al. 1984, belonged Nguyen to Udipsamment; the et al. Graycalm with small inclusions of the Rubicon soil series, a mixed, frigid Entic Haplorthod (Soil Conservation Service series are distinguished by being deep, soils formed in sand on Conservation Service 1979). glacial 1979, 1976). Both excessively drained drift materials On the oak site the (Soil Graycalm 13 series consisted of a weakly developed sandy glacial drift overlying a calcareous, indurated till, with textural bands of varying thickness occurring at fluctuating depths; features not unlike other soils in the immediate vicinity (Farrand 1982) . Regional Study Areas; The Forest, 29 stands 80-90 were miles located southwest in of the the Manistee sludge National study plots. Figure 2.1 shows the general locations of the two study areas. Soils on the two areas are similar, surface soils overlying glacial drift, series occurring in both areas. equal representation stand compositions. across predominantly sandy with analogous soil Stands were sampled to give a range of productivities and Specific stands were randomly selected from a list of stands provided by the Huron-Manistee National Forest. of Stands occupied by pioneer species, or showing signs disturbance sampling. in the past 40 years were excluded from The minimum basal area of stands sampled was 18.36 m2 ha‘1 (80 ft2 ac'1) . 14 85 W Montmorency f Manistee Co. ‘ “ ]' ~ ’1 ' W e x f o r d Lake Mason Co 40 miles 42 N Figure 2.1. Location of sample stands. Co Co Co. 15 Other stand selection criteria included: 1) the overstory must be at least 55 years old, 2) the stand must be normally stocked, canopy should be closed as far conditions will permit, i.e. the as site 3) stocking must be uniform throughout the stand with no extensive open areas, 4) the topography must be representative of upland conditions, 5) the 6) soils must be well drained, and no more than 30% multiple stems. of dominant overstory in Sample Collection on the Sludge Study Area: Site nutrient resources on the oak sludge study area were calculated from forest 1983, floor and soil samples collected in two years after application. Thirty points each on sludge and control treatments were sampled. Sample Collection on Regional Stands: Samples were collected for the regional nutrient resource study in the summer of 1983. A soil pit 150 cm in depth was located near the center of a homogeneous one hectare portion of the stand. classified points were to The the soils series randomly were level. located and understory samples were collected. then characterized Three additional soil, forest and sample floor and Forest floor samples were 16 collected from 6 sampling points (three at the pit and one at each of the three additional points) with a 30 x 30 cm metal frame and systematically separated into litter or fermentation and humus fractions. Litter (Oi) samples included recognizable and nearly entire leaf material not affected by decompositional processes, and woody material. and humus layer (Oe and Oa) samples Fermentation consisted of finely divided decomposed organic materials extending to the upper boundary of the mineral soil. Soil samples collected directly beneath the forest floor samples were divided into surface (SI, A and E horizons), and subsurface to a depth of 45 cm) layers. (S2, upper B horizon Twenty eight bulk density samples were collected with a hammer-type core sampler and were used to convert nutrient concentration data from the laboratory to a content by depth basis. Forest growth was measured at each sampling point using a 10 basal area factor prism. All trees at each prism point were measured for DBH, total height and merchantable height to a 10.2 cm top. values to Trees were sampled across the range of DBH determine total age at DBH. Stand growth was calculated by averaging measurements of the four subplots. Chemical Analysis: Forest floor samples were oven dried at 70°C, ground, and subsampled prior to analysis. weighed, Soil samples were 17 air dried, sieved, and subsampled prior to analysis. Total Kjeldahl micro-Kjeldahl N and digestion TKP were procedure and determined by a analyzed on a Technicon1 Auto Analyzer II system (Technicon 1977) . The data were used with forest floor weights and areas, and soil depths and bulk densities to calculate TKN and TKP contents. All chemical analyses were performed in the MSU Forestry Department laboratory using 10% sample replication to insure precision and bulk sample analysis with each sample set to insure accuracy. The quality assurance procedures confirmed with a probability of .95 the determination of mean TKN and TKP concentrations with a 10% confidence interval. Statistical Analysis: Scatterplots between MAI and stand nutrient contents of the eight forest floor and soil components revealed several nonlinear relationships. More linear scatterplots were obtained with log (base 10) transformations of the variables. A multiple regression equation was stepwise procedure available in the formulated using a MICROSTAT statistical package (Ecosoft 1984). 1 Use of a trade name does not constitute an endorsement by either Michigan State University or the EPA. 18 The equation was of the form: Y = B0+ B1X1 + B2X2 . . . + BnXn where: Y = growth (MAI, m3ha'1yr’1) , n = number of variables in equation Bn = regression coefficients, and Xn = nutrient contents in the particular soil horizon (kg ha'1) . The normal matrix coefficients (B) is: solution for the vector of regression (X'X)'1 X'Y (Draper and Smith 1981) Confidence intervals for the predictions on sludge and control plots on the sludge study were calculated using: Y +/" t(v, 1-1/2 )*s 1/g + X0'CX0 Where: Y = Predicted value v = Sample size of regional study minus the number of parameters in the regression equation including B0, (29-4) s = Standard error of estimate g = number of observations in X0 X0 = a (nxl) Data Matrix, and C = the (nxn) Inverse of the Variance-Covariance matrix (X'X)'1 (Draper and Smith 1981). RESULTS AND DISCUSSION The regression equation developed from the 29 regional stands was: MAI (m3 ha'1 yr'1) = 1.1883 + (-5.5847 * LOG TKN CONTENT IN SI) + (5.7873 * LOG TKP CONTENT IN SI) + (2.6515 * LOG TKN CONTENT IN 02) The standard error of the estimate was 0.5241 m3 ha'1 yr'1. Contributions of individual variables to the equation can be evaluated through their partial correlation coefficients (calculated with the effects of all other predictor variables removed). The partial correlation coefficients were 0.357 for TKN in the A and E horizons, horizons, and 0.519 layers. The three similar magnitudes contributions Kjeldahl N in to for TKN in the partial fermentation and humus correlation indicating the the 0.437 for TKP in the A and E that estimation fermentation they coefficients made had comparable of stand growth. and humus layers Total was the strongest contributor to long term growth, while TKN in the litter layer had the weakest association. If the Oe and Oa horizons are thought of as nutrient repositories for sludge 19 20 applied nutrients, then mineralization should result in greater nutrient availability in these horizons and greater growth. The N concentrations are total determinations, and do not reflect the availability or immobilization of sludge N applied to the forest floor. The residuals were normality were noted. was explained examined and no deviations from Nearly 70% of the variability in MAI (adjusted R2 of .691). The results of the regression analysis confirm that a site's nutrient resources have correlations with stand growth as measured by MAI. Table 2.1 presents mean TKN and TKP contents for forest floor and soil components on sludge and control plots. Increases from control existed with sludge treatment for all Table 2.1. Means of TKN and TKP contents for forest floor and soil components two years after treatment on the sludge fertilization study. xACi/ij.ncii'ij. KTTTnTDTT?MrP COMPONENT — CONTROL n=30 SLUDGE n=30 --------- Kg ha'1 — Nitrogen in SI 833.8 895.9 Phosphorus in SI 110.2 149.3 Nitrogen in 02 391.6 584.1 21 variables in the regression equation for MAI. control and sludge equation yielded the treatment means results presented into Insertion of the regression in Table 2.2. The predicted growth on the sludged plots of 4.62 m3 ha'1 yr"1 is greater than that predicted for the control plots (3.57 m3 ha' 1 yr'1) by 29%. This suggests that sludge application has resulted in nutrient changes in soils and forest floor that may have long-term growth effects on this site, assuming that the two year changes will persist or can be maintained by retreatment. A significant 44% increase in three year basal area growth, and a significant 63% increase in three year diameter growth were found using conventional fertilizer trial techniques (Nguyen et al. 1985). Table 2.2. Predicted mean annual increments over two years ■Pnv* a.w x TREATMENTS r»l u i i u w i % « •.< -»1 V X PREDICTED MAI 4- v Wi. C U W Q t 10% CONFIDENCE INTERVAL m3ha'1yr'1 Control Sludge 3.57 a (*) 4.62 b +/“ -3403 +/" -4998 (*)Predictions followed by the same letter are significantly different at an alpha =.10 level. not 22 The difference between treatments exceeded the confidence intervals for the sludge and control predictions and was statistically significant. The growth predicted for the sludge treatment was 8.5% higher than the maximum growth from the regional study (4.62 vs. 4.26 m3 ha'1 yr’1) . This is an extrapolation of the data and results should be cautiously interpreted. It should be pointed out that although the predicted sludge growth was beyond the range of the 29 regional stands, it was not greater than measured growth in the region. Stands in the region which had MAI measured in the summer of 1983 had growth rates of up to 5.11 m3 ha'1 yr'1. The small magnitude of the over-range, high statistical significance, and results of short-term interpretations observations reached here. tend to Namely, reinforce that there the was a significant difference between sludge and control treatments equivalent to 1.05 m3 ha'1 yr'1. The use of MAI-nutrient resource relationships to assess potential changes in long term growth from sludge is a new approach. Development of this approach and regression model are still being refined and verification of the technique must be performed. The MAI-nutrient resource relationship was also used to calculate changes hypothetical in potential retreatments three major assumptions. of the long-term oak site. growth This with requires First, the difference between sludge 23 and control nutrient contents of components (Table 2.1) were assumed to represent two nutrients on the oak site. were assumed equivalent to have to the year retention sludge added Second, subsequent reapplications loading initial of rates and retention loading and two rates year retention. Third, N and P contents from previous sludge additions have different retention rates than nutrients and organic matter from newly applied sludge. Figure application 2.2 presents and three MAI predictions retreatments over eight for initial years using three different retention rates of sludge more than two years old. Curve 'A' assumes a retention rate of 100% for sludge nutrients after the second year. The MAI predicted in the eighth year using this assumption was 70% greater than the highest MAI observed in the 29 regional stands and represents an extrapolation which must be interpreted with caution. The 100% retention rate would result in maximum accumulations with no further degradation of matter and nutrient release. nutrient applied organic This may not be a reasonable assumption for a forest system. Curve B 1 presents growth when sludge nutrients were assumed to have nutrient retention rates equivalent application. to those of the first two years after The MAI in curve 'B' in the eighth year was 23% above the maximum sampled MAI. However, retention rates for the first two years after application may be unreasonably low for ‘older1 (and probably more resistant) organic nutrient 24 • Years from initial ap p l i c a t i o n • C\J o LD CO o (T- ^ T_B4eW)lVW Figure 2.2. Hypothetical MAI curve. three different nutrient sludge nutrients > 2 yrs initial retention rate, Changes over time with retention rates for old; a. 100%, b. and c. 75%. 25 pools. For this study area, Hart et al. significant increases (1984) report that in nutrients and organic matter were found in the forest floor following sludge application. With this in mind use of a 75% retention rate, which falls between the two extremes, seems a reasonable stand response as represented by curve 'C '. From the first application onward the MAI's exceed the range of MAI's in the 29 stand regional sample from which the regression was extrapolations developed which may and, limit therefore, interpretations. constitute From the point of view of regional stand growth, curve 'B' is the only one which does not exceed MAI's measured in the field (5.11 m3 ha'1 yr'1) , although it does exceed the maximum growth rate of the regional study stands (4.26 m3 ha'1 yr‘1) . The figure should be interpreted on the basis of the general form of the curve from a biological basis rather than to strictly accept the accuracy of the MAI predictions. The shape of the curve is similar to many growth curves with nutrient additions and the model appears to be sensible from a biological point of view. If no retreatments are made to the stand, one would predict nutrient contents might decrease over time until they attain levels close to control levels. Stand growth should also decrease over time until it reaches a rate similar to the control. The length of this response period is not known. In summary, site nutrient resources explained almost 65% of the variation in stand growth for 29 regional stands, and use of the nutrient resource approach with sludge and control plots predicted significant increases in growth with sludge application. ACKNOWLEDGEMENTS Although the information in this document has been funded in part by the United States Environmental Protection Agency under assistance agreement No. S005551-01 to the Michigan Department of Natural Resources and Michigan State University, it has not been subjected to the Agency's publication review process of the and, therefore, may not necessarily reflect the views Agency and no official endorsement should be inferred. Mention of constitute trade names or commercial products does not endorsement or recommendation for use. Special appreciation is extended to Connie Bobrovsky, Barbara Kinnunen, Leslie Loeffler, Laurie Zwick, Michael and David Richmond, invaluable analysis, Robert Morel1, assistance in and data entry, and Steve Westin and Brenda Ellens for their sample preparation, laboratory and to Donald Zak, George Host, for their efforts in sample collection. Particular appreciation is extended to coauthors Dr. Hart Jr., Dr. P.V. Nguyen; and insightful review comments. . 27 Dr. C.W. Ramm for J.B. their LITERATURE CITED Brockway, D.C. 1983. Forest floor, soil, and vegetation response to sludge fertilization in red and white pine plantations. Soil Sci. Soc. Amer. Journal. 47:776-784. Draper, N.R. and H. Smith. 1981. Applied analysis. John Wiley and Sons, N.Y. 709 p. regression Ecosoft, Inc. 1984. Microstat, an Interactive General-Purpose Statistics Package. Release 4.0. Ecosoft, Inc. Indianapolis, IN. Farrand, W.R. 1982. Quaternary geology of Michigan (Map). State of Michigan Department of Natural Resources Geologic Survey. Hart, J.B.; P.V. Nguyen, C.W. Ramm; J.H. Hart, and D.M. Merkel. 1984. Ecological Monitoring of Sludge Fertilization on State Forest Lands of Northern Lower Michigan. Annual Progress report. Dept, of Forestry, Michigan State Univ., East Lansing, MI. Heiberg, S.O. and A.L. Leaf. 1961. Effect of forest debris on the amelioration of sandy soils. Rec. Adv. in Bot. 1622-1627. Koterba, M.T., J.W. Hornbeck, and R.S. Pierce. annl in a f i W A V * * Torrey, S. Techniques. i r» o M A 4W A . n>*r» W 44W A . A A V» a W 4 «« W W W * ■ ?+*a W .A .W W * (ed) Sludge Disposal Noyes Data Corp. 372 p. Bv 1979. v> • Sludge O K O - O CW OV *» W ** f T M »• Landspreadina Leaf, A.L., R.E. Leonard, J.V. Berglund, A.R. Eschner, P.H. Cochran, J.B. Hart, G.M. Marion, and R.A. Cunningham. 1970. Growth and development of Pinus resinosa plantations subjected to irrigation-fertilization treatments. p.97-118 IN: Tree growth and forest soils. Proc. of Third N. American Forest Soils Conf. August 1968, North Carolina State Univ., Raleigh. Oregon State University Press, Corvallis. 28 Nguyen, P.V., J.B. Hart, Jr., and D.M. Merkel. 1985. Sludge fertilization on oak forests in Michigan: Short-term nutrient changes and growth responses. Chapter 26 In: The Forest Alternative for Treatment and Utilization of Municipal and Industrial Wastes. Proceedings of the International Symposium on Forest Utilization of Municipal and Industrial Wastewater and Sludge., June 25-28, 1985. University of Washington, Seattle, p. 282-291. Soil Conservation Service. Soil survey. USDA. Soil 1976. Rubicon series. Conservation Service. 1979. Coop. Soil survey. USDA. Nat. Coop. Graycalm series. Nat. Technicon Industrial Method. 1977. Individual / Simultaneous determination of N and/or P in BD acid digests. Method No. 334-74W/B. Technicon Industrial Systems, Tarrytown, NY. White, E.H. and A.L. Leaf. 1964. Soil and tree potassium contents related to tree growth I: HN03-extractable potassium. Soil Science 98:395-402. Zinke, P.J. 1960. Forest site quality as related to soil nitrogen content, p. 411-418 IN: 7th Intern. Congress of Soil Science. Madison, WI. 29 Chapter 3. Distribution and Variability of Nutrients and Organic Carbon Across Glaciated Michigan Landscapes. INTRODUCTION The most influential soil nutrient pools assumed to be located in the surface soil. his 1937 article that (Coile 1937). and predictors of site quality Leaf Pawluk and Arneman 1959). exceeds in the This has led examination of nutrients in the upper solum as indicators 1956; often Coile noted in fine roots were concentrated upper 30 cm of the soil profile to extensive are 1961, (Broadfoot 1969, Youngberg and Scholz Most plants have a potential for nutrient uptake which their requirements if contact with the soil solution, nutrient supply the roots have sufficient and if there is an adequate (Mengel and Kirkby 1982). Trees with roots which utilize nutrient enhanced subsolum strata may support greater forest growth. pine The branching and proliferation of red (Pinus resinosa), jack pine (Pinus banksiana) , and big tooth aspen (Populus qrandidentata ) roots in subsolum textural strata are examples of adventitious root exploitation of deep soil features (Hannah and Zahner, 1970). Greater growth of red oak (Quercus rubra L. ) has been found on Kalkaska soils (Sandy, mixed, frigid typic Haplorthods) in northern lower Michigan which had subsolum textural strata within the rooting 31 32 range2. If the effects of factors known to influence site quality such as landform, climatic variability, depth Barnes (Spurr and 1973, are held may make determining site taxonomic classifications quality. This is particularly true for soils developed in potential drift more and occurrences of subsolum 1969) then glacial of Broadfoot constant, sandy, recognition and effective soil useful alluvial features in materials subsolum textural which strata. have It was hypothesized that inclusion of subsolum features as phases of soil series would reduce the variability which has made soil taxonomic classifications inadequate for separating sites by productivity differences Shetron 1972, (Grigal 1984, Esu and Grigal 1979; Spurr and Barnes 1973; Wilde and Leaf 1955, Wilde and Scholz 1934). Significant variations in the productive capacity of a site have been associated with subsolum features et a l . 1984, Hart et al. 1969, (Comerford White and Leaf 1964; White and Wood 1958, and Wilde and Leaf 1955). On a potassium (K) deficient glacial outwash soil in northern New York a two-fold difference in volume adjacent sites. growth at age 25 was found between Both sites had sandy surface horizons with one having a fine textured strata occurring at 1.8 m while the other had a similar strata of fine textured material 2D.T. Cleland, Deep Bands and Forest Growth on Kalkaska Sands. Unpublished Report. 33 occurring at 2.7 m (White and Wood 1958). The authors attributed the site quality difference to tree roots being able to tap into the greater nutrient and available water supplying capacity shallower site. of the fine textured strata on the Later studies on these sites found that mean basal area, mean forest floor weights, and the mean K content of the forest floor were all greater on the site with the shallower textural strata (Hart et al. 1969). The deeper roots penetrate into soils with subsolum textural strata, the more likely it is that subsolum sampling will identify nutrient reserves which influence site quality. This is particularly true when the solum consists of sandy silicious materials. Despite the recognition of potential increases in site quality due to subsolum properties there have been no nutrient regional contents northern hardwood by evaluations depth in of soils the distribution underlying oak of and forests to quantify nutrient differences which may be associated with these subsolum properties. Site quality influence conditions the it is growth can a combined of stands be effect of on a site. expressed by measurements of growth such as volume. factors which Under certain direct long Alternately, term it may be estimated by features which are strongly associated with stand growth, in which case it is important to measure site features which most directly influence stand growth (Carmean 1975, Coile 1952). Examinations of nutrient contents (e.g. 34 kg ha'1) have resulted in good correlations with stand growth of red pine (Pinus resinosa Ait.) ponderosa pine indicating the (Pinus (White and Leaf 1964), and ponderosa association of Laws.) nutrient (Zinke, contents 1960), with site guality. Collection and analysis of forest soil samples are often performed to group forest sites into classes of similar site quality and response to environmental perturbations such as clearcutting or classification acidic allows more precipitation. efficient Forest forest site management by estimating the productive potential of forested lands which are not suitable (Carmean 1979, for normal site quality measurements Coile 1952). The precision and accuracy of nutrient estimates may limit the utility of management information derived from soilsite relationships. In order to obtain precise estimates of soil characteristics, to calculate an knowledge of soil variability is needed adequate sample size. Nutrient variability of forest soils has been found to be greater than agricultural soils. the Greater increases in soil variability were found when entire 1963). zone of root proliferation was examined (Mader In addition, on forest soils of glacial and alluvial origin it has been observed that variability increases as one proceeds further into the soil profile Mollitor et al. 1980). (Hart et al. 1969, However, the variability in nutrient contents associated with depths greater than the surface 150 35 cm has not been thoroughly investigated in forested soils. OBJECTIVES The two primary objectives of this research were: examine and report distributions carbon contents and of nutrient and 1) to organic their variability to a depth of 450 cm, and 2) to determine whether the inclusion of a banding phase (indicating presence or strata) in the series taxonomic unit provides precise estimation alone. Both of the primary objectives are concerned with soil of absence soil of fine textured fertility than the soil subsolum a more series determining whether sites differentiated by physical presence of subsolum levels. textural strata displayed elevated nutrient The secondary objectives were to compare stand growth grouped by soil series and phased soil series, and to examine nutrient distributions and variability regionally, across soil series, and across phased soil series. MATERIALS AND METHODS Study Area: Study plots were located in the Manistee National Forest in the northern lower peninsula of Michigan between 85°30' and 86°151 west longitude and 45°52' and 44°30' north (Figure 3.1). The climate alternates between semi-marine depending on the direction of patterns. The 29 year (1940-1969) mean was 821 mm (32 in) and the 29 year mean was 5.8°C (42.5°F) (Strommen encountered on the sites, all developed in 1974). of frigid, Sandy, mixed, annual precipitation annual temperature The Typic Udipsamment; soil series and Grayling) Wisconsinian Conservation Service 1976, 1981, 1982). a mixed, continental and individual weather (Rubicon, Kalkaska, materials latitude age (Soil The Grayling soil is the Kalkaska soil is a frigid Typic Haplorthod; and the Rubicon soil is a Sandy, mixed, frigid Entic Haplorthod. All soils were well to excessively drained and were formed from sandy glacial drift parent materials. The more productive soils developed in or were underlain by till plain, ground moraine, or lake plain materials with textures of sandy loam or finer. Thirty stands stratified productivity were sampled. by low, moderate, and high Table 3.1 shows the pre-sampling 36 37 Wexford Co. Mason Co Figure 3.1. Location of sample stands. 38 Table 3.1. Pre and post sampling stratification for the 30 sample stands. Pre-Sampling Post-Sampling Stratifications Stratification n Low Productivity (hills and Plains) 6 Series n Grayling (Gy) 3 Moderate Productivity (hills and Plains) 10 Rubicon (Rb) 16 High Productivity (hills and Plains) Kalkaska (Ka) 10 14 Series Phase n Gy 3 Rb 10 Rb(Band) 6 Ka 4 Ka(Band) 6 stratification and subsequent post-sampling stratifications by soil series and soil series phase. of Stands occupied by pioneer species, or showing evidence disturbance were sampling. in the past 40 years excluded Stands originating from stump sprouts from (determined by more than 30% of dominant overstory having multiple stems) were also excluded from sampling. stands to be sampled was The minimum basal area of 18.36 m2ha'1 (80 ft2ac'1) . Other selection criteria were that the stand must be at least 55 years old, the stand must be normally and uniformly stocked, (i.e. the canopy should be closed as far as site conditions will permit), and that the topography be representative of well drained upland conditions. 39 Sample Collection: Forest floor samples, soil samples, and stand growth and yield data were collected during the summer of 1983 in conjunction with an ongoing ecological classification system project carried out jointly by the USDA Forest Service and Michigan State University. of one hectare or more At each stand a homogeneous area was selected for sampling. Four randomly located points per stand were measured for forest growth. Forest floor samples were metal frame at six points three additional points) collected with a 30x30 cm (three at the pit and one each at and systematically separated into litter or fermentation and humus layers (Figure 3.2a). Litter samples leaf (Oi horizon) material not included recognizable and nearly entire affected Fermentation and humus consisted of finely by decompositional layer samples divided processes. (Oe and Oa horizons) decomposed organic materials extending to the upper boundary of the mineral soil. A soil pit about 1.5 m in depth (which extended into the C horizon) was located near the 3.2b). center of the stand (Figure After the pits were dug the soils were classified to the series level and egual volumes of soil were collected from each horizon. Bucket auger samples, stratified by horizons, were collected beginning at the pit bottom and ending at 4.5 m (Figure 3.2c). Bulk density samples were collected to 40 b.) Bhs Pit bottom a. •• •••o- 450 cm pjm ,re 3 2 Figure 3.2. Detail and and soilc .sampling; a. Detei_ of forest ^ sfloor o U pit( bucket aUger. 41 allow conversion contents. of nutrient concentrations to nutrient The areal extent of subsolum textural strata was determined from observations at the pit bucket auger boring and from observations at three additional bucket auger borings at the randomly located points. Laboratory Analysis: Forest floor samples were oven dried at 70°C, weighed, ground, and subsampled prior to analysis. Kjeldahl nitrogen Forest floor total (TKN) and total Kjeldahl phosphorus were determined after sulfuric acid digestion 1982) using (Technicon, determined dissolution a Technicon3 1977) . by in Total dry 8 N ashing HC1 Auto-analyzer cations at (K, 500°C (Wilde et Mg, (TKP) (Page et a l . II procedure and Ca) for 4 hours al. 1972) . were with ash Nutrient concentrations were converted to nutrient contents by use of areas and weights of the forest floor samples. Statistical analyses were to be performed with the stand as the sample unit, therefore, forest floor data was averaged by stand. Soil samples were air dried, screen, and subsampled for analysis. sieved to pass a 2 mm Both coarse (>2 mm) and fine fractions weights were recorded for each sample and used to calculate horizon contents corrected for coarse fragment 3 Use of trade name does not constitute an either Michigan State University or the USDA. endorsement by 42 contents. A 2 g soil sample was boiled for 10 minutes with 10 ml of 0.5 N nitric acid to extract potassium (K) , magnesium (Mg) , and calcium (Ca) . This extraction has been shown associated with tree K uptake and growth in red 1958, White and Leaf 1964), levels in radiata pine pine and with critical (Pinus radiata D. Don) Since Mg and Ca are usually extracted and have many chemical properties to be (Leaf foliar Mg (Adams 1973). concurrently from soils in common (Page et al. 1982), Ca was also extracted by this method. Acid extractions are thought to represent cations both adsorbed and mineral which plant uptake and leaching replenish elements (Page et al. 1982). depleted by Metson, (1974) concluded that the rate of Mg release from soil reserves was correlated with Mg extracted attack and is a good by a moderately strong acid guide to available Mg in continuous cropping or forest systems. Aliquots of the boiling nitric acid extractions and of the dry ash total cation concentrations (K, Mg, and determinations had cation Ca) determined by plasma emission spectroscopy. The extracts were brought to a concentration of 1,000 ppm LiCl interferences. to stabilize TKN and the plasma and suppress TKP were determined after sulfuric acid digestion similar to the forest floor samples. Percent organic carbon (OC) in the soils was determined by the Walkley-Black method with .5 N ferrous sulfate and 43 1 N potassium dichromate (Page et al. 1982). Sample size varied from 1-5 g in samples with high and low organic carbon concentrations. Horizon contents concentrations per densities, horizon were using converted concentrations, and horizon depths. by 50 cm nutrient soil bulk The contents were corrected for percent coarse soil separates. accumulated into depths to Nutrient contents were circumvent problems with differences in horizon thickness. All chemical analyses were performed in the MSU Department forest soils laboratories using Forestry 10% sample replication, National Bureau of Standards specimens, and bulk sample analysis to insure precision and accuracy. Statistical Analysis: Nutrient contents were tested square goodness of fit test, rejected the nutrients. null With (Steel and hypothesis the for normality by of data exception Torrie 1980), which normality of Ca for most contents, log transformed data had no departures from normality. these a Chi- Based on findings log transformed nutrient contents were used in the statistical analyses. A one way analysis of variance (AOV) differences was used to contents between Duncan's multiple test series for and between comparison test in phased with mean nutrient series groups. corrections for 44 unequal sample sizes was differences (Steel and used to identify specific group Torrie 1980). The inferences of the AOV and multiple comparisons were confirmed using a non-parametric Kruskal-Wallis one way analysis on ranked nutrient contents with mean separation by Dunn's method (Dunn 1964, and Torrie 1980). slightly more Hollander and Wolfe 1973, Overall, conservative Steel the non-parametric methods were than the parametric methods, finding fewer significant differences between groups. Results between the two tests led to similar interpretations of the data. Since greater, and the sensitivity the log of the parametric transformed data tests satisfied was the assumptions of the analysis of variance, the results of the parametric tests are reported. (CV's) were chosen Coefficients of variability to compare the relative variability of series and species groups because of their ability to compare variances between group means differing in magnitude (Steel and Torrie 1980). RESULTS AND DISCUSSION Characteristics of Sampled Stands: For ease strata will of description soils be referred to as with 'banded', subsolum textural and soils without subsolum textural strata will be referred to as 'unbanded'. 45 Taxonomic grouping by soil series differentiation by site quality. resulted in a trend of As presented in Table 3.2, a gradient of increasing volume productivity corresponded with the gradient of increasing profile development < Rubicon (Rb) < Kalkaska (Ka)). However, (Grayling (Gy) no significant differences were found between soil series productivities by the one way AOV. The same correspondence between site quality and profile development was evident when presence or absence of subsolum textural strata descriptions (Table 3.2). was included in the series An analysis of variance performed on the soil series phased by subsolum textural strata found significant differences between group mean volumes. Table 3.2 Stand volume grouped by series and series phase with one way AOV results. Phased Series n Stand Volume m3 ha‘1 3 147a Grayling Series Grayling Duncan's n Stand Volume m3ha -1 3 147 a 10 149 a Rubicon 16 189 Rubicon Kalkaska 10 227 Kalkaska 6 211 b Banded Kalkaska 4 251 b Banded Rubicon 6 256 b ■■11 a ... . ^ - Column means followed by different letters are significantly different by Duncan's multiple comparison test at an alpha = 0.10 level. 46 mean separation procedure determined that the banded phase of the Rb series had significantly higher volumes than the unbanded phase. than the unbanded The Gy series had significantly lower volumes banded phases phase of of the the Ka Rb and Ka series. series, The low and the relative productivity of the Gy series and the unbanded Rb phase was similar to that reported by Shetron (1972). Although not significantly different, the banded phase of the Ka series had slightly greater volume productivity than the unbanded phase. Nutrient and Organic Carbon Distributions: Nutrient and organic carbon contents accumulated by 50 cm depths for all stands are presented in Table 3.3. The distribution pattern by depth varied with the nutrient examined. Total Kjeldahl N (presented in Figure 3.3a) and OC displayed patterns depth. while Total acid depths. of Kjeldahl extractable asymptotic decline with increasing P displayed a less abrupt decline, K showed Acid extractable Mg little deviation (shown in Figure 3.3b) between and Ca had patterns of increasing content with depth, opposite that of TKN and OC. These distribution patterns are consistent with known properties of soils and nutrients. Much of the N and P in soils is in organic form, or fixed in living or dead organic material and the distributions of TKN and TKP follow that of organic matter in the soil (Armson 1977, Kimmons 47 Table 3.3. Mean nutrient and organic carbon contents and coefficients of variability for ail sam FF 0-50 cm 50-100 cm kg/ha n=30 n=30 CV Z CV Z X n=29 n=30 kg/ha n 139 224 X kg/ha I i i Nitrogen CV Z X kg/ha 1 n=.29 CV V. X kg/ha I 1 1 CV Z Z> M u x kg/ha 250-3 l x kg/ha 200-250 cm 1 150-200 cm 1 100- 150 cm n=29 X 289 37 2688 43 698 50 294 47 226 75 262 Phosphorus 19 37 882 37 544 37 344 37 332 48 369 99 323 Potassium 18 35 537 35 473 35 587 72 599 71 687 76 815 Magnesium 27 45 629 59 983 84 1128 72 1326 94 2894 283 4910 395 45 1069 203 5591 396 8152 332 7034 325 24397 363 31156 43225 26 11077 51 5257 40 3910 52 3077 48 3379 Calcium Organic Carbon N..0. R.ti. - organic carbon 3eEermTnaEIons were noE made on ForesE Floor samples. ton contents and coefficients of variability for all samples. 100- 150 cm 150-200 cm 200-250 cm x kq/ha x kg/ha X kg/ha 250-300 cm 300-350 cm 350-400 cm 400-450 ca CV 7. CV 7. CV 7. X kg/ha CV 7. CV 7. X kg/ha CV X kg/ha 7. 1 va n . V C 1 1 i 1 •100 ca n=; 3tr ~ n=30 n=27 n=29 n=29 n==25... CV X kg/ha 7. n=24 18 50 294 47 226 75 262 139 224 89 196 78 158 80 182 97 t4 37 344 37 332 48 369 99 323 81 433 111 350 55 368 54 '3 35 587 72 599 71 687 76 815 123 931 110 719 65 746 68 13 84 1128 72 1326 94 2894 283 4910 223 5341 220 5912 212 7174 202 n 396 R1R? 332 7034 325 24397 363 31156 288 40729 240 31559 186 33672 162 '7 51 5257 40 3910 52 3077 48 3379 69 3660 105 2635 56 2879 72 is-ueri~noE raa3e on ForesE Floor samples. 48 3.000 2.BOO 2.600 2.400 2 .2 0 0 2 & A* m t *-* § U | g 2'D0° i.eoo I.soo 1,400 1^°° 1.000 *-! BOO Z 600 400 200 0 FF 50 100 150 200 Lower depth of Figure 3.3a. 250 300 350 400 450 50 c m a c c u m u l a t i o n TKN contents accumulated by 50 cm depths for all stands. . Magnesium contents 8 00 0 7.000 - 4.000 - . 3 000 2.000 - 1.000 - 100 15 0 200 250 300 350 L o w e r a e p t n of 5 0 c m a c c u m u l a t i o n Figure 3.3b. Forest floor total and soil acid extractable Mg contents accumulated by 50 cm depths for all stands. 49 1986). Acid extractable K, Mg, and Ca are released from soil minerals, readily accumulate deeper leached in the from surface soil profile. soils, and may Especially when a feature such as a subsolum textural strata impedes the flow of water through the soil. The nutrient contents in the forest floor (FF) comprised only a minor fraction comparison to the of the site's soil contents. nutrient Figure resource in 3.4 displays the relative percentages of acid extractable K contents by the various depths. As shown here the FF total K is only 0.3% of the FF - 450 cm acid extractable K content. 250 C11.250 Figure 3.4. Percentages of K contents accumulated by FF and 50 cm depths to 450 cm. 50 Nutrient and Organic Carbon Variability: As presented in Table 3.3 the variability contents in the FF, the surface 150 cm, of nutrient and the 350-450 cm depths of the soil were lower than at the intermediate depths. Figure 3.5 displays this pattern for TKP variability. figure shows three discrete variability regions, This the FF-200 cm depth which has the least variability; the 200-350 cm depth which has the maximum variability, and at depths of 350-450 cm the variability decreases. FF 50 100 150 Lower Figure 3.5. 200 d e p t h of 250 300 350 400 50 c m a c c u m u l a t i o n Variability of TKP contents for all stands. 450 51 An increase in variability was noted at depths greater than 150 cm for all extractable Ca (Figure nutrients 3.6). in Table 3.3 except acid Acid extractable Ca and Mg had C V 1s 2-3 times that of TKN, TKP, acid extractable K, and OC. This increased variation may be due to variability in limestone and/or dolomite composition of the glacial drift (Veatch 1953). As shown in Figure 3.6 acid extractable Ca had high CV's throughout the profile (e.g. 203% in the surface 50 cm) and displayed beneath 250 cm. a decrease in variability at depths From Figures 3.5 and 3.6 it is apparent that 450 400 - i x v x x v i FF 50 100 150 200 Lower depth of Figure 3.6. 250 300 350 400 450 50 c m a c c u m u l a t i o n Variability of forest floor total and soil acid extractable Ca contents for all stands. 52 a general pattern exists of increasing variability with depth to about 350 cm, then decreasing to 450 cm. Nutrient and Organic Carbon Distribution by Soil Series: Nutrient contents, to the nearest kilogram per hectare, were evaluated by soil series and are presented in Table 3.4. As determined by one way AOV and Duncan's mean separation procedure, the FF nutrients displayed the greatest sensitivity to series differences despite their relatively low nutrient contents (see Figure 3.4). Forest floor TKN, TKP, and total Ca contents displayed significantly different contents between all three series groups. This was the only depth/content component which differentiated between all three series. AOV found that the surface 150 cm showed the The greatest sensitivity to series differences, with only two significant differences occurring below this depth. TKN, TKP, and K contents had the greatest sensitivity to series differences. The AOV found that all accumulations to 150 cm had significant series differences. All nutrient contents, except for acid extractable Mg, reflected the degree of soil profile development (Kalkaska > Rubicon > Grayling) in the surface 200 cm. This pattern was well-defined for K contents presented in Figure 3.7. 53 Table 3.1. Mean nutrient and organic carbon contents, coefficients of variability, and results of one uay RQV for soi O—SO cm FF n X kg/ha CV y. n X kg/ha 100-150 cm 50-100 cm CV y. n X kg/ha CV n m *i X kg/ha 200-25 150-200 cm CV n X kg/ha CV •j n X kg? V. Nitrogen Gray1ing Rubicon Kalkaska 3 16 9 27 15? a 260 b 27 374 c 31 3 16 10 2083 a 29 2294 a 36 3423 1 b 41 3 16 lO 394 a 37 604 a 40 936 b 44 3 16 10 153 a 52 276 a 53 381 b 23 3 15 10 176 213 276 21 87 61 3 15 10 21 21 37 Phosphorus Gray1ing Rubicon Kalkaska 3 16 9 11 a 53 1? b 29 24 c 31 3 16 10 799 931 1048 50 42 24 3 16 10 369 a 36 495 a 32 629 b 30 3 16 10 223 a 17 339 b 35 37? b 38 3 16 10 219 295 401 17 41 49 3 15 10 21 37 39 Potassium Gray1ing Rubicon Kalkaska 3 16 9 10 a 1? b 20 b 11 33 29 3 16 10 391 516 59? 35 28 39 3 16 10 384 476 49? 24 3? 35 3 16 10 348 a 46 474 a 38 806 b 79 3 15 lO 402 529 759 53 57 80 3 15 10 50 77 61 Magnesium Gray1ing Rubicon Kalkaska 3 16 9 12 a 2? b 32 b 41 28 53 3 16 10 422 668 589 20 57 66 3 16 10 553 1104 904 21 94 62 3 16 10 486 1111 1364 29 65 75 3 15 lO 495 1446 1440 30 101 77 3 15 10 146 152 557 177 «s 16 358 b 27 501 c 41 16 10 -w131? 937 221 92 16 10 1 ~?f. 7737 4225 377 204 X 16 10 201 9239 9483 325 29? 15 10 746 5253 12299 7? 237 29? 3 15 10 247 339 189 N.O. 3 16 10 36796 39690 49050 19 25 23 10 8922 9701 14170 2 42 55 3 16 10 5484 4685 6099 11 61 28 3 15 lO 4771 3545 4520 11 52 51 3 15 10 355 309 515 Calcium oray a ing Rubicon Kalkaska Organic Carbon GrayIing Rubicon Kalkaska 5 16 9 3 16 Column means for nutrient or organic carbon contents followed by different letters ore significantly different N.O. - organic carbon detorminatons were not made on forest floor samples. iants of variability, and rmsults of ona uay ROV for soil sorios groups. 100—ISO cm OO cm x /ha CV fi n 195 a 32 3 £ \ 3 *2 • IS 5? .? x kg/ha 150-200 c. CV X 153 a a-m* - S 52 a-« ? \ S “223 3 .a 17 12 b 35 S S b 38 n 200-230 cm x C V kg/ha X 250-300 cm .. kg^ha CV -- n 3 3 iq 15 10 11 211 210 374 sK 30 30 02 87 8? - x kg/ha CV 33 15 15 10 is 125 125 219 219 276 - 10 10 72 72 159 - 33 13 13 10 >o 3o 15 £ 16 75 135 1 » IS X 33 .a 15 10 11 126 176 -i,T 213 276 32 21 21 a~> 0? 61 2; IS 10 33 16 219 295 2? 401 24 49 IS 10 212 371 394 ' 16 82 If 91 II 10 205 — 354 2 289 ^ 3 15 lO 496 91? ??5 1? ,2 41 lo 339 ™ 377 348 a 46 474 a 38 806 b 79 3 15 10 402 529 759 53 57 80 3 15 io 507 774 614 49 141 8? 300-350 cm 3 15 n X 61 80 72 x kg/ha 160 219 350-*l00 cm CV n X 36 66 i s ifl 3 12 12 3sf 22 355 61 593 103? 923 21 120 103 3 12 9 441 ?40 791 48 63 69 3 1495 13 2162 10 11046 51 117 165 3 8171 12 1713 9 11320 159 81 169 3 13 10 21 94 62 3 16 10 406 1111 1364 29 65 75 3 15 10 495 1446 1440 30 101 77 3 15 10 1461 1529 5572 ?9 179 15 lO 76 '37 25 51 377 204 3 16 10 201 9239 9483 82 325 297 3 15 10 246 5253 12299 77 237 29? 3 15 10 2423 33914 18989 169 303 159 3 15 10 13535 40082 26021 1S8 354 211 3 13 10 1981 48719 45G90 101 272 126 3 12 9 •22 Ol 70 2 42 55 3 16 10 5404 4685 6099 11 52 51 3 15 10 3552 3091 3154 39 60 30 3 15 lO 3122 3779 3124 46 51 43 3 13 10 3025 4163 3495 40 116 09 3 12 9 4771 »45 4520 - 9 153 04 04 3 15 10 36 64 V. 116 3 16 10 11 51 28 135 163 ; « 24 37 35 11594 110 1374 76 6612 250 CV *, 84 76 97 164 x kg/ha 9877 162 30257 223 43835 137 2391 2000 2543 7 66 43 ollooad by”S7f farant’Tattars aro significantly different by Ckincan*s mu lt-iplct comparison tost at an alpha - 0.10 laval. orost floor samples. 54 1, 1 D 0 900 800 O' 700 600 500 400 300 200 100 FF 50 100 150 200 Lower depth of X I G r a y I i ng I X X I Rubicon 250 300 350 450 50 c m a cc umula tion KaIkaska Figure 3.7. Forest floor total and acid extractable K contents grouped by soil series. Significant series differences were found in the FF, 100-150, and 400 450 Ciu accumulations. Tuc only significant differences beneath the surface 150 cm were for TKP and acid extractable K at the 400-450 cm depth. significant series differences found by Duncan's procedure were between the Gy and Ka series. between series nutrient The only consistent contents The poor discrimination was consistent with poor discriminatory power found with soil taxonomic units for stand productivity (Shetron 1972). 55 Nutrient and Organic Carbon Variability by Soil Series: The series variability and evaluations displayed redistributed distinct series nutrient poor Gy series was commonly less the overall contrasts. The variable than the Rb or Ka series, while the Rb series had a higher variability than the more productive Ka series for acid extractable Ca and OC at all depths greater than 200 cm. The results for TKN, TKP, acid extractable K, and Mg were inconsistent with only 33% of the nutrient/depth combinations showing this pattern. The Rb and Ka series displayed the greatest variability for all nutrients at a range of depths from 250-350 cm which overlaps the range that the frequently encountered apparent for TKP and finer textured materials were (150-300 cm) . is presented This pattern was most in Figure 3.8. This increasing variability with depth to 350 cm occurred even when nutrient contents declined with depth. Patterns of increasing variability with increasing soil profile depth have been found glacial and alluvial origin Mollitor et al. 1980). in eastern (Hart et al. forest 1969, soils of Mader 1963, This pattern existed, especially for the Rb and Ka series, to a depth of about 250-350 cm, below which variability decreased. The soils in the cited studies were not examined to the depth which they were in this study, 56 150 130 120 110 100 90 BO 70 60 50 /X] /X 40 30 20 10 FF 50 100 150 200 Lower depth of I /\ Gray Iing Figure 3.8. 250 300 350 400 50 cm acc um u l a t i o n l\/l Rubicon Ikaska Variability of TKP contents grouped by soil series. therefore patterns of subsolum soil variability have not been reported. Since only the Rb and Ka series occurring at depths ranging from 150-300 cm, had banding there is the implication that the variability increases apparent in Figure 3.8 are a result of this banding. This hypothesis is supported by the fact that the Gy series has no banding, and displays no variability increases with depth. 57 Nutrient and Organic Carbon Distributions by Phased Soil Series: Nutrient contents evaluated by banded and unbanded soil series phases are presented in Table 3.5. The inclusion of subsolum textural strata in the soil taxonomic unit resulted in a greater number of significant differences being found as compared to the series evaluations (Table 3.3). sensitivity to soil fertility differences This greater was despite reduction of group sample sizes because of the phasing. the The one way AOV found significant differences between mean TKN contents in the FF and surface 150 cm similar reported in Table 3.4 for the series groupings. by Duncan's multiple comparison procedure to those When examined no significant differences were found between TKN contents of the banded and unbanded phases of the Rb or Ka series. Of the 23 significant nutrient or OC /depth combinations found by the one way AOV, 16 were for TKP and total or acid extractable K and Mg. The Ka phases were separated by Duncan's procedure more frequently (11 of 23) than were the Rb phases (4 of 23). Eight of the ten significant differences between the Ka phases were found with TKP and acid extractable Mg between the depths of 100350 cm. cm and The phosphorus in the TKP digests at depths of 250 below may represent the dissolution of insoluble 58 Table 3-5. h«an nutriant and organic carbon contonts, coefficients of variability, and rasults of one way 80V for soil seriQS n Nitrogen Gy Rb RbB Ka KaB x kg/ha CV n 3 10 5 4 6 211 200 230 290 431 IO 63 92 81 179 3 10 5 4 6 176 195 249 245 296 21 80 101 51 68 153 a 283 abc 264 ab 344 be c 405 52 45 71 31 17 223 a 344 be 332 abc 284 ab 440 c 17 30 33 29 33 3 10 5 4 6 219 288 308 224 519 a a a a 17 41 43 19 b 32 3 10 5 4 6 212 398 320 225 506 16 85 25 31 135 3 10 6 4 6 391 478 580 507 65? 35 2? 26 44 3 10 6 4 6 384 455 512 338 563 24 44 25 9 35 3 10 6 4 6 348 a 408 ab 584 b. 417 ab c 1065 46 36 31 21 68 3 10 5 4 6 402 406 777 413 989 53 a a 42 a b 47 a t. 28 b 71 3 10 5 4 6 507 674 975 389 763 61 58 99 34 69 41 22 35 43 56 3 IO 6 4 6 422 654 691 438 689 20 60 56 24 73 3 IO 6 4 6 553 953 1356 588 1115 21 68 113 28 57 3 10 6 4 6 486 a 906 ab 1454 be 606 a c 1870 29 51 67 41 56 3 10 5 4 6 30 495 a 1111 a 116 2117 t» 60 29 565 a 2024 t► 54 3 10 5 4 6 1461 1190 220Q 522 8939 110 63 73 42 198 16 27 27 40 46 3 10 244 593 50 164 181 32 100 176 3 10 492 6 19811 4 317 6 6630 51 98 241 69 226 62 201 a 3 685 ab 110 10 c 207 6 23495 78 315 ab 4 6 15595 be 233 3 2473 10 47290 5 7162 374 4 6 31399 158 312 138 64 157 3552 3103 3066 4470 2276 46 50 57 15 36 RbB Ka KaB 3 10 6 4 5 10 a 15 ab d 21 1? be d 23 11 31 29 23 26 tlagnesiua Gy Rb RbB Ka KaB 3 10 6 4 5 12 a 25 b 29 b 2? b 36 b 3 10 6 4 5 177 a 333 b 399 be 425 be c 561 Calcium 6 2523 4 668 6 1103 3 36976 N. 0. CV ri X 3 10 6 4 6 6 4 6 Organic Carbon Gy Rb RbB Ka KaB 3 IO 6 4 6 X 38 24 46 41 25 53 34 18 12 29 Gy Rb RbB Ka KaB 37 25 65 46 27 n 369 491 501 585 658 11 a 16 b 20 be 19 be 28 c Rb 394 a 635 be 551 ab 1197 d 762 ed kg/ha CV 2 n kg/ha CV V. X kg/ha 3 10 6 4 6 3 Potassium Gy n 50 36 49 9 17 3 10 6 4 6 io CV ?i 200-250 cm 3 799 10 997 6 622 4 824 6 1198 27 28 21 23 29 10 39257 6 40411 4 53492 6 46089 a a a b ab X kg/ha ca 3 10 6 4 6 150 a 240 b 294 b 309 b 426 c 2083 2206 2440 3547 3340 n X 150-200 29 31 43 6 56 3 10 6 4 5 Phosphorus Gy Rb RbB Ka KaB CV X kg/ha 100-150 cm 50-100 cm 0-50 cm FF 18 19 24 29 16 26 3 8922 a 10 9325 a 6 10327 a 4 19569 b 6 10571 a 2 36 51 40 55 3 10 6 4 6 5484 5111 3975 7127 5414 11 52 40 27 24 246 3 10 3361 5 9034 349 4 6 20265 77 276 198 54 232 4771 3232 4171 5071 4153 11 48 57 15 72 3 10 5 4 6 3 10 5 4 6 ”coluii»n”meen3 for- notriants or- organic carbon contonts followed by diforont 1attars aro significantly difforont^by I Soil sariai abbreviations o r a : Gy-GrayIing sorias, RbcRubicon sorias; RbB= Banded phase of Rubicon series, Ka=Kall KaB=Banded phase of Kalkaska series. N.O. - organic carbon determinations uara not made on forest floor samples. ionts of variability, and results of ono way flOV for soil sorias groups with phasos150-200 100-150 cun 250-300 200-250 cm cm ------------ -— CV CV ti kg/ba 2 1 CV X kg/ha 3 10 5 4 6 176 195 249 245 296 38 24 46 41 25 3 10 6 4 6 223 a 344 be 332 abc 284 ab 440 c 17 38 33 29 33 3 10 5 4 6 219 288 308 224 519 a a a a 24 44 25 9 35 3 10 6 4 6 348 a 408 ab 584 b 41? ab c 1065 46 36 31 21 68 3 10 5 4 6 402 406 777 413 989 53 a 42 a a b 4? a b 28 b 71 21 68 113 28 57 3 10 6 4 6 486 a 906 ab 1454 be 606 a c 1870 29 51 6? 41 56 3 10 5 4 6 51 99 241 69 226 3 02 201 in 685 ab H O c 207 6 23495 78 315 ab 4 6 15595 be 233 11 52 48 27 24 CV X y. kg/h 262 ab 23 286 ab 22 756 b 114 170 a 45 602 b 99 3 7 5 3 6 203 a 318 ab 421 b 206 a 429 b 42 25 6? 15 54 3 7 5 3 6 441 555 999 475 949 48 52 58 43 64 51 1495 a 1354 b 10O 3508 ab 102 48 765 a 17901 b 119 3 8171 7 1741 5 1674 600 3 6 16660 159 89 78 33 131 101 271 3 9877 7 20491 162 216 170 3 6 612 35 97 40 76 114 26 95 2391 2379 3582 2761 2434 7 47 73 14 55 5 4 3 10 5 4 6 205 374 315 202 347 16 90 63 8 507 674 975 389 763 61 58 99 34 69 3 10 5 496 602 1548 402 1024 49 58 142 1461 1190 2208 522 8939 30 495 a 1111 a 116 211? b 80 29 565 a 2024 b 54 77 276 198 54 232 4771 3232 4171 5071 4153 11 4Q 57 15 72 97 0 5 4 6 3 8 593 ab 476 a 1934 b 401 a 1271 b 7? 5 4 6 110 63 73 42 198 3 11594 ab 164 58 10 1107 a 6 1910 ab 86 40 591 a 6 13960 b 132 3 8 5 4 6 3 2473 10 47290 5 7lb2 374 4 6 31399 158 312 13b 64 157 3 13535 10 47253 9 25741 4 487 6 43044 169 312 62 109 1981 15610 1C1C93 5681 4 6 72697 3552 3103 3066 4470 2276 46 50 57 15 36 3122 3581 4175 3678 2755 39 49 120 29 27 3025 2798 6340 2474 4174 3 10 5 4 6 3 lO 5 4 6 36 75 50 112 80 3 16 85 25 31 135 6 y. 6 398 320 225 506 5 4 CV 135 165 159 258 120 212 3 X kg/ha 3 7 5 3 6 3 10 y. 36 63 66 125 86 10 6 n X kg/ha 160 185 273 207 177 1? 43 IS b 32 CV n --------------------- 3 8 5 4 6 6 41 — 30 54 124 63 114 68 5 4 --- ----------- ----- ----- 125 202 251 338 236 200 230 290 431 246 3 10 3361 5 9034 349 4 6 20265 3 10 5 4 6 3 10 5 4 211 52 45 71 31 17 5484 5111 3975 712? 5414 10 63 92 81 179 3 10 153 a 283 abc 264 ab 344 be c 405 3 10 6 4 6 n 80 lOl 51 3 10 6 4 6 2 36 51 40 55 CV 2 21 37 25 65 d 46 cd 27 c x kg/Ha 350—400 cm 300-350 cm cm ----- ----- 4 6 3 10 5 4 6 28 3 8 21 34 88 56 87 as axa-w S S t 5 _ F 5 n 5 5 5 5 ' b ^ ' ' d I f ^ ^ t " T 5 t t ^ : r ^ o signif i5ar . t l y ~ d t F f ^ « n t 5 5 - D = S c = ; - , - = T t i p l S c o ^ p ^ T s o n ' t a ^ ~ a t _ « n a l p h a = 0 . 1 0 i», Rb=Rc±>icon sari os; Rt>B= Bandmd pbasm of Rubicon sarias, Ka=Kalkaska series; of Kalkaska sarios. ka on forest floor samples. l«vel 59 calcium phosphates and it is not clear whether significant differences found at these depths represent any differences in plant available phosphorus. Figure 3.9a presents a comparison of K contents between soil series phases without subsolum textural strata. When compared to the soil series examinations in Figure 3.7, series differences in acid extractable K contents were reduced in the 100-200 cm and 300-400 cm depths. The nutrient poor Gy series shows little distinction from the unbanded phases of either the Rb or Ka The series. inclusion of subsolum textural strata resulted in noticeable contrasts between nutrient contents in Rb and Ka 700 600 r-X 500 px /X 400 300 200 100 FF 50 100 150 Lower | X I G r a y I in g Figure 3.9a. 200 depth of | X X | R u b ic o n 250 350 300 400 50 c m acc umula tion < a Ik a s K a Forest floor total and soil acid extractable K contents grouped by unbanded soil series phases. 60 phases. The banded phases had greater nutrient contents than the unbanded phases in 71% (42/59) of the nutrient or OC/depth combinations for Rb, and for 76% (45/59) of the nutrient and OC/depth combinations for Ka. These contrasts were most pronounced for TKP, acid extractable K, Mg, and Ca in depths below 100 cm. of the Ka This trend of contrasts between banded phases series for K contents is shown in Figure 3.9b. Greater acid extractable K contents were found in the banded phase across all depths from 0 to 450 cm, extending both above and below the range of banding depths (150-300 cm ) . Significant differences were found at 100-200 and 300-400 cm depths. 1, 3 0 0 1, 100 900 700 600 500 300 200 100 0 FF 30 10 0 150 Lower I Figure 3.9b. y \ Kalfcasfca 200 d e p t h of 250 300 350 400 430 50 c m a c c u m u l a t i o n Banded Ka Forest floor total and soil acid extractable K contents grouped by banded and unbanded phases of the Kalkaska soil series. 61 Since K, Mg, and Ca are readily leached from soils they may show illuvial increases in deeper soil horizons. It was suspected that the greater nutrient contents in the banded phases, especially at depths greater than 100 cm, was due to the parent materials having a higher nutrient status rather than to illuviation. Although the Gy series in Figure 3.9a displayed a slight increase with depth, the unhanded Ka phase in Figure 9.b showed no increase with depth. Nutrient and Organic Carbon Variability by Phased Soil Series: Variability of nutrient and OC contents were redistributed when examined subsolum by textural phased strata soil in series. the The series inclusion taxonomic of unit decreased the variability among the soil series phases without banding. The Rb and Ka series displayed different variability patterns when phased by subsolum textural strata. The Rb phases nutrient and showed variability. an inconsistent pattern of OC The Ka phases had a more consistent pattern, with the banded phase having a greater variability than the unbanded phase for all nutrients and OC contents and depths except TKN at 50-150 and 300-400 cm and OC at 100-150 cm. Figure 3.10a displays K content variability series phases without subsolum textural strata. between soil When compared with the soil series evaluations in Table 3.4 it can be 62 110 100 go & W 80 .1 70 >» a « Za > 60 o /x /x 50 S 40 t 30 'o s ,_x /X XX /X XX XX XX 20 XX XX 10 0 FF 50 f XI Gray Iino Figure 3.10a. 100 150 200 250 300 350 400 450 Lower depth of 50 cm accumulation |XX) Rubicon Kalkaska Variability of forest floor total and soilacid extractable K contents grouped by unbanded soil series phases. observed that the variability was reduced when the influence of subsolum textural strata variability occurs from was removed. This the surface to 450 reduction in cm, but ismost evident in the 300-350 cm depth. Variability of acid extractable K contents was greater in the banded phase of the Ka series than in the unbanded phase as shown in Figure 3.10b. A pattern of increased variability in the banded phases at all depths was observed. Other nutrients and OC displayed less distinct contrasts between phases with and without subsolum textural strata. pattern of increasing variability occurred, A general reaching maximum at a depth of 250-300 cm and then declining. a 63 90 00 70 60 50 40 30 20 10 FF 50 100 150 200 250 3QO 350 Lower depth of 50 cm accumulation I X | Kalkaska Banded Ka Figure 3.10b. Variability of forest floor total and soil acid extractable K contents grouped by banded and unbanded phases of the Kalkaska series. Harradine (1949) found a decrease in variability with increasing stage of profile development in the surface 270 cm. The low variability of the Gy series as compared to the Rb and Ka series as presented in Figure 3.8 contradicts these findings. Harradine assumed that the variability was due to soil maturity or age. factors of material, soil and No mention was made of any of the other formation; topography climate, (Jenny 1980). organisms, The parent strength and intensity of banding were also suspected as influencing the variability. Ka series, In Figure 3.10b it can be observed that for the the phase with subsolum textural variable than the phase without strata. strata is more The Gy series has no occurrences of subsolum textural strata, which may be a major 64 factor in its low variability. When the variability of the unbanded Rb and Ka phases are inspected, the Gy series has a greater variability in five of the 10 depths (Figure 3.10a). The parent materials of all three mineralogy due to glacial activity, soils however, have a mixed the Gy series resulted from deposition of highly washed, sandy material from glacial meltwater streams or spillways. As a result the Gy series has an inherently low nutrient status when compared with the Rb and Ka series which may have till, morainal, lake plain materials among their horizons. subsolum textural strata are removed or If influences of from the Rb and Ka series, then patterns of variability with profile development generally agree with those of Harradine for the surface 250 cm (Figure 3.10a). The increased variability in the banded soil series phases may be due to differences in texture between the band and the surrounding soil matrix (banding 'strength') or to continuity of the banding. The strength of banding is related to the thickness of the band and its texture. defined bands with striking textural Some sites had well differences from the surrounding soils, while others had poorly defined bands with only slight textural differences. The continuity of the banding was determined from the four bucket auger borings and relates to the areal extent of the banding on a site. ECS project 'banding intensity and continuity' On the was used to quantify strength and areal extent of the banding. Strength 65 was coded by a scale of zero to determined by averaging the bucket auger borings. codes. five and strength values continuity was from the four Table 3.6 lists the banding intensity Stands having mean banding and continuity indicies ranging from zero to five were found in both Rb and Ka banded phases. Figure 3.11 presents a comparison of acid extractable K contents by 50 cm depth accumulations across three stands with different banding intensity and continuity values. The soil pit descriptions are listed next to the figures along with horizon designations, thicknesses, and textures. It can be clearly seen that as banding intensity increases, so do acid extractable K contents in the region of the banding. Table 3.6. Banding intensity designations. 0 - No varves or textural stratifications. 1 - Thin varves or LS texture, but no bands. 2 - Bands of SL < 5 cm thick. 3 - Bands of SL between 5 & 15 cm thick or, bands of SCL < 10 cm thick. 4 - Bands of SL > 15 cm thick. 5 - Bands of SCL > 10 cm thick. Designations were determined for each bucket auger. Stand level banding intensity was average of four bucket auger intensities. 66 3.11a Gv Series; Banding Intensity = 1.25 Horizon Depth Field (cm) Texture •00 •00 A AB Bwl Bw2 BC Cl C2 C3 C4 C5 3.11b 0-3.5 3.5-9 9-34 34-47.5 47.5-70 70-230 230-260 260-375 375-410 410-450 LS LS S S S S CS CS CS FS 700 b •00 1 300 t s 8 I 400 300 100 0 900 930 300 330 400 ’ Daotn of 30 c« Accumulation Rb Series; Banding Intensity = 2.00 Horizon Depth Field (cm) Texture AE BE Bsl Bs2 Bs3 Cl C2 C3 C4 C5 0-5.5 5.5-13.5 13.5-31 31-91 91-131.5 131.5-210 210-275 275-350 350-375 375-450 3.11c S c’ MS LS FS FS FS MS SL S S S 200 330 300 330 400 * Oaotn of 30 cm Accumulation Rb Series; Bandinc Horizon Depth Field (cm) Texture A E Bsl Bs2 BC Cl C2 C3 C4 C5 C6 0-7.5 7.5-26 26-46.5 46.5-70 70-97.5 97.5-180 180-240 240-310 310-340 340-420 420-450 Figure 3.11. MS MS MS MS FS S LS SiL LS MS S 30 100 130 900 930 300 330 400 430 Lomr Daotn of 30 an Accusation Pit descriptions and acid extractable K contents for three stands across a range of banding intensities. 67 Figure 3.11a presents the Gy series K contents by depth. There were no textural strata found on this site, however, there is a slight increase in K contents with depth until 250 cm. Since the effects of weathering decrease with depth and the parent materials of this soil series are uniform, any increases are likely due to illuviation or residual organic materials horizon from roots. descriptions Figure for intensity and continuity. a 3.11b Rb soil shows with K contents moderate and banding The band has a sandy loam texture and extends from 210-275 cm. The surface 0-200 cm K contents are similar to those found in Figure 3.11a, however, there are sharp increases at depths of 200-300 cm. K contents found at these depths are nearly double that found in the these depths in the Gy series. Figure 3.11c displays a intensity and continuity. Rb soil with a strong banding On this site a silt loam band at 240-310 cm is sandwiched between sandy loam bands at 180-240, and 310-340 cm. A dramatic increase in K contents is evident at 250-300 cm coinciding with the depth of the silt loam band. Figures 3.11a, b, and c show proportionate increases in K contents with increasing banding intensity. Although Figure 3.11a suggests illuvial increases with depth, Figures 3.11b and c do not distinguish between illuvial increases and those due to the inherent fertility of the parent materials. CONCLUSIONS It is provide acknowledged all the considered for ecosystems. soil information classification systems. that topographic, that chemical needed data for may useful not site Pregitzer and Barnes (1984) reported vegetational and soil factors need to be successful classification of upland The sample size of this study was too small to adequately examine nutrient contents by ecological units which combined vegetation and soil factors. Successful site classification tools for forest management. system must take into systems are important In order to be successful, account factors accurately predict forest growth. which will a most How the soils information will be mapped can be crucial to the accuracy and utility of derived forest land inventory systems. mapped by soil series. Soils are commonly A pivotal shortcoming of Soil Taxonomy is that many facets of the landscape are not incorporated in soil taxonomic classifications. lowest level inclusion of of classification, phases of soil important landscape features or degree of This erosion) characterization to the the series (e.g. which series 68 is true even at the soil allows series. recognition The of subsolum textural strata would level be alone. overlooked by In addition, 69 description of the soil landscape is incomplete unless data on impurity of soil mapping units distinctness of soil boundaries are (heterogeneity) and included and (Hole Campbell 1985). A continuum found. TKN, of nutrient distributions with depth were TKP and OC displayed exponentially decreasing contents with depth, acid extractable K was fairly uniform in its distribution across depths, and acid extractable Mg and Ca displayed patterns of exponential increase with depth. When the soils were grouped by soil series the Gy series had consistently series. lower contents than either the Rb or Ka CV's for the Gy series generally increased with depth to 450 cm, while the Rb and Ka series displayed patterns of increasing variability to 200-350 cm, then decreasing. This was attributed to the influence of banding. The results reported in this chapter support the use of banding phases to produce homogeneous groupings increase the detection of group differences. banding phase for the Rb and Ka series and to Evaluation by resulted greater differentiation, with higher nutrient contents being found in the banded phases in 75% combinations. significant. Fifteen (89/118) of the 89 of the nutrient-OC/depth banding increases were Eleven of the 15 significant banding differences were found between the Ka series phases. Banding strength and continuity were thought to be the reason that variability of the unbanded phases was consistently lower than that of the 70 banded phases. phases at Increases in the variability of the banded 200-300 cm depths coincides with depths where subsolum textural strata occurred. Subsolum variability of banded soil phases strikingly different patterns than were found reported elsewhere) that of by for the solum. displayed (or have been The general pattern was variability increase to about 300 to 350 cm, followed a decline. Inclusion of a banding phase reduced the variability and would allow a more precise estimation of the site nutrient resource than use of only the soil series. Stand yield showed no significant differences when grouped by soil series, however, when the banding phase was introduced significant differences were found. The link between the nutrients contained in these bands and stand growth will be examined in the following chapter. This study research. phases the provides several directions for future To further reduce the variability of the banded intensity and continuity of strata need to be further investigated. subsolum textural Also, a consistent definition of banding intensity needs to be established. sample size examination of of this study banding questions such as: was strength too and small to allow continuity and The closer answer "Is a site banded if a single occurrence of strong banding is found on that site?", or "Would three occurrences of weak banding indicate a higher site quality than two occurrences of strong banding?". 71 Another direction for research involves utilization of the phased soil series evaluations with vegetational and ecological factors in an ecological classification system. Continuing work with ECS projects by MSU and the USDA Forest Service could provide such an evaluation. The results from partitioned nutrient variabilities more Soils evaluated a this study contents satisfactorily 450 cm found and than potential soil series nutrient soil tree series phases content alone. rooting depth accounted for more variability and, therefore would allow a more precise estimation of site nutrient resources than surface soil features, and should be considered in soils with sandy surface textures and similar subsolum features. The use of phased soil series in forest land evaluation systems has a potential value that should not be ignored. ACKNOWLEDGEMENTS The Ecological Classification System Project was funded by the USDA Forest Service. Special thanks are extended to Mr. Scientist, David T. Cleland, Soil Huron-Manistee N.F. whose tireless efforts in behalf of the ECS project allowed this research to be undertaken, and to Dr. James B. Hart Jr. for his ideas, reviews and comments. The efforts of Dr. Carl W. Ramm for calculation of stand volume and his statistical consultations are greatly appreciated. Zak, Dr. George Host, and Mr. Thanks to Dr. Donald Steve Westin for their assistance in sample collection, and to the many individuals who assisted in field work and laboratory analysis. Special appreciation is extended to Dr. Phu V. Nguyen for his invaluable advice and assistance. 72 LITERATURE CITED Adams, J.A. 1973. Critical soil magnesium for Radiata pine nutrition. N.Z.J. For.Sci. 3:390-394. Armson, K.A. 1977. Forest Soils: properties and processes. University of Toronto Press. Toronto. 390 p. Broadfoot, W.M. 1969. Problems in relating soil to site index for southern hardwoods. For. S c i . 15:354-364. Carmean, W.H. 1975. Forest site quality evaluation in the United States. Adv. Aqron. 27:209-269. Carmean, W.H. 1979. Soil and site relations for northern hardwoods. In:Proceedings of joint convention of SAF and Can. Inst, of For. p. 324-328. Coile, T.S. 1937. Distribution of tree roots Carolina piedmont soils. J . F o r . 35:247-257. in Coile, T.S. 1952. Soil and the growth of forests. Aqron. 4:329-398. North Adv. Comerford, N.B., G. Kidder; and A.V. M o n i t o r . 1984. Importance of subsoil fertility to forest and non-forest plant nutrition, p.381-384 In: Forest soils and treatment impacts. Proc. of Sixth N. American Forest Soils Conf. Tnnr t w uiiw 1 OOO TT» VUXV t •! w W ^ -P 11^ I Ml WAVl l l C i UKZfc/ U • - r Ui. Forestry, Wildlife and Fisheries. Dunn, O.J. 1964. Multiple comparisons using rank sums. Technometrics 6:241-252. Esu, I.E. and D.F. Grigal. 1979. Productivity of quaking aspen (Populus tremuloides Michx.) as related to soil mapping units in northern Minnesota. Soil Sci. Soc.Amer. J. 43:1189-1192. Fortescue, J.A.C. 1986. Environmental Geochemistry; A Holistic Approach. Ecological Studies; v. 35. Springer-Verlag. New York. 347 p. 73 74 Grigal, D.F. 1984. Short coinings of soil surveys for forest management, p. 148-166 In: Forest Land Classification: Experience, Problems, Perspectives. Proceedings of the Symposium, March 18-20, 1984. Madison WI. Hannah, P.R. and R. Zahner. 1970. Non-pedogenic texture bands in outwash sands of Michigan: Their origin and influence on tree growth. Soil Sci. Soc. Amer. Proc. 34:134-136. Harradine, F.H. 1949. The variation of soil properties in relation to stage of profile development. Soil Sci. Soc. Am e r . Proc. 14:302-311 Hart, J.B. Jr., Variation in outwash soil. A.L. Leaf, and S.J. Stutzbach. 1969. potassium available to trees within an Soil Sci. Soc. Amer. Proc. 33:950-954. Hole, F.D. and J.B. Campbell. 1985. Soil Landscape Analysis. Roman and Allenheld. N.J. 191 p. Hollander, M. and D.A. Wolfe. 1973. Non-Parametric Statistical Methods. John Wiley and Sons, New York. 502 p. Jenny, H. 1980. The Soil Resource: Origin and Behavior. Ecological Studies v. 37. Springer-Verlag. New York 377 p. Kimmons, J.P. New York. Leaf, A.L. 1 hivxx 1986. Forest Ecology. 531 p. 1956. -P V X Macmillon Inc. Growth of forest plantations on different TP 4 1 -»•»-> .>4 1 XllXUllUt TP I 'W i . » 4 fcJWX • *■> . *■» *■% *1 lo/r Ct 9 Leaf, A.L. 1958. Determination of available potassium in soils of forest plantations. Soil Sci. Soc. Amer.Proc. 22:458-459. Mader, D.L. 1963. Soil variability A serious problem in soil site studies in the northeast. Soil Sci. Soc. Amer. Proc. 27:707-709. Mengel, K. and E.A. Kirkby. 1982. Principles of plant Nutrition. Third Ed. International Potash Institute, Bern, Switzerland. 655 p. Metson, A.J. 1974. Magnesium in New Zealand soils. Jour, of Exp. A a r . 2:277-319. N.Z. 75 Mollitor, A. V. A.L. Leaf, and L.A. Morris. 1980. Forest Soil variability on northeastern flood plains. Soil Sci. Soc. Amer. J. 44:617-620. Page, A.L. ; R.H. Miller, and D.R. Keeney (Eds.) 1982. Methods of Soil Analysis Part 2: Chemical and Microbiological Properties. #9 Agronomy series, American Society of Agronomy. 2nd Ed. 1159 p. Pawluk, S. and H.F. Arneman. 1961. Some forest Characteristics and their relationship to Jack growth. For. S c i . 7:160-173. soil pine Pregitzer, K.S. and B.V. Barnes. 1984. Classification and comparison of upland hardwood and conifer ecosystems of the Cyrus H. Me Cormic Experimental Forest, upper Michigan. Can. J. of For. R e s . 14:362-375. Shetron, S.G. 1972. Forest site productivity among soil taxonomic units in northern lower Michigan. Soil Sci. Soc. Amer. Proc. 36:358-363. Soil Conservation Service. Soil Survey. USDA. 1976. Rubicon series. Soil Conservation Service. Soil Survey. USDA. 1981. Kalkaska series. Soil Conservation Service. 1982. Coop. Soil Survey. USDA. Grayling Nat. Coop. Nat Coop. series. Spurr, S.H. and B.V. Barnes. 1973. Forest Ecology. Ed. John Wiley and Sons. New York. 571 p. Nat. Second Strommen, N.D. 1974. Climate of Michigan by stations, 1940-1969. Michigan Department of Agriculture and Michigan Weather Service in cooperation with NOAA. US Department of Commerce. Technicon Industrial Method. 1977. Individual / simultaneous determination of N and/or P in BD acid digests. Method No. 334-74W/B. Technicon Industrial Systems, Tarrytown, NY. Veatch, J.O. 1953. Soils and Land of Michigan. State College Press. East Lansing. 241 p. Michigan White, D.P. and R.S. Wood. 1958. Growth variations in a Red Pine plantation influenced by a deep-lying fine soil layer. Soil Sci. Soc. Amer. Proc. 22:174-177. 76 White, E.H. and A.L. Leaf. 1964. Soil and tree potassium contents related to tree growth I. HN03-extractable soil Potassium. Soil S c i . 98:395-402. Wilde, S.A. and A.L. Leaf. 1955. The relationship between the degree of podzolation and the composition of ground cover vegetation. Ecology 36:19-22. Wilde, S.A. and H.F. Scholz. 1934. Subdivision of the upper peninsula experimental forest on the basis of soils and vegetation. Soil S c i . 7:383-399. Wilde, S.A.; G.K. Voigt, and J.G. Iyer. 1972. Soil and Plant Analysis for Tree Culture. 4th Ed. Oxford and IBH New Delhi 172 p. Youngberg, C.T. and H.F. Scholz. 1950. Relation of soil fertility and rate of growth of mixed oak stands in the driftless area of southwestern Wisconsin. Soil Sci. Soc. Amer. Proc. 14:331-332. Zinke, P.J. 1960. Forest site quality as related to soil nitrogen content, p.411-418 In: 7th Int. Congress of Soil Science. Madison, WI. Chapter 4. Relationships of Soil Fertility and Stand Growth in Glaciated Michigan Landscapes INTRODUCTION This study jointly examines several topics which have not been previously considered together in soil-site studies on oak and northern hardwood stands in. the lake states. topics include examination textural strata, or bands, of the influence of These subsolum on stand growth; the use of acid extractable nutrients as a measure of site nutrient resources; and a soil sampling scheme which collects soils throughout the rooting depth of the forest overstory. site fertility contents magnesium, (kg and ha’1) calcium, stand of growth nitrogen, Relationships between were examined phosphorus, utilizing potassium, and organic carbon accumulated by 50 cm depths as measures of soil fertility; and mean annual volume increment (MAI) to represent stand growth. The yield of forest products and therefore, the management of forest climatic, stands are influenced by the sum of edaphic, and biological factors which affect its long-term ability to accumulate biomass. referred to as site quality. This sum of factors is often Increasing demand for forest products and a decreasing land base for timber production have led to a situation where a more agronomic philosophy of yield maximization is being applied to forestry. This philosophy 79 is evident in more intensive forestry practices such as short rotation forestry and whole tree harvesting. and productive when sites forest management techniques can be applied are grouped into classes of similar management potential on the basis of stand growth. of stands More efficient (e.g. disturbed stands, Increasing numbers uneven-aged stands, and stands in which the desired species is not present) also occur in which direct measurements of stand growth cannot be made. Use of soil-site relationships allows estimation of stand growth on these sites (Carmean 1975, Coile 1952). Soil-site examinations have had a long and useful history determining site quality and resulting forest growth, and in conjunction ecological factors with examination (Broadfoot 1969, of alone vegetative Cajander 1926, and Carmean 1975; Hannah and Zahner 1970, Leaf 1956, Lowry 1970; Pawluk and Arneman 1961, Pregitzer and Barnes 1984, Wilde and Scholz 1934). Many soil-site studies have examined the associations of features such as surface soil depth, slope with stand growth goals of prediction collected these of and (Carmean 1975). soil-site stand subsoil texture and growth analyzed. examinations with However, factors these This reflects the which which were are factors the easily are only indirect indices of more basic growth factors such as nutrient and moisture availability. The practical aspect of prediction was valued more than the inclusion of more fundamental growth factors. Broadfoot (1969) cautions against using variables 80 which are effects rather than causes of tree growth and he attributes the low accuracy of his soil-site study of seven southern hardwood species to this. There is no agreement whether soil physical or chemical properties provide better predictions of stand growth. Some researchers have found soil fertility (expressed as nutrient concentrations (ppm)) to have a more stand growth than soil physical 1970; Pawluk and Arneman 1961, limited influence on properties (Leaf 1956, Lowry Youngberg and Scholz 1950). Relationships between stand growth and site fertility were found to be stronger when soil fertility was expressed as nutrient contents instead of nutrient concentrations (Hart et a l . 1969, Mader and Owen 1961; Viro 1961, White and Leaf 1964; Youngberg and Scholz 1950, Zinke 1960). Zinke (1960) a positive correlation between nitrogen (N) contents found in the top 1.2 m (4 ft) of the soil and site index of ponderosa pine (Pinus ponderosa Laws.). thick soil layers Potassium contents of 30 cm (1 ft) to a depth of 210 cm (7ft) were found to be significantly different from one another, to red 1964). pine (Pinus resinosa Ait.) growth and correlated (White and Leaf Viro (1961) suggests that site nutrient contents may be more appropriate for reporting soil chemical properties in soil-site studies. Researchers growth have associated sandstone strata, noted with significant variations subsolum groundwater, soil and in stand features fine textured such as strata 81 including bands or lenses of fluvial materials (Comerford et a l . 1984, White and Leaf 1964, White and Wood 1958; and Wilde and Leaf 1955). For soils with variable subsolum properties the examination of only the top 30 cm may not be sufficient for estimating stand growth. This is especially true for deep rooted species such as red and white oak which have effective rooting depths of 450 cm (15 ft) or more in sandy soils (USDA, 1965). In their evaluation of oak sites in southern Michigan, Gysel and Arend (1953) found that moist subsolum strata at depths of 120-300 cm (4-10 ft) were important factors in oak site productivity. The importance of glacial outwash soil in northern New York. adjacent shown sites on a in deficient on was fertility of plantations growth soil estimation pine stand subsolum potassium displayed a two the (K) Red fold difference in volume growth after seven years of equally poor growth (White and Wood 1958). A band of fine textured soil was discovered at a depth of 180 cm (6 ft) and 270 cm (9 ft) on the good and poor sites, respectively. The authors attributed the growth difference to the greater K and water supplying abilities of the band within reach of the roots. Subsequent studies revealed that of 91-213 cm acid extractable K at depths (3-7 ft) had the highest correlation to total tree height (White and Leaf 1964). On sandy glacial outwash soils it in increase northern lower Michigan in growth of red oak was found that the (site index and mean annual 82 diameter growth) on soils which had subsolum textural bands within the rooting range was highly significant as compared to unbanded soils with similar stand characteristics4. The growth response of red oak was greater than that of sugar maple, presumably because of a more efficient exploitation of the subsolum bands by the deeper rooted red oak. Few soil-site studies have been conducted northern hardwood stands in the lake states Pregitzer and Barnes 1984, Shetron 1972; on oak and (Carmean 1975, Wilde and Scholz 1934, Wilde et al. 1948; Youngberg and Scholz 1950), and none have examined the contributions of soil fertility beneath the surface 150 cm to stand growth. OBJECTIVES The objectives of this study were to determine if there is an association between soil fertility and stand growth; and if there is an association, to determine the contributions of soil fertility to stand growth of upland oak and sugar maple stands in northern lower Michigan. find the strongest set of linear site fertility association Other objectives were to variables which with stand growth has the and 4 D.T. Cleland, Deep Bands and Forest Growth on Kalkaska Sands. Unpublished Report. to 83 determine whether soil sampling to a depth of 450 cm, which includes subsolum banding, has a stronger association with stand growth than sampling only the surface soils to a depth of 45 cm. Specific hypotheses to be tested were: 1) H0: r(fertjlity & HAI) = 0 H1: Not H0, and 2) Hg. ^(fertility below 45 cm & H A D — ® H1: Not Hg. A final objective was to use information gained above to make recommendations on soil sampling depths for forest land inventory programs. MATERIALS AND METHODS Study Area: Study plots were located in the Manistee National Forest in the northern lower peninsula of Michigan between 85°30' and 86°151 west longitude (Figure 4.1). and 45°52' and 44°30' north latitude The climate alternates between continental and semi-marine depending on the direction of individual weather patterns. The 29 year (1940-1969) mean annual precipitation was 821 mm (32 in) and the 29 year mean annual temperature was 5.8°c (42.5°F) on the sites (Strommen 1974). (Rubicon, The soils series encountered Kalkaska, and Grayling) were all of Wexford Co. Mason Co Figure 4.1. Location of sample stands. 85 Wisconsinian age (Soil Conservation Service 1976, 1981, 1982). The Grayling soil is a Mixed, frigid, Typic Udipsamment? the Rubicon soil is a Sandy, mixed, frigid Entic Haplorthod; and the Kalkaska soil is a Sandy, mixed, frigid Typic Haplorthod. All soils were well to excessively drained and were formed from sandy glacial drift parent materials. The more productive soils developed in or were underlain by till plain, ground moraine, and sandy loam or finer. lake plain materials with textures of Stand growth increases along a gradient of Grayling, Rubicon, Kalkaska. Thirty stands stratified productivity were sampled. by low, moderate, and high Stands occupied by pioneer species or showing evidence of disturbance in the past 40 years were excluded from sampling. The minimum sampled was 18.36 m2ha’1 (80 ft2 ac’1) . basal area of stands Other stand selection criteria were that the stand must be at least 55 years old, the stand must be normally and uniformly stocked, and that the topography must conditions. be uniform and representative of upland Tree species found on the stands were sugar maple (Acer saccharum Marsh.), white ash (Fraxinus americana L.); red maple (Acer rubrum L . ) , red oak (Ouercus rubra L . ) ; white oak (Ouercus alba L.) , and the black oak group (Ouercus spp.) . Sample Collection: Samples were collected during the summer of 1983 in 86 conjunction with an ecological classification system (ECS) project being carried out jointly by the USDA Forest Service and Michigan State University. stand. The sampling unit was the At each randomly selected stand a uniform area of one hectare or more was selected for sampling. Four randomly located points per stand were measured for overstory growth and yield. Two sets of soil samples were collected, the first was collected at the soil pit to a depth of 450 cm, and the second was collected to a depth of 45 cm. Forest floor samples frame at six points were collected with a 30x30 cm metal (three at the pit and one each at three additional points) and systematically separated into litter, or fermentation and samples (Oi horizon) leaf material not humus of (Figure 4.2a). Litter included recognizable and nearly entire affected Fermentation and humus consisted layers finely by decompositional layer samples divided processes. (Oa and Oe horizons) decomposed organic materials extending to the upper boundary of the mineral soil. A soil pit about 150 cm in depth was center of each stand (Figure 4.2b). located near the After pits were dug the soils were classified to the series level and equal volumes of soil samples, were collected from each horizon. Bucket auger stratified by horizon, were collected beginning at the bottom of the pit and terminating at 450 cm (Figure 4.2c) . Bulk density samples were collected to allow conversion of nutrient concentrations to nutrient contents. 87 b.) Bhs Pit bottom a. 450 cm 45 cm Figure 4.2 Detail of forest floor and soil sampling^ a. forest floor and surface soils, b. soil pit, c, bucket auger. 88 Surface soil samples were collected directly beneath the forest floor samples. These were divided into surface (A and E horizons) and subsurface (upper B horizon to 45 cm. depth) layers (Figure 4.2a) and composited for analysis. Stand Growth Measurements: Forest growth was measured at four random points per stand using a 10 basal area factor prism. At each sample point all trees with a diameter greater than 8.89 cm (3 in) were measured height, for diameter at and merchantable height breast to a height 10.2 cm (DBH), total (4 in) top. Increment cores were taken from two trees per sampling point, across the range of DBH values, to obtain total age. Stand means were obtained by averaging measurements from the four prism points. Mean annual volume increment (MAI) was computed from the inventory data and was chosen as the site quality variable by which to express stand growth. Laboratory Analysis: Forest floor samples were oven dried at 70°C, weighed, ground, and subsampled prior to analysis. Kjeldahl nitrogen (TKN) Forest floor total and total Kjeldahl phosphorus (TKP) digests were determined after sulfuric acid digestion (Page 89 et a l . 1982) using a Technicon5 Auto-analyzer II and Technicon procedures (Technicon, 1977). Total cations (K, Mg, and Ca) were determined by dry ashing at 500°C for 4 hours with ash dissolution in 8 N HCl (Wilde et al. 1972) . Nutrient concentrations were converted into contents by use of areas and weights of the forest floor samples. Soil samples were air dried, sieved screen, and subsampled for analysis. to pass a 2 mm Both coarse (>2 mm) and fine fraction weights were recorded for each sample and used to calculate horizon contents corrected for coarse fragment contents. Commonly used agronomic soil testing procedures in which mild extractants are used to quantify 'available' nutrients have been found to be poorly related to tree growth (Adams and Boyle 1982, 1964). Leaf 1958, Thompson et al. 1977, White and Leaf Tree roots have been found to assimilate nutrients directly from unweathered soil minerals by contact feeding (Boyle and Voigt 1973, Viro 1961) and exploit the nutrient resources of a site for decades at a time. therefore, meaningful that more estimates of caustic site It is reasonable, extractants nutrients will important derive to tree productivity. Acid extractions are thought to represent cations which replenish elements depleted by plant uptake and leaching 5 Use of trade names does not constitute an endorsement by either Michigan State University or the USDA. 90 (Page et al. of Mg 1982) . release extracted by therefore, a Metson from a soil (1974) reserves moderately good concluded that the rate was strong guide to correlated acid attack available Mg in with and Mg was, continuous cropping or forest systems. A 2 g soil sample was boiled for 10 minutes with 10 ml of 0.5 N nitric acid to extract potassium (K) , magnesium (Mg), and calcium (Ca) . This extraction has been shown to associated with tree K uptake and growth in red pine 1958, White and Leaf 1964), levels in radiata pine and with critical be (Leaf foliar Mg (Pinus radiata D. Don)(Adams 1973). Since Mg and Ca are usually extracted concurrently from soils and have many chemical properties in common (Page et al. 1982), available calcium was also determined by this method. Cation concentrations (K, Mg, and Ca) of the boiling nitric acid extractions were determined by plasma emission spectroscopy. The extracts were brought to a concentration of 1,000 ppm LiCl to stabilize the plasma and interference. to suppress TKN and TKP were determined after sulfuric acid digestion similar to the forest floor samples. Percent organic carbon (OC) in the soils was determined by the Walkley-Black method using 0.5 N ferrous sulfate and 1 N potassium dichromate (Page et al. 1982). Sample size varied from 1-5 g to allow greater accuracy in samples with high and low organic carbon concentrations. Horizon concentrations were converted into nutrient 91 contents per horizon using concentrations, densities, and horizon depths. soil bulk The contents were accumulated by 50 cm depths to circumvent problems with differences in horizon thickness. Statistical analyses were performed with the stand as the sample unit, therefore, forest floor and surface soil samples were averaged by stand. All chemical analyses were performed in the MSU Forestry Department forest soils laboratories using 10% sample replication, National Bureau of Standards specimens, and bulk sample analysis to insure precision and accuracy. Statistical Analysis: Before regression analysis was performed several exploratory data analyses were performed on the data. Variables were checked to verify compliance with regression assumptions of linear X-Y relationships between independent and dependent variables and normal distributions of the dependent variable (Draper and Smith 1981) . Scatterplots were generated plotting site fertility variables against MAI to inspect variable linearity (MAI) of was the tested relationships. for normality The by a dependent Chi-square goodness of fit test. There are no "best" methods for choosing among several independent variables to find the regression equation which reveals the strongest association with the dependent variable 92 (Draper and Smith 1981). Variable selection was complicated by the fact that the number of independent variables exceeded the sample size for the pit soil fertility samples. Variable reduction was accomplished in several stages. First, any variables linear associations not in displaying linear scatterplots were or transformed discarded. Second, the correlation matrix of the remaining variables was examined and variables not having a significant r value with MAI were removed. Any pair of fertility variables whose correlation was greater than r = 0.80 were noted and not entered together in a regression due to multicollinearity effects (a problem in multivariate regressions when there is more correlation among independent variables dependent variables remaining multiple soil between (Draper and Smith fertility regression than variables equation and independent 1981)). were Third, entered variables not into and the a having significant regression coefficients (at an alpha = 0 . 1 0 level) were removed. In addition to multiple regression analysis on linear and transformed linear soil fertility variables, regressions were performed on standardized soil fertility variables to generate partial regression coefficients which allow direct evaluation of relative variable contributions to the regression (Netter and Wasserman 1974). soil fertility When r values > 0.80 existed between variables, the variable with the greater contribution to the regression (as determined by adjusted R2, 93 SEE, and partial regression coefficient values) was chosen. After the regressions were formulated, residuals were examined for any specific violation of the regression assumptions (Draper and Smith 1981, Steel and Torrie 1980). Principal components analysis (PCA) has the ability to overcome many of the limitations of correlated variables. The principal components generated by PCA are uncorrelated linear functions of the between the original variables, principal and multicollinearity components (McDonald 1980, Morrison 1976). (PC's) is eliminated PCA is effective at variable reduction and can reveal hidden factors which account for a large percentage variables. of the variation in the independent (McDonald 1980, Morrison 1976). Criteria for selecting principal components from the set of generated PC's were determined, a priori. to be all PC's with Eigenvalues greater than 1, or that set of PC's which explains at least 80% of the variation in site fertility. The correlation matrix was employed in the PCA because its use is more appropriate in data sets where some variables have standard deviations of a higher magnitude, as was the case in this investigation. were used components to for The Eigenvectors developed by the PCA generate each a set stand. regression analysis with MAI. of These standardized PC's were principal used in a RESULTS AND DISCUSSION Exploratory Data Analysis: When soil fertility variables of the pit samples were plotted against MAI to examine linear associations, of the pairings displayed non-linear associations. several Sigmoidal transformations improved the linearity of soil fertility and MAI associations. variables and Scatterplots for the surface soil fertility MAI also displayed non-linear associations, however, they had less scatter than the pit samples. the pit samples, As with sigmoidal transformations made many of the pairings more linear. The results of a Chi-sguare goodness of fit test on MAI found that the null hypothesis of a normal sample distribution could not be rejected (Steel and Torrie 1980). Soil fertility relationships with stand growth: Soil sampling to 450 cm Of the 64 possible soil fertility variables, 48 exhibited linear associations with MAI 94 or could be transformed to 95 variables with linear associations and are presented in Table 4.1. The collected sample size was too small to analyze this variable pairs set by multiple of the remaining regression methods. variables had values greater than or equal to 0.80. Twenty correlations five with r Of each pair of highly correlated variables the one with the highest correlation to MAI was first chosen for the regression equation. This resulted in a set of 17 variables for the initial regression analysis (Table correlated variables 4.1). variables with linear The (25), discrepancy variables associations between selected (48) was highly (17), because and some variables were correlated with more than one other variable. A series of regressions were performed, being examined before the subsequent the results of one regression was run. Highly correlated variables were randomly substituted for one another and determine adjusted the coefficients better were R2 and SEE overall examined at values were regression. each stage utilized to Regression for statistical significance and variables with non-significant coefficients were discarded. Evaluation by adjusted R2, residual mean square, and SEE values resulted in the selection of a regression equation with seven fertility variables presented in Table 4.2. of the residuals assumptions. revealed no violations of Examination regression Partial regression coefficients identified total K contents in the F&H layers as having the greatest 96 Table 4.1. Reduction of forest floor and pit variable size for regression analysis. Initial Variables N-La N-F&H N-50 N-100 N-150 N-200 N-250 N-300 N-350 N-400 N-450 Mg-L Mg-F&H Mg-50 Mg-100 Mg-150 Mg-200 Mg-250 Mg-300 Mg-350 Mg-400 Mg-450 P-L P-F&H P-50 P-100 P-150 P-200 P-250 P-300 P-350 P-400 P-450 Ca-L Ca-F&H Ca-50 Ca-100 Ca-150 Ca-200 Ca-250 Ca-300 Ca-350 Ca-400 Ca-450 N-L N-F&H N-50 N-100 N-150 N-200 N-350 P-L P-F&H P-50 P-100 P-150 P-200 P-250 P-300 P-350 P-400 P-450 K-L K-F&H K-50 K-100 v— i cn V— T j-j K-F&H K-50 K-100 K-150 K-200 K-250 K-300 K-350 K-400 K-450 Variables with Linear or Transformed Linear associations OC-50 OC-lOO OC-150 OC-200 OC-250 OC-300 OC-350 OC-400 OC-450 Mg-L Mg-F&H Mg-50 Mg-100 Mg-150 Mg-200 Mg-250 Mg-300 Mg-350 Mg-400 Mg-450 Ca-L Ca-F&H Ca-50 Ca-100 Ca-150 Ca-200 Ca-250 Ca-300 Ca-350 Ca-400 Ca-450 w Preliminary Regression Variables N-F&H N-200 P-L P-150 P-200 P-350 K-F&H K-50 K-100 K-150 K-350 Mg-100 Mg-200 Ca-50 Ca-100 Ca-200 Ca-250 a c r\ n w K-200 K-350 K-400 “ - L = Litter organic layer; F&H = Fermentation and Humus organic layers; Number following nutrient symbol is lower depth of 50 cm accumulation. 97 Table 4.2. Regression summaries for forest floor and pit nutrient contents accumulated by 50 cm depths and descriptive statistics. Ad j . R Standard Error of Estimate 0.86 0.357 Independent Variables Reg. Coeff. (Bf) N P P K K K K 0.314 0.304 -0.328 0.091 -1.6E-3 0.424 0.307 B0C = 0.759 150-2008 L 100-150 F&H 0-50 50-100 100-150 F Ratio = 24.74 S.E. of Reg. Coeff. Partial Reg. Coeff. 0.065 0.087 0.138 0.018 6.2E-4 0.105 0.091 0.424b 0.283 -0.262 0.526 -0.325 0.408 0.351 27 df Descriptive Statistics: Variable n Mean MAI 30 2.84 N P P K K K K 29 29 29 29 29 30 30 226 6 344 12 537 473 587 150—200a L 100-150 F&H 0-50 50-100 100-150 Minimum Maximum - m3 ha'1 yr'1 ______ 1.19 4.33 ---Kg ha'1 — 15 3 182 5 275 238 205 649 12 665 24 1138 776 2400 Standard Devia- 0.97 169 2 127 5 185 163 421 - Numbers represent depth of nutrient accumulation; L = litter layer; F&H = fermentation and humus layer. b - Partial regression coefficients derived from regressions utilizing standardized variables (z = (x - x bar)/s). c - B0 = Y intercept of regression equation; df = total degrees of freedom for regression 98 contribution of the variables entered, followed by TKN in the 150-200 cm soil depth and acid extractable K in the 50-100 cm soil depth. TKN, TKP, significantly and total, associated intermediate depths, or acid with extractable MAI. K were Accumulations in all the between the depths of 100 and 200 cm, were significant for all three nutrients. No variables beneath the surface 200 cm were entered into the regression. The strong association of total and acid extractable K contents agrees with published results finding significant associations with K contents and growth of southern hardwood and red pine stands (Broadfoot 1969, White and Leaf 1964). In the previous chapter it was found that acid extractable K began to show differences depth. in This content was textural strata. and variability especially true in at soils the 100-150 with cm subsolum The regression in Table 4.2 was interpreted as confirming the importance of intermediate soil depths in supplying nutrients for forest growth. In a previous examination of soil fertility distribution and variability it was found that high nutrient contents were associated with the presence of subsolum textural strata. was an expectation, therefore, that deeper soil fertility variables would be entered in the regression equation. regression data set subsolum textural included stands both with strata. It The and without The inclusion of stands without subsolum banding may have obscured the influence of strata 99 fertility on stand growth. As a check of this hypothesis the data set was divided into two fractions, those with and those without subsolum textural strata and simple linear correlations with MAI were calculated for both fractions. Correlations of MAI with total or acid extractable K contents are presented in Table 4.3 . The r values decrease rapidly below the surface 100 cm for the unbanded stands, while they remain strong at greater depths in the banded stands. Although the sample size was too small to perform separate regressions on banded and unbanded stands, the correlation patterns suggest that different regressions would be formulated. Table 4.3. Correlation analysis is sensitive to Correlations between MAI and forest floor total or soil acid extractable K in soils grouped by presence or absence of subsolum textural strata. Correlations with MAI Total and Acid Extractable K Accumulations Soils w/o subsolum textural strata n=12 Litter Layer Fermentation & Humus Layer 0-50 cm 50-100 cm 100-150 cm 150-200 cm 300-350 cm 350-400 cm Soils w/ subsolum textural strata n=ll 0.233 0.017 0.555 0.333 0.481 0.013 -0.113 -0.368 -0.109 0.336 0.430 0.695 0.571 0.161 0.469 0.593 100 sample size, therefore, the data in Table 4.3 are only meant to display general trends. Principal component analysis was performed on the set of 48 variables which displayed linear associations with MAI. The a priori selection criteria of 80% explanation fertility variance was used to select principal (PC's) from the set of generated PC's. of components This approach was adopted over selection by eigenvalues greater than one because it resulted in a smaller set of PC's for analysis. Eighty- two percent of the explained variance was consolidated in the first seven PC's. dimensionality variables. This represents a dramatic reduction in the of the data from the initial 48 fertility Standardized PC's were computed from the first seven eigenvectors. The standardized regression equation. regression PC's were entered into a multiple Elimination of PC's with non-significant coefficients left an equation with three PC's. Regression summaries for this relationship are presented in Table 4.4. This regression representation of the MAI was judged to be a poorer - fertility relationship with an adjusted R2 value 19% lower, and a SEE 56% higher than was obtained by direct examination of the pit fertility variables. Inspection of partial regression coefficients revealed that although the fourth PC explained only 8% of the variation in fertility (Table 4.5), it was a greater contributor to the regression than the second PC (which explained 15% of the 101 Table 4.4. Ad j . R Summaries for forest floor and pit principal component regression analysis. Standard Error of Estimate 0.67 Independent Variables Reg. Coeff. S.E. of Reg. Coeff. (B«) 0.556 PC #1 PC #2 PC #4 B0b = 2.846 0.168 0.083 0.207 Partial Reg. Coeff. 0.026 0.042 0.056 F Ratio = 20.79 0.678a 0.209 0.394 29 df a - Partial regression coefficients derived from regressions utilizing standardized variables (z = (x - x bar)/s). - B0 = Y intercept of regression equation; df = total degrees of freedom for regression fertility variation). Table 4.5 presents the eigenvectors significant PC's in the regression equation. of the three The first PC had low loadings for TKN contents at all depths except the organic fermentation nutrients and humus in the layers, litter layer. and The low loadings for all interpretation of the first PC was that of a balanced representation for the overall nutrient status of the stand. The second PC displays a pattern of decreased or more negative loadings with depth for all fertility variables except soil TKP, which has a uniform loading across all depths. The second PC was interpreted as representing the importance of the forest floor and surface 100 cm of the soil. The fourth PC had major loadings for TKN, 102 Table 4.5. Eigen values and vectors for principal components with significant regression coefficients: Forest floor and pit samples. 50 cm Nutrient accumulation Eigen Values % Variance Principal Component #1 17.6 37 Principal Component #2 Principal Component #4 7.3 15 4.1 8 Eigen Vectors: N-La N-F&H N-50 N-100 N-150 N-200 N-350 0.018 0.166 0.026 0.053 0.052 0.034 0.064 0.015 0.186 0.205 0.199 0.128 0.146 0.058 0.173 0.101 0.226 0.144 0.303 0.224 -0.025 P-L P-F&H P-50 P-100 P-150 P-200 P-250 P-300 P-350 P-400 P-450 0.054 0.180 0.113 0.122 0.150 0.170 0.122 0.143 0.185 0.187 0.185 -0.020 0.099 0.143 0.120 0.137 0.130 -0.026 0.024 0. 069 0.065 0.127 0.273 0.126 0.005 0.014 0.076 -0.019 -0.191 -0.252 _nV/ ■Xt o o -0.205 -0.205 K-L K-F&H K-50 K-100 K-150 K-200 K-350 K-400 0.047 0.200 0.132 0.130 0.141 0.118 0.169 0.183 0.147 0.127 0.056 0.056 -0.042 -0.096 -0.158 -0.154 0.112 -0.010 0.287 0.198 0.253 0.249 0.052 0.088 103 Table 4.5. Continued. 50 cm Nutrient accumulation Principal Component #1 Principal Component #2 Principal Component #4 Eigen Vectors: Mg-F&H Mg-50 Mg-100 Mg-150 Mg-200 Mg-250 Mg-300 Mg-350 Mg-400 Mg-450 0.152 0.117 0.157 0.198 0.192 0.172 0.173 0.152 0.160 0.144 0.229 0.135 0.082 -0.021 -0.064 -0.205 -0.201 -0.250 -0.234 -0.160 -0.059 -0.212 -0.127 -0.032 0.013 0.015 0.033 0.005 0.053 0.005 Ca-L Ca-F&H Ca-50 Ca-100 Ca-150 Ca-200 Ca-250 Ca-300 Ca-350 Ca-400 Ca-450 0.053 0.161 0.124 0.146 0.165 0.184 0.185 0.172 0.185 0.147 0.140 0.217 0.236 0.186 0.139 0.022 -0.031 -0.179 -0.182 -0.179 -0.219 -0.138 -0.029 0.019 -0.007 -0.168 -0.082 0.005 0.013 -0.014 0.017 -0.002 -0.015 OC-450 0.061 0.049 -0.224 a 11 L -r = Litter .■j __ organic layer; F&H - Fermentation and Humus a organic layers; Number following nutrient symbol is lower depth of 50 cm accumulation. 104 TKP, and acid extractable K. for the depths. litter layer, the The TKN loadings were greatest 0-50, 100-150; and 150-200 cm The prominent TKP loading was in the litter layer. Acid extractable K had its greatest loadings from soil depths of 0-150 cm. The loadings of PC #4 corresponded well with the fertility variables selected by the regression in Table 4.2 and was interpreted to represent the soil fertility resource at intermediate depths (50-200 c m ) . Plotting the PC's with the largest partial regression coefficients revealed a reasonable separation of banded and unbanded stands. Figure 4.3 shows that stands with subsolum textural strata occur above the diagonal line which nearly bisects the plot, while stands below this line lack subsolum textural fertility strata. Stands with bands have a high overall (indicated by larger values for PC #1) regardless of fertility in the intermediate depths (indicated by values for PC #4) . If the stands have a higher fertility in the intermediate depths (PC#4) then they are likely to be banded regardless of their overall fertility (PC#1). Soil fertility relationships with stand growth: Soil sampling to 45 cm The selection of the "best" regression proceeded in the same fashion with the surface soil samples as for the pit 105 CD HSCL a m n cx □ □ CD □ CD □□ □ CD - "T □ Q cr £! □ □ cx #1 id v D C X CD £) □ cl □ CD -t Vo- £) □ CD □ CCD ■a □ □ .Q cr □ jQ cr C\J I £> □ Principal Component mco wraro (0 V £3 d) < cr □ JD cr □ _ (d >N ID □ □ v I _ ID >s (J □ CD I irtt Figure 4.3. aueuodujoo iQdiDuiJd Plot of forest floor and pit principal components with strongest association with MAI. 106 samples, however, analysis. the fewer variables allowed a more direct Inspection of the MAI - fertility scatterplots found linear and transformed linear associations for 14 of the 22 possible fertility variables. Six pairs of fertility variables had correlations with r values greater than or equal to 0.80. Random substitution of highly correlated variables and examination of adjusted R2 and SEE values led to the final regression equation residuals revealed in Table no 4.6. violations Examination of the of the regression assumptions. Site fertility expressed by the surface soil samples (45 cm) represents only 10% of the soil volume represented by the pit samples (450 c m ) . The depth limitation of surface soil sampling may be balanced by their greater sample size which gives a better representation of landscape fertility than a single pit sample. The surface soil regression had a SEE which increased by 35% and an adjusted R2 which decreased by 12% when compared to the pit regression in Table 4.2. The surface soil regression had three fertility variables from the forest floor, of its four implying that the forest floor may reflect soil fertility of the intermediate depths in the (which were found to have strong association with MAI pit intermediate regression). soil depths Cycling to the of forest nutrients floor occurs from by vegetative nutrient uptake and subsequent deposition on the forest floor. The negative regression coefficient of total 107 Table 4.6. Ad j . R Regression summaries for forest floor and surface soil nutrient contents and descriptive statistics. Standard Error of Estimate 0.74 Independent Variables Reg. Coeff. (Bf) 0.481 P La K L K F&H Ca A&E B0C = 0. 747 0.666 -0.425 0.082 0.258 F Ratio = 21.04 S.E. of Reg. Coeff. Partial Reg. Coeff. 0.144 0.187 0.019 0.067 0.622b -0.329 0.472 0.436 28 df Descriptive Statistics: Variable n Mean Minimum — MAI 30 2.84 P La K L K F&H Ca A&E 29 29 29 29 5 6 12 603 m3 ha'1 yr’1 1.19 --- Kg ha'1 — 3 3 5 28 Maximum Standard Deviation 4.33 0.97 12 14 24 8146 2 2 5 1589 a -■1L ^ —= litter layer; F&H = fermentation and humus layer; & A&E = A and E horizons. b - Partial regression coefficients derived from regressions utilizing standardized variables (z = (x - x bar)/s). c - B0 = Y intercept of regression equation; df = total degrees of freedom for regression. 108 K in the litter layer may be a result of K not being released through decomposition to the F&H layer or the soil. extractable regression Ca in the coefficient A & E which horizons agrees had with a the Acid significant findings of Bowersox and Ward (1972) who found Ca in the A and E horizons (which were sampled separately) was an important contributor to oak site quality in Pennsylvania. TKP in the litter had the greatest contributions of the entered variables as judged by the high partial regression coefficient. Acid extractable K and Ca were about equal in their contributions. Principal fertility components variables analysis found that of 82% the of the surface variation fertility could be explained by the first four PC's. standardized analysis, PC's were subjected to soil multiple in When the regression only two had significant regression coefficients. The results of the surface soil PC regression are presented in Table 4.7. The PC regression had an increase in SEE of 19% a decrease in the adjusted R2 of 11 %as compared to the direct surface soil fertility regression (Table 4.6) and was considered to be a poorer expression of the MAI - fertility relationship. Inspection of the eigenvectors of the significant PC's in Table 4.8 revealed that the first PC had its major loadings from all nutrients in the organic fermentation and humus layers and the A and E mineral horizons, and Ca in the B 109 Table 4.7. Ad j . R Summaries for forest floor and surface soil principal component regression analysis. Standard Error of Estimate 0.63 Independent Variables 0.572 PC #1 PC #4 Reg. Coeff. (Bf) S.E. of Reg. Coeff. 0.267 -0.204 0.039 0.094 F Ratio = 25.37 B0b = 2. 794 Partial Reg. Coeff. 0.777a 0.249 29 df a - Partial regression coefficients derived from regressions utilizing standardized variables (z = (x - x bar)/s). b - B0 = Y intercept of regression equation; df = total degrees of freedom for regression horizon to 45 cm. The first PC was interpreted to reflect the importance of forest floor and surface soil to site fertility. The fact that the litter layer does not have major loadings may imply that nutrient uptake and subsequent deposition are not greatly dissimilar between stands which differ in subsolum ■ Pa v4- ■ {1 i U4.XX vj v>4 > » MUi. • >v Uiijr jr ^ > U X . / J. U llUV^ * > iwii 1* > + • ■ ?r » > * ir « U W U i U U X U are a process that becomes more apparent over time. This would explain the high loadings of the F&H layers and the A & E horizons. The fourth PC is more problematic in its interpretation. Positive loadings were observed for total Ca in the litter layer, for Total Mg and Ca in the F&H layer, and for TKN in the B horizon to 45 cm; regression in Table 4.6. none of which appeared in the PC Negative loadings were noted for TKP 110 in the litter layer, acid extractable K in the A&E horizons, and for TKP and acid extractable K in the B horizon to 45 cm. Only TKP appeared in the regression equation in Table 4.6. The interpretation of the fourth drawback to principal components easily interpreted multicollinearity PC analysis, variables for illustrates that uncorrelated that may variables a major is trading exhibit which are difficult to interpret (Jackson 1983). Plotting of the significant PC's in Figure 4.4, reveals a poorer discrimination of subsolum banding was achieved when compared to the pit P C 1s in Figure 4.3. assumption that the forest This strengthens the floor and surface soils give distorted representation of subsolum fertility. a Ill Table 4.8. Eigen values and vectors for principal components with significant regression coefficients: Forest floor and surface soil samples. Nutrient accumulation Principal Component #1 Principal Component #4 Eigen Values % Variance 7.9 44 1.4 8 Eigen Vectors: N-L P-L K-L Ca-L 0.129 0.160 0.208 0.177 0.071 -0.206 0.119 0.361 N-F&H P-F&H K-F&H Mg-F&H Ca-F&H 0.295 0.281 0.293 0.271 0.299 0.106 -0.133 -0.017 0.259 0.259 N-A&E P—A&E K-A&E Mg-A&E Ca-A&E 0.205 r\ oti V • 6 /X 0.221 0.198 0.265 -0.424 -0.066 0.106 N-B to 45crn P-B to 45cm K-B to 45cm Ca-B to 45 cm 0.188 0.146 0.207 0.308 0.215 -0.492 -0.365 0.011 0.090 _.n * i ~r a - L = Litter organic layer; F&H = Fermentation and Humus organic layers? A&E = A and E soil horizons; B to 45 cm = B horizon to 45 cm. 112 m a> < □ CD (0 CD (0 V CD (0 □ Pr i nc i pal Component #1 (0 □ □ V(0 CO □ m v(0 CD n O □ CD -B- _JX_ “ter jD CD □ □ ID CD □ CO _ C\J t CD n CD □ S cr cc a c r n □ >£ SD >, □ t^# Figure 4.4. s□ >> O □ i luauodwoo iE d !d u ]J d Plot of forest floor and surface soil PC's having the strongest associations with MAI. CONCLUSIONS It is acknowledged that factors such as temperature, light, tree genetics, and stand history play important roles in tree growth and site quality Spurr and Barnes 1973). (Carmean 1975, Grigal 1984; In addition colloidal soil separates have a great influence in the moisture and nutrient supplying power of soils making effects due to nutrients and available water difficult to separate (Coile 1952, Voigt et al. and White and Wood 1958) . 1957, No attempt was made to try and separate the effects of moisture from those of fertility, or to inspect other influences on stand growth. Results from both the pit and surface soil regressions rejected the null hypothesis of no association between MAI and fertility. Results of the pit regression rejected the null hypothesis that soil fertility below 30 cm has no association with MAI. Carmean (1975) states that a soil-site examination can be considered successful explained. surface if 65 to 85 % of the variability By this criterion both the pit soil (to 45 cm) accumulations (to 450 cm) were is and successful estimators of MAI with adjusted R2 values of 0.86 and 0.74, respectively. With a SEE of 113 0.357, the pit nutrient 114 accumulations estimated MAI with more precision than the surface nutrient accumulations (SEE of 0.481). The principal successful component estimators of MAI regressions as were determined marginally by Carmean's yardstick, with adjusted R2 values of 0.67 for the pit samples and 0.63 for the surface soil samples. significant regression forward to interpret. coefficients The pit PC's with were fairly straight The eigenvector of the fourth PC had loadings which corresponded strongly with the variables chosen in the pit regression analysis. Although this PC explained less of the variation in soil fertility than the other two significant PC's, it had a strong relationship with MAI. The pit samples had a much greater soil volume which could potentially be exploited by tree roots than the surface soil samples. However, the regressions utilizing the pit samples were only slightly stronger than those with the surface soil samples. The SEE of the surface soil regression was 35% greater than that of the pit regression, but this represents an increase of only 4% when compared to the mean MAI value. From a practical standpoint there was not a great deal of difference between the pit and surface soil regressions and the collection of surface soil samples is by far less time consuming while still providing a reasonable estimate of stand growth. It was unclear whether the performance of the surface soil samples was because they gave a better representation of site fertility due to their larger sample size, or whether 115 they were results however, a nutrient pool of the critical correlation to analysis stand growth. in Table 4.3 The suggest that stratification of soils by banding phase may result in different and possibly more precise regressions. Correlations of MAI with soil fertility in banded soils revealed stronger associations as depth increased, while unbanded soils had the strongest correlations in the surface depths. Splitting the samples into banded and unbanded fractions would likely result in two different regressions being developed. Unfortunately the sample size of this study did not allow separate regressions to be formulated for banded and unbanded soils. This study is only a first influences of subsolum soil step in evaluating the fertility on site quality. The reliability of the regression estimates needs to be verified before they are used for forest land evaluation 1976). Further relationship performed. more with investigation regard to of the banding MAI phase (Me Quilkin - fertility needs to be The results presented in this paper suggest that attention needs to be paid to subsolum features if optimal forest land evaluation is to be obtained, particularly on soils with subsolum textural strata. ACKNOWLEDGEMENTS The Ecological Classification System project was funded by the USDA Forest Service. Mr. Special thanks are extended to David T. Cleland whose tireless efforts of in behalf of the ECS project allowed this paper to be undertaken, to Dr. Phu V. Nguyen for sharing his expertise on laboratory methodologies; to Dr. Donald Zak for his help in soil sample collection, and to Dr. George Host and Mr. Steven Westin for collection of measurements, individuals understory respectively. who assisted vegetation and productivity The contributions of the many in laboratory analysis and preparation is greatly appreciated. Particular Associate appreciation Professor of is Forest extended to Dr. C.W. Ramm, Biometrics for MAI data, statistical consulting, and review comments. 116 LITERATURE CITED Adams, J.A. 1973. Critical soil magnesium for Radiata pine nutrition. N.Z.J. For.Sci. 3:390-394. Adams, P.W. and J.R. Boyle. 1982. The quantity and quality of nutrient cations in some Michigan spodosols. Soil Sci. 133:383-389. Bowersox, T.W. and W.W. Ward. 1972. Prediction of oak site index in the ridge and valley region of Pennsylvania. For. Sci. 18:192-195. Boyle, J.R. and G.K. Voigt. 1973. Biological weathering of silicate minerals: Implications for tree nutrition and soil genesis. Plant and Soil 38:191-201. Broadfoot, W.M. 1969. Problems index for southern hardwoods. Cajander, A.K. 1926. Fenn. 29:1-108. in relating soil to site For. S c i . 15:354-364. The theory of forest types. Acta For. Carmean, W.H. 1975. Forest site quality evaluation in the United States. A d v . Aaron. 27:209-269. Coile, T.S. 1952. Soil and the growth of forests. Aaron. 4:329-398. Adv. Comerford, N.B., G. Kidder; and A.V. Mollitor. 1984. Importance of subsoil fertility to forest and non-forest plant nutrition, p.381-384 In: Forest soils and treatment impacts. Proc. of Sixth N. American Forest Soils Conf. June 1983, Univ. of Tennessee, Knoxville. Dept, of Forestry Wildlife and Fisheries . Draper, N.R. and H. Smith. 1981. Applied Regression Analysis. John Wiley and Sons, New York. 709 p. Grigal, D. 1984. Shortcomings of soil surveys for forest management, p 148-166 In: Proceedings of the symposium, Forest Land Classification: Experiences, Problems, Perspectives. March 18-20, 1984 Madison, WI 117 118 Gysel, L.W. and J.L. Arend. 1953. Oak sites in southern Michigan: Their classification and evaluation. USDA forest Service Lake States Exp. Sta. Tech. Bull. #236. 57 p. Hannah, P.R. and R. Zahner. 1970. Non-pedogenic texture bands in outwash sands of Michigan: Their origin and influence on tree growth. Soil Sci. Soc. Amer. Proc. 34:134-136. Hart, J.B. Jr., Variation in outwash soil. A.L. Leaf, and S.J. Stutzbach. 1969. potassium available to trees within an Soil Sci. Soc. Amer. Proc. 33:950-954. Jackson, B.B. 1983. Multivariate Data Analysis: Introduction. Richard D. Irwin Inc. Homewood IL. 244 p. An Leaf, A.L. 1956. Growth of forest plantations on different soils of Finland. For. S c i . 2:121-126. Leaf, A.L. 1958. Determination of available potassium in soils of forest plantations. Soil Sci. Soc. Amer. Proc. 22:458-459. Lowry, G.L. 1970. Soil and site evaluation in forestry. Activities of Pulp and Paper Res. Inst, of Canada 16:13-15. Mader D.C. and D.F. Owen 1961. Relationships between soil properties and red pine growth in Massachusetts. Soil Sci. Soc. Amer. Proc. 25:62-65 f t t u « iJ t 1 non W • * n ^-p W X 1 V W U Q b W W C U U I C D 4^ XI i multivariate analysis. p. 242-244. In: Proceedings of the workshop, The Use of Multivariate Statistics in Studies of Wildlife Habitat. April 23-25. Burlington Vt. McQuilkin, R.A. 1976. The necessity of independent testing of soil-site equations. Soil Sci. Soc. Amer. J. 40:783785. Metson, A.J. 1974. Magnesium in New Zealand soils. Jour, of Exp. A q r . 2:277-319. Morrison, D.F. 1976. Multivariate Statistical Methods. Ed. McGraw-Hill, New York. 415 p. N.Z. 2nd Netter, J. and W. Wasserman. 1974. Applied Linear Statistical Models. Richard D. Irwin Inc. Homewood, IL. 842 p. 119 Page, A . L . ; R.H. Miller, and D.R. Keeney (Eds.) 1982. Methods of Soil Analysis Part 2: Chemical and Microbiological Properties. #9 Agronomy series, American Society of Agronomy. 2nd Ed. 1159 p. Pawluk, S. and H.F. Arneman. 1961. Some forest characteristics and their relationship to Jack growth. For. S c i . 7:160-173. soil pine Pregitzer, K.S. and B.V. Barnes. 1984. Classification and comparison of upland hardwood and conifer ecosystems of the Cyrus H. McCormick Experimental Forest, upper Michigan. Can. J. For. Res. 14:362-375. Shetron, S.G. 1972. Forest site productivity among soil taxonomic units in northern lower Michigan. Soil Sci. Soc. Amer. Proc. 36:358-363. Soil Conservation Service. Soil Survey. USDA. Soil 1976. Rubicon series. Conservation Service. 1981. Coop. Soil Survey. USDA. Soil Conservation service. 1982. Coop. Soil Survey. USDA. Kalkaska Nat. Coop. series. Nat. Grayling series. Nat. Spurr, S.H. and B.V. Barnes. 1973. Forest Ecology. Wiley and Sons, New York. 571 p. John Steel, R.G.D. and J.H. Torrie. 1980. Principles and Procedures of Statistics:____ A Biometrical Approach. McGraw-Hill, New York. 633 p. Strommen, N.D. 1974. Climate of Michigan by stations, 1940-1969. Michigan Department of Agriculture and Michigan Weather Service in cooperation with NOAA. US Department of Commerce. Technicon Industrial Method. 1977. Individual/simultaneous determination of N and/or P in BD acid digests. Method No. 334-74W/B. Technicon Industrial Systems, Tarrytown, NY. Thompson, G.R., M. Behan, J. Mandzak, and C. Bowen. 1977. On the evaluation of nutrient pools of forest soils. Clavs and Clav Minerals. 25:411-416. USDA Forest Service. 1965. Silvics of Forest Trees of the United States. Agriculture Handbook No. 271. 762 p. 120 Viro, P.J. 15:2-8. 1961. Evaluation of site fertility. Unasvlva. Voigt, G.K., M.L. Heinselman, and Z.A. Zasada. 1957. The effect of soil characteristics on the growth of quaking aspen in northern Minnesota. Soil Sci. Soc. Amer. Proc. 21:649-652. White, D.P. and R.S. Wood. 1958. Growth variations in a red pine plantation influenced by a deep-lying fine soil layer. Soil Sci. Soc. Amer. Proc. 22:174-177. White, E.H. and A.L. Leaf. 1964. Soil and tree potassium contents related to tree growth I. HN03-extractable soil potassium. Soil S c i . 98:395-402. Wilde, S.A. and A.L. Leaf. 1955. The relationship between the degree of podzolation and the composition of ground cover vegetation. Ecology 36:19-22. Wilde, S.A. and H.F. Scholz. 1934. Subdivision of the upper peninsula experimental forest on the basis of soils and vegetation. Soil S c i . 7:383-399. Wilde, S.A.; P.B. Whitford, and C.T. Youngberg. 1948. Relation of soils and forest growth in the driftless area of southwestern Wisconsin. Ecology 29:173- 180. Wilde, S.A.; G.K. Voigt, and J.G. Iyer. 1972. Soil and Plant Analysis for Tree culture. 4th Ed. Oxford and IBH New Delhi 172 p. Youngberg, C.T. and H.F. Scholz. 1950. Relation of soil forf it jfy cincl irsts of growth of inixsd Qair stsnds in the driftless area of southwestern Wisconsin. Soil Sci. Soc. A m e r . Proc. 14:331-332. Zinke, P.J. 1960. Forest site quality as related to soil nitrogen content, p.411-418 In: 7th Int. Congress of Soil Science. Madison, WI. Chapter 5. Conclusions Soil fertility was investigated across a range of regional stand productivities and the distribution and variability of nutrient and organic carbon contents to depths of 45 cm or 4.5 m were evaluated. The influence of subsolum textural strata (or banding) upon soil fertility was of particular interest. In addition, various aspects of the relationship between soil fertility and stand growth were examined. These examinations led to several notable results. Regional examination of MAI and TKN and TKP contents in forest floor and surface soil samples found that almost 70% of the variation in MAI was explained by a regression equation with TKN and TKP contents in the A and E horizons and TKN in the fermentation and humus layers as significant independent variables. The partial correlation coefficients for all three fertility variables were similar, indicating comparable contributions of each towards explanation of variation in MAI. Investigations northern lower of sludge Michigan application to found fertility oak stands increases in in the forest floor and surface soil (to 45 cm) persisting two years after application. When TKN and TKP means for sludge and control stands were entered into the regional regression, significant of sludge (29%) a increase in MAI was predicted as a result application. This prediction was supported by increases in basal area and diameter growth of 44% and 66%, respectively. The duration of the fertility increase due to sludge fertilization is not known, however, if site fertility 122 123 increases can be maintained, long-term site quality increases are predicted. Examination of the distribution of nutrient or organic carbon contents accumulated by 50 cm depths to a depth of 4.5 m displayed two distinct patterns. organic matter (TKN and TKP) Nutrients associated with rapidly decreased with depth, while those associated with the mineral soil (acid extractable K, Mg, and Ca) increased with depth. Forest floor contents of all nutrients were only a fraction of those contained in the mineral components to 4.5 m. Grouping samples by soil series resulted in general patterns of increasing fertility with increasing stand growth from the forest floor to 200 cm. Below this depth the slightly less productive Rb series displayed higher nutrient contents than the Ka series. Nutrient increases beneath the 200 cm depth, most notable for the Rb series, were thought to be due to banding. The only significant series differences consistently found by Duncan's mean separation procedure were between the Gy and Ka series. The AOV on forest floor fertility variables found significant differences between soil series for all forest floor revealed series. The three nutrients. TKN, TKP, significant AOV by series and total differences groups Ca in the between revealed all a poor separation of soil series by soil fertility with only 22% of the nutrient or organic significantly different. carbon/depth combinations being 124 The use of subsolum textural strata (bands) to phase soil series revealed peaks in soil fertility which coincided with banding depths, implying an increased nutrient supplying power of these bands. The inclusion of the banding phase resulted in a noticeable increase in nutrient contents of the banded phase of a soil series as compared with its unbanded phase. This difference was found at all depths, most noticeably at the 300-350 cm depth. The Ka series was most successfully separated with the inclusion of the banding phase. Although the forest floor AOV again revealed significant differences for all nutrients, the mean separation by Duncan's procedure found only one significant phase difference each for the Rb and Ka series. Differences in forest floor contents may be due to differential "pumping" of nutrients from the soil by vegetation. On sites with greater soil fertility vegetatative uptake and deposition pumps more nutrients. this way the forest floor can reflect soil fertility. the In The poor separation of series phases by the forest floor may be an indication that the processes of uptake and deposition may result in a loss in the contrast between the phases. Variability of depth to about 350 soil cm, fertility generally increased with below which variability decreased. Grouping samples by soil series resultd in a general pattern of increasing variability of soil fertility with increasing productivity. Inclusion of the banding phase partitioned nutrient content variability more successfully than did soil 125 series evaluations. Unbanded soil series phases had a pattern opposite that of the soil series groupings for the surface 250 cm, the highly productive Ka series being less variable than either the Gy or Rb series. The banded phases displayed marked increases in variability over their unbanded counterpart. Maximum differences in variability between banded and unbanded phases were found at the 300-350 cm depth. It was expected that phasing the soil series would have resulted in more homogeneous groups, however, this was not the case. The banded phases showed a much greater variability than the unbanded phases. An explanation for this disparity consistent with the data was that the intensity and continuity of the subsolum textural strata was highly variable. Further evaluation of this aspect of banding is needed. MAI was successfully estimated by both pit (to 450 cm) and surface soil (to 45 cm) fertility variables. TKP in the litter layer and total K in the fermentation and humus layer were common to the two fertility regressions. The pit regressions had slightly stronger associations with MAI than the surface soil regressions. The similarities in precision of the pit and surface soil regressions may be a result of differences sampling in effort points per sampling of the stand) intensity. surface may represent soil The samples site landscape better than the pit samples more intensive (six sampling fertility across the (one pit sample per stand) . Further investigation of banding influences utilizing 126 a greater sampling intensity needs to be done. From a practical standpoint the determination of presence or absence of banding is quicker and easier than the analysis of FF samples and makes their utilization for routine forest land evaluations more likely. Regressions satisfactory with estimation nutrient of site regressions estimated MAI within reliability of before these regression contents quality variables. are needs used to in in Both 10% of mean values. estimates relationships resulted The be verified site quality prediction. Soil fertility variables from the forest floor to 200 cm were found to have significant associations with MAI. subsolum increased textural strata fertility and were used to variability phase were soil found in When series, banded phases as compared to unbanded phases of the same soil series. On soils with sandy surface textures subsolum textural strata can have important influences on site quality and should not be ignored in forest land evaluations.