Fore" Sci.. Vol. 29, No.3, 1983, pp. 469-477 Copyright 1983, by the Society of American Foresters Soil Nitrogen, Sulfur, and Phosphorus in Relation to Growth Response of Western Hemlock to Nitrogen Fertilization M. A. RADWAN J. S. SHUMWAY ABSTRACT. Sixteen sites of westem hemlock (Tsuga heteroph.vlla) located in both the coastal and Cascade zones in western Washington were examined to determine the relationships of soil N, S, and P to growth response of the trees to N fertilization. The sites were chosen from among nineteen N fertilizer-test installations established in Washington in 1969 by the Regional Forest Nutrition Research Project of the University of Washington. The sites varied in productivity, but site index did not correlate with growth response. Concentrations of total N and extractable P were much higher in the forest floors than in mineral soils to a depth of 15 cm. Amounts of total N, mineralized N, and sulfate S were higher in mineral soils of the coastal sites than in those of the Cascades. Concentrations and amounts of extractable P of both the forest floors and mineral soils, however, were higher in the Cascades than on the coast. The nutrients studied and some of their ratios were significantly related to growth response to N fertilization. Overall, the strongest correlation was with extractable P in the forest floor (r = 0.77, P < 0.001) and the best correlations (r = 0.67, 0.66, P < 0.01). The data strongly suggest that low involving nutrients in mineral soil were with the ratios of extractable P/mineralized N P < 0.005) and extractable P/total N (r = levels of P alone or in combination with high soil N may be important factors in the reported lack of growth response of hemlock to ADDITIONAL KEY WORDS. N fertilizers. FOREST SCI . 29:469-477. Tsuga heterophylla. site index, forest floor, mineral soil. nutrients. total N, mineralized N, mineralized S, sulfate S, extractable P. WESTERN HEMLOCK (Tsuga heterophylla (Raf.) Sarg.) is one of the most productive conifer species in the PacificNorthwest. Increased demand and prices for hemlock products during the past decade have spurred interest in management of this important resource. Application of nitrogen (N) fertilizer to increase productivity has not been consistently successful. Growth responses reported so far range from increases of 50 percent or more to apparent reductions of about 20 percent. In general,N fertilization has been more successful in stands located on the lowlands west of Puget Sound and on the west slopes of the Cascade Range than in forests of coastal Oregon and Washington where N fertilization is now believed unprof­ itable (DeBell and others 1975, Webster and others 1976, Olson and others 1980). Presently, the causes of lack of growth response of hemlock to N fertilization are unknown. Olson and others (1979, 1980) concluded that factors other than gtowing space may govern the potential for response in natural stands. Similarly, recent experiments have shown that source of fertilizer N may not be a factor The authors are, respectively. Principal Plant Physiologist, Forestry Sciences Laboratory. USDA Forest Service, and Soil Scientist, Forest Land Management Center, Washington Department of Natural Resources, Olympia, Washington. They thank the PacificNorthwest Regional ForestNutrition Research Project and the University of Washington. Seattle, for making the studied sites available for collection of forest floor and mineral soil samples and for provjding the data for growth response and the data used to calculate site index. Manuscript received 4 December 1981. l About Th Th is fi e is File Wa s cre '. ated by Misscan . scann' Ing the s Identi V LUME 29, NUMBER 3, 1983 I 469 printed fi ed by t . h0.\J\l.e p U bl'Icat he Soft ver, so /on. ware h rne rnis ave bee tak s rn n correc a y rern ted' a'In. , affecting response of hemlock seedlings (Radwan and DeBell 1980a). On the other hand, high supplies of soil N or low levels of extractable phosphorus (P) or sulfur (S) seem to be likely factors. This view is based upon recent work in our laboratory and investigations by others elsewhere. For example, we have recently reported that foliar N was higher and foliar P and sulfate S were lower in coastal than in Cascade forests (Radwan and DeBell 1980b). Our unpublished data also show rapid depletion of foliar sulfate S after N fertilization. In addition, greenhouse tests have shown positive growth response to P fertilization by hemlock seedlings grown in coastal soil (Anderson and others 1979, Heilman and Ekuan 1980). We conducted this study, therefore, to further examine the relationships of soil N, S, and P to growth response of western hemlock to N fertilization. MATERIALS AND METHODS The Sites. -Sixteen natural stands of western hemlock on sites of different pro­ ductivity were used. Sites, approximately S acres each, were chosen from among 19 installations established in western Washington in 1969 by the Regional Forest Nutrition Research Project (RFNRP) of the University of Washington. Stands were 10 to 40 years old when the plots were installed, and growth responses to N fertilization were obtained from the RFNRP. Eight sites occur in the coastal hemlock zone within about 40 km of the Pacific coast, and the other eight are located inland on the west slopes of the Cascade Range (Fig. 1). On average, the coastal zone is believed to be more productive than the Cascade zone. Estimation o/ Site Index.-Site index values are based on data from 24 dominant trees at each site. Heights and breast-high ages of the trees were measured in 1970 by RFNRP personnel; we used this data and Wiley's tables (1978) to estimate site index at SO-year breast-height age. Determination 0/ Growth Response.-Seven-year radial-increment growth re­ sponse to application of 224 kg N/ha was determined by RFNRP personnel, using a tree-pairing method. Fertilized and unfertilized trees were paired according to similarity in size, competitive status, and past diameter growth. Percent response was calculated using 30 pairs of-trees (Olson 1979, Olson and others 1980). Briefly, the main findings were: (I) response was generally small and extremely variable, ranging from -20 percent in the coastal zone to 38 percent in the Cascades; (2) some sites in both geographic zones showed good response, but average response in the Cascades (S.9 percent) was somewhat higher than that on the coast (3.S percent); and (3) response was not related to site class. Forest Floor and Soil Sampling a nd Processing.-In 1979, three 30-m transect lines were laid out at each of the 16 sites. The starting point and direction of each transect were selected at random from 28 grid points located in the unfertilized strips between RFNRP plots. Samples of forest floor materials and mineral soils were collected at 3-m intervals along the transects. Forest floors were obtained using a 78.S-cm2 template and a cutting knife, and underlying mineral soils were sampled to a IS-cm depth. Sampling points were moved to avoid stumps and logs. Twigs and other material < 6 mm in diameter were discarded, and forest floor samples included both the litter and humus. There were 10 samples per transect, and samples of forest floors and soils were composited separately for each transect. About 0.004 m3 of soil was collected per site, and volume of forest floor collections varied with its thickness. Samples were airdried at room temperature; and roots, large stem material, and rocks were removed. Forest floor samples were weighed and ground to a fine powder in a pulverizer equipped with ceramic plates. Soil was passed through a 470 / FOREST SCIENCE i Western hemlock • o Coast Cascades 240 N 023 1090 1080 o o 111 58 Seattle Olympia o 18 0117 ---___ • 42 Kelso , 30km FIGURE I. Approximate geographic location of study sites in western Washington. Sites 100. 15. 84. 80. 9, 4, 3, and 42 occur in the coastal hemlock zone. and sites 24. 23. 109. 108. III. 58, 18. and 117 are located on the west slopes of the Cascade Range. 2-mm sieve, and the resulting fractions were weighed. For determination of total N, a subsample of the sieved soil was pulverized to pass through a O.S-mm sieve. All samples were individually stored in plastic containers at -1SoC until analyzed. Chemical Anal.1'sis.-Moisture in forest floor and soil samples was determined by drying to constant weight at 65° and 10S°C, respectively. Total N (including nitrate) in the sample was assayed by the micro-Kjeldahl procedure (Bremner 1965a). Mineralizable N was determined, in duplicate, under anaerobic conditions for soil samples and aerobic conditions for forest floor materials. Anaerobic condi­ tions were obtained by waterlogging. according to Waring and Bremner (1964). and all incubations were run for 2 weeks at 28°C, Aerobic conditions were achieved by mixing subsamples of forest floor with 30- to 60-mesh acid-washed quartz sand in widemouth vials and moistening the mixture with a small amount of VOLUME 29. NUMBER 3. 1983 I 471 TABLE I. Zone and installa tion number Approximate geographic area Site and stand characteristics. Elevation Weight of forest floor m Soil parent material Site index at SO years I mTlha m percent 49 Sedimentary 31 . Coastal 90 Growth response to N fertilizerl 17 IS Sappho 42 Cathlamet 360 74 Volcanic 35 16 80 Littleton Creek 240 48 Sedimentary . 30 10 84 60 28 Sedimentary . ISO ISO 38 Glacial 29 Neah Bay 90 40 opalis 60 22 Mt. Gunderson Burnt Hill 9 Humptulips 4 100 3 Average 36 7 37 3 Glacial 37 3 Sedimentary 34 Glacial 33 34 ISO I - 20 5 Cascade 117 24 Landers Creek 900 48 Volcanic 34 38 Deming 660 82 Glacial 20 20 9 23 Bold Mountain 330 39 Glacial 35 18 Orting 480 44 Volcanic 36 5 111 Duvall 240 29 Volcanic 29 2 108 Sap Lake 120 32 Glacial 36 -2 109 Beaver Lake 900 52 Glacial 33 - 12 270 31 Volcanic 32 - 12 Echo Glen 58 488 Average 32 6 I Based on heights and ages of 24 trees at each site measured by RFNRP in 1970. 2 Se,ven-year radial-increment growth response to a pplication of 224 kg N/ha determined by RFNRP using a tree-pairing method (Olson 1979). distilled water to approximate field capacity. Ammonium produced under an­ aerobic conditions was determined by steam-distillation in presence of KCl and MgO. Similarly, ammonium andnitrate production under aerobic conditions were assayed as ammonium by steam-distillation in presence of KCl, MgO, and De­ varda's alloy (Bremner 1965b). Mineralized N in all samples was calculated by subtracting initial ammonium and nitrate contents from post-incubation concen­ trations. Mineralizable S was determined in triplicate, using aerobic incubation for both forest floors and mineral soils. Subsamples were mixed with 30- to 60-mesh acid­ washed quartz sand in widemouth plastic jars containing a small volume of distilled water. Soils and forest floors were incubated for 2 and 3 months, re­ spectively, at 30°C (Williams 1967). Water lost by evaporation was periodically replaced. Sulfate, extracted from incubated and unincubated samples with Ca(H2 P04h according to Fox and others (1964), was determined by the turbidimetric method of Butters and Chenery (1959). S mineralized during the incubation period was estimated by subtracting the initial from post-incubation sulfate concentra­ tions. Extractable P, extracted from all samples with Bray-Kurtz solution 2, was determined colorimetrically (Bray and Kurtz 1945). Kilograms per hectare of total and mineralized N, sulfate and mineralized S, and extractable P were calculated from concentrations found in the samples and dry weights of forest floors and the < 2-mm fraction of mineral soil in the 0- to 15-cm layer. Statistical Analysis. 472 / - Data were subjected to analysis of variance, and means were FOREST SCIENCE TABLE 2. Chemical characteristics of forestjloor materials from the study sites.· Zone and inst all at ion number Coast al TotalN Mineralized N Sulfat e S ...................................................................................... kgl ha Miner alized S Extract able P ...................................................................................... 15 391 a I a Oa I a 42 616 a 12 a Oa 5a 9 a 80 421 a 3 a Oa 3 a 13 a 84 237 a 8a Oa 1a 6a 9 316 a 9a Oa 1a 5a 4 232 a 4a Oa I a 9 a 3a 19 a 100 345 a 9 a Oa 5a 3 215 a 2a Oa 2a 2a 347 x 6x Ox 2x 8x 117 212 a -2 b Oa I a 41 a 24 513 a 1b 1a 1a 25 ab 23 326 a 9 ab Oa 2a 7 b 18 360 a 6b Oa 3 a 6 b 111 299 a 6b Oa 1a 8b 11 b Average Cascade 108 286 a 11 ab Oa 1a 109 559 a 24 a 1a 4 a 8 b 58 246 a 6b Oa I a 12 b 3 50 x 8x Ox 2x 15 y Aver age J Within zones, values in t he same vertical column which are followed by the same lett er, a or b, and zone averages in t he same vertical column followed by t he same lett er, x or y, are not st at ist ically differ ent (P < 0.05) by T ukey's t est. compared according to Tukey's test. Correlation coefficients (r) were caJculated according to Snedecor (1961). RESULTS Site Index.-Site index at 50-year breast-height age ranged from 30 m to 37 m on the coast, and from 20 m to 36 m in the Cascades; it averaged 33 m over the 16 sites (Table I). The average site index for the coastal sites, 34 m, was somewhat higher than that for the sites in the Cascade zone, 32 m. This result is in agreement with earlier findings (Radwan and DeBell 1980b). Also, site index was not sig­ nificantly correlated with growth response to N fertilizer. Using 26 western hem­ lock RFNRP installations, Olson and others (1980) reported a similar result. Total Nitrogen. Total N in the forest floor averaged 0.83 percent and 348 kgl ha over the 16 sites (Table 2). Highest level of N, 1.07 percent, was in site 109; and the lowest, 0.48 percent, was in site 117, both in the Cascades. Amounts of total N ranged from 212 to 616 kglha; it averaged 347 kglha on the coast and 350 kglha in the Cascades. Differences within and between zones, however, were not significant. Also, total N in the forest floor was significantly correlated with mineralized S; it was n.ot significantly related to growth response to N fertilizer (Table 3). Concentration of total N in mineral soil (range, 0.05 to 0.55 percent) was lower than that in the forest floor, but the average amount ofN was much higher in the surface 15 cm of soil than in the forest floor ( l ,830 vs. 348 kglha) (Tables 2 and 4). Amount of N varied significantly within and between the two geographical - VOLUME 29, NUMBER 3, 1983 / 473 TABLE 3. Simple linear regressions among soil nutrients and between nutrients and groll'th response a/ hemlock to N/ertili=er. Forest floor Correlation coefficient Regression (r) . P-valuel Mineral soil Correlation coefficient (r) P-valuel 0.47 0.066 0.82 0.001 -0.01 0.971 -0. 56 0.023 Total N vs. mineralized S 0.62 0.011 0.83 0.001 Mineralized N vs. mineralized S 0.52 0.041 0.68 0.004 Mineralized N vs. extractable P -0.51 0.043 -0.30 0.261 Mineralized S vs. extractable P -0.35 0.181 -0.22 0.401 0 . 15 0.581 -0.50 0.050 -0.47 0.067 -0.43 0.093 -0.34 0.203 Total N vs. mineralized N Total N vs. extractable P Total N vs. growth response Mineralized N vs. growth response 0.00 Sulfate S vs. growth response Mineralized S vs. growth response -0.13 0.639 -0. 60 0.014 0.77 0.001 0.44 0.084 0.68 0.004 0.66 0.006 -0.06 0.825 0.67 0.005 0.70 0.002 0.28 0.285 Extractable P vs. growth response Extractable P/total N vs. growth response Extractable P/mineralized N vs. growth response Extractable P/mineralized S vs. growth response I Correlations are considered significant at P s 0.05. zones, and the average total N in the soil was significantly higher in the coastal zone than in the Cascades. Soil N was also strongly correlated with mineralized N and mineralized S; it was negatively related to available P and growth response to N fertilizer (Table 3). Mineralized Nitrogen. Incubations of the forest floor under aerobic conditions for 2 weeks resulted in net mineralization of N in 15 sites and net immobilization of N in I site (Table 2). The N mineralized in the forest floor ranged from I to 24 kglha (11 to 440 ppm); it averaged 7 kglha. with no significant difference between zones. Mineralized N in the forest floor was also correlated with min­ eralized S and negatively related to extractable P; it was not significantly correlated with growth response to N fertilizer (Table 3). More N was mineralized in the surface 15 cm of mineral soil than in the forest floor (38 vs. 7 kglha). and mineralized N varied significantly among and within zones (Table 4). It ranged from a low of 4 kglha (6 ppm) for installation 24 in the Cascades to a high of 88 kglha (139 ppm) for installation 84 on the coast. with a significantly higher average value in coastal soils than in the Cascades. Soil N mineralized was also significantly correlated with mineralized S but not with growth response to N fertilizer (Table 3). - Sulfate Sulfur. Forest floors were very low in available S as measured by levels of sulfate S (Table 2). Sulfate was detected in only two samples of forest floor, indicating that sulfur was present mostly in the organic form. Mineral soil contained much higher concentrations (up to 79 ppm) and amounts (up to 50 kglha) of sulfate S (Table 4). Also, sulfate S differed significantly within and among zones, with greater amounts occurring in coastal soils than in those of the Cascades. Average amount of sulfate S in the soil did not correlate with , growth response to N fertilizer (Table 3). - 474 / FOREST SCIENCE TABLE 4. Zone and installation number Coastal Chemical characteristics of mineral soils from the study sites.' Total N Mineralized N Sulfate S ............................................................................................. kg/ha Mineralized S Extractable P ............................................................................................. 15 1,735 cd 15 e 42 2,390 abc 41 cd 50 a 13 a 17 a 63 a 80 1,057 d 21 de Od 6a 33 a 22 a 4 cd 150 a 84 2,683 ab 88 a 4 cd 14 a 9 2,224 bc 43 c 15 bed lOa 9a 4 2,422 abc 56.bc 5 bed 13 a 48 aa 100 3 Average 2,096 bc 42 cd 30 ab 15 a lOa 2,977 a 72 ab 26 abc 20 a 16 a 2,198 x 47 x 17 x 14 x 44 x 146 ab Cascade 606 de I I cd Ob Oe 24 337 e 4d Ob I bc 23 1,104 ed 43 ab Ob 16 ab 117 69 ab 213 a 18 2,962 a 54 a 17 ab 21 a 17 b III 2,187 b 32 bc l Ob IS abc 36 ab 108 1,469 be 30 be 13 ab 13 abc 134 ab 109 1,597 be 28 be 13 abc II b 58 1,442 be 23 bed 38 a 2b I I abc 87 ab 1,463 y 28 y l Oy ll x 89 Y Average I Within zones, values in the same vertical column which are followed by the same letter. a to e. and zone averages in the same vertical eolumn followed by the same letter, x or y, are not statistically different (P < 0.05) by Tukey's test. Mineralized Su/fur.-As with N, S was mineralized upon aerobic incubation of the forest floor (Table 2). Mineralization of S, however, was much slower than that ofN; and incubations had to be run for 3 months before sulfate levels became easily detectable. Still, mineralized S was consistently lower than mineralized N; amounts ranged from I to 5 kglha (15 to 128 ppm), but differences among and within zones were not significant. In addition, mineralized S in the forest floor was not significantly related to growth response to N fertilizer (Table 3). Similar to N, more S was mineralized in soil than in the forest floor (average. 12 vs. 2 kglha) (Tables 2 and 4). Mineralized S varied greatly among the 16 sites, but average amounts in the two zones did not differ significantly. Additionally. soil S mineralized was negatively correlated with growth response to N fertilizer (Table 3). Extractable Phosphorus.- Extractable P in the forest floor ranged from 2 kglha (89 ppm) on the coast to 41 kglha (746 ppm) in the Cascades (Table 2). The average amount of extractable P in the Cascades, 15 kglha. was significantly higher than that in the coastal sites, 8 kglha. More importantly, extractable P in the forest floor, the ratio of extractable P/total N. and the ratio extractable P/min­ eralized S were all significantly correlated with growth response to N fertilizer (Table 3). Concentrations of extractable P in mineral soil (18 to 333 ppm) were much lower than those found in the forest floor. Average amounts of extractable P in the surface 15 cm, however, were higher than these found in the forest floor (66 vs. 12 kglha) (Tables 2 and 4). Amounts of extractable P varied significantly VOLUME 29, NUMBER 3. 1983 I 475 among the Cascade sites, and the average amount was significantly higher in the Cascades than on the coast. The ratios extractable P/total N and extractable PI mineralized N were strongly related to response to N fertilizer (Table 3). DISCUSSION AND CONCLUSIONS Hemlock productivity at the 16 sites used in this study varied greatly. Growth response to N fertilization, as determined by Olson (1979), also varied consid­ erably. High-quality sites and good growth response to fertilizer occurred in both geographical zones, but response was not significantly correlated with site index. Similar results were recently reported by Olson and others (1979, 1980) using 26 RFNRP hemlock installations which included the 16 sites used in this study. The data are also in general agreement with earlier findings by Webster and others (1976). Chemical properties of forest floor materials differed markedly from those of the surface mineral soil. In general, concentrations of total N and extractable P were much higher and contents of sulfate S were much lower in the forest floor than in mineral soil. Compared with mineral soil, however, the forest floor con­ tained much lower quantities of total N, sulfate S, extractable P, and mineralized N and S. Nevertheless, the forest floor still accounts for considerable quantities of nutrients, and hemlock is known to have many of its "feeder" roots in the forest floor (Ross 1932). Moreover, the forest floor recently has been suggested to be important for P nutrition in the coastal hemlock forests (Heilman and Ekuan 1980). Amounts of some nutrients in the forest floor and in mineral soil varied sig­ nificantly between the two geographic zones. For example, total N, mineralized N, and sulfate S were significantly higher in mineral soils of the coastal sites than in those of the Cascades. On the other hand, extractable P levels of both forest floor and mineral soil were significantly higher in the Cascades than on the coast. High levels of total and mineralizable N and low amounts of extractable P in coastal soils could account for the reported unfavorable growth response of hem­ lock to N fertilizer on the coast. In addition, the low P status of the coastal sites is in agreement with recent results by Heilman and Ekuan (1980). The nutrients studied and some of their ratios were significantly correlated with growth response of hemlock to application of N fertilizer. Overall, the strongest significant correlation was with extractable P in the forest floor. The best corre­ lations involving nutrients in mineral soil were with ratio of extractable P/min­ eralized N, followed closely by the ratio of extractable P/total N. These data suggest that, in general, positive growth response to N fertilizer by hemlock is most likely to occur when extractable P is high and total and mineralizable N are low. This conclusion is in agreement with our discussion above; it also supports earlier results implicating P as a possible factor in the erratic response of hemlock stands to N fertilization (Anderson and others 1979, Heilman and Ekuan 1980, Radwan and DeBell 1980b, Gill 1981). Conversely, results with S, especially the negative correlation between mineralized S in the soil and response, indicate that S is probably not involved. Field experiments involving application of P with and without N at carefully selected sites are necessary before a definite and practical regime can be recommended to obtain a consistent positive growth response of hemlock. LITERATURE CiTED ANDERSON, S. J., R. J. ZASOSKI, and S. P. GESSEL. 1979. Gr eenhouse nutritional studies of conifer seedlings in two coastal Washington soils. Northwest Sci Abstr 1979:36. 476 I FOREST SCIENCE BRAY, R. H., and L. T. KURTZ. 1945. Determination of total, organic, and available forms of phosphorus in soils. Soil Sci 59:39-45. BREMNER, J. M. 1965a. Total nitrogen. In Methods of soil analysis, Part 2 (c. A. Black, ed). Agronomy 9: 1149-1178. BREMNER, J. M. 1965b. Inorganic forms of nitrogen. In Methods of soil analysis, Part 2 (c. A. Black, ed). Agronomy 9: 1179-1237. BUTTERS. B., and E. M. CHENERY. 1959. A rapid method for the determination of total sulphur in soils and plants. Analyst 84:239-245. DEBELL, D. S., E. H. MALLONEE, J. Y. LIN, and R. F. STRAND. 1975. Fertilization of western hemlock: A summary of existing knowledge. Crown Zellerbach Forest ResNote 5, 15 p. Crown Zellerbach Corp, Cent Res, Camas, Wash. Fox, R. L., R. A. OLSON, and H. F. RHOADES. 1964. Evaluating sulfur status of soils by plant and soil tests. Soil Sci Soc Proc 1964:243-246. GILL, R. S. 1981. Factors affecting nitrogen nutrition of western hemlock. Unpubl Ph D thesis, Oregon State Univ, Corvallis, Oreg. 98 p. HEILMAN, P. E., and G. EKUAN. 1980. Phosphorus response of western hemlock seedlings on Pacific coastal soils from Washington. Soil Sci Soc Am J 44:392-395. OLSON, J. 1979. Phase I, western hemlock fertilizer response analysis. Report to Regional Forest Nutrition Research Project. Coli of Forest Resour. Univ Wash. Seattle, Wash. I I p. OLSON, J., W. ATKINSON, and M. RINEHART. 1979. Response of western hemlock to nitrogen fertil­ ization and thinning in the Pacific Northwest. In Forest fertilization conf proc (S. P. Gessel. R. M. Kenady, and W. A. Atkinson, eds), p 67-77. Univ Wash, Seattle, Wash. OLSON, J., W. ATKINSON, and M. RINEHART. 1980. Radial increment response of western hemlock to nitrogen fertilization and thinning. Regional ForestNutrition Research Project Tech Rep, 9 p. Coli of Forest Resour, Univ Wash, Seattle, Wash. RADWAN, M. A., and D. S. DEBELL. I 980a. Effects of different sources of fertilizer nitrogen on growth and nutrition of western hemlock seedlings. USDA Forest Serv Res Pap PNW-267, 15 p. Pac Northwest Forest and Range Exp Stn, Portland, Oreg. RADWAN, M. A., and D. S. DEBELL. 1980b. Site index, growth, and foliar chemical composition relationships in western hemlock. Forest Sci 26:283-290. Ross, C. R. 1932. Root development of western conifers. Unpubl MS thesis. Coli of Forest Resour, Univ Wash, Seattle, Wash. 63 p. SNEDECOR, G. W. 1961. Statistical methods applied to experiments in agriculture and biology. Iowa State Univ Press, Ames, Iowa. 534 p. WARING, S. A., and J. M. BREMNER. 1964. Ammonium production in soil under waterlogged con­ ditions as an index of nitrogen availability. Nature 201:951-952. WEBSTER, S. R., D. S. DEBELL, K. N. WILEY, and W. A. ATKINSON. hemlock. In 1976. Fertilization of western Western hemlock manage conf proc (W. A. Atkinson and R. J. Zasoski. eds), p 247­ 252. Univ Wash, Seattle, Wash. WILEY, K. N. 1978. Site index tables for western hemlock in the Pacific Northwest. Weyerhaeuser For Pap 17, 28 p. West For Res Cent, Weyerhaeuser Co, Centralia, Wash. WILLIAMS, C. H. 1967. Some factors affecting the mineralization of organic sulphur in soils. Plant and Soil 26:205-223. Reproduced by USDA Forest Service For Official Use VOLUME 29. NUMBER 3. 1983 I 477