Crown Development and Site Estimates in a Douglas-Fir Plantation Spacing Test ROBERT O. CURTIS DONALD L. REUKEMA Abstract. Relationships among stem and crown dimensions of Douglas-fir were examined 43 years after planting on site IV land at initial spacings of 4 X 4 through 12 X 12 feet. Average dbh, height, and crown dimensions of the largest trees (largest 20 percent or 100 trees per acre by dbh) and of comparable crown classes all increased consistently with increase in initial spacing. Trees of similar dbh or total height were currently quite similar in crown dimensions, although they had arrived at this condition by somewhat different routes. Striking differences among spacings in apparent site indices are attributed mainly to restriction of height growth by competition rather than to real site differences. Average heights of several stand components were compared as bases for site index estimates; heights of a fixed number of the largest diam eter trees were most nearly consistent among spacings, although no procedure eliminated spacing effects. Comparisons suggest possible usefulness of live-crown length as one criterion for acceptable site trees in stands of abnormal density. High initial density in low-site stands can lead to serious underestimates of potential productivity. Forest Sci. 16:287-301. Additional key words. Pseudotsuga menziesii, site index. IN 1925 a Douglas-fir plantation spacing test was established on site IV land on the Wind River Experimental Forest near Carson, Washington. Following stand closure, diameter and height growth and volume growth per acre have all increased from close to wide initial spacings (Isaac 1937; Munger 1946; Eversole 1955; Reukema 1959; Reukemal). Differences have increased with advancing age. Unlike many spacing tests in other species, initial spacing appeared to strongly affect height growth of dominant and codominant trees. The area has been considered quite uniform in site quality, and increasing differences in heights and in estimated site indices based on these heights have been viewed as effects of spacing. Today, these plantations of equal age and believed equivalent site present striking contrasts in tree size, apparent vigor, and crown development. Crown and stem dimensions were measured in 1967 on a sample of trees in each 1 Reukema, Donald L. Forty-year development of Douglasfir stands planted at various spacings. (In preparation for publication, Pacific Northwest Forest & Range Exp. Sta., Portland, Oregon.) spacing, and an an analysis analysis w was as m made ade of of rrelationships elat lationships ionships bbetween etween ccrown rown aand nd sstem tem dimensions. di Objectives dime mensions ions. O bjecti tives ves were: were: 11.. To To compare compar co are re ppresent resent rel rrelationships elattiionsh ships bbetween etween crown dimen dimensions, sions, ddbh, bh, aand nd tree height, and and rrelate elate tthese hese ttoo ppast ast ddevelopment. evelopm lopment. t. 2. 2. To To compare compare average average stem stem tem and and crown crown ddimensions, di imensions, bby y sp sspacings, pacings, ooff tr ttrees rees ccomposing omposin ing conventional conventional ccrown rown cclasses lasses aand nd ooff a nnumber umber ntly uused sed ttoo of alternative mean trees2 frequently ccharacterize haracteriz ize sstand tand de ddevelopment. evelopm pment. t. 3. To 3. To compare comp mpare hheights eights ights ooff aalternative lternative m ean mean ttrees rees ffor or cconsistency onsistteency ency aand nd rreasonable easonablenness ess as as bbases ases ffor or es timation of site index in tthese hese estimation ssttands of widely diiffferin ffering density. T his ppaper his aper re rreports repo eports rresults esults ooff th tthese hese This ccomparisons. omparisons. A more detailed de scription description 2 As used in this paper, “mean tree” is the tree of quadratic mean diameter of a specified stand component. Quadratic mean diameter = diameter of tree of mean basal area (Curtis 1968). The authors are, respectively, Principal Mensurationist and Silviculturist, Pacific Northwest Forest and Range Exp. Sta., Forest Service, U. S. Dep. of Agr., Portland, Oreg. Manuscript received Sept. 22, 1969. volume 16, number 3, 1970 / Reprinted from F OREST SCIENCE, Volume 16, Number 3, September, 1970 Purchased by the USDA Forest Service for official use. 287 of the experiment, volume production comparisons, and a discussion of stand development will be given in a subsequent paper (see footnote 1). The Study The Study Area. The spacing test plantations occupy a site IV alluvial flat at an elevation of about 1,350 ft. The soil is a loose sandy loam with sporadic admixture of basaltic gravel and cobble. This juvenile soil is developing on pumiceous alluvium, generally 4 to 7 ft deep and underlain by lightly fractured, basaltic rock.3 Felled timber on the area was accidentally burned in 1920, following which all usable material was salvaged. In 1924 the area was reburned by a very intense fire that destroyed all reproduction and burned all duff and debris down to mineral soil. In the spring of 1925, 1–1 stock was planted at spacings from 4 X 4 through 12 X 12 ft, with different spacings contiguous but without replication. Subsequently, three ¼acre plots were established in each 2.8-acre plantation, excepting the smaller 12 X 12 plantation which contains a single 0.4-acre plot. Recent measurements were confined to these plots. Basic Data. Stand tables were available from the 1965 remeasurement. Height and crown dimensions were measured in 1967 for a sample of 214 undamaged trees— approximately 36 trees per spacing, equally distributed among plots, and across the range of diameters present on each plot. Values measured or calculated included: 1. Dbh outside bark (D). 2. Total height (H H). H). 3. Height to base of live crown (HLC), defined as the lowest whorl with live branches in at least three quadrants, exclusive of epicormic branches and whorls not continuous with the main crown. For a few trees with very lopsided crowns, heights to each half of the crown were measured and averaged. 3 Personal communication with Richard E. Miller, Soil Scientist, Pacific Northwest Forest and Range Exp. Sta., Olympia, Wash. 288 / Forest Science 4. Average crown width (C CW), W), defined defined as twice the mean length of the longest live branch in each of eight consecutive 45° sectors. Length was determined as the measured distance from center of stem to vertical projection of branch tip. 5. Crown length (CL), calculated as the difference, H — HLC. 5. Volume of crown, considered as a paraboloid. 7. Surface area of crown, considered as a paraboloid. Measurements of height and crown dimensions in 1945, based on a different sample, were also available, as well as some data on height and crown length taken in 1951. Analysis. The general plan of analysis was: 1. Calculate regressions of total height on dbh and of crown dimensions on dbh and on total height, for spacings and for individual plots. 2. Compare regressions for possible differences among differences among spacings. spacings. 3. Use the regressions for each spacing to calculate stem and crown dimensions of a number of possible mean trees, including (a) average of all trees, (b) average of each crown class, (c) average of dominants and codominants, (d) average of the 100 largest diameter- trees per acre, and (e) average of the largest 20 percent by diameter. 4. Compare heights of mean trees specified by alternative site tree selection rules for consistency among spacings and apparent reasonableness as bases for estimation of site index. Comparisons of regressions were based on both inspection of curves and on analysis of covariance (Table 1). In this analysis, the “among spacings ” F-value tests the hypothesis of no difference in regressions among locations occupied by plantations of different initial spacings. Since the experiment lacked replication, spacing is confounded with location. Interpretation of a “significant difference” and associated trend in the curves as “caused by spacing ” is a subjective judgment, dependent on the untestable but apparently reasonable assumption of homogeneity of site across the range of spacings. T h e a r e a h a d b e e n chosen for its apparent homogeneity, and subsequent examinations of soil and topography have given no indication of any gradient across spacings, although there is some indication of minor local site differences irregularly distributed within the area. Regression models used for tree stem and crown dimension relationships were based on inspection of scatter diagrams and arbitrary choice of functions which appeared to fit the data. Subsequent plotting and trial of a number of alternative equations indicated that most of these relationships could have been satisfactorily represented by the wellknown “allometric” equation, y = axb, often observed to hold for the dimensions of growing organisms. Results and Discussion Average tree dimensions and stand productivity differ among plots within spacings; presumably these differences result from some combination of localized variations in site quality and in past injuries influencing tree and stand development. Heavy mortality following initial planting, amounting to 36 percent in the first season, may have contributed to variation among plots. Dead trees were replaced each year for 5 years after initial planting, but location of replacements is unknown. Numbers of trees have been substantially reduced by subsequent mortality. Relationships which differ significantly (Table 2) within spacings have evidently been influenced by such location effects. Relationships which differ among but not within spacings almost certainly reflect true spacing effects. Differences both with and among spacings indicate probable confounding of spacing and location effects. Height Growth and Height-Diameter Curves When first measured, 5 years after establishment, average heights of all trees were greatest in the 4 X 4 and 5 X 5 spacings (Isaac 1937). At the 10-year remeasurement, this difference had disappeared; and since then, average heights (both all trees and largest 100 (by dbh) per acre) have been greater at the wider spacings, with differences consistent and steadily increasing with age. Differences among spacings in the heightdiameter curves have always been rather slight, indicating that the influence of stand density on height growth has been roughly proportional to its influence on diameter growth. Despite this similarity, there has been a clear trend of change over time. In 1945, when crowns were just closing in the wider spacings, differences in heightdiameter curves were small except for the 4 X 4 spacing (Fig. 1). The greater elevation of the curves for close spacings probably reflects initial restriction of diameter growth following stand closure, which occurred first in the 4 X 4. Contrary to the belief that height growth is less affected by competition than is diameter growth, the subsequent upward displacement of these curves was greatest at the wide spacings. By 1967, a quite different and fairly consistent pattern with respect to spacing had developed, with heights for a given dbh increasing slightly with spacing (Fig.2). When height-diameter curves for plots within each spacing are compared, the volume 16, number 3, 1970 / 289 higher curves are associated with greater stand diameter and greater height of dominants and codominants, suggesting that position of the curves is affected by minor site differences and possibly by mortality differences as well as by initial spacing. Average curves for the 5 X 5 and 8 X 8 spacings have always been slightly lower and those for the 4 X 4 and 10 X 10 slightly higher than might be expected by comparison with the others, and these differences are consistent with differences in stand diameter and other measures. 290 / Forest Science Crown-Stem Relationships Regressions of 1967 crown dimensions on tree diameter and height show that trees of given dbh or total height but in different spacings are quite similar in other stem and crown dimensions. Although crown widths for trees of given dbh increased with spacing, no clear and consistent trends of the curves of crown length, live-crown ratio, or height to live crown are evident. Crown Width. Since stand density expressions are frequently based on the crown volume 16, number 3, 1970 / 291 width-diameter relationship, the finding that among spacing differences in crown width-diameter regressions are significant 292 / Forest Science while those within spacings are not, isexpected. Figure 3A compares spacing regressions calculated under the assumption of common slope, since differences in slope were not significant but those in elevation were. Differences in crown width for trees of given dbh are not large, but the increase in crown width with spacing is consistent and undoubtedly a real effect of spacing. In the crown width-total height regressions, the 12 X 12 had considerably greater crown widths than other spacings, probably reflecting more recent stand closure. Otherwise, there was little indication of any trend with spacing. Height to Live Crown. The 1967 regressions of height to live crown on dbh differed significantly within spacings but not among spacings. Curves for individual plots within each spacing were quite variable and, like the height-diameter curves, those of higher relative position appeared to be associated with greater stand average diameters and heights. Heights to live crown of stand mean trees were slightly greater in wider spacings, due to greater mean tree diameters. Although data do not permit calculation of comparable curves for earlier measurements, relationships among spacings have differed substantially in the past. In 1945 live crowns were just beginning to recede in the closer spacings, and by 1952 average heights to live crown ranged from 6.7 ft in the 12 X 12 spacing to 19.5 ft in the 4 X 4. Following stand closure, live crowns receded rapidly in the wider spacings also. Crown Length. Although corresponding differences in crown lengths among spac- ings existed during the period of stand closure, differences among spacings in crown lengths of trees of similar dbh or total height had largely disappeared by 1967, except in the 12 X 12. Differences in regression slopes were not significant, and significance of differences in elevation is due almost entirely to the 12 X 12 spacing (Fig. 3B) and reflects its more recent stand closure. Differences among spacing regressions of crown length and of live crown ratio on total height were not significant, although nearly so; and there was no indication of any trend with spacing aside from the greater elevation of the 12 X 12 curve. Crown Volume and Surface Area. Regressions of these values on dbh differed significantly among spacings but not among plots within spacings. The corresponding curves showed clear trends, with estimated values for a given dbh increasing with spacing in a manner consistent with previously discussed trends in crown widths and crown lengths. Stand totals of crown volume and crown surface area were obtained by applying these regressions to the 1965 stand tables (Table 3). Total crown volume increased as spacing increased from 4 X 4 through 12 X 12 ft, consistent with trends in volume growth noted earlier. If suppressed trees are excluded from the comparison, a similar trend also exists with crown surface area. Comparisons of Crown Classes Among Spacings Average trees of a given crown class but in different spacings differ widely in both volume 16, number 3, 1970 / 293 294 / Forest Science volume 16, number 3, 1970 / 295 crown development and height (Fig. 4). Trees of crown development similar to dominants in the closest spacings are 296 / Forest Science assigned to progressively lower crown classes with increasing spacing but are quite similar in dbh and total height. This is consistent with Baskerville’s (1965) observation that in balsam fir stands of equal age, widely varying density, and assumed equivalent site, dominant trees of given dbh in dense stands appeared identical in other dimensions to intermediate and suppressed trees of the same dbh in low density stands. “Crown class” has been defined (Soc. Amer. Foresters 1958) as “A designation of trees in a forest with crowns of similar development and occupying similar povolume 16, number 3, 1970 / 297 sitions in the crown cover.” But, in these stands of widely differing density, trees of similar crown development do not occupy similar positions in the crown cover. Since classification is on a relative scale peculiar to the individual stand, trees of the same crown class but growing under different stand conditions do not in general represent equivalent conditions of past or present competition. Comparisons of Stand Mean Trees by Spacings Stands are often compared in terms of some “mean tree,” most commonly the average (quadratic mean dbh) of all trees above a fixed minimum diameter (1.5+ inches in this study). Other frequently used mean trees are average of dominants and codominants, average of the 100 largest diameter trees per acre, and average of the largest 20 percent by diameter. In these plantations, all the above mean trees show pronounced differences among spacings in both stem dimensions and crown development (Fig. 5). 298 / Forest Science Site Index Estimates and Competition Effects Site index estimation procedures are commonly based on the belief that average height of the larger trees in a stand is little affected by differences in density and associated competition, although exceptions have been observed (Lynch 1958, Holmes and Tackle 1962, Collins 1967). Such an assumption is clearly untrue in these plantations. Site estimates for Douglas-fir have long been based on average height of dominants and codominants (McArdle et al. 1961). In these plantations, estimates of site index by this procedure range from about 80 for the 4 X 4 and 5 X 5 spacings to 120 for the 10 X 10 and 12 X 12 (Table 4). Neither early height measurements nor topography nor any known soil characteristics suggest the existence of real site differences of anywhere near this magnitude. Differences in apparent site index among spacings are attributed principally to effects of differing intensity of competition on height growth of trees classed as dominants and codominants. Site index estimates can also be based on average heights of other stand components, including dominants only, the largest 20 percent by dbh (King 1966), and a fixed number of the largest trees per acre. Although the optimum procedure cannot be determined from comparisons for a single site condition, sensitivity to stand density effects is clearly an undesirable attribute. A satisfactory procedure should give approximately equal average heights and corresponding site index estimates for each spacing in these plantations of equal age, widely differing density, and approximately equal site. From this standpoint, alternative procedures can be ranked on the basis of relative variability of site estimates among spacings, as expressed by the coefficient of variation of mean site tree heights. Such a computation gave for the 1967 data: Mean tree Average of all trees 1.5+ inches Average of largest 20 percent Average of dominants and codominants Average of dominants Average of largest 100/acre bbyy ddbh bh Average of largest 40/acre1 by dbh Coefficient of variation 0.24 .20 .19 .18 .16 .16 .16 “Top height” as currently used in Great Britain (Bradley et al. 1966) and elsewhere in Europe. 1 Although none of the above rules eliminates the trend of average site tree height with spacing, a fixed number4 of the largest trees per acre gives the most nearly consistent estimates. For the averages of trees selected by each of these rules, there is a pronounced trend in relation to spacing of not only total height but also dbh and both absolute and relative crown dimensions (Fig. 5). Both the average of dominants and co-dominants and the average dominant tree in the 4 X 4 spacing are markedly inferior in both stem dimensions and crown development to the same crown classes in the 12 X 12 (Fig. 4). Differences are even greater for the largest 20 percent, which consists of 285 trees per acre in the 4 X 4 spacing, including many codominants, compared with 53 trees in the 12 X 12 consisting of the larger dominants only (Table 5). The basic difficulty is that these rules classify trees on scales which vary with stand density and are relative to the existing stand rather than to any standard condition. Clearly, these rules do not choose trees of comparable development in these stands of widely differing density, and they cannot give comparable estimates of site index under the conditions existing in these plantations. Averages used in this study were based on the N largest trees per acre on each plot, without regard to location. An average based on the single largest tree on each of a series of 1/ 1/N /N acre subplots would be somewhat more representative. Although a comparison by Reukema (see footnote 1) indicated little difference in corresponding diameters for these plots, differences between the two methods could be important when applied to larger and more variable areas. 4 volume 16, number 3, 1970 / 299 Fully consistent site estimates under conditions comparable to these plantations would require either (1) a site tree selection rule choosing trees subject to equivalent cumulative competition effects in stands of different density (if such trees exist) or (2) some means of adjusting conventional site index estimates for density effects. MacKinney et al. (1937), Lynch (1958), and Alexander et al. (1967) have made such density adjustments in other species, and if such effects are indeed widespread in young Douglas-Fir, similar procedures might be applicable. In stands of comparable density, such as those used as the basis for McArdle et al,’s site curves, crown development and relative position of trees within the stand (crown class) are closely associated. This is not the case for trees of similar relative position growing in stands of widely differing density. In these plantations, crown development of trees of similar relative position but in different stands, like development in diameter and in height, has been progressively reduced with increasing density, reflecting the cumulative effects of differences in competition. Competition effects on crown development and on stem development have been associated since crown closure. At present, the regression estimates of heights corresponding to either a dbh or crown length equal to that of the largest trees present in the closest spacing are quite similar for all spacings, though not identical. Thissuggeststhatcriteriaforselectionofsite trees including some standard of acceptable crown development in addition to relative position in the stand might reduce errors in site index estimates arising from differences in stand density. Or, that abnormal height development due to cumulative competition effects should be associated with deviations fromthenormalpattern of crown development, and that the latter might be used in a manner analogous to Lynch’s or Alexander’s procedures for adjusting site estimates for differences in stand density. In either case, ease of measurement suggests expression of crown development in terms of crown 300 / Forest Science length and total height rather than the crown area-diameter relationship often used as the basis for stand density measures. These possibilities emphasize the desirability of systematic crown measurements in spacing and thinning studies in young stands. These plantations have both (1) relatively poor site and (2) uniformity of age and spacing characteristic of plantations. Stagnation is often thought to be associated with poor site, and crown differentiation is slower and less pronounced in uniform plantations than in many natural stands, which may accentuate this tendency. It is not definitely known whether similar effects of competition on average heights and on site index estimates exist over the range of sites present in the Douglas-fir region nor whether manipulation of stand density after initial stand establishment has similar effects. It does seem clear that competition effects on height growth and on site index estimates are important in poor site stands of high initial density, and could lead to serious underestimates of the potential productivity of such lands. Literature Cited ALEXANDER, R R.. R R., ., D D.. TACKLE, and W. G. DAHMS. 1967. Site indexes for lodgepole pine, with corrections for stand density; methodology. USDA Forest Serv. Res Pap RM-29, 18 p. Rocky Mountain Forest Range Exp Sta, Ft. Collins, Colo. BASKERVILLE, G. L. 1965. Dry-matter production in immature balsam fir stands. Forest Sci Monogr 9, 42 p. BRADLEY, R. T., J. M. CHRISTIE, and D. R. JOHNSTON. 1966. Forest management tables. Great Brit Forest Comm Booklet 16, 219 p. COLLINS, A. B., III. 1967. Density and height growth in natural slash pine. USDA Forest Sery Res Pap SE-27, 8 p. Southeast Forest Exp Sta, Asheville, N.C. CURTIS, R. O. 1968. Which average diameter? J Forest 66:570. EVERSOLE, K. R. 1955. Spacing tests in a Douglas-fir plantation. Forest Sci 1:14-18. HOLMES, JJ.. R R.. B B., ., aand nd D D.. TACKLE. 1962. Height growth of lodgepole pine in Montana related to soil and stand factors. Mont State Univ Bull 21, 12 p. ISAAC, L. A. 1937. 10 years’ growth of Douglas-fir spacing-test plantations. USDA Forest Serv Pacific Northwest Forest Range Exp Sta Forest Res Notes 23:6. Portland, Oreg. KING, J. E. 1966. Site index curves for Douglas-fir in the Pacific Northwest. Weyerhaeuser Forest Pap 8, 49 p. LYNCH, D. W. 1958. Effects of stocking on site measurement and yield of second-growth ponderosa pine in the Inland Empire. USDA Forest Serv Intermountain Forest Range Exp Sta Res Pap 56, 36 p. Ogden, Utah. MCARDLE, R. E., W. H. MEYER, and D. BRUCE. 1961. The yield of Douglas-fir in the Pacific Northwest. U S Dep Agr Tech Bull 201 (rev.), 74 p. MACKINNEY, A. L., F. X. SCHUMACHER, and L. E. CHAIKEN. 1937. Construction of yield tables for nonnormal loblolly pine stands. J Agr Res 54:531545. MUNGER, T. T. 1946. The spacing in plantations. USDA Forest Serv Pacific Northwest Forest Range Exp Sta Forest Res Notes 34:3-4. Portland, Oreg. REUKEMA, D. L. 1959. Some recent developments in the Wind River Douglas-fir plantation spacing test. USDA Forest Serv Pacific Northwest Forest Range Exp Sta Res Note 167, 7 p. Portland, Oreg. SOCIETY OF AMERICAN FORESTERS. 1958. Forestry terminology. Ed 3, 97 p. volume 16, number 3, 1970 / 301