Forestry Departmental Series 
No. 4
May 1971
FACTORS
AFFECT I
Height an imtrGrowth 
and Site Index
of &icalyptus 
glhls (Labill.
in the
HUANACO 
VALLEY of 
PE-RU
Ag CjUQ Fprr~ o Auburn 1->iversit' 
Auburn, .A Wbao 
E. V. Smith, 
D irw t_~
Arr cuitura! Experiment 
Stat
LIST OF TABLES
1. Basic statistical data from 20 plantations of Eucalyptus globulus
in the Huianaco Valley of Peru.
2. Analysis of variance of the E. globulus data when tree height
was the dependent variable and site index was not included as an
independent variable.
3. Analysis of variance of the E. globulus data when tree height
was the dependent variable and site index was included as an inde-
pendent variable.
4. Adjusted mean total height by significant independent variables
when site index was included in the analysis.
5. Adjusted linear regression coefficients of significant covariables
when the dependent variable was tree height.
6. Analysis of variance of the E. globulus data when site index
was the dependent variable .
7. Simple correlation coefficients expressing relationships between
variables in the E. globulus data.
8. Adjusted linear regression coefficients of significant covariables
where the dependent variable is site index.
9. Analysis of variance of the E. globulus data when the mean di-
ameter breast high (d.b.h.) of the stand was the dependent variable.
10. Adjusted linear regression coefficients of significant covariables
when the dependent variable is d.b.h.
Appendix Table. F values from analyses of variance of the tree
height data computed using aspect values obtained from different
index directions.
LIST OF FIGURES
1. Map of Peru showing location of the Hianaco Valley.
2. Scene in the Hianaco Valley showing typical terrain.
3. E. globulus pit props and timbers.
4. Seven-year-old E. globulus plantation.
5. Road cut showing nature of the soil in the study area.
6. Site index curves for E. globulus in the Huanaco Valley.
Appendix Figure. Trace of the value of the sine function when
used as in this study.
Cover photo. Scene in the Hianaco Valley.
ACKNOWLEDGMENT
The authors express thanks and appreciation to Ronald
Rochis and Milton Cone who collected the data used in this
study while serving in Peru as members of the Peace Corps.
They were attached to the Peruvian Forest Service and under
the direction of the senior author, who was a member of
United Nations Development Program Special Fund Project No.
116, FAO, Forestry/Peru. Mr. Cone was killed in action in
Viet Nam in 1969 and this paper is dedicated to his memory.
Factors Affecting -leight and Diameter Growth and Site Index of
Eucalyptus globulus (Labill.) in the l-luanaco Valley of Peru
SEYMOUR I. SOMBERG, EVERT W. JOHNSON, and L. E. DEBRUNNER*
INTRODUCTION
THE GEOGRAPHY AND ECONOMIC STRUCTURE of Peru are
dominated by the high and rugged cordillera of the Andes
mountains that separates the country into three distinctly
different regions. The largest of these regions lies east of
the mountains, is sparsely populated, has a tropical cli-
mate, and is dominated by dense rain forests typical of
the Amazon basin. West of the mountains is a relatively
narrow strip of coastal desert that harbors approximately
two-thirds of the nation's population and exhibits the coun-
try's most advanced economic development. The moun-
tains, along with their associated highlands (Altiplano),
form the third major subdivision of the country. Few
transportation links join these three regions and those that
do exist are poorly suited to heavy traffic. This is particu-
larly true of the eastern region that is almost completely
isolated from the others. These geographic factors have a
profound influence on the timber economy. Although the
country has enormous resources of usable timber, Peru is
a net importer of sawn lumber and other forest products.
In addition to its forest wealth, Peru has extremely large
reserves of copper, gold, silver, and other mineral re-
sources. Most of these minerals are found in the central
mountain region where the mining industry is most active.
Most mining in this region is of the shaft type that requires
large quantities of both sawn and round wood for pit
props. Because of the transportation problems mentioned
earlier, the great supply of wood east of the mountains is
not economically available to satisfy the demand for min-
ing wood. Transportation links from the coast to the cen-
tral area are adequate for supplying wood to the mines,
but this wood is extremely expensive. Because of these
conditions, extensive plantings of Eucalyptus globulus
(Labill.) have been established in the mining area. This
species has been favored since it exhibits rapid growth, an
ability to coppice, and resistance to drought. However,
little is known concerning the silviculture or management
of this species under conditions existing in the central re-
gion of Peru. This report provides some information rela-
tive to this subject.
* Formerly Associate Professor, 
Department of Forestry, now
Director of Research and Professor, School of Forestry, Stephen
F. Austin State University; and Professor and Assistant Pro-
fessor, Department of Forestry, respectively.
Site of this investigation was the Hiianaco Valley, Fig-
ure 1, which lies in the midst of the mountains, cover and
Figure 2. The valley floor ranges in elevation from 2,500
to 4,000 meters (8,000 to 14,000 feet) above sea level and
the surrounding mountains rise to altitudes of approxi-
mately 6,000 meters (20,000 feet). Valley walls are steep
and rocky. Relatively little land is suitable for growing
trees. Consequently, large amounts of pit prop material
must be grown in short rotations on small areas. E. globu-
lus, because of its rapid height growth, has great potential
in this area.
In the central region of Peru, wood for pit props is not
FIG. 1. Map of Peru showing location of Hianaco Valley.
the1 da8(a XXcr (' no t complefIIte tor XIomei pIlots, onI I 2 were
suiItable]( tor all phasesc oIt the stud(IX Eaclh plot X as 2(0 X
2(0 meters (65.6 XK 65.6 ft, 1 25 he1ctareI, oI atlmos~t ex-
act ly 1 19 actre) a11d was soXI located wXithinl the stal d that
('(l
14
('lc 1111( not 8lit a actor
tation:
Scen n1 1 t t 1 (11ac XIICXX
1. Age of plantation, in gars (\yhicl should have becu
the Same as tree age since, 6th this species, only about
:3 months clapsc betmull seed germination and outplant-
2. Planting method used (bare root or in cans);
3. Whether fertilizer had bean applied (type imd quail-
tits of fertilizer it, \ycll as tinting of application wcm till
known);
t picnl terrain.
1. Whether the site had been prepared for planting
(site preparation consisted of plo\yin(or hand cult iyatiug
sites prior to pltmting);
5. Whethui the plantation had been irrigated at any
time in its evidence (no inlormatioti was available Mil-
ceniiiig amount of water used or till, timing of its applica-
lion. )
till the iliid1111l piece, XX lichi is boughit tol satisf tX 8 pIe-
cific minit; need. Couise(Iuenitl, IctigthX arc1 X aiabil',
avX Craginig abou)(t 2 meter s (6 feI't), XXhereas.I topJ dliameiIter
to (6 ices), XXithi 13 cen'Itimtet('rs (.5 itiches ) pleired( (.
P~ieces XXith top
1 
diameters greater' thtat 1.5 cenItimterstI' are1
someI(timesc used2(, but tl)C'X ilC( less5 valable tol thie mIintcs
becaulse thecir add~ittinal XXeight 111(d volume intcr ease him-
LarIgel pieces alXso a1re- Ituilcirale to produIic(rs b l)causeX'
hieightt as IrapidlX as poss~ile. B1cause51 iot this, kiXlmd(lgI
growXthl XXul be(1( of~ great po~tenItiaIl valu1e tol producII'('
DATA COLLECTION
l XXettt no11n-cop[pi'edI p~l.Ittins, Figurel 1, XXr I'l'e-
lectedl tatndoltyi tflt stIudv from11 ~il] p1111 tiolsI ill till vat]1
5'
It. 2.
,.\ 1 ac (. amp X l eI jblot thle ft iIi ll t 11( 1 i to tatio w1 Xas oh-
L Altitudec altove XI'a l'eel in meter X, IIiug ai
t
i aromet-
2. Slope, iln (t(gr'I'X (usingi an1 \lbteX levX l
Appenidix);
5. Numer of (Isestl pcr etar hig time1. olf datacllc-II
IlXti; oatl('t~l Xa eta . ctltlc'tai I:
6. Batal X ra nsur etprhcae
7.Si hdtrmnd uig i ied kto nu w
t,4] C
t..._
Otlcr l~ti(" b~nlishz
ii 1,1able
A\gi
Alt itude I
Slt)e
Numiiibir of tirexs
lSixil 1AIet
Pecin t Suitd
Pit cit silt
Deipth ot A&Bl hoizonit
D).bi.h. oft all tee" ini standii
Arti it hmticsli
2.728n. 8:,317tft.
5.21 iii. 56.1 It.
2(19(5ha. 848 it ace
27.60) iti./ 12
0
.
2 
ft-
hai. a
:39.1 enli. I5.) ini.
2:3.3 ai. 76.)f ft.
1:.5 cmi. .5.3 iii.
Othet Millis ii ~tigli',)
10.4 xi.
1:3.9
-'515.8 u. -?-1,573 ft
-'-12.:39
in. hut..
?426/
acrei 1
--51(.0
ft/ tce
IS8.:3 vi '-7.2 inI.
-- 7.1 1 +3: '-3.0 ft.
-4.2 emil. -1-).67 ill
-- 1(0.3
0'1.910
-- 7.5
-'4.9
-'5.8
11.91)
1:3.510 
iii.
5761-
7250/hat.
:3.-17-6(5.015
i .- ]lit.
9.7-
1415.:3 ft.
2:33-
29:34 ten'r
15.1-287.7
ft.-/acre
O thier
(-1 7 xi.
0-:35
4..5-8.0
34.0)-(6.1)
22.11-18.1)
8.0-:39.1)
9-61) nt. :31)-197 ft.
6.15-24.9 2.6-9.8 in.
tcii.
andt B3 sli)oili toizonis xii(re cidixteeleid c iitaftes. bTe stuitis-
Itil iode wtti s ai s foItllowxxs:
(I -r
ma
I1I. T'e niatuii i' i t'e p)airent mtiait ii) xits ireti tet) ax
pie ploit, Figuire .5, uid m(1 ittetiiieullx ,iialx ,ed in tlie lab-
DATA ANALYSIS AND DISCUSSION
tuinied ini t'e coel of i t'e tiitlx
axs the pritiat x tdepi'iili'nt xaiable xiiice it is ft'e citil
iiiiformiatin, litwxevxer, effeets a) tlie meaisured ariabiuult' on
d~lbli. alit) siti' indtex uilsoi xxetr' inxi'tigatt't. Da)tui xx ret
Dautuia wetc ,italxze siniitg tlie ''Leaist-Squpareis idii \Mai-
iimuim Likelihoiiot Gciiera) Puipoise Pitogrami" teetltopi't lhx
W.' 13. fIaritx ofit Ohioi State 1, iux rxutx. IPlaiitiig mtetlitid
xxeci contasideired to be the xai ablt's gixviiig riset' fth nii'itiii
effects, wh etreas age, ultitde, sltope, uspi'tt sit' XAppx'i
tdix), plaiiting xpuce, iiumber'i tr(ies pit ii ii1 itt i run arei.
basal area per iitit girouinit area, sil pll, perci 'tit sndt,
i I
Tree Height
eff ects ,idi inIt e ttti s, S, F, Px, I. IS, INS, uiiit IJix x werc
'I' xiii 2. \N.-~v~ oi VAMAi', iii \xEt F~ i l . 910ttliti l 'xi
XX iii N 'lii IrIiir AV'sti I X xlii I)~nx i',lw NiriXxiixt
xAmi Satr IitiN AV'xs Niii INI t iUtn it xx x
I~Niti i Ni)NiX xi x V IABA
Sit it.
Pitiii' i metho (i P')-
lii tili,,ttiitiiF)
l'xS
l
t
x I'
i\S
Xi, ( lintn i
ltitiili' ( litea i
Slop,1 ( linea )
N iiii)ii ot tiveis/hai.
( linear)
pI ( liniiar)
Per ci it tsanx ( linea jti
Pcili cut sit ( liit')i
Pcr-t cclitlus in r
Itisidna
I .'x l
ti. XIMS. 12- of sig-
ii Ii. ,iii
361.192
59.1011
13.5.:388
t99.048
96:3.920
6(1.8'9)
:38(5.8201
81.699
57.51:3
:3 (5.4 37
:3 15.476
6.533
6:3.267
:3501.170)
:3570 ..50)
01.115
65.822
101.8:35
4.544
7.765
722.275
16.975
:3.476
7.976
15.534
:3.58:3
22.788
1.81:3
.3.:390
2.147
20.352
0.:385
:3.727
21).629
2101.31:3
0.00t 7
0.-1)2
01(.38
0.3268
0.457
42.551)
N.S.
N S.
01.0015
01.025
01.025
01.0015
N .S.
01.00(5
0.0501
N.S.
N-S.
0.005
N.S.
N.S.
11(1(15
11.00(5
N.S.
N .S.
N.S.
N.S.
N.S.
01.0015
TABLE I. BA',R SIAIIYtICAt. DATA rnosi 211 I't.s iaruoxs OF E',walyp(n.s _lobribis i\ my Ill A\seo 
or Pi:,m, 1966-67
1,000- 6,200-
3,650 nt. 12,000 ft.
1; k1, _v f (P)lanting 
method;
(I)rrigationi + (S)itc, + (F)crtiliz.cri
(Pl)u + (PS);,, 4 (PF)~ + (IS)O,, +
Total tree
height, meters 
Site index
40
3030
24
21
20 18
10
0 6 7 8 9 10 II 12 13 14 15 6 7
Total tree age, years
FIG. 6. Site index curves for E. globulus in 
the Hiianaco
Valley of Peru.
significant, along with the covariables altitude, number of
trees per hectare, basal area per hectare, 
and aspect.
Results of the analysis of variance in Table 2, which 
in-
dicate that stand density was the strongest 
independent
variable and that age had no effect, are contrary 
to the
usual assumptions on which site index is based. 
This sug-
gested that some powerful variable was not being con-
sidered in the analysis, and this variable was deduced 
to
be site quality. Consequently, the family of site index
curves in Figure 6 was developed from the available data.
The procedure used to develop these curves is that de-
scribed by Chapman and Meyer1. This procedure is rel-
atively crude, since it does not recognize polymorphism,
but sketchiness of the data precluded use of more sophisti-
cated procedures. The guide used in construction of the
curves is the trace of the equation:
H = 12.3266 + 3.3069A
1
/2
Where: H = total tree height (in meters), and
A = tree age (in years).
The index age was set at 10 years and the interval be-
tween curves was set at 8 meters. Despite crudeness,
these curves may be useful to forest managers in the
Huanaco Valley since site index curves for E. globulus in
that area apparently are unavailable.
These site index curves were used to obtain site index
values for each sample plot in the study. These values
were used as an additional covariable and the tree height
1 CHAPMAN, H. H. AND W. H. MEYER. 1949. Forest Men-
suration. McGraw-Hill Book Company. New York (522 pp.).
data were reanalyzed on this basis. No attempt was made
to transform other independent variables in this analysis.
Utilization of this, or any other, set of site index curves
in a study of this type undoubtedly will cause adverse
comment. Site index actually refers to the height attained
by dominant or codominant trees at a given age. Con-
sequently, site index is closely correlated with tree height.
Inclusion of site index in an analysis of variance of this
type would show it to be a powerful contributor to varia-
bility of the dependent variable. Furthermore, if any other
independent variables were correlated with site quality,
and many of those used in this study would be expected
to be so correlated, their roles in tree height growth could
be obscured effectively in the analysis of variance. How-
ever, since these interrelationships could be evaluated in a
subsequent analysis of variance of the site index data, the
decision to reanalyze tree height data with site index as a
covariable was reasonable.
Results of the analysis of variance that included site
index as a covariable, Table 3, are far more conventional
than those obtained in the first analysis. As expected, site
index and age have strong influences on tree heights, and
basal area no longer appears as a significant variable. How-
ever, the appearance of soil depth and aspect as significant
variables was unexpected. These are site quality variables
and ordinarily would be expected to be obscured by site
index.
Insofar as the main effects and their interactions are
concerned, only site preparation (S) and the interactions
PxS, IxS, and IxF were significant, Table 3. Table 4 gives
the adjusted mean total heights associated with these ef-
fects. Included in Table 4 are results of a series of Dun-
can's new mutiple range tests made to isolate the combina-
tions of effect that were acting differently from the others
and causing the interactions to appear significant.
The information in Table 4 indicates that site prepara-
TABLE 3. ANALYSIS OF VARIANCE OF THE E. globulus DATA
WHEN TREE HEIGHT WAS THE DEPENDENT VARIABLE
AND SITE INDEX WAS INCLUDED AS AN
INDEPENDENT VARIABLE
Level
Source d.f. M.S. F of sig-
nificance
Planting method (P) 1 38.408 3.326 N.S.
Irrigation (I) 1 2.368 0.205 N.S.
Site preparation (S)------ 1 69.770 6.044 0.025
Fertilization (F)--------- 1 0.143 0.012 N.S.
PxI 1 27.130 2.350 N.S.
PxS 1 233.382 20.216 0.005
PxF 1 5.403 0.468 N.S.
IxS 1 82.188 7.119 0.010
IxF 1 67.085 5.811 0.025
SxF 1 38.323 3.320 N.S.
Age (linear) 1 136.553 10.962 0.005
Altitude (linear) 1 7.641 0.662 N.S.
Slope (linear) 1 2.650 0.230 N.S.
Planting space (linear) 1 8.543 0.740 N.S.
(linear) . 1 24.174 2.094 N.S.
Basal area/ha. (linear) 1 1.632 0.141 N.S.
pH (linear) 1 21.887 1.896 N.S.
Per cent sand (linear) 1 1.446 0.125 N.S.
Per cent silt (linear) 1 4.218 0.365 N.S.
Depth of A&B (linear) 1 108.127 9.366 0.005
Aspect (linear) 1 222.317 19.257 0.005
Site index (linear) 1 2541.952 220.185 0.005
Residual 465 11.545
[6]
TABLE 4. ADJUSTED MEAN TOTAL HEIGHT BY SIGNIFICANT
INDEPENDENT VARIABLES WHEN SITE INDEX WAS INCLUDED
IN THE ANALYSIS (INCLUDED ARE RESULTS OF A SERIES
OF DUNCAN'S NEW MULTIPLE RANGE TESTS)
Comparison level
Variable and level 
Mean of significance
0.05 0.01
Site preparation
Site prepared 24.10
Site not prepared 2.0.90
Planting method x site preparation
Bare rooted & plowed__ 25.00 I 1
In cans & not plowed ------- 24.26
In cans & plowed --- 23.20
Bare rooted & not plowed ............. 17.55
Irrigation x site preparation
Not irrigated but plowed 24.77
Irrigated & plowed -- - 23.43
Irrigated & not plowed 22.19
Not irrigated or plowed 19.61
Irrigation x fertilization
Irrigated & fertilized 24.36
Not irrigated nor fertilized ........ 23.60
Irrigated & not fertilized- ---- 21.26
Not irrigated but fertilized ....... 20.78 I 1
tion by "plowing" prior to planting had, in general, a ben-
eficial effect on tree height growth.
The effect of site preparation (compared to no site prep-
aration) on trees planted in cans was negligible. However,
trees that had been planted with bare roots on plowed
land did somewhat better than trees planted in cans on
either plowed or unplowed sites. Bare-rooted trees planted
on plowed land did much better than trees planted with
bare roots on unplowed land.
Irrigation alone had no discernible effect on height
growth. However, trees responded differently to irrigation
when other factors were considered in conjunction with it.
Irrigation and site preparation showed evidence of an
interrelationship only when a site was not plowed prior to
planting and the trees were not irrigated following plant-
ing. This combination of no soil disturbance and no irri-
gation yielded poorer results than any other treatment or
set of treatments. This suggests that plowing and/or ir-
rigation would be required for good results.
The pattern of response to irrigation and fertilization
was unusual in that it made little difference whether, on
the one hand, the trees had been irrigated and fertilized
or, on the other hand, the trees had been neither irrigated
nor fertilized. However, application of fertilizer in the
absence of irrigation had a negative effect. It is possible
that the application of fertilizer created a situation where
uptake of water normally available was reduced because
of an unfavorable osmotic potential. Addition of fertilizer
can result in osmotic potentials so low as to retard plant
growth
2
. Watering alone was better than fertilizer alone,
but yielded results poorer than those associated with
either irrigation and fertilization or with no irrigation or
fertilization treatment. This situation probably was pro-
duced because the limited watering encouraged herbac-
eous weed growth that utilized the shallowly penetrating
2 KRAMER, P. J. 1969. Plant and Soil Water Relationships:
A Modern Synthesis. McGraw-Hill Book Company. New York
(482 pp.).
TABLE 5. ADJUSTED LINEAR REGRESSION COEFFICIENTS OF
SIGNIFICANT COVARIABLES (DEPENDENT VARIABLE
IS TREE HEIGHT )
Covariable b
Age + 0.678 meter of tree height per year
Depth of A&B horizons.- 0.0357 meter of tree height per
centimeter of soil depth
Aspect + 0.000040 meter of tree height per
sine unit (0.00001) of the azi-
muth from N 45
? 
W
Site index + 1.325 meters of tree height per
meter of site index
water along with natural precipitation and thus decreased
available soil water for the deeper rooted Eucalyptus.
Adjusted linear regression coefficients of age and site
index are positive, Table 5, indicating that tree height in-
creases with increasing age and site quality, which is to
be expected. However, increasing soil depth apparently
has an adverse effect. Though this was unexpected, since
good sites usually are associated with deep soils, it is not
unreasonable when the character of the soil (mostly coarse,
gravelly sand) is considered. The deeper this soil, the
easier and faster the rainfall or irrigation water can per-
colate beyond the root system of the trees. The effect of
increasing azimuth of slope from northwest (aspect) was
to increase height growth. This conforms to the theory
that better sites and, consequently, taller trees occupy the
cooler, wetter slopes. However, the appearance of soil
depth and aspect as significant variables when site index
was included as a variable was not expected and raised
many questions.
Site Index
Inclusion of site index into the analysis of variance of
tree height data effectively obscured effects of most vari-
ables that actually control site quality (and, inexplicably,
TABLE 6. ANALYSIS OF VARIANCE OF THE E. globulus DATA
WHEN SITE INDEX WAS THE DEPENDENT VARIABLE
Level
Source d.f. M.S. F of sig-
nificance
Planting method (P)-- 1 42.203 0.097 N.S.
Irrigation (I)t n (...... 1 449.348 1.035 N.S.
Site preparation (S) ...... 1 308.125 0.709 N.S.
Fertilization (F)----- 1 482.952 1.112 N.S.
PxI 1 387.242 0.892 N.S.
PxS................ 1 101.726 0.234 N.S.
PxF 1 470.985 1.084 N.S.
IxS 1 1185.951 2.730 N.S.
IxF ................ 1 63.894 0.147 N.S.
SxF 1 151.923 0.350 N.S.
Age (linear) 1 253.805 0.584 N.S.
Altitude (linear) 1 2713.156 6.246 0.025
Planting space (linear) 1 1007.155 2.319 N.S.
Number of trees/ha.
(linear) . . 1 6750.411 15.541 0.005
Basal area/ha.
(linear) 1 36321.356 83.620 0.005
Per cent sand (linear) 1 199.066 0.458 N.S.
Per cent silt (linear) 1 24.128 0.056 N.S.
Depth of A&B
(linear) 1 139.762 0.322 N.S.
Aspect (linear) 1 1141.092 2.627 N.S.
Residual -.. .. 50 434.362
[7]
did not obscure two of the site quality variables). Thus,
it was necessary to study effects of these variables on site
index itself. This was done in the analysis of variance
given in Table 6. It should be noted that this variance is
based on 72 rather than 489 observations since, in this
case, only one observation per plot was available. Because
of this loss in degrees of freedom, analysis of the site in-
dex data is considerably less sensitive than was the tree
height analysis.
Results obtained from the analysis were not as expected.
Of the variables usually associated with site quality, only
altitude was significant. Furthermore, altitude appeared
to have less effect on site index than did stand density. In
an attempt to clarify these results, a study of the pattern
of correlations between the continuous variables was made,
based on information in Table 7.
Of the variables under discussion, altitude, slope, soil
pH, per cent sand, per cent silt, per cent clay, soil depth,
and aspect usually are considered to be associated with
site quality. Site index was correlated.signfficantly and
relatively strongly with two of these, altitude and slope,
and significantly but relatively weakly correlated with soil
pH, per cent silt, soil depth, and aspect. Site index was
not correlated significantly with the other variables.
Site index increased as both altitude and slope increased,
which is logical since at higher altitudes the sites are sub-
jected to lower temperatures and shorter growing seasons
and soil on steeper slopes retains less water for use by
plants. These conditions are compounded by the strong
positive correlation between altitude and slope, which in-
dicates that slopes become steeper as altitude increases.
The strong correlation between altitude and slope results
in only one (altitude) appearing significant in the analysis
of variance in Table 6. Only a minor change in the data
probably would have resulted in slope being significant
and altitude not significant.
Correlations between soil depth and altitude and soil
depth and slope are significant but not strong. In general,
soil depth decreases as altitude and slope increase, follow-
ing the pattern of site index. As soil depth decreases, the
per cent of clay decreases and per cent of sand increases.
This is logical since clay particles are transported down-
ward by water more easily than are sand particles. Clays
are deposited on the more gentle lower slopes while sands
remain above on steeper slopes. High pH values are more
likely to be associated,'with clays than with sands. The
data in Table 9 are consistent with this pattern. Thus, the
signfficant correlation between soil depth and pH is a re-
flection of the correlation between soil depth and per cent
clay and between per cent clay and pH.. This also explains
the relationship between pH and altitude and slope. In
short, one would expect to find deep, clay soils with high
oon
.COD
I-z
CO
ala
U,
stH
:dz
L~i0
ti0
t z
e0
siH
tlH-
'CO
.e~S
10
La"a
o'S
0)
00
0
ncO
0
0
4-
~P-l
a
CO
0-
;lop
U
COC
CO
z
roont50)byA
Lopao
5
CO
IgeI
CO
co H-
H- CO
O O O
O Ic I
bil01
CO CO 10
000M
ncec
1 COH
m 10
t666
I Ip
o00
i)0100H
Co00
t666
-I C "-I
jO O O
1001{00
H- 10 C
0010
clI Lt)
CO 00
10 
O 
0
eH- 
01
CD
10
00T6
di
CO
n6
eccq
ite-
6
.d7
C
00
CO
co
6
CO
3eCo
tY0
,a6
0
CO
10
sif
6
C"
r--
01
1?
iin
6
CO
01
te6
~~00
rr10
I6
t~H-
~v10
an0
IttI- -qCC 1001000C
C)1-C00I 1-I-0 m
G 0o I'll OCOcc0
CO ->--1:r--0000
6666 566614
1 1
N01Co CO01 -CO?
COr- 0r-0 r-000
OH-OH-0000--
Co 101 - - 00C:
000 r--CO 0 0--
H- 010 001C 0 -{
6666666C
OCOs~-
0- 100Co e)COC
010000
10y0 F 0O d 0
0O0001 Co 0
'-O0 CO
}- Co OH- 0
0 10 d r1
010H-00
O 0
- c 01
6664
I 
I
oo
0Co
i00
LCo
-Ico 0 10co O L 0
00co 10 ) cCo 1
Coq 00 0 O - C0
t666 6 6 0-
1001 CO C0 0
S000 0 H- 0
--+Co C ICO O0
0000 10O
0 H- 0C
Co00 C 0r-1
H- 00 0
COO~0
co 000
OoO
COO
010
O6
000,
010j
010 I I I d
t I 
a
-z z p w vx
c0)r4-i
0)
0
6Tdd
O
PIc
CZCl
O"" 
U
0)d
- 0)
cco
10U
0 U
"0
00C
m0
S4-J
c c
O0
00
U CO
SC
M U,
0
TABLE 8. ADJUSTED LINEAR REGRESSION COEFFICIENTS OF
SIGNIFICANT COVARIABLES WHEN THE DEPENDENT
VARIABLE IS SITE INDEX
Covariables b
Altitude -0.0038 meter of site index per
meter of altitude
Number of trees per hectare... --0.0017 meter of site index per
tree per hectare
Basal area per hectare ________________0.3232 meter of site index per
square meter of basal area
per hectare
As aspect changes from northwest to southeast, however,
ages decrease, and this is a relatively strong correlation.
This series of correlations indicates that poorer sites faced
northwest and were the first planted. Therefore, a logical
correlation between age and site index should have been
negative. However, the actual correlation was weak and
not significant even with 489 observations. In short, while
relationships do exist among these variables they are so
feeble that an analysis of variance based on only 72 ob-
servations was not sufficiently sensitive to detect the ef-
fects.
Number of trees per hectare could be considered a
stand density variable, but when considered without ref-
erence to tree size it is difficult to explain the effect of
number of trees per hectare in a stand density context.
Usually number of trees per unit area is inversely propor-
tional to stand age, and this is generally true in this case,
Table 8. Furthermore, site index is inversely proportional
to the number of trees per hectare. This should indicate
that site index is related to the number of trees per hectare
through age. However, the correlation between site index
and age is not significant. Consequently, the relationship
must exist through some other (and perhaps obscure)
channels. There simply is no good explanation for this re-
lationship. Its cause may be happenstance, but the re-
lationship seems too strong for such an explanation.
Basal area per hectare is clearly a stand density variable,
but the influence of this factor on site index is not easily
explained. According to the assumptions on which site
index is based, height growth of dominant and codomi-
nant trees (such as those in this study) is dependent only
on age and site quality. Stand density in reasonably closed
stands has little or no influence on height growth. This is
not precisely correct with some species, but the error in-
troduced by ignoring this effect when determining site in-
dex usually is quite small. Consequently, the very pro-
found effect of basal area per hectare on tree height and
site index in this study is surprising. The only logical ex-
planation is that E. globulus is highly sensitive to stand
density insofar as height growth is concerned. As stand
density increases, trees grow faster in height. If this be
the case, use of site indices with this species would be
invalidated and some other measure of site quality would
have to be used.
Another problem was lack of significance of the effects
of irrigation and fertilization on site index, Table 6, par-
ticularly since some effects of these variables on tree
height growth had been detected, Table 3. Again the ap-
parent effects were so small that they could not be de-
tected in an analysis of variance based on a sample of only
72 observations. It is possible, however, that the irrigation
and fertilization were applied at the wrong time and/or
in improper quantities, causing the effect to be negligible.
Further study in this area might lead to more precise and
effective treatments.
D.b.h. Data
To complete the study, an analysis of variance of the
mean plot d.b.h. values was performed using the same in-
dependent variables (including site index) that were used
in the tree height analysis, Table 9.
The three stand density variables (planting space, num-
ber of trees per hectare, and basal area per hectare) were
significant as expected. Regression coefficients associated
with these variables are given in Table 10. The more
space available to recently planted seedlings, the faster
these seedlings grew in diameter. As the number of trees
per hectare declined through mortality, there was a cor-
responding increase in mean diameter of the residual
stand. The regression coefficient in Table 10 for basal
area per hectare indicates that d.b.h. increased as basal
area per hectare increased. This can be explained by the
fact that average d.b.h. increases at a rate faster than
basal area because of higher mortality in the smaller size
TABLE 9. ANALYSIS OF VARIANCE OF THE E. globulus DATA
WHEN THE MEAN DIAMETER BREAST HIGH (d.b.h.) OF
THE STAND WAS THE DEPENDENT VARIABLE
Level
Source d.f. M.S. F of sig-
nificance
Planting method (P) 1 
36.185 1.278 N.S.
Irrigation (I) 1 22.985 
0.812 N.S.
Site preparation (S)...... 1 9.953 0.352 N.S.
Fertilization (F)-------- 1 4.546 0.161 N.S.
PxI 1 0.019 0.001 N.S.
PxS 1 0.332 0.012 N.S.
PxF------------------. 1 75.161 2.655 N.S.
IxS .----- _........... 1 28.584 1.010 N.S.
IxF 1 19.751 0.698 N.S.
SxF -1 2.373 0.084 N.S.
Age (linear) 1 76.207 2.692 N.S.
Altitude (linear)-------- 1 57.828 2.043 N.S.
Slope (linear) 1 0.051 0.002 N.S.
Planting space (linear)... 1 706.307 24.948 0.005
Number of trees/ha.
(linear) 1 149.174 5.269 0.025
Basal area/ha. (linear) ---.. 1 534.068 18.865 0.005
pH (linear) 
1 1.995 
0.070 N.S.
Per cent sand (linear).... 1 0.327 0.012 N.S.
Per cent silt (linear)-... 1 4.925 0.174 N.S.
Depth of A&B (linear)... 1 141.706 5.005 0.050
Aspect (linear)- 1 1.780 0.063 N.S.
Site index (linear) .............. 1 354.419 12.519 0.005
Residual 50 28.311
TABLE 10. ADJUSTED LINEAR REGRESSION COEFFICIENTS OF
SIGNIFICANT COVARIABLES (DEPENDENT VARIABLE IS d.b.h.)
Covariable b
Planting space ...
Number of trees per hectare_
Basal area per hectare
Depth of A&B horizons
Site index
+0.3003 inch of d.b.h. per
square meter of planting
space
-0.0003 inch of d.b.h. per tree
per hectare
+0.0641 inch of d.b.h. per
square meter of basal area
per hectare
-0.0097 inch of d.b.h. per cen-
timeter of soil depth
+0.1277 inch of d.b.h. per me-
ter of site index
[91]
- -~------, -,
classes. These results are reasonable and agree with exist-
ing theories.
The appearance of depth of A and B horizons 
as a sig-
nificant variable in the analysis of d.b.h. 
data in the pres-
ence of site index as an independent variable was not 
an-
ticipated because depth of these horizons is normally 
ex-
pected to be an integral contributor to site index. 
The A
and B horizons are zones that contain the greater portion
of the available nutrient capital and the most readily avail-
able water supply. Thus, it apears that the deeper these
horizons, to a point, the better the site quality. This, then,
indicates that the harmonic site index curves developed
in this study are inadequate. The use of soil depth as an
independent variable, along with height and age, 
prob-
ably would improve the predicting power of the site 
index
values.
SUMMATION
This attempt to establish the importance of certain sil-
vicultural and ecological influences on height growth of E.
globulus has been limited to factors for which data were
available. It is only a foundation for further study and
understanding of the culture of this species in the Hianaco
Valley of Peru. In spite of this, certain general observa-
tions can be made that should be of value to persons
working with this species in Peru.
Site preparation significantly contributes to increased
height growth. Planting seedlings in cans or on prepared
sites improves height growth. Irrigation plus fertilization
on prepared sites also improves tree height growth.
Special note must be made of the stand density char-
acteristics (number of trees per hectare and basal area)
that apparently affect height growth more strongly than
would be expected from experience with most forest spe-
cies. More study into this subject is warranted.
Site index and its components (aspect, soil depth, pH,
soil type) also were shown to have effects 
on height
growth, as was expected. However, the strong 
effect of
altitude, even when site index was included 
in the analysis,
warrants mention as an important and powerful restraint
on height growth. The provisional site index curves de-
veloped in this study obviously need refinement 
so as to
recognize the unexplained aspects of these factors. 
It is
possible that the site index curve equation should include,
as independent variables, altitude and expressions of 
soil
characteristics, in addition to age.
Diameter growth, not unexpectedly, was found to 
be
affected most powerfully by stand density. What 
is sur-
prising, however, is that all these expressions 
of stand
density (planting space, number of trees per hectare, 
and
basal area per hectare) were significant contributors in the
presence of one another. Wider spacing, fewer trees per
hectare, and greater basal area per hectare all showed sig-
nificance at the same time. One would expect that any
one of these would mask the others. The presence of all
three as significant variables needs study.
Site index also showed a strong positive relationship
with diameter growth. In addition, one of the compon-
ents of site index, the depth of A and B soil horizons, 
had
a significant effect. As the depth increased diameter growth
decreased. The reason this factor was significant at 
the
same time as was site index must be determined.
No attempt has been made to fully explain why certain
relationships exist as were found in this 
study, although
an attempt has been made (speculatively) to explain most
of them. However, this effort hopefully may serve as a
guide to further the study of this valuable species to the
Peruvian forest economy.
APPENDIX
Determination of the Index Direction
for the Measurement of Aspect
Since it is not logical to assume that height growth
would vary according to direction of slope when meas-
ured using the azimuth from north, the aspect data col-
lected in the field had to be modified. In temperate por-
tions of the northern hemisphere the most productive up-
land sites usually are found on northeast-facing slopes.
These slopes usually are relatively cool and 
damp and
thus provide a more favorable water regime than 
slopes
facing in other directions. Conversely, the poorest site
(those that are hottest and driest) usually are found on
southwest-facing slopes. Extending this principle to the
southern hemisphere would lead one to expect that the
best upland sites would be found on southeast-facing
slopes and poorest upland sites would be found on north-
west-facing slopes.
To describe this relationship mathematically, aspect is
expressed as the sine of the azimuth of the slope, meas-
Sine of aspect
angle from NW
1.o0I-
.9
.7
.6
.5
4
.3
.2
I I I I
0L
180s
site
450
Azimuth of
900
aspect from
APPENDIX FIG. Trace of the value
when used as in this study.
135
?
poorest
of the sine function
[10]1
IP c+rl-\nrr ntnnr\+ rr+ I r
L
APPENDIX TABLE. F VALUES FROM ANALYSES OF VARIANCE OF THE 
TREE HEIGHT DATA COMPUTED USING ASPECT VALUES
OBTAINED FROM DIFFERENT INDEX DIRECTIONS
F values with aspect from
Source Due W N80?W N70?W N60?W N50?W N40?W N30?W N20?W N1O?W
Planting method (P)--- 0.481
Irrigation (I)---------- - 0.014
Site preparation (S)- - 1.639
Fertilization (F)- --- 0.015
PxI 0.659---------------------
PxS . 10.656------------------
PxF0.13----------
IxS 2.860----------------
IxF 3.396-------------------
SxF--------- - ---- 0.198
Age (linear)---------- - 9.361
Altitude (linear) ----------- 1.421
Slope (linear)-------------- 0.850
Planting space (linear)------ 3.074
No. tree/ha. (linear)--------- 1.992
Basal area/ha. (linear)------- 0.343
pH (linear)--------------- 2.283
Per cent sand (linear)-------- 0.053
Per cent silt (linear)-------- 0.380
Per cent clay (linear) ------- 0.078
Depth of A&B horizons
(linear) ------------ ----- 10.784
Site index ----------------- 256.112
A spect-------------------- 0.857
0.523
0.020
1.775
0.069
0.653
10.237
0.111
2.857
3.942
0.209
8.203
1.569
0.671
2.710
1.832
0.323
1.683
0.034
0.319
0.063
0.761
0.251
2.266
0.156
0.753
10.573
0.148
3.150
4.922
0.361
7.016
1.147
0.395
2.018
1.688
0.260
1.021
0.026
0.274
0.057
1.563
0.728
3.680
0.180
1.235
13.098
0.257
4.347
6.082
1.145
6.854
0.256
0.227
1.216
1.586
0.214
0.751
0.019
0.207
0.045
11.660 12.598 12,658
250.741 240.798 227.270
0.029 0.832 6.044
2.934
0.534
5.683
0.017
2.145
18.229
0.426
6.433
6.164
2.869
9.339
0.137
0.256
0.806
1.789
0.194
1.411
0.045
0.225
0.056
10.612
220.094
15.6463
3.321
0.000
6.304
0.145
2.450
22.201
0.476
7.334
4.195
3.964
13.695
0.633
0.820
1.054
2.357
0.320
3.529
0.110
0.329
0.090
7.923
230.186
22.795
1.653
0.665
4.152
0.386
1.495
19.727
0.282
5.247
1.857
2.455
16.223
0.003
2.843
2.567
2.768
0.769
6.065
0.086
0.343
0.073
7.088
261.688
18.370
0.651
0.550
2.339
0.134
0.786
14.513
0.144
3.403
1.770
0.824
13.610
0.840
2.316
3.458
2.548
0.762
4.880
0.060
0.354
0.068
0.435
0.214
1.825
0.012
0.618
12.079-
0.108
2.838
2.249
0.381
11.485
1.662
1.777
3.540
2.256
0.645
3.688
0.038
0.324
0.057
ured clockwise or counterclockwise to a maximum 
of 1800
from the direction of the poorest site. The Appendix 
Fig-
ure shows the reason for using this function. The 
sine
value is lowest when the slope is in the direction associ-
ated with the poorest site and is greatest when 
the slope
is in the direction associated with the best site.
The proximity of the Equator to the study area 
is not
considered in the preceding discussion. However, 
it is
possible that slope directions associated 
with the poorest
and best sites are not northwest and southeast, 
respec-
tively, in an area such as the Hianaco Valley, 
which is
only 100 south of the Equator. To 
give aspect a proper
weight in the analysis of height data, measurements 
would
have to be taken from the appropriate 
direction. To do
this, the data were analyzed first when 
the index direction
was N80 ?W, then when it was N70 
?W, and continuing
at 100 intervals to N 1
0
W. F values associated with these
analyses are given in the Appendix 
Table. From this ta-
ble, the F value associated with 
aspect was greatest when
the index direction was N40 ?W. This 
indicates that aspect
has its greatest effect when measured 
from this approxi-
mate direction. Consequently, it 
was, assumed that the
initial hypothesis (that the poorest 
site was associated with
a northwest-facing slope) was substantially 
correct and
the subsequent analysis was made 
using this base for the
aspect variable.
[ 11]
8.488 9.818
270.101 267.348
8.726 3.759
i,
I1fI V.1V. V em . v .~v.. ." -