CORRELATION OF TREE GROWTH AND RAINFALL Range of tests. — Fairly extensive correlations were made between

tree

growth

and

Chacon and

Santa

Fe

rainfallboth for groups

and

for individual trees.

The

trend

method was

applied in its complete

form

until it

was

determinedthat variationsof i or 2 years did not unduly distortthe results. In addition to the

more

extensive corre- lations, selectedtestswere

made

betweentree

growth and

therecords ofotherrainfall stations.

Tree growth (groups) and

Chacon

rainfall.

— ^Groups

^{i to} 9 were correlated with

Chacon

rainfall for the month-intervals

shown

in table 15.

However,

the tableincludes only thosegroupswhich

were

mostsignificant.

The

table shows that correlations with March-July

and

January-

August

rainfallare the highest,andof these

two

intervals

March-

July is the

more

important. July rainfall is necessarily included as is indicatedbythelowercorrelationsof March-June.

Of

the 5months,

March-

July, the rainfall of

May- June

is

more

important to tree growth than that of March-April

and

the rainfall of April is of less importance than that of

May,

June, or July. Apparently tree growth,as represented bythetrees selected, respondsdirectly tothe rainwhichfalls during

and

the several weeks immediately preceding the actualgrowingseason.

The

moststriking factobvious at first sight isthe correlation be- tweengroup₇ (alltrees) and

March-

July rainfall,the trend coeffi- cient being 0.965

and

the ratio of opposed trends 0.12.

A

trend ratioof 0.12

means

thatthetrend ofrainfall

was

oppositetothetrend

^1Bot. Rev.,vol.7,pp.651-655, 1941.

NO. l8 TREE

GROWTH AND RAINFALL— CLOCK

₂₉ oftreegrowth for4years out of32ofvariation.

Of

these opposite trends that for 1910contains 84percent of the numerical disagree-

ment

and, if 1910be eliminated, the trend coefficientrises to 0.994.

A

comparison of the rainfall

among

all seven stations with thetree growthofgroup_{7 for the}4years of oppositetrend, 1910, 1912, 1913,

Table _iS-—Correlation of treegroups mid Chacon rainjail

Trendcoefficientsandratios ofopposed trends

G4 G s G 7

Go

G 10 G II

^^^'"•-J^Iy 0-93 0.82 0.96s 0.88 0.06 0.92

(0.22) (0.31) (0.12) (0.19) (0.16) (0.22)

Jan.-Aug 0.92 0.73 0.95 0.88 0.94 0.91

(0.16) (0.31) (0.19) (0.25) (0.22) (0.28)

Mar.-June 0.85 0.82 0.89 0.77

(0.28) (0.28) (0.16) (0.28)

May-Aug 0.80 0.68 0.80 0.66

(0.25) (0.37) (0.28) (0.28)

May-June 0.73 0.70 0.72 0.49

(0.31) (0.34) (0.28) (0.34)

Tan.-May 0.72 0.32 0.70 0.62

(0.25) (0.44) (0.28) (0.41)

Nov.-May 0.71 0.26 0.68 0.59

(0.23) (0.42) (0.26) (0.39)

Mar.-Apr 0.50 0.44 0.55 0.52

(0.41) (0.44) (0.37) (0.50)

May-July 0.84 0.71 0.85 0.71

(0.22) (0.31) (0.19) (0.25)

April 0.04 0.20 0.09 0.14

(0.47) (0.47) (0.56) (0.56)

May

0.75 0.34 0.64 0.27

(0.25) (0.44) (0.34) (0.41)

June 0.42 0.79 0.59 0.54

(0.47) (0.25) (0.37) (0.44)

July 0.62 0.48 0.66 0.63

(0.41) (0.50) (0.37) (0.37)

August 0.39 0.33 0.35 0.27

(0.37) (0.44) (0.41) (0.41)

and 1931, shows that

from

one to three stations disagree with the remainderineachcase.

The

₅years of greatest ^parallelvariation in the complete record have only one station disagreeing with the re- mainderforIyear.

Where

theparallelvariationsare of small

amount

the different rainfall stations are

much

at variance with each other forall years. Therefore,it ispossible to speculate that the rainfall at the siteof the trees actuallv agreed with tree growth; however,

30

SMITHSONIAN MISCELLANEOUS

COLLECTIONS VOL. Ill the data at

Chacon

are the closest legitimate record

and must

be retained as they stand.

The

quality of the correlationbetweentree

growth

and March-July rainfall at

Chacon

forthe 33-year interval (table 15) ^isallthat can be ecologically expected considering the distance between

Holman

Pass

and

Chacon, andconsidering the quality of the correlations be- tween the rainfall of

two

stations approximately as far apart as

Holman

Pass and Chacon. This suggests that the trees as a group follow witha high degreeofaccuracy the fluctuations ofrainfall at theimmediatesite.

In general, group 7

shows

slightly higher correlations than the othersand group5 slightly less^;otherwise thereis littlechoice

among

them.

Group

4,theclosest toChacon, hasaveryslightadvantage over group 9, and both have higher correlations than group 5, which is ecologicallylesssimilartothe other

two

thanthey arebetween them- selves. Inthecase ofgroups 10and 11,theformer

(from

the drier sites) has a slightly higher correlation than thelatter although not sufficiently so to justify any conclusions.

Group

7, containing all trees, possesses slightlybetter correlation than group ₇ (restricted), thevalues for

March-

July being 0.95 and (0.28) and for January-

August

0.94

and

(0.19).

Figure

4 shows Chacon

rainfall for

March-

July

compared

with treegrow^thof the several pertinentgroups.

The

charted correlations ofgroup7with

Chacon

rainfall infigure6 indicateingeneral that theabsenceof

summer

rainfall

and

the pres- ence ofwinterrainfall militateagainsthighagreement. Itisneither springrainfallalonenor springcombined with winter rainfallwhich gives highest correlations but springaddedto early

and midsummer

rainfall.

Tree

growth

(individual trees) and

Chacon

rainfall.

—

Individual trees were correlated with the

two

rainfall intervals of

March-

July and January-

August

(table 16).

The

resultsare to be expected, no doubt,inviewof theformergroupcorrelations. In general, thetrees agree alittlebetterwith

March-

Julythanwith January-

August

rain- fall. Tree

HPC

3, a foxtail pine, has thehighest correlation and

HPC

9, a ponderosa pine, has the lowest.

However, HPC

5, a Douglasfir,runsaclosesecondto

HPC

9.

As

a matteroffact, tree

HPC

3,whichstandsbetween

HPC

ⁱ

and

2, could beusedasafair substitute forgroup7. Ponderosa pineshave no advantage overthe otherspecies.

On

thewhole, thetrees eastof thePasscorrelatebetter than those on the Pass and these latter slightly better than those

NO. l8 TREE

GROWTH AND RAINFALL CLOCK

31 Table 16.—CorrelationofHolmanPasstreesand Chaconrainfall

Trend coefficientsandratios ofopposed trends 1909-1941

March-July

HPC

I 0.92

(0.25) 0.89 (0.25)

0.92 (0.19)

0.89 (0.25)

0.81 (0.34)

0.78 (0.31)

0.90 (0.22)

0.88 (0.22)

0.51 (0.28)

January-August 0.90 (0.31)

0.90 (0.25)

0.92 (0.12)

0.83 (0.2s)

0.68 (0.41)

0.77 (0.31)

0.87 (0.34)

0.90 (0.22)

0.62 (0.34)

Holman

Pass

Tree Growth

and Chacon Precipitation

Percent ^PerCent

Group

4

Precipitation March-tIuly

Precipitation March-July

Group

1910

^0^

30 ^{i^'*^}

Fig4—Graphs^{of treegrowth andrainfall 7miles}distant, inrawpercentages.

Group4,eastofPass;groupS,on Pass;group9,westofPass;group_10,nor- malordry-site trees:group11,wet-site trees;and group7,alltrees.

32

SMITHSONIAN MISCELLANEOUS

COLLECTIONS VOL. Ill west of the Pass althoughthereare individual exceptions.

The

most striking contrastappearsbetweenthetrees

from

the driersites,group

10.andthose

from

the wetter,group ii.

In

summary,

it is rather clearly evident, first, that a group is superiorto single treesfora record ofrainfallvariationsand, second, thatthe variations

shown among

thetrees in table i6, especially inthe ratios of opposed trends, emphasizethe influence of

what

has pre- viouslybeenreferredtoas micrositefactors.

A

unionof several^tree records apparently generalizes the record of response to rainfall.

When

consideration is given the factsthat the trees do dififer

from

each otherby an

amount

tobe expected overa short term, inviewof the variations

among

^different rainfall records themselves; that the treesare several miles

from Chacon

;that rainfall isbut one growth factor in a complex;

and

that rainfall itself is rather remote

from

itsincorporation into the hydrostaticsystemoftheplant,the correla- tions not only between rainfall and tree groups but also between rainfallandindividualtreesare surprisinglyhighfortheperiod 1909- 1941.

Tree

growth

(groups)

and

therainfallof otherstations.

—

^Certain

groups

were

correlatedwiththerainfallof thestations atBlack Lake,

Taos

Canyon, Taos,

and

Albuquerque.

The

results for four of the groupsare

shown

in table 17. Beforecontinuingitshouldbe mentioned thatthese particular correlations

were

not included todemonstratethat tree

growth

can be

compared

to distant rainfall with significant re- sults or to indicate favor for such correlations.

They

are

shown

rather because they appear to indicate that detailed influence of specific rainfall subsides with distance and only general variations

common

totheregionremain.

With

ratiosof opposedtrendsranging

from

0.22 to 0.50, tree

growth

in one locality gives a poor picture ofrainfallvariations atadistance.

On

the one hand, correlations with

March-

July rainfall, the best inthe case ofChacon, are

mixed and

poor; itisdifficultto readany significanceintothem.

On

theotherhand, correlationswiththe

more

general interval of January-

August

rainfall are higher

and more

consistent and emphasize the regional regime.

Even

so, the

number

of instancesin whichthetreesrespondina direction oppositeto the rainfalltrendsmilitatesagainsttheuse of treegrowth,as exemplified by the

Holman

Pass collection, for an accurate gauge of regional rainfall variations

from

season to season. This is not to say that smoothing

would

notbring out general trendsifthe influence of other

NO. l8 TREE

GROWTH AND RAINFALL — CLOCK

₃₃

\o ^{c^} o\ (^

oooooooo

o o o o

o o O o o

1—> flO lO

6 6 6 6 6

1^PDt^fO^O ro00 -^

66666666

o,^1- N ^{-^o\00} 6 6 6 6 6 6

^

•I 0\00 »o^1-1VO00

for>.Nvo f?00 N

6 6 6 6 6 6 6

i-H

o

i^ -^"^t^ 00 rfj^fOr^^P) l>^CV|

6666660

34

SMITHSONIAN MISCELLANEOUS

COLLECTIONS VOL. Ill factors wereeliminated

and

if thetreeswereproperlyselected

from

the regionand

from

theproperzone.

Furtherto test thegeneralrelationships,the

March-

July

and

Jan- uary-Augustrainfallof Chacon,

Las

Vegas,

Taos

Canyon,

and

Black

Lake were

combined for the period ₁909-1941

and

correlated with group7.

The

results follow.

Raw Smootliefl

percentages percentages

March-July 0.95 0.96

(0.19) (0.22)

January-August 0.96 0.86

(0.16) (0.31)

Although these values are high no advantage results

from

the use of the combined rainfall.

The raw

percentage values for group ₇ (restricted) correlated with

combined

rainfall of

March-

July are 0.955

and

_(0,22), andofJanuary-Augustrainfall0.955

and

(0.12).

An

analysis of table 17 shows, further, that altitude in general has

some

effect: the correlations are slightly higher for group 9,

which isclosertotheaverageelevationof therainfall stations.

Table 18.

—

Correlation of^treegroupsafuiLas Vegas rainfall Trendcoefficientsandratios ofopposedtrends

March-July

1893- igio-

1941 1941

G

4 0.78 0.93

(0.33) (0.28)

5 0.73 0.81

(0.31) (0.27)

9 0.59 0.64

(0.47) (0.44)

7 0.79 0.91

(0.35) (0.31)

Trees

HPC

3, 5, 7, and 9

were

correlated with the stations listed in table 17.

The

results are similar to those for the groups in the tableexcept for

somewhat

lowervalues.

Tree growth

and Las Vegas

rainfall.

— ^With

the exception of SantaFe,

Las Vegas

has thelongestrainfallrecord of any station in thegeneral area butit is

some

37 miles distant

from Holman

Pass

and

3,000 feet lower. Table 18 gives the trend coefficients

and

the ratiosofopposedtrendsbetweentreegrowth

and

Las

Vegas

rainfall for the intervals

and

years noted.

On

the whole, the correlations

January

NO. l8 TREE

GROWTH AND RAINFALL CLOCK

₃₅ withMarch-Julyrainfall slightly exceedthose with January-August.

They

decreaseinqualitywithdistance;thatis,correlationsofgroup 4 (east of Pass) are highestandthose ofgroup 9 (west of Pass) are lowest.

The

most striking feature of the table is the decided increase in correlation of theperiod1910-1941 over the period 1893-1941. Dur-

mg

the later period (1910-1941) the trees follow

more

closely the variationsinrainfallas recordedatLas Vegas.

Tree

growth

and Santa

Fe

rainfall.

—

Although Santa

Fe

isdistant

some 40

miles

from Holman

Pass it isworthwhile, because of the length of record, to compare Santa

Fe

rainfall with treegrowth in ordertodetermineifthe quality ofcorrelationvariedthroughoutthe length of that record. Nine tree groups were correlated with all rainfallintervalsfor the periods 1850-1897 and1898-1941 separately.

Data most pertinent to the studyappear intable 19, which ^gives the trend coefficients and ratios of opposed trends for the periods mentioned above.

The

remainder of the data, not shown, simply corroborate

what

thetable itself shows.

On

the whole,tree growth correlates considerably better with March-July than with January-

August

rainfall. Here, however, in contrast with Chacon rainfall, groups4, 5,and7 agree

somewhat

betterwithMarch- Junerainfall.

General correlations are fair; they possess litde value except to

show

aregional tendency toward similarity during a portion of the years. This appearsin table20 wheretrendcoefficientsfor the period of 1850-1941 vary

from

0.52to0.67andtheratiosofopposedtrends

from

0.24to 0.37.

The

values for group 7 are 0.65 for the trend coefficient and0.35 for the^ratioof opposedtrends. Thus,a case of 35 opposite trends against65 ^parallelgives neither highnor depend- ablecorrelation. Surprisingly, the^trees fromthe wettersites, group II,

compare

most favorably with Santa

Fe

rainfall for the period 1850-1941. In view of the quality of correlation between

Holman

Passtree growth and

Chacon

rainfall on^the one hand and between

Chacon and

^Santa

Fe

rainfall ontheother, the correlation between tree growth and Santa

Fe

rainfall possesses values consistent with therelative distances.

Figure 5 shows Santa

Fe

rainfall for March-July comparedwith tree growth of the several pertinent groups. Figure 6 shows ^the correlationsincharted form.

For

the period 1898-1941 the trend of the graph resembles that for the Chacon correlations. It contrasts notablywiththegraph for the period 1850-1897,wheretheemphasis seems to be onspringrainfall.