TREE GROWTH AND RAINFALL CLOCK 43 within the Transition Zone up into the Canadian ; the number of trees

sampled

was

Hmited to nine; the collection contained trees of four different species

and

comprised three groups geographically and hypsometrically;thetrees,inthefield,divided themselvesecologically into

two

groups, the one (group lo) representing normal water relationships without excess drainage to or from the trees, and the other (group ii) representing slightlyabove-normal water relation- shipsfor the generallocality;thesamples consisted of increment cores

—

essentiallyone radius to represent theentire volume growth of a tree; site factors were judged solely by inspection on the spot; the terrain

was

mountainous; the nearest rainfall station, Chacon,

was

7 miles distantat an elevation intermediatebetween the highest and lowesttrees;andanalyseswerebasedon

raw

(i.e.,unsmoothed) data.

The

present studybrings out

many

points in

summary

which are distinctly secondaryto the

main

objectives. In the present stage of investigationsofthistype,allpoints,secondaryas well as primary, are highly suggestiveonly. It remainstobedeterminediftheprinciples

and

methodshereusedinthefieldandinthelaboratoryare of

more

general application.

Only

then can growth-layer sequences beinter- pretedintheabsence ofnearby rainfall stations. Obviously,

we

must

know how

trees reveal their ecologic information before

we

can determine

what

theytell.

Study

of grozuth^layers.

—

i. Cross-dating of high quality is not a necessary prerequisite to the correlation of growth-layer thicknesses

and

rainfall,

and

its nearlytotalabsence does not indicate a lack of significantresponseonthe part of thetrees to rainfall variations.

2.

The

presence or absence of high-quality cross-dating does not necessarily constitute the criterion wherebya treerecord is included inagroup averageorexcluded

from

it.

Some

othercriterionshould beapplied for the elimination ofcertaingrowth-layer sequencesafter the collection isbrought tothe laboratory, if such eliminationis at- tempted withjustification.

3. Partial disagreement

among

the various trees, growth layer to growth layer, emphasizes a definite localization of ^{site factors to} eachtree.

4. Disagreement

among

thetreesincreased with increasingdistance, distancemeasuredinyardsrather thanin miles.

5. Intergroupcorrelations (ofgroups_4,5,and9)were merely^fair, notatallstriking.

They show

a dual influence of^siteanddistance.

6. In so far as the^collection

from Holman

Pass isconcerned, the nature of thespeciesisclearlysubordinatetotheinfluenceofsite.

44

SMITHSONIAN MISCELLANEOUS

COLLECTIONS VOL. Ill 7. Correlations

among

different trees

and among

dift"erentgroups were distinctly lower for the period 1850-1897 than for the period 1898-1941.

8.

A

simultaneous comparison of trend

among

all trees yielded 9 complete agreements

among

the trees during the

48

years of the period1850-1897

and

15duringthe

44

years of the period 1898-1941.

9.

For

group 10 (dry sites) average year-to-year variation, aver- agedeparture,

and

average departure fromi

mean

variation increased whereasforgroup11 (wetsites)they decreased for the period 1898- 1941 in contrast with the period 1850- 1897.

However,

the average departure of

two

of the trees in group 10 actually agreed with group II.

10.

The

average departure

from mean

variation of group ₇ (re- stricted) and group 7 itself decreased for the period 1898-1941 in contrastwith theperiod 1850-1897.

11.

A

study and comparison of the growth-layer sequences

em-

phasize the role of site factors local to each tree and the striking contrast of characteristics between the

two

periods, 1850-1897

and

1898-1941.

Study

ofrainfall characteristics.

—

i.

Chacon

rainfall

was

correlated withthatof the other sixstationsfor eight different month-intervals.

Trend

coefficients ranged

from

0.61 to 0.99

and

ratios of opposed trends

from

0.41to0.09.

2.

No

^clear-cut pattern

emerged from

this correlation between

Chacon and

the other stations.

However,

the values declined with thepresenceof

summer

rainfalland with an increasein the

number

of

months

inthe month-intervals.

Within

the area

from

whichrain- fallstationswere drawn, distance

from Chacon made

little difference inthe variations

among

the several stations.

3.

The

average trend coefficient between

Chacon and

the other stations

was

approximately_0.89

and

theratioofopposedtrends0.23.

Ifthe trees

were

responding directlyto the rain falling at theim- mediate site,they

may

be expectedto correlatewith

Chacon

rainfall to a degree equaling or slightly exceeding (because of the distance involved) the average of the correlations between

Chacon

and the other rainfall stations.

4. Correlations

among

the eight differentmonth-intervalsat

Chacon

ranged

from —0.33

to 0.99 for the trendcoefficients

and from

0.66 to0.06 for theratio of opposedtrends.

Such

divergences

demanded

that treegrowthbe tested against the full series of month-intervals.

NO. l8 TREE

GROWTH AND RAINFALL CLOCK

₄₅

5. If tree

growth

shows high correlation with a certain rainfall interval,as

March-

July,

and

ifthatintervalhas highcorrelationwith asecondone, asJanuary-August,thentreegrowth

may

be expected to

show

high correlation with the second interval even though part of therainfallof the longerinterval

may

not influence growth.

6. Within limits,

maximum

correlation combined with

minimum-

length month-intervals should be the focus ofcriticalinformation on the response oftrees to rainfall.

7.

The

average March-Julyrainfall atSanta

Fe was

higher during the period 1898-1941 than during the period 1850-1897 and higher during the period 1909-1941 thanduring theperiod 1898-1941.

8.

Average

year-to-yearvariation, averagedeparture, and average departure

from mean

variation of March-July rainfall at Santa

Fe was

lessduringtheperiod1898-1941thanduringtheperiod 1850-1897.

9.

For

thecontrasted periods 1850-1897 and ¹898-1 941, the char- acteristicsof the dry-sitetreesran countertothose ofrainfallwhereas those of the wet-sitetreesran parallel.

10.

When

acriterion of conformity, based on average departure,

was

applied and the

two

trees notconforming were eliminated, the characteristics of the resultant group ₇ (restricted) followed those ofrainfall.

Correlation between free groivth and rainfall.

—

^i. Correlations betweentreegrowth andrainfallofChacon,the neareststation, were highest for therainfallof theMarch-July^intervalof the

same

year.

This is consistentwith the principle of

maximum

correlation with minimum-length month-interval.

The

next best correlation, with January- August,

was

also high, but the reason

was

held to be the ratherhighcorrelationbetweenthatintervaland

March-

July.

2.

The

growthof thetreescomposingthe

Holman

Passcollection correlateddirectlywiththeprecipitationwhichfellimmediately before and duringtheseason of greatestgrowth.

3. Correlationbetweenthe

Holman

Passtreesand Chacon March- Julyrainfall,based on

raw

(unsmoothed) data,attainedthe following remarkably high values: a trend coefficient of 0.965 and a ratio ofopposedtrends of0.12.

4.

The

accumulated evidence points rather ^clearly to the conclu- sion that thetreesrespondvery nearly 100 percentto fluctuationsof rainfall at the immediatesite.

5. Correlationsbetween ^{individualtrees} and Chacon rainfallwere lower thanthose for groups.

A

^{few were}surprisinglyhigh.

6.

The

nature of the species appearedto

make

little difference in the qualityof correlations.

46 SMITHSONIAN MISCELLANEOUS

COLLECTIONS VOL. Ill 7.

The

variations

among

the trend coefficients

and

ratios of op- posedtrends of individualtrees emphasizedagain the locaHzed influ- enceof sitefactors onthe singletree,the so-called microsite factors.

8. Correlations betweentree

growth and

rainfall of stations other than

Chacon

gave

mixed and

rather poor results. Certain regional tendenciesremained,but they are oflittle or

no

value.

9. Correlations betweentree

growth and

Las

Vegas

rainfall were higher for the period1910-1941 thanfor that of 1893-1941.

10. Generalcorrelationsbetweentreegrowth and Santa

Fe

rainfall were fair to poor and havelittle value as regards season-to-season fluctuations.

Such

results

were

tobe expectedinviewof the distances involvedandthe arealdififerences in rainfallasmeasured

from

station to station.

11.

The

higher the correlations

were among

the treesthemselves, the highertheircorrelation with rainfall.

An

increasein

amount

of rainfall

was

accompanied bygreateragreement

among

thetrees.

12.

The

most important information broughtout

by

the correlation of tree

growth

and Santa

Fe

March-July rainfall for the periods 1850-1897, 1898-1941,

and

1910-1941

was

this:

The

quality of the correlations

was

lowest for the firstperiod

and

highest for the last.

All treegroupings conformed.

The amounts

of

March-

July rainfall atSanta

Fe showed

a similar increase for the three periods.

13. In the above correlations, the trees

from

the wetter sites

showed

not only the highest correlations but also the greatestincreases.

14.

A

changeintemperature thataffected thetimeof

growth

ini- tiation in the spring probably shifted the month-interval of rainfall towhichthe treesresponded.

15. Trees

from

drier sites, as a group,

were

poor recorders of changes in rainfallcharacteristics; individually,

two

out of the four

conformed

in partto the wet-site group.

16.

A summary

of changes

from

the period 1850-1897 to the period 1898-1941 follows:

In March-Julyrainfall

:

Average variation, averagedeparture, and average departure from mean variationdecreasedwithanincreaseinaverage rainfall.

Intreegrowth:

Among

all trees,internal agreementincreased.

For dry-site trees, group 10, average variation, average departure, and averagedeparturefrommeanvariation increased.

For alltrees, group 7,averagevariation and average departure increased whereas averagedeparture from mean variationdecreased.

Forwet-sitetreesand group7 (restricted),averagevariation,averagede- parture,and averagedeparturefrommeanvariationdecreased,thusagree- ingwithchangesinMarch-Julyrainfall.