GEOGRAPHIC VARIATION IN SLASH PINE
(Pinus elliottii Engelm.)
ANTHONY E. SQUILLACE
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The author gratefully acknowledges the assistance of his
supervisory committee: Drs. A. T. Wallace (Chairman), W. O.Ash,
A. D. Conger, R. E. Goddard, C. M. Kaufman, and G. R. Nocgle.
Special gratitude is also due the following organizations for
assistance in collecting seed and foliar samples used in the study:
Buckeye Cellulose Corporation, Foley, Florida; Continental Can
Company, Incorporated, Savannah, Georgia; Florida Forest Service;
Georgia Forestry Commission; International Paper Company, Bainbridge,
Georgia; School of Forestry, University of Florida; South Carolina
Commission of Forestry; West Virginia Pulp and Paper Corporation,
Georgetown, South Carolina.
Statistical computations were done at the University of
Florida Canputing Center and financed jointly by that unit, the
School of Forestry of the University, and the Southeastern Forest
Experiment Station. The author is Grateful to these orCanizations
and to personnel who helped with the work. In this effort,
special thanks are due Ir. John C. Barnes who did most of the
proLramning and Dr. A. E. Brandt for statistical counseling.
Finally, the author is very grateful to the Southeastern
Forest Experiment Station for providing facilities, materials,
and technical help and to the many persons of the Olustee Uaval
Stores and Timber Production Laboratory who helped in various
ways on the project.
TALU OF CUMO
LIST OF TABLM . .... ...... ............ iv
LIST OF FInFW ............. ............ v
IQnU erTUl ... .. .. .... ........ 1
IVIVW OP LnUTtJ ....... ............ 4
81uBhn P . . .... . ..... .. .. . .. .. 8
BASIS FO YTAIATIOK IN KAU PIB . . . . . . . . 15
PMrogm tera . . . . . . . . .. . . 2
PaantalNet a........................... 24
Single rate analys ........ . .. ... 29
tivariate analysis . . . . . . . . . 35
5RWMB yAeNldU .. . . . . . . . . . . .. 37
Sgle ariMte Anale *...................... 37
ene dimens nss...... ............. 37
e ielt .......... l. . . . . .... . 3
SeeB A.b dimaiOa . . . . . . . ..... 37
Sl eed Y.Led* A. L. . a a a a a . a a a * * 37
GerSiudility and s ed of gft intion . . . . . 46
Getyledemin ber . . . . .. . . . . . 54
Total height . . . . . . . . . 56
St d e r .... . . ... .. . ...... 63
Needlesper faile . . . . .. .. . . 68
Deirlength ................ ....... 72
suffer of resin ducts ... .... ......... 85
ThldKnee ef h peB ...... .............
Mltivawlnae Analys ... . . . . . . . 106
NDvecaitury Amh ndirtions Within Sta nd .. * *. 117
DivTAsity cmogie s . . . . . . . . . 119
Nomenclatural Gansihdrations ................. -17
SUMMY AIM CdMMIOM.....................0 0 0 0 0 0 0 0 0 0 0 119
LIMIAIM CITED . . ...... . . . . . . . 122
A . . . . . . . . ...... 130
LIST OF TABLES
1. Sumary of slash pine seed source tests which
sampled a relatively narrow latitudinal zone . . . .. 11
2. Summary of slash pine seed source tests which
sampled a relatively wide latitudinal zone . . . .. 12
3. Means and ranges of variation for parental data . . . .38
4. Mean squares and estimates of variance components
obtained from analyses of variance of parent tree data . 39
5. Means and ranges of variation for progeny data of
Nursery Test 2 . . . . . . . . . . . 47
6. Mean squares and estimates of variance components
obtained from analyses of variance of progeny
data of Nursery Test 2 .................. .48
7. Means and ranges of variation for progeny data of
Nursery Test 1 . . . . . . . . . . . 57
8. Mean squares and estimates of variance components
obtained from analyses of variance of progeny
data of Nursery Test 1 ......... ...... 58
9. Coefficients of variation for parental data--per cent . . 99
10. Coefficients of variation for progeny data of
Nursery Test 1--per cent . . . . . . ... 100
11. Coefficients of variation for progeny data of
Nursery Test 2--per cent . .... ...... .. . 101
12. D values (x 10), with stands arranged in order of
decreasing similarity to 8 stands in the north-central
region .. . . . . . . . . . . . .. 107
13. Average within- and between-cluster D values (x 10),
clusters formed as described in text and arranged in
order of decreasing similarity to "North-central
(west)" cluster . . . . . . . .... ..... 111
14. D values (x 10) for stands in a transect going from
stand 24 (north-central region) southward through
the center of Florida to stand 47 (south Florida) . . .. 116
LIST 0' FIfUM
1. Ranges of the two varieties *f slash pinE ad
locatil sf otf t lan... led . . . . . . 9
2. Z JIn y teqr tn ................. 16
3. Averg diffeumee between mu mudiw ad owM
indamr towmeratue (Or.) during athsl of April
thtre'h gtember. .. ................. 17
4. MMm amuel preeipitatiE .* . . . . * . 19
5. Preciydtatle from etsbesr %waeieh M -a pe r aet
of Sma mmal pleiitAribi. 0 0 . . . . . 20
6. TheM s ef precipitatia-ovgeortie ratios for
mths el Fbruay tlughe April. ... ... . .. .. 22
7. em patem of stadL veriatioa in teae length .* . . . 40
8. The pttera of stant vyriatiLe in er s a m err. .. .. . 41
9. The ptter of st ad variation in soad n sa yield
par OGe. a * a . a a * * * 44
10. ihe pattern o stand variation in seed night. . . . * 45
11. Te pttehrn of stead variation in s-e od dmability. . 49
12. The pattern of stead aristion in sped of se*a
13. The patter of stat ve atien in nue of
otyLeasi p er sNdling. . . ............ 55
14. The pattern of riat variatlia in heights at
needlias.................... ..... 59
15. On-yer-eld slsh pine seedlings, showing
diffrencs in total height and stm dimtr. . . . 61
16. Th pattern of sat variat ion in state dwieter. . . . 6
17. The pattern of stai variatiU in wrongs nuber
of o edles per facile in parest trees. . . . . 69
LIST OF FIGUtSB (continmd)
18. The pattern of stand variation in average number
of needles per fascicle in progenies.. .* * * *
19. Th pattern of stand variation in needle length
in parent trees. ... . . . . . ** *
20. The pattern of stand variation in needle length
in progenies. . . .. . . * . . **
21. The pattern of stand variation in fascicle sheath
length in parents. ....... *. * * *
22. The pattern of stand variation in fascicle sheath
length in progenie.. . . . . * . .
23. The pattern of stand variation in number of
rows of stomata per a. of needle width in
parents* e e e e e e * * * * * * * * *
24. The pattern of stand variation in number of rows
of staata per m. of needle width in progeniee. . .
25. The pattern of stand
stoeta per m. of
26. The pattern of stead
stmata per m. of
variation in number of
needle length in parent.
variationea number of
needle length in progenies.
. . .
27. The pattern of stand variation in number of
smatat per sq. m. of needle surface in parents.
26. Th pattern of stand variation in number of
stoata per sq. m. of needle surface in pregenies.
29. The pattern of staad variation in number of
resin ducts per needle in parents. .......
30. The pattern of stand variation in ntuber of
resin ducts per needle in progenies. . . . .
31. The pattern of stead variation in number of
layers of hypoderm in parents. . . . ....
32. The pattern of stand variation in number of
layers of hypedier in progenies. ........
LIST Or FIG0W (testiand)
33. Average of 3 valY e between each s4kad at eight
stwema within the Mnrth-central regiai, awing
the egree of similarity that regie.. . . . . 109
3k. Delimeatio of clusters of stols for ie in
determinga relative ips. . . . . . . . . 113
35. agrawatic rpreentation of the appey iuate
degree f similarity mes clusters of *stas
aeeording to average between-cluster D valws. . . . 114
When a plant species occurs over a vide geographic range,
individuals or populations growing in different localities frequently
display differences in one or mere traits. This phenotypie variation
associated with locality geographice varition) me be dme to eviren-
mantal or genetic factors, or interaction between thee.
lnviroamntal difference are a consequence of modifications
caused by habitat factors. Genetic variation associated vith locality
(racial variation), on the other hand, is due to such meahenism as
mutation, natural selection, hybridization, or combinations of these
factors. It bsieally results from the fact that the individuals
within populations differ genetically. The genetic heterogeneity
between individuals is caused by rtatioa or hybridization. It is
maintained by intricate macha-ims inherent in mest species, enhancing
chances of survival of the species in a constantly changing envireimat.
This genetic variation meng individuals is the basis for racial
If the localities are characterized by different envirommats,
and if sem degree of reproductive isolation is presents racial
variation will occur. Pleats that are genetically mest suited to
their particular habitat will wsrvive and reproduce in greater aubers
than these not so well ended. Sea degree of reproductive isolation
is necessary because if interbreeding occurs readely throughout a
species range, natural selection in a given locality would merely
result in a change in the men of the whole species. In forest trees,
sufficient isolation is provided by the limited distance of pollen and
Although natural sMlectia is the mat iwpertat meme of racial
variatie, it is belleed that umh variatiom my als result fro
chme. fluetwttoms in gwe frequo cies (emetic drift) losing to
fixatsxe of umes. Gretie drift is most pt to occur in mall,
isolated populaties ard oevirie-mei diffiraees ee aot be prreet.
(IG gr* aic variation oceurs in ckeracteristic pattern, depeding
upo the asture of the ferees that eamed it. Siae climatic factors
are eftm important natural seaetion forces, aft smine climate eftb
eh. Ms Cgratlly ever a species rage, the pattern of racial variation
frequently is oestinuous or elial. However, relatively uiform and
diasntinwuas hkbitatA mW cause relatively disfeete pepulatia s or
eeetypes. Likewise, prosnt or past iseatioe ay ease ecotypes or
acembitioAn of beth clinal d eeotypic variation.
Needles to ay, pogriphbic variation in foest trees is coma,
at it is of great interest to forest lad -age ad forest
setemtists. 'E aeture of geographic variation (i.e., the proportion
of evrommiter l ma agnetic cmpoments) is lpeortet to lad Masaers
beease if differences in eseamielly importat tracts are genetic
they met we eare in seaeeting srrees of se m for fewest plating.
Likeowse, forget otisists ae keenly Mwee o tie peesibilities
of espitalising en reial vTariateu in deveely-nt at se erior strains.
Tamommists m interest* in patterns of variatite in their attempts
to classify trees a both the species ad sur species level.
The present study was dsign*d mainly to investigte the nature
ad patt eus of geographie ad racial variatieo for a nWber of
chareoteristios in slash pine (Pinus elliettii enela.), me of the
more important commercial trees of the Southeast. Secondary objectives
were (1) to search for causes f patterns of variation that might be
found, and (2) to cpware the magnitude of variation associated with
localities against that associated with individuals within localities.
REVIEW OF LITERATURE
It is probably safe to say that geographic variation has been
studied in all commercially important forest tree species and in
many of the noncomnercially important ones. Langlet (1938)
summarized much of the early work. Several recent publications
include brief reviews of much of the past literature: Dorman (1952),
Critchfield (1957), Echols (1958), Squillace and Bincham (1958),
Callaham (1962), and Langlet (1953).
These studies have demonstrated that racial variation is
prevalent in forest trees, although some species such as red pine
(P. resinosa Ait.) chowed no, or relatively small, variation in
some traits (Buchnan and Buchman, 1962; and Wright et al., 1963).
As might be expected, differences were found to be Greatest, or
most prevalent, where the species range covered a large geoGraphic
area, such as ponderosa pine (P. ponderosa Lais.) and Scotch pine
(P. sylvestris L.). Uowvver, variation has been found even in
trees having a relatively small geographic range, such as sand pine
(P. clausa (Chapm.) Vasey)(Little and Dorman, 1952a), and western
white pine (P. nonticola Dougl.) (Squillace and Binjlam, 1958).
Mny of the patterns reported contained an elent of continuous
or clinal variation. Where the variation is a result of gradual
changes in climatic or geograhic features, and where complete repro-
ductive isolation is absent, one might, of course, expect the variation
in plant characteristics to be continuous. Stebbins (1950, p. 44)
expressed the opinion that most species with a continuous range,
enco~assing changes in latitude or cliaate, will be found to possess
lines for physiological characteristics adapting the to conditions
prevailing in various parts of their range. NHerous patterns showing
continuous variation associated with rainfall have been reported
(Larson, 1957; Thorbjornsen, 1961; Goddard nd Stricklad, 1962; aad
Squillace ad Silen, 1962). Blevational trends were reported by
Callahan and Liddicoet (1961) end Critchfield (1957). N* erus
istances of gradual changes associated with latitude or length of
photoperiod hxve been found (Latglet, 1936; and Schoenike ad Brevn,
One frequently also see in the literature evidences of ecotypic
patterns of variation (for exples, see Wright, 1944; Pauley and
Perry, 1954; Vaertaja, 1954; Squillace ad Binghm, 1958; ad Wells,
1962). However, same of these anthers used the term broadly, applying
it to patterns which re genetic ad atWptive but not neceserily
discontinuous. Too, there is often sem question as to whether the
ecotypic variation occurs exclusive of other types.
Theoretically, distinct ecotypes with no element of
continuity can occur in a species having geographical isolation,
and in which genetic adaptation to a uniform habitat (such as soil
or exposure) has occurred. However, since the habitat within a
species range or within parts of a species range often varies
continuously, combinations of patterns are more likely. Thus, it
is possible to visualize a situation in which a species occurs in
geographically isolated groups, with ecotypic variation occurring
among groups as a result of adaptation or genetic drift, or both.
But with the climate varying continuously through the range we
could have clinal variation occurring both within and between the
ecotypes. This may indeed be the situation in snme species such as
ponderosa pine, in which elevational gradients were reported by
Callahan and Liddicoet (1961), and in which ecotypes were
delineated by Wells (1962). In this same species, Squillace and
Silen (1962) pointed out apparent clinal variation associated with
climatic variables but acknowlveded that likelihood that
discontinuities also occurred; irregularities in a clinal pattern
were illustrated by Callahan and Hasel (1961). Clausen et al.
(1948) found clinal trends for height of plant between climatic
races of Achillea lanulosa. In Scotch pine, Wright and Baldwin
(1957) and Wright and Bull (1963) delineated broad ecotypes within
the species range, hlile Langlet (1936) pointed out that clinal
variation for certain characteristics occur both within and
between ecotypes of this species.
The existence or nonexistence of the two 1-inds of variation
often becomes a matter of degree, with interpretation highly
subject to the opinions of the investigator and confused by
terminology. It is no wonder that considerable discussion and
debate have resulted on this problem (Turesson, 1936; Faegri,
1937; Langlet, 1936, 1959, and 1963; Kriebel, 1956; and Callahan,
1962). Until more concrete terminology and guidelines for
classification are available (if indeed ever) the wise investigator
will describe his pattern of variation as best he can without
attempting to classify it categorically (Langlet, 1963).
Another type of variation noted rather frequently in the
literature is random variation. Here differences among stands
sampled within the species range may be real but exhibit no
distinctive Gcographical trends or patterns such as lines or
ecotypes. This type of variation is likely to occur where the
species range is discontinuous in the present or had been so at
some time in the recent past, as exemplified by the random pattern
found in the major portion of the range of European black pine (Pinus
nicra Arnold) by Urigit and Bull (1962). However, random differences
have been found for seed Cermination in slash pine by Hergen and
Iloelstra (1954). Likewise, Thorbjornsen (1961) reported random
variation for ving length, seed length, cone length, needle length,
and frequency of serrations on needle margins in loblolly pine (P.
taeda L.). Both of these species have rather continuous ranges.
The cause of random variations in such cases is obscure, although
partial reproductive isolation which is believed to be coamon in
most trees may have a bearing (Wright, 1943).
Slash pine, like m y pine species, has suffered a oacfused
neemnclature (Little a d baeam, 1954). Recently, these authors (1952b)
subdivided it into two varieties, P. elliettii angel var. elliottii,
typioel slash pine, and P. elliottii var. d4Mee Little ad Bormsn, South
Florida slamh pine, formally publishing a deseription of the latter.
The range of the two vrietles, as given by Little and Bonn
(1954), are shown in Figure 1. The authors showed the varieties as
being allopetric, the boundary between them being indieaed by the heavy
dashed line in central Florida. At a later date, Lagden (1963) published
a revised rage of the dean variety, extending it northward a considerable
distance a shown by the dotted line in Figure 1. He indicated that
trees of both varieties occur in the arma of overap. lash pine dees
not extend into the Caribben Islids.
Peatwes which, according to Little and IDoma (195k), distinguish
the two varieties are as follows:
Var. elliottii: Needles in fascicles of two and three, ad on ml
seedlings with erect, slender, pencillike stems.
Var. dena~ Needles in faseicles of two (infrequntly three);
seedling with grasslike, alaost steless stage with nry crowded needles,
and thick tap root. The vood of this variety is also heavier ad has
thicker o utnrood than the typieel variety.
Mature trees of the two varieties also differ srawwhat in general
ppiermee. Variety densa is norally shorter, with its stem often
foeuing into large branches sad its crown being generally flat-topped
at open, cared to the ui aly taller and relatively narrow-crwned
typical variety. However, according to mny foresters, these differeaese
ad ev the more Arttisettive sedling ehwaaeteristies beeme obscure
in the portions of the species rage whre the two varieties mt,
making it difficult or impossible to separae the tw vaTrieties.
lash pin*, being elatlvely sweeptible to fire inawy, was
originally caflined to pomns, ped orgias, mrl ether wrt aem
(Ceeper, 1957). With te avedet of white ma md fire protection it
hMa inaAed drier arems, whee it grows ?yatrielly with the relatively
fire resistant laongeL a pin (P. pal trlA Ml1.).
Seoth Florida slash pine oeeur in pue stads an flatweeds si es
in the southern part of its rage, while to the mort it is eofimet
to the wetter sites along stroms an in other porly drained or mwqy
area (Lemg9on, 1963). In the sMothern portie of its re there is
nm degree of geogrrepic ie latim betwom the two seestel ees,
eaMse by the 1verglades. The tw "preFes" along the eeest, however,
met in Polk a seeemla Seittee.
A nmber of seed suree studies studiess in which seeds were
selleeted f tress growing in different portion~ of the species ra~ e
sar plante in a s eem eavimmatt) have ben emdateted with slash
pine. Saom of these ampled only the eorther portiae of the species
rage (Tble 1), thile others namlet a relatively breed latitudinal
moe (Table 2). The studies were resigned minly to deteemine variatim
within the rage of *eliettli ealy. Newer, smling in some studies
of the latter greW (Table 2) eKtnled as far south s P rlk Ceosty,
Florida, wvteh is in the area brdert th tm varieties (treittioa
meo). In the "Fleridoe-a ergla" experiment (Table 2), a sinle mree
fell within the rage ef dmes (Collier Cesty, Florida) wv included
o -0 O S 0) >a
H.3 a p *QO .? -
; O 43 c0 *0 r 00
09 *1 0 O 0 y U W
N 0 *is 0 En C) ^g ) a
Hq -0 0 0* *U 0 a 0 1- 4
S*r > O H f 0 03
~I ( D T M > r .i S
44-43O H %14-i zrj ,
O aci M e 0 C1'o a
o 0 0 0 0) I4*)
0d 0 >C
A 0 *r-H
rd H 4
H rl o $
c N NO N
S0 -5 %
43 H H 0)
43I r-' 43
04 0) C ;4 0
43 4i o , I 0)0
4-3 r -, % 9 5a
0 0% ab S
H 3 0, $ j0 4- 4 d0
rg | 43 a)
ca -- i $
0 0 L El C In
El r O 8 '1
O ? Cr 0 0 U
S^'- c3 /3 ,C Do a
e1 ca 0C 431 A
HA 0 H
V3 a 43 44-
43 C *r -0 A md 0 rg 0
U BL* 0 0 C CO
S --' l OO Tl l 6u >0 40 C.
l (1) 4-) -P r
i iS Cs MO O [r c3 0
In C 0 *q 0H C s-
H1 4-) oi 0 4 I
a z 0 0 4.r4 4 4
+' O i 0 : T 4- d
1-1 Hi rf c k B C o
11 0 0 Cr -H $ r
) 4 ) C *
(U F 14 -9 U 0 WU 4 )H
H w 0 O C O O
ca 0 O4 -' Ta\
) + C c O f
S-i LC 0 1 o Q cL '
A pO 'd o
2C 4-e -' 0 * '- O O O
*r 0 -A
(D -P r
\ o 3 'd
o 4 -
4 d 0 0C A 0
+ G\ \ G a
0 J 0\ 0 0
0C to I..
0 ,d to LO ccL
O I r- 0 H -H
U ) O
*i < c1 0 r S *
0 v 5(12 ri 0
O 0 H t U- 7N -
E-1 4 4) O O O 0
SH H : H CM E
0i h (1 H
O H C)
0 H ^i 0 / g 0
H 0 0 0 H
OU H COP
H i r4' 4' 3 5- ,.
along with elliottii sources in cne of the seven plantations in the test,
but was not included in the statistical tests indicated.
As seen in Tables 1 and 2, significant differences were found more
frequently in those studies sampling a bread latitudinal zone than in
those sapling only the northern part of the species range. This was
especially true for growth rate. In one experiment, latitudinal growth
rate differences were mainly due to a sample from Polk County, Florida,
in the transition zone and results suggested the existence of natural
hybridization between varieties in that area (Mergcn, 1954; and
Wakeley, 1959). (For further evidences of hybridization see Mergen, 1958.)
In still another experiment, Growth rate was usually moderately superior
among sources from extreme south Georgia and north Florida ( north-central
region ), and it decreased both to the north and south ef this area
(Squillace and Kraus, 1959). These authors suggested that climatic
conditions may be optimum in the north-central region, where superior
growth rate mlay have resulted from relatively strong natural selection
for this trait. Resistance to cold damage in the northern fringe and
unfavorable distribution of rainfall in he southern areas may have
been relatively more important than growth rate in natural selection
in these areas.
The results for survival were similar to those for growth rate--
differences were found more frequently when a broad latitudinal zone
was sampled than when only the northern portion of the species range
was sampled. In both the "Southwide" and the "Florida-Georgia" studies
early survival was usually greater among northern sources than among
southern ones. Some traits, such as stomatal frequency (Mergen, 1958)
ad fusifem rust resoistmee (Henry, 1959), shaed evidenees of
loagltudiaal variaion in the nerth.
several studies other than seed esore tests, hae also provided
information on go gaphic varition in slash pine. A plutatiu user
Gstameville, Florids, coeabSing cloea from phemtypically superior
trees selected in vaeris pertious o the ra ge of elliottli (eTrry
and Vang, 1955) osowe diffieemes ia Sm yielding ability at ahoat
7 yeas of ae (Anowamous, 1962, p. 124). A eaktle &im test in
south Florida, cqparing the two varietUes ef ssh pine, showed
siiafient differemesa in grpeth at survival (Hilum, et al., 1962).
Stemsod specific cavity ma/or smerrood per ceot wre studied in
elliottii tres grwiag in their nmbtral abitatu by several investigators
(Larsom, 1957; Perry and Was, 1958; Wheeler Mlat itehell, 1959 md 1962;
and Geaderd at stricklmd, 196). These studies aread in showing that
specific gravity ad samrweued per ceat increase in gling frm north
to senth through Georgia ad Florida, md fro wet to east through the
northern portion of the species rage. Ike elinal pattern of variatiOb
was shown to be closely associated with seaemal distribution of rainfall,
in admitiae to latitiw au d leegitde. ewever, the two experiaJts
reported 1oa1 by Ecuels (1960), sham in Table 1, acsg t that the
pattern in these weod properties is largely eviremmtal rather tha
gantic. Variation in tim ef poll. ad msed ripeidg bha been
reported by BomsI ad aerer (1956).
As neotd srlier, the Soe stacties de6 t minly with variety
elliottli. The possibility of variation within variety deA~i mr
to have eseeed rtudy.
BASIS FOM VARIATION IN SLASH PIE
This section contains an examination of the environmental
factors thich ay have been instrumental in causing geogr qhic and/or
racial variation to be reported. Information on climate was freely
drawn from U.S. Weather Bureau reports (Weather Bureeu, 1956 and 1959).
Climate within the range of slash pine varies from a eone of
transition between taperate and subtropical conditions in the north to
tropical conditions in the Florida Keys. Taerature variation and
other factors are strongly affected by latitude and proximity to the
Atlantic Ocean or Gulf of NIxice. Smers ore relatively long, wamn,
and humid; winter are relatively mild due to the sextherly latitude
and vra adjacent sea waters, but periodically cool and cold air from
the north invades the region.
Mean January temperatures increase gradually from a lov of about
50F. at the northern extreme in South Carolina to a high of about 70o?.
in the Florida Keys (Fig. 2). No such gradient occurs in suer,
however, man July teeratures averaging about 800-60F. throuaeut
the region. Length of frst-free season increases from a low of
about 240 days at the northern extremes to a high of 365 days in south
Flerida. The spread between daily vmxims and miniam taiperature is
greatly affected by proadaity to the sea, especially during the growing
season. Per example, the een spread for the anths of April through
September varies froa ea little as 140. along the coasts to as high
as 260r. in interior portions of the species rage (Fig. 3).
C Ea r-I
r 0 0k\1
_I Q, O j
I, 00 ro
Hem annual precipitation varies fro as high as inches in
seutheest Florida and southar Loudsisma ad Misisiippi, to a low
as 14 inches at the northern limits in east Georgia (Fig. 4). Although
the pattern is somewhat erratic there is a general tendmcy for decreasing
rainfall from southern Louiiae5a, eaet and northeast to South Carolina
md from south Florida northward.
Seasonal distribution of rainfall skews distinctive patterns.
Precipitation is distributed favorably in the northern portion of the
species range, with highs occurring generally in February and March,
and July ad August. In the south, meit of the total rainfall occurs
in the midsamer months and wintertime drouths are rather ceioon. The
variation expreeed in these terms produces contimuos patterns. These
are well illustrated in maps dmn by Squilllae sad Krams (1959) which
show patterns of rainfall for Janary through April, and June through
September. The same situation is also expressed in Figure 5 which
shows isogras for rainfall frem October through May as a per cent of
ampual. Note that it is low in extreme southwest Florida sad increase
rather uniformly to the north ad northwest.
Estimates of precipitation-evaporation (P-E) ratios were determined
for weather stations within the range of slash pine, using the aethed
described by Thornthwite (1931.) These ratios are measures of
precipitation effectiveness at are estimated from mea meathly pre-
cipitatien and mea~ monthly tea~ nature, utilizing Tho mtlwaite's
formula er his nomogr. (he latter, a grapbieel atMd, was w M
for the present study). eP-1 es wane wonI at
Februwy, March, and April, .io.. ....
because effective rtMwaia
-J ,d l.
C p e0 01
H 0 'd
oA -4 a
i- z C
i CM m
-P (8 h
i- + (
associated with growth of slash pine than rainfall during other periods,
as reported by Coile (1936). The data showed a distinctive, continuous
pattern (Fig. 6), mach like tht for October-May precipitation per eent.
Hurricanes are comon along the coastal areas (Weather Bureau, 1959).
Chances of hurricane foroe winds are greatest at the southern tip of
Flrida, and the probabilities generally decrease to the north along the
Atlantic coast to southeast Georgia where they increase slightly. On
the Gulf coast, probabilities decrease northward to the Tapa region
but then become high again in west Florida and south Alabea.,
Soils within the range of slash pine are for the most part sandy
in texture, and low in mineral nutrients and moisture holding capacity.
They are often underlain with hardpans 18 to 2~ inches below the surface.
Coastal areas are low and flat while the interior portions are generally
rolling, with gentle hills and ridges mostly under 200 feet in elevation
but reaching as high as 3k5 feet in Florida, and 600 feet in Georgia.
local variations in soil characteristics, frequently associated with
aall differences in elevation (as little as several feet), are eoamen.
These variations strongly affect tree growth (Cooper, 1957).
Forest geneticists are concerned as to whether or not racial
differemees associated with local variations in soils are present.
Edaphic races have been reported fer ame species of plants (Snaydon
and Bradshaw, 1961). obet workers feel that this type of variation
has not developed in slash pin. Until recently, slash pine occurred
only on pond nargis. *Natural selection probably has not had sufficient
time to cause appreciable changes in gene frequencies on the higher areas,
especially since these areas frequently are interspersed with flatweeds.
t 0 \
"- 0 0 0
0 0 -\
. 4Drl H
0 ?-t %
- 0 -&C)
z c -'
I $- *\
S*r4 -P 0\
M 0 f
Geological changes during the Pleistocene period (beginning about
3/4 million years ago) undoubtedly had ane bearing on the development
of variation in slash pine. Following the Kansan glaciation,the Florida
peninsula was reduced to a group of small islands extending from Hamilton
County in the north to as far as Highlamds County in the south (MacNeil,
1950). The second shoreline recognized by MacNeil, following the Illinoian
glaciation, shers a similar group of isluads but they were larger and
the mainland extended as far south as Alachua County. During the mid-
Wisconsin glacial recession, much of Florida occurred as part of the
mainland, the peninsula extending as far south as Glades County, with a
number of islands mostly along the east and southwest coasts. The final
and most recent shoreline recognized by MacNeil was of post-Wisconsin
origin. Although the degree of inundation was relatively mall at this
time, a number of islands occurred along coastal regions.
the conventi eal seed owee teelique ws used for this study
but with the additional features of: (1) empling paretal materials
to me us ge qambhle varl aie, a (2) milxteiag individ l mother
tree identity in order to study Ibber te variation within stands.
In the fall of 1960, abu comes ad follage emple were collected
from eah ef five (in a few irnstaces le) mother trees at each of 55
stsms scattered lthreegot the rage of lsth pine. Proposed stead
locations were prendesiated -aLly by gridding the are on a ma, with
a spacing interval of ebut 50 miles. Neoever, rimegaeity of the
species rang, nn-feoeeted areas, amd other aeosirations aeseesitated
moving may of the proposed locations so that the actual distribution
of the stuids only faintly resembles a grid (Fig. 1).
It should be noted that systematic amsling of stads leds te a
bia in veriace nd the um idtude of this bla is 1w iMm. An
alternative proedwue weald have been to sa le stesl eaepletely at
randeh or to startify ad emle raRdoly within strata. System ic
sampling wa eboeen beeaue ef a strong desire to inetwe the extremities
of the rage, ad becoeee it vas felt that this mthed would be meet
suitable for elucidating patterns of variation.
Materials were collected through the aid of cooperators. Instructioas
included selection of accessible, natural stas as We as feasible to
the prodesigeated point., with the rCeuirmats that they (1) be at
lest 00 feet y from flowering slash pine platatioma, (2) be of
fruiting age, ad (3) not be selected for aay particular traits.
Within each stand, mother trees were selected randomly but with
restrictions that (1) they be dominants or codoinants, possessing mature
cones, (2) they be at least 200 but not more than 1,300 feet apart, and
(3) they have one or more neighbors within 100 feet. In those areas where
the two varieties meet or overlap (transition zone), no attempt was made
to select one or the other variety, because (1) identification of the
varieties in the mature stage is difficult, a noted earlier, and (2) it
was felt that attemptcd selection would prevent the possibility of
determining the population structure of lhe transition sone. Mother
trees within stands were designated "A" through "E". These letters,
combined with stand s (1 through 55), served to identify all
From each tree, 10 to 15 cones and 5 branch shoots were collected
frm the upper and outer portions of the crowns. Most of the materials
were obtained by shooting the out of the trees with a rifle. Plant
materials were seat to Olutee, .Florida, for processing.
Collections were highly successful but, upon receipt of the
materials, the *e le from stand 51 as found to be loblolly pine
rather than slash pine (identification was verified upon sowing of
seed). Hence, this stand was discarded. Also, materials for three
mother trees (291, 48A, sad 48C) were missing. Finally, it was later
determined that mother tree 21D was apparently a hybrid (or backeross)
between slash ad longleaf pines, and hence data from this tree were
eliminated from analyses. These circumstances reduced the number of
stands to 54, and mother trees to 266.
a!iiii nR ........
In the late fall of 1960, after eone collection, seven additional
stands were designated (nmaers 56 through 62) ad used for collection
of foliage sales (see Fig. 1 for location of these). Thawe s pplemntery
sales were taken mainly to check em what appewed to be unusual results
from the main sales and to increase sapling intensity in north Florida.
Data from the supplemental sales were not used in statistical anlyses
but were inclated with data from aain sale in elucidating patterns
Upon receipt, the unopeed coae were counted and 10 (or less when a
shortage occurred) were selected from each mother tree ad photographed.
The negatives were then projected on a microfilm reader ed the lengths
and dimeters (across broadest portion) of each come were measured.
Cones were dried in the open air; then the seeds were extracted aad
wimaowed with a seed blower which removed practically all empty seed.
Full seed were then counted, weighed, and stored in a refrigerator at
approximately 400F. until plated.
Branch shoots were handled as follows: Eight fascicles were taken
randomly freo the central portion of the first flush of the 1960 incremnt
of each branch shoot (40 fascicles per mother tree). The number of
needles per fascicle was determined on each of them. Then 3 fascicles
were selected randomly from each group of 8 sales (15 per mother tree),
and on these the lengths of fascicles and the lengths of the fascicle
sheaths were measured. Finally, 2 additional needles were selected from
each shoot, again from the central portion of the first 1960 flush of
growth (10 per mother tree) ad the uppemost 2 inches f each was cut
and preserved in foramlin-aceto-ethyl alcohol fluid.
The preserved needle specimens were then used for additional
measurements as follows: The lower 1/8-inch of each section was cu
and examined under a binocular dissecting microscope (45X) and the
following measurements taken: (1) Width of the needle, measured across
the flat surface or surfaces (biaate needles had one flat surface while
ternate needles had two), using an eyepiece micrometer; (2) the number
of rows of stomata on the flat surface or surfaces; ad (3) the number
of staoata in two rows, ea 1.68 millimeters long (the length of the
micrometer scale); for binate needles the second row nearest each edge
of the single flat surface was used, while for ternate needles the
second row nearest the rounded surface van taken from each of the two
flat surfaces. The number of rows of stomata was divided by the total
flat surface width in millimeters to obtain numberr of rows per millimeter
of width." The number of ros per m. of width was then multiplied by
number of stomata per am. of row to obtain number of stomata per square
am. of needle surface.
Freehand cross sections were then cut from the lower end of each
of five needle segments (one per shoot) and wmunted in water on
microscopic slides. These were then examined under a microscope (100X)
and the number of resin ducts and number of layers of hypoderaal cells
determined. The latter measuremeat proved difficult. Invariably there
as a well defined, thin-walled, outer layer of cells. Inside of it
occurred one or more "layers" of thick-walled cells, but these were not
always in true layers, the inneraost frequently containing sporadic,
single cells. However, four points were systematically predesignated
on each section (always between staata) and the number of "layers"
counted at each, to obtain an average for the needle.
Seeds were soe on March ~1-15, 1961, in a nursery at Oltee,
Florida, in two nursery terts. Nursery Tet 1 wa xdigned to obtain
Mcn develgeat of foliage, md for this roemen needs wnee me
in plastic pets 6 inleb in diameter a 6 inews deep. luh design
vs a rmadmied bloek type, with individual tree plots m five
repliationm. FroP oeo to three se*s were som per pot, dApading
wiom the amber wnalable, al t the seedlings r thine to me per
pet son after grmiation.
Nursery %at 2 was designed ma ily to pehduo seedlings in
quatity for outplating, vhish is not rmeeesSw o in this report.
Hwever, the material provide a epportialty to obtain mre liable
sta ,a oned g emindwti an oetyledea n n r then cauld be obtained
froa Nursery Test 1 a hbero vwa used for this pury e.
In Nursery est 2, seds of each their tre ware sewn in ar
plots of 44 sdees eh, with 3 replications. Nut in order to minimize
acpetitie effects, the five =rher trees of each stma were renWtdoed
within sted plets, mnd std pleot were radmined within repliesties.
Seds were awn at a spaciag of 1 inAh within rows ml raen were speed
6 inehes apart.
OaGi ntion wa eeowntd in Nirsery Test 2 a March 29, 1961, ad
again on April 10, 1961. The first count divided by the wneMed eat,
x 100, gwae a inded of the speed r rbe of geamiatioa in per aent,
while the latter coumt (e ireed in per aent of seeds son) alme Wa
sed as a nosure of gelmi6ability. Alms, estylode co ts wear obtained
on up to 10 randeay choee seedlings per row in April, 1961.
Total heights and stem diameter outside bark at ground line were
measured on the seedlings of Nursery Test 1 on November 3, 1961.
In the late fall of 1961, foliar samples and measurements were
obtained from the potted seedlings of Nursery Test 1 as follows: First,
counts of the ln r of needles per fascicle were obtained on each of
10 fascicles taken from each seedling. Fascicles were chosen randomly
froa the upper portion of tih first flush of growth. The foliar
material was then handled in a armer similar to that from the parents.
However, here fascicle length and fascicle sheath lengths were measured
on three fascicles obtained from each seedling and the stomatal, resin
duct, and hypoderm measurements were obtained for two needles per
Single variate analyses
Statistical analyses consisted mostly of two types, single variate
and multivariate. In the single variate analyses the stands e divided
into three groups as follows:
Group i. Stands within the range of the elliottii variety, excluding
those close to the limits of the densa variety, as follows: Nubers 1
through 26, 31 through 40, 52, 5, and 55. Total, 39.
Group 2. Stands arbitrarily considered to be within the transition
zone between the two varieties: Numbers 29, 30, 41, 42, iA, and 45.
Group 3. Stands within the range of South Florida slash pine as
delineated by Little and Dorman (1954): hKubers 27, 28, 43, 46 through
50, and 53. Total, 9.
Note that the aaaigment of borderline stands in the transition
zome Qepaws inconsistent in om instanees, according to limits of
the varietal ranges shown in Figure 1. The reaso for this is that
the asigment of stands into groups vas made aeerding to the mall-
scale map in Little and DoIn (1954), the most recent available range
arp at the time. fie northern limits of var. dansa shown in Figure 1
were reproduced from Lagdon' (1963) are recent ad detailed asp,
revealing what appears to be inconsistencies.
The purpose of grouping the stands vam to provide a mean for
determining the presence or absenoe of significant atand differences
within varieties. To this extent, limitations imosed by the arbitrary
nature of the groping should be recognized.
The analyses of variance for data from parent tree samples were as
Source of Variation D.F. Expected Mean Squares
2 2 2
Groups of stands (G) 2 7 + kl2 O + kU 0 1
Stands within roups (S) 51 oT4 + k22 0
Mother trees within stands (M) 209 CO-
In the above analyses the deficiency in degrees of freedom for
mother trees was due to seven "missing" trees (9D, 21D, 29E, 36B, 48A,
48B, and 48c). Tree 21D wva dropped because of evidence that it was a
hybrid, while the rsaaining missing trees were due to lack of samples.
Coefficients for the variance components for all analyses of
variance were computed using the technique outlined by Gates and Shiue
(1962). For the parent tree analyses the coefficients were as follows:
k12 = 4.870 kkl = 56.464
k22 = 4.869
The analyses of rimoie for progeny data of Nursry Teat 1 were
Source of Variation D.F. t peeted Mesa i a re
keplications (1) 4
2a 2 22
Groups of stands (G) 2 Oj + k13 Ui + k1l2 IC + k1 Cg
2 2 2
Stads within grpeps (s) 51 u j + k23 aM + k22 C
Another trees within stands (M) 209 dj + k33 0jj
Error (i) 1043 OR
In the above analyses the deficiencies in degrees of freedom for
mother trees nd error were due to seven "missing" mother trees (21D,
221, 29W, 42B, 4i B, ad 48C) ad five miningg" seedlings (7A-4,
9C-., 38D-1, 46C-4, ad 46D-4). M-ther tree 211 we dropped for reasons
noted earlier, while the reiaing missing items were due to laek of
Ooefficieats oeeputed for the oemponents of variance estimates,
were as follows
k13 = 4.982 kl2 23.746 kU = 2TT.504
k23 4.984 k22 24.278
The aaalyses of variance for progeny data of Nursery Test 2 were as
Source of Variation D.F. Expected Mean Squares
Replications (R) 2
2 2 2
Groups of stands (G) 2 Uj + kg12 O- + kU G-
Stands within groups (s) 51 1+ k22 'S
Error 1 (z1) 106
Mother trees within stwads (M) 202 112 + k33 O-M
Error 2 (E2) 404 0
In the above analyses the deficiency in degrees of freedom for
mother trees was due to 14 "missing" mother trees (17D, 21), 22E, 25D,
29A, 29C, 29E, 330, 4lB, 413I 48 48C, 53A, and 530). Mother tree
21D wva dropped for reasons noted earlier while the remaining trees were
dropped because of lack of samples.
Coefficients cemputed for the ccponents of variance estimates
for progeny data of Nursery Test 2 were as follows:
kl2 14.061 ku 159.140
The main purse of the malyses of variuee w to obtain objective
estates of the degree of variatieio Mseciatd with the freters studied.
To aid in doing this, estimates of compeaets of varier were obtained
using the sen squures comuted in the aalyse of varime and the
"eapeeted oma square" shew above (Saeeer, 1956, p. 261). The
estimated coempoaets obtained in this m r vere finally expressed in
per cent of the total of all eempmrets (xKeludiaR the "replicatima"
eemReat in progeny data).
The c ompoaet of variaoee assocIated with groups as eo idered to
be expressive of the division of the species iant the two varieties ad
the tradition smea. That esuseiate with stands within gro s expresses
the degree of geographic varieties within varieties. These tw oompments
taken together are expreesive f geographic variation for the species as
a whole. If either or both of these ecampats were statistically
siaifieant and appreciable in maaituie, isegrp were druw in an
attempt to eluidate the pattern of geographic variatia for the trait
Nete that the above analy ses ws hm bgemoe vari ances. As
vill later be seen, variation was freuetly fead to be greater
in sea pertioes of the species range thra i ethers. This eirematece
affects the validity ea the estimates of variaee eapements and the
significance tests. Hence, the estates and tests should be considered
tiltivariate aclysis as employed to eamine the pattern of
geographic variation oonaidering a group of traits asiftaeously.
ghaulanbia' generalizedd distance function" was eboen. (For
discussions of this and other aultivariate techbiqces see Rao, 1952;
B~e~w 1960; Wells, 1962; Wrigt and Ball, 196e; an Mmtieng, 1963.)
This function, B2, presses the Legree of relationship between two
populations, considering sailtanemm uly the group of traits chosen.
The formula for two traits (X1 and X1 is am folles:
D2 (,u It)2 + 1u _112) ya ya2) + (r.-X2 a)2
812 S12 S
in whieh i and 112 are the ewns of trait I for the first and second
populations, respectively; X1 and Zg2 the means of trait 2 for the
see two populations; 512 and S22 the peeled estimates of the variances
of traits 1 and 2; iad 812 the oovariaee of traits 1 and 2.
As can be seen, the manitude of 2 for any two populations
increses with increasing difference in the mnsa for each trait, and
dereases with increasing variane and eovarianee within populations.
Per mare than two traits the formula is more conveniently expressed
as follows: Z S 1 a d
i j ij i j
where i i the een population difference for trait i
and d& the mean population difference for the Jth variable
and Si the element in the inverse of the corariance matrix
oerreMpnding to the ith and Jth variable.
Using procedures outlined by Rlo (1952, pp. 345 ad 357), D2 values
were cauted for 17 traits, including 4 fro the parent tree data
(coem legth, come dimeter, seeds per eem, d seed weight), ad 13
from progeny data (total height, stem difmer, amber of termate fascicle,
needle length, sheath length, owv of stomata, steta per m., strmta
per sq. m., resin ducts, hypodemr thickness, gemnability, speed ef
geannation, ad cotyledea manber). Since theee were 54 tands or
"populations" a total of (54) (53) 1,431 values of D2 had to be coe uted.
The wek was dore vith IBM 709 electronic computer at the University
ef Florida (Gaouting Cater.
RESULTS AND DISCUSSION
Results of the single variate analyses ad patterns of variation
for indivdual traits will be presented first. Following will be a
recapitulation of the individual trait patterns along with a discussion
of possible causes of variation. Next will be an analysis of the
degree of variation (diversity) among individuals within stands and
among stands within varieties sad their implications. Then follows
the results of the nultivariate analysis, and finally a discussion
of taxonemic considerations.
Single Variate Analyees
Mother tree naws of come length varied from 7.0 to 15.5 ca.
(Table 3). Most of the variatien was associated with mother trees
within stnds but stads within group aecownted for a considerable
proportion (22 per cent) of it (Table 4). Since little of the variation
was associated ith groups of stands (6 per cent) the trait was net
distinctive for varieties. The stand-to-stad variation exhibited a
fairly distinctive pattern, however. Cones were relatively short in
southeast Florida and increased to the north (Fig. 7). An east-west
waxiam occurred near the Georgia-Florida beudary (Walton County,
Florida, to Daval County, Florida), abeve which cone length decreased
Variance components for come diameter were rather similar to these
for come length, with stands accounting for a sizable proportion
(37 per cent) and with groupings of stands accounting for none of it.
Although te variation mng stands was net associated with varieties,
a fluctuating clinal pattern was apparent (Fig. 8). Cones were thickest
** ** ** *
U V 0o
** ** *
51 ( c r
* * I
- r 1 I
UN UNL ^--
oj cnn cn
R i i
c1 CM CU
H U \De
* * I
C8 p O l
I I I
IP\ ir\ \'
cm A rl,
* S 6
I I I
* S S
CU CU CU
en Cu co
lI i 00
C\J O 0\
0 CO cu
0 L- m
d N 00
V) L J
~U N~I ~tt vCd
in the collection from Collier County, Florida, ad they deeresed in
diameter toward the north, east, ad south. An eat-veet trough seemed
to occur in the neiglberhood of Pelk Coaty, Florida, ad either extending
seothwest-northeat through the northern portion of the species rage,
with a ainimu at Brantley County, Georgia.
The cone dimnsiom foed here (Table 3) agree fairly well with
values reported by others, seen by the tabulatiae of "eomOr" ranges
below. However, it is dbvlous that these eome dimension re not
particularly useful for identifying varieties.
Sml (1933, p. 4)
Caker and Totten (1937, p. 19)
Little ad Dorma (1954)
Waseley (1954, p. 198)
west d Arnold (1956, p. 5-6)
Present study (ranges among
mother tree mans)
Little ad Dorman (1954)
Wakeley (1954, p. 198)
Present study (ranes mosg
mother tree nea a)
elliottii dmeas varieties
8-12 8-15 --
Seed yield was extremely variable both mong mother trees (1 to 127
seeds per cone) and ong stands (3 to 97 seeds per cone) (Table 3 and
Fig. 9). Much of the variation among either trees was associated with
groups (22 per cent) ad stmads within groups (32 per cent) (Table 4).
Variation mong stead means fell into a irregular clinal pattern
(Fig. 9). Som of the irregularity may be due to differences in stand
density or similar factors not studied. A high occurred in an area
centering at Thomas County, Georgia, with a moderately high ridge
extending to the east sad vest. Yield usually decreased from this
ridge both to the north and south, reaching a extremely low point at
Big Pine Key, Florida.
Since seed crops generally vary from year to year, and since
locality by year interactions are probable (Tousey and Korstian, 1912,
p. 105), one should not assume that the pattern of seed yield per cone
found here would be consistent in time.
The mean sound seed yield found for the whole species, 51 seeds
per cone, is lower than a reported by Wakeley (1954), 60-70 seeds
per cone. The discrepancy may be due to yearly effects as noted above,
or to differences in the degree of winning.
The eems of seed weight for mother trees were extremely variable
(10 to 51 ag. per seed) (Table 3) Much of this variation was associated
with stands ad it exhibited a clear, costly clinal pattern (Table I and
Fig. 10). A nertseast-seuthwest trough occurred in southeast Georgia
extending from Pierce County to Evan County. Seed weight increased in
all directions from this area. To the srath, a aertheot-southmest high
occurred extending freo Dixie Comty, Florida, to DIu.i Ceaty, Florida.
It then decreased irregularly to the south. ote that the rate of Shage,
however, was not mifora, the drp being the shampest in south Florida.
The mea seed weight for all treeo, 30.6 ag. (vbick eeonerts to
about 14,800 needs per lb.) agre we ll with the reges for slash pine
giWve in the Forest Servioe Woody Plant Seed Meaial (Anonymous, 1948,
p. 269), 13,000 to 16,000 seeds per lb. ad else with the rages of the
mess of 100-ned aplee, 2.8-3.5 grm, givew by W eley (1954, p. 198).
Germiability and speed of gepriaetiea
GeriLnbility of seed varied highly among thker trees (6 to 100
per cent) (Table 5). Sianiflomst amoats of the variation were
asociated with stads and groups (17 ad 6 per cent, respectively)
(Table 6). Geriinability averaged highest in the daeme variety, next
highest in the transition soe, ad lowest in the typical variety.
However, the pattern seemed to contain a large elemt of radeumes
ad sa ilogre were drum (Fig. 11).
The result agree with Mergen ad Hookstra's (1954), in that
significat differeaes a seeed lots from different portics of the
rage of the typical variety were foand ad that no distinctive pattern
occurred. However, the differeces i geaeinability of seed from
comarable areas in the two studies showed little agreement.
Germinability of seed ma of oeone be affected by maturity at
time of collection ad other factors. Althagh attempts were mde to
collect only mature comes, there is no asuramee that all lets were of
the see degree of maturity. Hence, even tbui siptficant stand
Table 5.--Meuws and ranges of variation for progeny
data of Nursery Test 2
a Speed of Cotyledon
Group : Germinability : pee io b :Cotyledon
Per cent Per cent Number
1 60.7 67.1 7.43
2 66.7 75.3 7.29
3 73.2 89.4 6.83
All groups 63.3 71.4 7.32
RANGES AMONG SEEDLIN S
1 -* -- 4-12
2 -- 4-13
AMES AMNDG OTH5R TRIED A3E
1 6-96 0-99 6.0-9.4
2 23-91 7-100 6.2-9.3
3 14-100 53-100 5.5-8.0
Per cent of sound seed geminating within 27 days after
b 15-day g nation x 100.
Table 6.--Wemn sqwrtees n esti be at variance components obtained
from anamlyes t varianee of progeny data or Nursery Test 2
: Geriability : o Cotyledon
Replisations 5,027** 135 .199
groups 8,271*- 25,163** 17.743"*
Stanids/G 1*l14* 1,8146* 2.106*
Mirror 1 131 0o .067.
Mother trees/S 833" 978@ .*597**
Error 2 86. 192.. .065--
ETIMItTED CQMPOnH OF VAIANCE--PR CUBT
Grn-pu 6 13 17
Stands/G 17 9 24
IrrOw 21 37 15
Mother trees 43 4 29
Errr 2 13 17 15
* Significant at the 5 per cent level.
- Significant at the 1 per cent level.
2 (n -5 "
differeacee were found they were not neceaearily genetic in ntwre.
Speed of geminstion also vaeed greatly mog mother tees
(from 0 to 100 per cent) (Tale 5). Sipifieemt properties of the
variation were accountd for by goupes ad stemds (13 ad 9 per cent,
respectively) (Table 6). The stand aviation edhibited a distinctive
clinal pattern (Fig. 12). A lav occwurd in Wsre Cotaty, Georgia, which
also tended to ext d westrwd to Holmes Comty, Florida, ad Catalina
Islmd, Mississippi, and ertheo twrd to Georgetoa Conty, South
Carolie, as wll. Speed of gezmintion imeremed both to the merth
ad to the south of the trough.
Evidence of racial variation in speed of geminatimo has also beea
found in lodgepole pine (P. contorts Dougl.)(Critchtield, 1957), eastern
halock (Tsuga medeais (L.) Garr.) (Sterns aRd Olson, 1958), spruce
(Picem) (Schell, 1960), ad ponderosa pine (Callham, 1959 ad 1962).
Like gexriuability, differences in maturity of seed could have had
scm effect upon the differences in speed of germination amog stdts.
However, the nature ad distinctiveness of the treads practically rule
out the possibility that such extremeous factors oould have caused the
pattern. Mere likely it was due to geetic differoenes in the mdi,
brought abont by natural seleetie ad causing differential response to
It is of interest to speculate en the nature of the gemetic
differences that were apparently present, ad on the particular
envircmsatal factors to which the eeds responded at the planting site.
paft studies suggest that teperature is a ajor enviroma~tal factor.
According tb Callaha (1962), the speed of germination of tree seeds is
governed primarily by toeprture given adequate moisture ad light, with
geainatie prooeeding mot rapidly at meo optiam toemeratwe. Iuperi-
seats by Jones (1961) suggest that pheeoperied was not a preominuat factor
in casing the differences in rate of gerinatie. he shoed that a
single expeware of slash pine seeds to 15 aamite of daylight Saouled the
total garminatieo per ent ever that obtained u r complete darkness.
But illumination periods of 8-, 12-, en 16-hours used ne differences
in either speed of germinatieo or total glerniatieo per seat.
Assuming that temerature was a mjor eavirommetal factor, one
might speculate that the seeds pesesM ed different geetieally-fixd
optimum teperatures ed this would be reflected in different rates of
gemination when the seeds were planted in a eemn environment. Such
was feod to be the case through laboratory tests by Callaha (1959 ad
1962) for pondeessa pine. Hmwer, this seme would net explain why
seeds brought north from south Florida and south freo the northern
limits te Olustee, Florida, garmiated early.
Preeeaee or abeeee of seed dormacy my here been iwmprtat. In
ex ining this possibility, it is well to review what is knom aohut
factor that my be involved. Moet slash pine *eed are shed in October
(Cooper, 1957). Under natural editions, seed tend to gerinsate in
spring, ut when soil moisture is adequate oesierable germinatieo my
oceur in early atmn (berr, 1959). In esoth Florida, eeodities for
early fall gsemination would see to ccur rather frequently because
Oetober rainfall there nerages about 6 inches. In eeatrast, October
Sriafall arerages about 2 inches in the north. In the south, the
winter months are dry (average rain about 2 inches per math) ad
relatively vern, vhile in the north they re wetter (about 4 inches
per maath) and considerably cooler.
Stored slash pine eeds show a mild degree of dermancy, germination
being abetted by stratification, while fresh seed do not (Anonymous, 191 ).
These findings on dormncy were most likely based upe work with the
typical variety of slash, although this int is not certain.
It is possible that dormacy may be more characteristic of northern
seeds than southern seeds. In the north, if the seeds.do not germinate
promptly in the fall, there would likely have to be a mechanism built
int the seeds to prevent germination over winter, because of te danger
of cold temperatures to newly germinated seedlings. In the south, e
the other had, there would not seem to be a need for dormacy, because
of the wra winters. In fact, it would seem that gerAination as early
as possible after seed fall would carry a high selective advatage--
propt germination to avoid mortality fros severe winter drouths.
The feet that northern seeds will germinate promptly under favorable
conditions in the fall suggests that oaset of dormancy (if it actually
occurs) is delayed. Prempt fall germination undoubtedly carries a high
selective advantage--trees germinating in the fall obtaining "a heed
start" on those germinating in the spring in regenerating denuded lands.
However, prompt fall germination under suitable weather conditions plus
dormacy when weather conditions fail would seem to be the best ocibination
for the variety. Theee cojectures on dorancy are feasible in view of
the findings with several forage species a Europe, in which it was shown
that germination characteristics of species inhabiting different climates
were closely tied in with dormancy mechenisms (Cooper, 1963).
Aaeunng both differential daormn y ad different optiman temperature
requirements, we might attempt to explain the results of the present study.
South Florida seeds gerxmiated earliest because they lacked doracy.
Seeds frso south Georgia ad orth Florida g rainoted late became the
stored seed possessed a mild degree of dormey--hac the seed been
stratified differences m y not have bees foead. Seeds from the extreme
northern limits of the speeles rag garminated promtly beoeese, although
they also posses. moderate dormcy, their optimi teeratwr was
attained sowr, having been moved fron a northerly to southerly direction.
The latter conjecture asaes no differamse in optira typcuratre
requtirents within the northern regnia. Of course these are little
ern them gueses, further eaperimatilon being neesoeery on this
The number of cotyledons per seedling varied from low as 4 to -
high as 13 (Table 5). Much of the vaiartiom ws oasoclated with stmda
(24 per cent) and graVe of stamds (17 per cemt).
Stad average displayed a distinctive clinal pattern (Fig* 13) tuch
like that for seed weight (Fig. 10). On the average, otyleden numbers
wee higher in the north then in the soath (Table 5). However, as sen
in Figae 13, the pattern is neh mnre subtle tha this, with a law
occurring in the north m well in t the south.
The means ad rges agree fairly well with pervirely reported
values, a indicated in the following tabulation (mees are folleved by
rages. in parentheses) .
- NuersI of cetyledBns
tneelmm (1880, pp. 174, 186)a
Butts ad Buchholz (1940)
Little ad Dorms (1954)
Daeoto atiemol FPrest, Mms.
Clinch CeOty, Ga.
HOadry Ceouty, Fla.
Pr eet study (rags are 7-43(4-12) 6.83(4-10) 7.32(4-13
a Cited by Little ad Borm (1954)
b Origin not specified
he eerrelatice between cotyledon amer amd seed eight n a s tand
mm basis was .7, highly signifio t; the pooled correlation for mther
trees within stads as .42, alms highly asiifiomt.
Reil variation in re pet to cotyledon mbear hs aLe been found
in loblolly pine (ThoaJbjomsen,1961). The positive correlation between
seed weight and eetyledo number agrees vith findings by Buchhels (1946)
for penterosa pine.
One-yew-old seedling heights varied greatly ad the majority of the
variation (66 per est) vas asseiated with gro~ ings of the stads.
Seedlings in the Derthern portion of the species rage were tallest
(Tables 7 aTd 8, at Figs. 14 anB 15). Variation in the north was
relatively sL1- but heights deceased rapidly going frn north to meuth
through Florida. Thus, the pttera is largely ruat in the morth ad
clinal through Florida. There was also a modest eat-vet gradient
* o I \D
rlm l C C
r- L-I C -
*~ * I *
0\ D Iko
4 m, C
H H rH
cJ' ( CMj
I I I
t I \
r\ i-\ r-i
co 0 0
CM I (
I I Iw
Cj CU CM
* U C
f rH C
0 0 C
r cu A
H CM o
m .. ) 0 & % m CC)
(D N O -0 M61
403 5- 4 *
5 0 r-1
4a 9 .... omm(S&
S prii com
In 0 8
+3 W4 to 0
4 4 -t a m
4o aas u
0 8O n1 I C
G G3 * ** g HO
PA 0 i 6\0 \D A"\
H g 0 r 2 (n
03 O 030 C 0 3 \I
0 *rd H l
4 r 43 jr
a a'2 s s g
aN fn Lrf\o _l t
0 "l T
00 > I ^ I m m
t33o 8 "j 1
5c a1 *^ ''
Figure 15.--0a-year-oldA las pime smalings, baring differmees
in total height a stem dimiter. Upper $tlre reprenumt a
latitudimal tramset thrum k the species rag., the -me a the
exteme left beig from lig Pine Key, Frlrin a e me- a
the extreme ripit from SmC er Corsty, Qergia. Lewr pwhte
shrws diffeuremas betrW a tree from the vwt mst (the tw
trees a left), the interior (e3 ter two), d the east eeo t
(tie two Om riet) of eantraul Florid.
iniii: .. ':: ii :" ":':i E :iii :i i mi iiliii;E:::: ..:iiiiM
::: ........~ ~ ~ ~ ~ ~ ~ ~ ~~~~~~.. .........................::E iE:::::::::EE:.. .:E
... ....... '" !!" "
..... iii.................. .............. ..... ............. ... ......;; ; ii~ i ~
......E ... 50 1 E
tbeosu central Florid seedlings being tallet in the center of the
state, eat shortest oeang the sets.
In a gnral way these reeouts are in haamay with Little and orams's
(19 ) ume of st height as a diaosttie fetifw for ideatifying varieties.
Iwever, beewme of the gradient in rForida it paemntly would be difficult
to clasify seedling in the transition sme.
The faet that seedlings in the north-eentral region were not
particularly taller thae those at the extraities of the north, seems to
diagree with finding by Sqtillace wd Kraus (1959). Ever, seeds
wre relatively enl m ad germinatian relatively lbte in the north-central
region. fshe two factors apparently had sem effect upon heights. The
within-stand pooled correlation eeefficient between seedling baight mad
seld eight as .31 (sipifilent at the 1 per cent level) msd between
seedling height nd rate of gerdnation, .17 (sigpifient at the 5 per cent
On the ether hrd.n, the serfiority nl early height rowth of tree
fta the north to those of the sacth is great enough to be rel in spite
of soed weigt ad rate of yrerLnute effects. RBeses for this
differesee probably lie in the fact that the seth generally suffers
from artrems of climatic ad other enviramtenl cneiti es more so
tiMh doe the north. Such factors could include poor rainfall distri-
buttio with frepqnt dreaghts in spring nd flooding in samr, disagim
toupietl terms, ad paolbly frequency of fires. In the sewth, natural
selection is probably relatively strong for resir~ta e to these factors,
htbeh mnr came relatively weaer selection for rapid grorh thai in the
Admittedly there are alse climatic extremes in the peripheral
portions of the north. For exmle, relatively oeld temperatures nad
frequent ice stems are characteristic of the area Just south of the
northern limits; tropical sUtor are relatively frequent along the Gulf
coast; rainfall distribution is relatively amfaverable along the coasts
of Georgia ad South Carolina; eeoditions conducive to fusifera rust
dae seem to be ast favorable at the northern extremities (McCulley,
The east-vet gradient through mch of Florida ay be associated
with the difference between men amxim and mem liniuzm daily
temperatures (Fig. 3)-trees tend to be tall where the temperature
difference is relatively high. This possible association is supported
by findings reported by Krmer (1957) and Mellers (1962)--in laboratory
tests leblolly pine and northern red oak (Quercus rubra L.) grew fastest
under the greatest day-night temperature differential tested
Variation in stem diameter showed a moderately high racial compement
(25 per t for groups and 6 per cent for stand within groups) aad the
stand ieans exhibited a cliaal pattern (Table 8 ad Fig. 16). Stems
were thickest in the South Florida seedlings ad they decreed rather
uniformly to aertheast-southwest low exteading from Taylor County,
Florida, to Liberty County, Georgia. North of this trough, dimeters
increased slightly, but were not as large as those freo south Florida.
Stema usually were thicker (especially relative to height) along the
coasts of Florida than in the interior.
Thick steps are a indication of a carrot-like taproot. Thus, in
a general vay, the results aree with Little ad Doran's (1954) se of
this trait as a diagnostic feature. Stands in groups 1, 2, and 3,
averaged 7.1, 7.4, and 8.5 an., respectively. Trees from stands near
the northern limits of the species range had moderately thic stems but
they were taller than South Florida seedlings and hence would not detract
from diagnostic utility of this trait. However, like tetal height, the
difficulty is that because of the clinal nature of the pattern it would
apparently be difficult to classify trees or stands in the transition
Thickness of stem in slash pine seedlings has undoubtedly been
important in natural selection. South Florida seedlings, which charae-
teristically have thick stems, are more resistant to fires t north
Florida seedlings (Ketcha and Bethuan, 1963). Apparently, this
thickening of the hypecotyl, which is mostly dead outer bar also
inner bark and weed, imparts a degree of insulation against heat (Little
and Drnaa, 1954). The thick stem also probably provides a means for
string food, utilizable for sprouting when the crown buras. Hence, the
trait is assunid to have resulted a an adaptive response to fire (Little
and Boman, 1954).
If the trait is an adaptive response to fire, one would expect that
the frequency of natural fires, or the extent of damage from fires,
increases gradually from north to south, following the pattern of variation
in steA thickness. No concrete and reliable data could be foeid to check
this possibility. However, as noted earlier, slash pines in the north
were originally restricted to pends, pond margins, and ether wet areas.
Hence, it is possible that fires in the south invaded slash pine stands
more frequently, and perhaps were more intense, than in the north. Extended
late winter and early spring rth.s id ba d w inter taorstn es, ieem
in the mokth, my be faeters affecting the frequmey ad intensity of
egressieas wer eaulated te tetertnde factors that mijkt hawe
beea involwd in the qaparm t natural seletles a stem dim etr. Ste
dieter (stuad mws in oatitiJrs, Fig. 16) was u as the depaledt
variable. Indepaaatt variables used were s follows: (1) latitude
( sltm vales in degres); (2) the se of preeipitwitem-evtapMrtion
(P-E) ratios for maths of Febrary throv April (stend values, Fig. 6);
ard (3) mee Jaiwry tmperature (etat vales it F., Fig. 2). P-I
ralos (used as a mease of late winter-erly spring drouth) and
Jansry tamperstwe wrre eemsmeered as possible enviremamal faetrs
eaing netwall selection. Laitdle in itself weevl not, of seewse
Ceoe natural seleetilm, but the variable wa inelued to test the
apparently strong latitudinal trea and to see if effects of P-I ratios
at tdmpergture, iantpedientD of latitWue, ooulA be shAm. The aalyses
included simple, mIltiple, ad eurvilinear regressio Results are
Simple Iragraion Analyses
StIe diameter (T) m: Regrusie aoeffielante detetinatie
Latitude (Xl) -.0232 0.l**
Feb.-Apr. P- ratios (X2) -.o02 15.6
Ja. te erature (x3) .00 36.T**
Multiple and Curvilinear Analyses
Standard partial Coefficients of
Ste diameter (Y) on: regression coefficients deteramiation
X amd X2 -.615, -.031 40.2**
X aad X3 -.702, -.072 40.1*
Xl, X2, and X3 -1.111, -.205, -.631 40.8*
xI and X12 -6.649, 6.000 48.1*
Salad X32 -2.765, 3.385 46.8**
** Significant at the 1 per cent level
In the simple regression analyses latitude shoved the strongest
relationship to steal diameter, as indicated by the coefficients of
determination. This suggests that some envireanntal factor, correlated
with latitude, was instrumental in causing the stem diameter pattern.
The regression coefficient for temperature was almost as strong as
latitude, while that for P-E ratios was considerably weaker, but still
highly significant. Multiple regressions showed no significant increase
in the variance accounted for (indicated by the coefficients of determi-
nation) over and above that aceounted for by latitude alone. This wae
due to high intercerrelations between the independent variables. Therefore,
there is n proof that either temperature or P-E ratios had effects
independent of latitude. Because of the reversal in trend ef stem diameter
in the north-central area, curvilinear regressions were tried for latitude
and teaerature. Both regressions accounted for significantly (1 per cent
level) more of the variace above that accounted for by respective linear
regressions. However, latitude still was superior to temperature.
Frm the malysis we em oly oaeel te that the latitudinal tread
in sle diameter, with a reversal in the arth-eetral ves, was
sipifieant. Sperature ma P-L b le my hme had am real asoeiation
with the trmA, but sum other emviremmital factor rmt also be involved.
Jedles per fascicle
Both blamat ad ternate faeielcas were fomd In the parental m loes,
but the relative freqacies varied emsiidew ly a- indiested by average
ambers of needles per fscicle (Table 3). Strm differrmees displqed
a very distietive pattern, with a north-south high in eKtrme southeast
ealgia and northest Florid, and another n rthwest-seothent high in
nrth-eentral Florida (Fig. 17). Needles per fascilee iually decreased
grahially war frm thee hi~g A notable feare was that, although
ealers wer lo in a uth Floriae, they wer also usually low at the
emttrities of the species zra Thus, the results do not gee well
with Little and orrma's (195) reeemmedi u-e of this eharector for
sweoating varieties--diff mnees in sepling teetlique mr have eased
the dtaepement. Average ~mer of needles per faeilee in the proenies
w- g-erely higher tha in the premts (Table 7). his my be due to
an effect of tree age, or to the fact that the progmies, being prw in
a nwsery, had a are favoreble eviremmat the trees tader natural
osaditims. A very few proegmy faselles eantained four needles ald o
The pattern of vsrictim moeg stads in the progenies was somewhat
similar to that in the parents (Fig. 18). However, the twe prewmamet
hirhs fold in the parts were less notieable in the progmies and
alse the difference between sefth Florida ad the rimat r of the species
rage was more proMomood in the progelies.
The pattern of variation in both parents and progenies seems to be,
in some respects, associated with severity of environment. The low in
south Florida coincides with unfavorable distribution of rainfall and
the low in the extreme north is associated with cold winter temperatures.
Somewhat similar trends have been reported for ponderosa pine. Needles
per fascicle in ponderosa pine tend to be low in eastern portions of the
species range (Weidman, 1939; Haller, 1962; and Wells, 1962), where the
climate is relatively severe and the trees are generally slower growing.
The results agree with Shaw's (1914) statement that in some species of
trees the number of needles per fascicle is dependent upon climatic
conditions, smaller numbers occurring in colder regions.
The apparent relation of needles per fascicle and severity of
climate may be associated with photosynthetic efficiency. It can be
shown that a ternate fascicle has about 20 per cent more leaf surface
area per unit of needle volume than a binate fascicle of the same
diameter and length. Thus, a ternate fascicle, having more surface
area for absorption of light and for exchange of gases per unit of
chlorophyll-bearing tissue, may be more efficient photosynthetically
than a binate one. A binate type, on the other hand, would seem to be
an adaptation for conserving moisture loss or for frst hardiness, at
the expense of growth efficiency. High frequency of ternate fascicles
then may be an adaptation to vigorous growth in optimum climate while
a tendency toward a preponderance of binate ones an adaptation to less
favorable climate. These possibilities would seem to be worthy of
Needle length in the part trees exhibited a rather complicated
pattern of variation mO sat s (Fig. 19). In eneiral, needles avraged
leader within the rages of variety daen the in the north (Table 3).
However, the tendency wea not unifeim highs recurring in the earth as
wvll as in the seth. Needles tended to be relatively long in the
coastal wree, suggesting a possible tie-in with the difference between
mewn minial-neen unmwa toperatures (Fig. 3). But the corelation
eofflcint between these tne variales w u iemagifioat (r -.23).
The pattern in the preoenies was slyler, eeatait ng a strong
slamnt of clina.l variation (Fig. 20). N edles vere generally long in
smutkh Forida (secepting at the mtrmew tip) and they tedLr sed mrthwwd
te a northest-southwest lov through south Georgia, and then increased
above th.s area. The pattern vaguly resembles that in the parent in
*tat needles were, on the average, longest in the ieuth (Table 7).
The ranges in lengths ef needles for parent material re cepared
agitast those shewn by ethers below.
Author elliottil deas varieties
- Centimeters -
9erler (1931) 15-30
Eaml (1933, p. 4) 18-30
Ceker mnd Totten (1937, p. 19) 15-23a
West sn Arnold (1956, p. 5-6) 18-30 18-30
Present study rangess meug 15-27 18-31 15-31
other tree mans)
a Rarely, 10-25
-- \ ,
I o to
0 0 0 <
l ^- z
Fascicle sheath length
Variation in fascicle sheath length in the parental data was strongly
associated with stands, none of it being associated with groups (Table 4).
But the pattern of stand variation was rather intricate (Fig. 21). A
significant feature was that a pronounced north-south low occurred
through the center of Florida and southeast Georgia.
In the progenies the stand component of variation was significant
but rather small, 11 per cent (Table 8). Stand means displayed no
particular trends, with a large element of randomness (Fig. 22).
The ranges of variation in sheath length found in the parental data
do not agree very well with reports by others as seen below. The dis-
crepancies my be due to differences in maturity of the foliage sampled,
or to differences in technique of measurement (such as inclusion or
exclusion of frayed ends).
Authors elliottii densa
- Centimeters -
De Vall (1941a) 0.8-1.3 1.0-1.4
West and Arnold (1956, p. 5-6) 1.3 and under 1.6
Present study (ranges are among 1.2-2.3 1.1-2.3
mother tree means)
De Vall (1940) considered fascicle sheath length to be very diagnostic,
it being unaffected by climate, soil type, tree age, etc., and that the
character was useful to separate slash and longleaf pine.
Results of the three measures of stomatal frequency were similar in
that (1) in the parental data only mall amounts of variance were associated
l a *
>- C- d
S- d i
with groups or stands, with the patterns of the stand meons being largely
randc; and (2) in the progenies it was possible to show patterns for the
stand means, although they were samewhat erratic (Figs.23 through 28).
A camon feature vas a tendency for stomatal frequency (r.1 three types
of measurements) to average relatively high in the north and low in the
south, and also soue tendency for a high to occur in the north-central
Mergen (1958) found a clinal pattern for stomta per ma. increasing
from est to east in slash pine progenies from 12 sources encompassing
ruch of the northern part of the species range in Georgia and Florida.
The pattern was curvilinear, however, with most of the variation occurring
in the east. His pattern is only vaguely apparent in the progeny data
of the present study--a high occurred in east Georgia but another high
occurred in the extreme western portion of the species range.
Thorbjornsen (1961) found geographic variation in stomata per m.
in natural stands of loblolly pine. His pattern ws somewhat similar to
Mergen's, frequency tending to be highest in the eastern part of the range.
But the trend was not uniform, the pattern appearing to be somewhat random
eat of the Mississippi river. He also found a rather strong positive
correlation of stcaSta per m. with a drought index, the ratio of May-
August precipitation over average suaer temperature. A check for a
similar relationship was sought in the present data for slash pine, with
no success--if anything there was a slight negative trend. Apparently
the relationship Thorbjornsen found was mainly due to the very low
sumer rainfall west of the Misisissippi being coincident with low stomatal
frequency in that area. If so, the lack of a relationship for slash pine
is not surprising.
to oo d to
cli, d r
re c~ aw
~I io 1 ) 0
Thames (1963), sapling loblolly pine seedlings originating from
areas in Caldwell and Cherokee Counties, Texas, northwest Georgia, and
Crosett, Arkansas, found stomatal frequencies (both stomata per mrm. and
stomata per sq. nm. of needle surface) to be lowest in the two Texas
sources, which agrees with Thorbjornsen's results. Although there were
only two sources east of the Mississippi the two traits showed no
consistent east-vest trend in this region.
Thames (1963) found no significant racial difference in number of
rows of stoata in loblolly pine and this was also found to be true for
provenances of European larch (Larix decidua Mill.) (Gathy, 1959).
Low stomatal frequency a y be an adaptation to xeric conditions as
suggested by Thames (1963). High stmnatal frequency may be associated
with photosynthetic efficiency as found in Ribes by Bjurman (1959).
Number of resin ducts
The number of ducts in parental foliage averaged 6.90 per needle,
ranging froa 2 to 13 among individual needles, and from 3.0 to 10.2
among mother tree means (Table 3). Trees of the densa variety averaged
slightly more ducts than those of the elliottii variety or those in the
transition sone, but the differences attributable to such groupings were
not significant (Table 4). Stands-within-groups was significant but
accounted for only 9 per cent f the variance. The pattern among stand
means was rather intricate, highs occurring in south-central Georgia,
and also along the coasts of Florida (Fig. 29). The low in extreme
southeast Florida agrees with d reported by De Vall (1941b).
The high mother tree compenent (89 per cent) may be largely due to
environmental edification rather than to genetic differences among
trees. White and Beals (1963) showed that resin duct frequency in pond
pine (Pinus serotina Michx.) was related to tree age, growth rate, vertical
position in crown, and "crown exposure side." Their findings suggest
further that even the stand variance may be due to environmental modifi-
cation rather than racial effects.
In the progenies the numbers of ducts were uch fever, averaging
2.40 and ranging from o0. to 5.0 aong seedling means (Table 7). Complete
absence of ducts was extremely rare, being found in the sample of two
needles from a single seedling. "Twos" and "threes" were the most common.
Very little of the variation in progenies was associated with groups
or stands, error accounting for most of it (Table 8). The pattern of
variation among stand means was largely random (Fig. 30). These results
do not agree well with those of Mergen (1958), who found that slash
pine seedlings fro the central and northeastern counties of Florida
and southeastern Georgia had the fewest ducts.
The absence of a distinct difference in number of resin ducts in
parental foliage between the varieties of slash pine agrees with Little
and Dorman's (1954) findings, but not entirely with those of others as
indicated in the tabulation below.
Author elliottii densa
- Numbers of ducts -
De vail (1941a) 3-5 4-9
De Vail (1945) 2-3a 4-9a
Little and Dorman (1954) 2-8b 3-9b
West and Arnola (1956, p. 6) 3-4 5-10
Present study (ranges among 3-10 4-9
mother tree means)
a Resin droplets visible with a hand lens on a cut surface
in this case.
b For natural stands; the authors showed generally fewer ducts
for plantations, whch may have been an a effect.
Thickness of hypoderm
Although the thickness of hypoderm in the parents averaged only
slightly greater in the densa variety than in elliottii the differences
were significant, 37 per cent of the variance being associated with
groups of stands (Tables 3 and 4). The stand means displayed a clinal
pattern, increasing frno north to south, through much of Florida and a
random one in the north (Fig. 31).
In the progenies the results were completely different. Groups and
stands accounted for relatively small (although significant) portions of
the variation, 7 per cent each (Table 8). North Florida progenies had
slightly thicker hypoderms, on the average, than south Florida ones
(Table 7). But the over-all pattern of stand means showed no clear cut
trends, and contained a large element of randomness (Fig. 32).
The outer, thin-walled hypoderm layer was invariably present in both
parent and progeny material. In the parents at least one fairly continuous,
inner, thick-valled layer we preset. In the pregenies, horver, the
inner "lyer" often oemsisted of sporadic thiek-wlled cells.
Ike results for parent toees agree fairly vll with Dexma and
Little (195I), although the mrait k of the differences they reported
between elliettii (twe, rarely thr layers) and dema (three to four,
rarely two or five) were greater the i f d here (Table 3). This mw
hwve been due to the fact that only current yea's needles were used
in the prm t study. The poorly developed hypodem fond in seedlings
is probably a ag effect. Because of this ea should aot oenelude
that thm variation in thickness of hypedem in mtw trees is net
gEaetic in nature. In a racial variation study vith pederesa pine,
Veidman (1939) did find that geographi difference in this trait were
inherited to a large eteat.
Little and Dorwa (195~), vh studied Caribbem pine as wll -
slash pine, suggested a peeeible tie-in with cllate, thick hypedF r
being assoeeited vith a deemt dry sees for these subtropical ma
trpical pines. In ponderose pine thick hypoder seem to be seeciated
with sevre climtes (WeidLa, 1939).
Diseissien of Individual Trait Variation
At this point the individual trait patterns ad the omyonmts
of vaerice found in the analyses hall be suamrimel, ad the causes
and nature of the patterns sehll be explored frme the genetic standpoint.
Six of the 12 trait stwuAed in the parents ad 11 of the 13 studied
in the progenies showed siificant difference (either at the 5 or the
1 per cent level) mang gr ps of stads. The prevalence of these
difference vuw not surprising stoee they enempn e the whole species