THE SIGNIFICANCE OF POPULATION SUCCESSIONAL STATUS TO THE EVOLUTION OF SEEDLING MORPHOLOGY IN PINUS CONTORTA VAR. LATIFOLIA (PINACEAE)

14 Pages·2001·12.3 MB·English

by  Timothy J Brady

Madrono, Vol. 48, No. 3, pp. 138-151, 2001
THE SIGNIFICANCE OF POPULATION SUCCESSIONAL STATUS TO THE
EVOLUTION OF SEEDLING MORPHOLOGY IN PINUS CONTORTA VAR.
LATIFOLIA (PINACEAE)
Timothy Brady'
J.
Department of Biology, Harney Science Center 342, University of San Francisco,
2130 Fulton Street, San Francisco, CA 94117-1080
Abstract
The objective ofthis research was to testthe hypothesis thatthe successional role ofaplantpopulation,
because of its imphcations for the nature of the selective regime experienced by regeneration cohorts,
determines, in part, the course ofautecological evolution in a lineage ofpopulations. A provenance study,
which involved the raising of seedlings of lodgepole pine (Pinus contorta Loudon var. latifolia) from
seed under uniform conditions in a greenhouse, provided a test ofthis hypothesis. The seeds came from
serai, climax, and persistent lodgepole pine populations indigenous to the Blue Mountains region of
northeastern Oregon and southeastern Washington. Data about shoot and root system features, collected
at the end ofthe first season ofgrowth, proved useful in evaluating competitive competence, the relative
ability of a plant, or group thereof, to compete successfully for essential resources such as light, water,
and mineral nutrients. Analyses of variance and discriminant function analysis facilitated the search for
correlations between population successional status and seedling morphology. The total leaf area, pho-
tosynthetic potential, degree of subdivision ofthe root system, and total root length ofa typical seedling
derived from a serai population are smaller than such quantities for the other population types. In climax
populations, seedhngs tend to exhibit the largest total leafareas and total photosynthetic potentials among
lodgepole pine seedlings. They are equipped with more elaborate and larger root systems. Despite their
exceptional heights, seedlings belonging to persistent populations not only have slightly lower total pho-
tosynthetic potentials than those from climax populations, they also possess reduced lateral root densities
and total root lengths. Greater competitive competences, with reference to life in the subcanopy, char-
acterize seedlings from climax lodgepole pine populations compared to seedlings from serai orpersistent
populations. The results of this study support the hypothesis that successional status influences the evo-
lution ofautecological attributes in a population lineage.
Forest succession is the undisputed exemplar of sequently, a plant almost invariably must partici-
vegetation change through time. The formation of pate in intense intra- and interspecific competition
a canopy as the crowns of adjacent trees expand in when a resident of the subcanopy. Any new aut-
size represents the most consequential structural ecological trait appearing through mutation or gene
change occurring during forest succession. In co- flow that improves the "competitive competence"
niferous forests, canopy closure may not occur for of a subcanopy plant will increase its fitness, i.e.,
over a century after stand establishment (Feet the likelihood that it will become a part ofthe can-
1981). The appearance of a canopy results in the opy and contribute to the genetic constitutions of
elimination of an open site and the creation of a future generations.
subcanopy. Gradual alterations in stand structure Competitive competence is a relative expression
and abiotic environmental conditions occur as an ofthe ability ofa plant, or group thereof, given the
opeAnnsoitpeengivseitse,wtahyoutgohaitcainsoapnieadresatacndh.aracterized foefagteunreetsitchsa,tpchhyasriaocltoegryiz,eainta(tionmcyl,udimnogrpmhaonlyogays,peacntds
by physical extremes, is a place where a plant, due breeding behavior), to compete successfully for
to the absence or paucity of other plants with sim- those resources that limit maintenance and growth
pileatritrieoqnuifroremeesnstesn,tipaalrtriecsiopuartceessinsulcithtlea,silfigahnty,,wcatoemr-, activities. It is an heuristic tool that provides a
means of comparing plants according to their abil-
and mineral nutrients. In contrast, the canopy in-
sulates tree seedlings and other inhabitants of the ities to survive, grow, and reproduce under a com-
subcanopy from the severe abiotic environmental petitive regime. Competitive competence recalls the
conditions that prevail on an open site. Neverthe- general version of the concept of tolerance em-
less, because resources are highly accessible to, and braced by most silviculturists during the first half
vigorously exploited by, the fully developed trees of the twentieth century (e.g., Buhler 1918; Baker
composing the canopy, at least one resource ordi- 1937, 1950; Toumey and Korstian 1937). Compet-
narily limits the survival ofsubcanopy plants. Con- itive competence permits the comparison of plants
growing in naturally complex settings, not under
simplified garden, greenhouse, or laboratory con-
E-mail: [email protected] ditions; it avoids the artificiality of the tendency to
2001] BRADY: SEEDLING EVOLUTION IN LODGEPOLE PINE 139
rank individuals, populations, or species by single-
factor tolerances.
To ensure the continued existence of its popula-
tion lineage (a temporal sequence of conspecific
populations related as ancestors and descendants),
the members of a population must beget juveniles
that are able to compete successfully for resources
within the subcanopy, or they must produce prop-
agules that disperse and give rise to individuals ca-
pable of surviving, growing, and reproducing in a
different location (seed dormancy and cone serotiny
may permit the continuity of a population lineage
in the absence of emigration). The autecology of a
population represents the product of evolutionary
history, acting as a phyletic constraint that limits
the types of successional role that a population can
assume in a given environment. In view of its con-
sequences for regeneration dynamics, successional
status certainly contributes significantly to the se-
lective milieu experienced by the members of a
population. Consequently, succession potentially
influences the course of autecological evolution
within a population lineage. Populations of lodge-
pole pine indigenous to the Blue Mountains region 0 200km
ofthe Pacific Northwest, which occupy an immense 'hi
habitat island, represent consummate candidates for
studying the relationship between successional sta- t
tus and the evolution of competitive competence.
The successional roles ofBlue Mountain lodge- Fig. 1. The Blue Mountains Province, located in north-
polepinepopulations. The range ofPinus contorta eastern Oregon and southeastern Washington. Modified
var. latifolia (hereafter referred to by its vernacular from Franklin and Dyrness (1988:6).
name "lodgepole pine") encompasses the Rocky
Mountains, the Washington Cascades, the area be-
tween the Rockies and Coast Range in Canada, and Dyrness (1988) defined the Blue Mountains Prov-
the Blue Mountains. Due to its rapid early growth ince as the 2.5 million hectare region of mountain
rate and its tolerance of exposed conditions and ranges and intervening valleys that extends from
poorly developed substrates, lodgepole pine estab- central Oregon, east of the Cascades, to southeast-
lishes a foothold on many open sites throughout its ern Washington (Fig. 1).
range (Pfister and Daubenmire 1973; Volland Serai populations. Given the availability of a
1985). Hence, lodgepole pine often is a colonizer. nearby seed source, a lodgepole pine population as-
In most cases, because of a dearth of nearby seed sumes a serai role in an essentially unidirectional
sources, it becomes only a minor constituent ofthe successional sequence upon colonizing an open site
stands on these sites. Lodgepole pine adopts a serai characterized by more-or-less moderate abiotic en-
role in such situations: It does not regenerate suc- vironmental conditions. Lodgepole pine becomes
cessfully beneath the canopy; species having great- the dominant canopy element because of its rapid
er shade tolerances, viz., grand fir {Abies grandis early growth rate. In the subcanopy, lodgepole pine
(Douglas) Lindley) or subalpine fir (A. lasiocarpa seedlings, which represent the offspring of canopy
(Hook.) Nutt.), replace it in 50-200 years. On some dominants, participate in a futile fight for essential
sites, lodgepole pine populations achieve domi- resources with individuals of species that exhibit
nance and assume unique successional positions. superior competitive competences (grand firor sub-
Pfister and Daubenmire (1973) identified three gen- alpine fir). In the absence of stand disturbance, the
eral successional roles for populations oflodgepole lodgepole pine population is unable to replace itself
pine in stands that it dominates: (dominant) serai, in situ. Its adversaries attain dominance in 100-200
climax, and persistent. These three population types years. Perpetuation of the population lineage to
do not represent different temporal components of which the lodgepole pines belong depends upon lo-
a common sere. Rather, each is a particular element cal disturbance or the abilities of the seeds of ma-
of a unique sere. F. C. Hall (USDA Forest Service, ture individuals to reach, and give rise to viable
unpublished) classified lodgepole pine populations seedlings on, other (open) sites. As they produce
according to successional role in communities that highly vagile propagules and seedlings capable of
they dominate in the Blue Mountains. Franklin and tolerating the harsh abiotic environmental condi-
MADRONO
140 [Vol. 48
tions of open sites, these serai lodgepole pines are the site. In 60-80 years, the dominant lodgepole
exceedingly fit desite their inferior competitive pine trees reach maturity (dbh > 14.5 cm) and be-
competences. come susceptible to attackby mountain pinebeetles
i
{Dendroctonus ponderosae). The deaths of these
Climax populations. Because of the exceptional trees produce a high fuel load within the stand, j
frost resistances and shallow root systems of its which inevitably leads to a hot fire 5-10 years later.
seedlings, lodgepole pine is capable of becoming The fire consumes all live and dead woody material
established in a topographic depression where cold on the site, including canopy members and poten-
eaisrtahcecsuomiullsautrefsacaen.dAthleodwgaetpeorlteabplieneofptoepnulaaptpiroonacahs-- ttihael rdeigsetnuerrbaetdiosniteinbtehceosmuebscasnuoiptyab.leInfo1r0-t2he0 eysetarasb,-
sumes a climax role in a unidirectional, but trun- lishment of lodgepole pine seedlings. Following
cated, sere in such a harsh location. Individuals of dispersal from some external seed source, the se-
other tree species appear on such bitterly cold, and quence begins anew. Succession proceeds until a
in some cases, periodically inundated, sites only on beetle kill and fire again destroy the stand. As a
occasion and in insignificant numbers (though, Pic- persistent population cannot replace itself in situ,
ea engelmannii, Engelm. is quite abundant on some the appearance oflodgepole pine after each fire de-
sites). Even if potentially competing seedlings of pends upon seed dispersal from another site. The
other species are present, the superior initial growth temporal continuity of a persistent population lin-
rate of lodgepole pine ensures its rapid domination eage depends upon the success with which the
of the canopy. In the subcanopy, the offspring of seedlings ofits constituent populations can colonize
canopy lodgepole pines generally remain free of open sites. Hall's view of the successional role of
competition for essential resources from species a persistent Blue Mountain lodgepole pine popu-
having greater competitive competences (grand fir lation is controversial. The details of his proposed
and subalpine fir). Although intraspecific competi- successional scenario will likely change once stand
btiloenainsd,fiienrcfea,ctl,ocdgoempmoolen.piPneersroengaelneorbasteirovnatiisopnosssiin-- dviesvieolnaolplmye,ntthatstsuedieedsvaagrielictoympalnedtes.eedIlainsgsutmoel,erparnoc-e
dicate that climax populations usually contain a ofharsh abiotic environmental conditions, not seed-
mixture of size classes. Seedlings and saplings ling competitive competence, contribute most sig-
crowd many light gaps. Stuart et al. (1989) dem- nificantly to the fitness of a persistent lodgepole
onstrated that water/mineral nutrient gaps are nec-
pine tree.
essary for seedling establishment in climax popu-
lations of Finns contorta Loudon var. murrayana
(Grev & Balf.) Crichf. in south-central Oregon. Re- Materials and Methods
cruitment and the ascendancy of suppressed lodge- Cone collection. A lodgepole pine population
pole pines into the canopy in the Blue Mountains represented a dominant canopy element in 61
may require gap formation. Unlike a serai popula- stands inspected by F. C. Hall (USFS unpublished
tion lineage, a climax population lineage can persist data) as part of a plant association analysis of the
through regeneration in situ, provided that it takes Blue Mountains region. Twenty of these popula-
place with sufficient frequency (at intervals no tions proved appropriate for inclusion in this study:
greater than the longevity of lodgepole pine, about In August 1994, each had at least one dominant,
250 years, according to Franklin and Dyrness codominant, or intermediate lodgepole pine tree
[1988]), and in adequate quantities. In view of the that was cone-bearing, accessible, and apparently
ilnotdegnespeolientpriasnpeecciafincopyc,ompseetliecttiioonn bweilllowcoanfecrliemlaex- fsrceheeomfeinisnetcrtoddaucmeadgeab(oTvaeb.leH1a)l.lUsiidnengtitfhieedtrispiaxrtiotfe
vated fitnesses on individuals that produce seed- these populations as serai (species with superior
lings with greater competitive competences. competitive competences present with height
Persistentpopulations. Ten to twenty years after growth rates that exceed those of lodgepole pine
a crown fire destroys a portion of a mature grand trees), seven as climax (topographic depressions;
fir stand, lodgepole pine seeds originating else- species with superior competitive competences ab-
where effect colonization ofthe resulting open site. sent), and seven as persistent (species with superior
The lodgepole pine population takes on a persistent competitive competences present, but lodgepole
role in a cyclic sere on such a site. Due to intense pine trees exhibit greater height growth rates).
competition with grasses and shrubs (and some- Within each successional type, the populations are
times grand fir seedlings), the growing lodgepole widely distributed throughout the Blue Mountains
pines are widely spaced on the site. Nevertheless, with respect to latitude, longitude, and elevation. To
because oftheir rapid early growth rates, they soon reduce the chances that gene flow, not common
form and dominate a canopy. Before long, the sub- site-specific selective pressures, might produce a
canopy becomes filled with lodgepole pine seed- correlation between successional status and com-
lings (the progeny ofcanopy dominants) as well as petitive competence, I admitted no contiguous pop-
seedlings having greater competitive competences, ulations possessing identical successional roles into
i.e., the offspring of grand fir trees that surround this study.
''''
2001] BRADY: SEEDLING EVOLUTION IN LODGEPOLE PINE 141
Table 1. Blue Mountain Lodgepole Pine Populations Sampled. The number of participating families indicates
those submitted to analyses ofvariance and discriminant function analysis.
Participating Latitude Longitude Elevation
Population families (N) (W) (m)
oeral populations
Mount Pisgah 2 44°28' 120°14' 2,100
Little Kelsay Creek 1 44°54' 118°45' 1,900
North Fork WolfCreek 2 45°08' 118°08' 1,900
Bingham Spring 1 44°30' 120°30' 1,900
Thompson Spring 2 44°28' 120°15' 2,200
Little Phillips Creek 3 45°42' 118°03' 1,700
Bingham Prairie 2 44°31 120°32' 1,900
Jackson Creek 2 44°27' 119°58' 1,800
Crowsfoot Creek Edge 1 43°54' 119°30' 1,800
Summit Prairie Edge 2 44°11 118°30' 1,800
Wickiup Creek 2 44°1 1 119°14' 1,800
Ditch Creek Edge 1 45°07' 1 19°21' 1,600
Myrtle Creek 2 43°59' 119°05' 1,900
Persistent populations
Stove Spring 2 44°31 120°33' 1,800
Summit Prairie Slope 3 44°11' 1 18°30' 1,900
Camp Creek 2 44°03' 1 19°07' 1,800
Tribble Creek 1 45°10' 119°02' 1,800
Indian Springs Butte 1 44°15' 118°42' 2,100
Dixie Butte 1 44°33' 118°37' 2,000
Winom Creek 1 45°01' 118°38' 1,700
Several assistants and I collected cones from one Seed extraction. Cone processing took place in
or more lodgepole pine trees belonging to each September 1994 at the Wind River Nursery Seed
population during August 1994. Since minimum Extractory operated by the USPS near Carson,
distances of 40-50 m separated the sampled trees, Washington. After transferring the cones from the
each most likely resides within a distinct genetic paper bags to loose-weave nylon sacks, we im-
neighborhood. We recovered a minimum of 50 mersed them in hot water (about 60°C) for 4-5
cones from each tree. We picked the cones off minutes. We dried the cones in a kiln dryer at 35°C
branches removed from each crown by gunshot or for 24 hours. This treatment effectively opened
through the use of loppers. To maximize the pro- nearly all of the cones. We shook the seeds out of
portion within each family ofprogeny derived from the open cones by use of a manually-driven tum-
crosses between members of the same genetic bler. A Clipper cleaner permitted us to partially de-
neighborhood, we generally collected cone-bearing wing the seeds and remove debris from each seed
branches from the sides of crowns (the infiltration lot. We used an x-ray machine and Polaroid film to
of pollen grains from afar to branches below the generate images of the contents of 19-200 seeds
tops of crowns probably is limited, especially for per cleaned sample.
trees belonging to serai and climax lodgepole pine
populations where crowns frequently are contigu- Seedling production. After soaking 39 seed lots
ous or overlapping). We used positional criteria so (each representing a single family) in water for 72
as to collect only those cones that matured in re- hours, I stratified them without media in polyeth-
sponse to pollination events occurring in the spring ylene bags at 2°C for 33 days (Bonner et al. 1974;
of 1992. Although most lodgepole pine trees in the Krugman and Jenkinson 1974; Owens and Molder
Blue Mountains possess nonserotinous cones (Lo- 1984; J. McGrath, USDA Forest Service, personal
tan and Critchfield 1990), we were careful to avoid communication). We sowed the seeds immediately
older closed (serotinous) cones. At the time of col- after completion ofthe stratification process in con-
lection, most cones were light brown in color, tainers in a fiberglass, unhealed greenhouse at the
which is indicative of ripeness (Krugman and Jen- University of California, Berkeley. We sowed 2-5
kinson 1974). Many were beginning to open. We seeds in a 1:1 mixture of sphagnum peat moss and
stored the cones in paper bags, loosely packed in vermiculite in each of 50 Ray Leach Pine Cell
cardboard boxes, in the open bed of a truck for the Cone-tainers per family. The sowing rate, calculat-
duration ofthe fieldwork (up to four weeks prior to ed by reference to the x-ray image of a subsample
seed extraction). of seeds produced at the time of seed extraction.
MADRONO
142 [Vol. 48
varied in direct proportion to the percentage of Table 2. Descriptions of 12 Attributes Pertainingto
filled seeds. We subsequently placed each container Shoot and Root System Morphology Assessed for
in a randomly chosen slot in a clusterofrectangular Lodgepole Pine Seedlings on which Terminal Resting
trays in the center of the greenhouse. We raised Buds had Developed within 151 Days of Growth in a
Greenhouse.
over 1,000 lodgepole pine seedlings through a sin-
gle three-stage season (151 days) under uniform SHL (shoot length)
conditions in the greenhouse. To encourage seed The distance between the cotyledonary node and the base
germination (1-24 days after sowing), we main- ofthe terminal resting bud on the main stem as measured
tained high air temperatures andrelativehumidities, with a ruler to the nearest millimeter.
suspended a 50% shadecloth near the greenhouse STC (stem caliper)
rgoroofwitnogremdeudcieupmhointoenacfhlucxondtenasiinteiresc,onasntdanktelpytwetth.e The maximum diameterofthe main stem at its midlength
As necessary, we thinned each container to a single 0i.S0H2L/52)m,m.as measured with a verniercaliperto the nearest
seedling. During the free growth phase (25-115
days after sowing), we operated an evaporative NPL (number of needle-like primary leaves)
cooling system for 12 hours per day, illuminated The number of needle-like primary leaves attached to the
the seedlings with fluorescent lights for 18 hours main stem, disregarding cotyledons.
perday, and injected a Plantex 20:20:20 macro- and NSS (number of axillary short-shoots)
micronutrient solution (100 ppm of nitrogen) into The number ofaxillary short-shoots attached to the main
the irrigation system during every other watering stem.
(about once each week). We created drought stress NLS (number ofaxillary long-shoots)
conditions and terminated the use of daylength ex- The number of axillary long-shoots attached to the main
tension lights at the beginning of the budset stage stem.
p(e1r1i6o-d15o1f sdeaeydslianfgtegrrsoowwtihn,g)w.eTuhsreodugahnouetxhtahuisstfifnaanl BLL (blade length)
for 24 hours per day to encourage lower tempera- Pertaining to the intact and fully developed primary leaf
tures and relative humidities, and we halved the closest to the midlength of the main stem {SHL/2), the
distance between the base and tip ofthe leafblade along
pniptmrogoefnnictornotgeennt).of each fertilizer application (50 mitestemri.drib, as measured with a ruler to the nearest milli-
Collection ofdata on seedling morphology. The- TPC (taproot caliper)
oretical predictions and empirical evidence suggest The maximum diameter of the taproot at its midlength
that, regardless of sibling relatedness and heritabil- (actually, halfthe distance between the cotyledonary node
ity, an acceptable approximation of a family mean and the tip of the taproot), as measured with a vernier
for a given quantitative morphological feature is caliper to the nearest 0.025 mm.
obtainable by sampling 10-20 progeny (Brady un- NLR (number of lateral roots)
published). Consequently, within three weeks of The number of lateral (secondary) roots attached to the
termination of the growth period (i.e., the end of proximal half of the taproot (the region between the cot-
the bud-set stage), we harvested ten randomly-cho- yledonary node and the midlength ofthe taproot).
sen seedlings per family on which terminal resting LRL (lateral root length)
b1u2dsdishtaindctdemvoerlpohpoeldogiacnadlaactqturiibruetdes.inAfordmeastciroinptifoonr Tlehnegtlhenogftthheoftatphreoolta,tearsalmreoaostuartetdacwhietdhcalorsuelsetrttootthheenmeiadr--
of each trait appears in Table 2. est millimeter
Statisticalprocedures. Following the recommen- A^77? (number oftertiary roots)
dations of Gould and Johnston (1972) for identi- The number oftertiary (absorbing) roots connected to the
fying patterns of geographical variation within and lateral root that is attached closest to the midlength ofthe
among species, I adopted different approaches, viz., taproot (we consideredatertiaryroot andallofitsbranch-
analyses of variance and discriminant function es, ifpresent, as a single unit).
analysis, in my attempt to discover a correlation SHB (shoot biomass)
between lodgepole pine seedling morphology and The weight of the air-dried shoot system (epicotyl), as
population successional status. The analysis ofvari- obtained with an electronic balance to the nearest 0.01 g.
ance (ANOVA) furnishes a way to evaluate the ob- ROB (root biomass)
served differences among three or more (statistical) The weight of the air-dried root system (hypocotyl), as
population means (Winer 1971; Tabachnick and Fi- obtained with an electronic balance to the nearest 0.01 g.
dell 1983; Lindman 1992; Bogartz 1994). In the
context of the present investigation, successional
status functions as the sole independent variable
with three levels, or groups (serai, climax, and per- make use of a set of independent variables to dis-
sistent). The morphological features act as depen- cover the dimensions along which the differences
dent variables. Discriminant function analysis in- among groups are greatest, to test the statistical sig-
cludes an array of multivariate techniques that nificance of those differences, to predict group
2001] BRADY: SEEDLING EVOLUTION IN LODGEPOLE PINE 143
Table 3. The Mean and Standard Deviation of the Family Mean of each of 12 Morphological Features of
LoDGEPOLE Pine Seedlings by Successional Class for 34 Families. Units ofmeasurement appearin parentheses. All
numbers were rounded to two digits to the right ofthe decimal point for display purposes.
SHL STC BLL TPC LRL SHB ROB
(mm) (mm) NPL NSS NLS (mm) (mm) NLR (mm) NTR
serai (total number of families = 11)
mean = 58.36 1.87 106.86 4.66 2.02 35.56 0.66 27.20 92.38 26.49 35.56 28.05
sd = 8.17 0.11 10.41 4.10 0.64 2.70 0.11 2.62 5.33 3.79 7.40 4.69
climax (total number of families == 12)
mean = 61.00 1.95 1 17.54 1.48 2.26 36.68 0.65 29.66 98.23 29.14 37.13 33.33
sd = 8.00 0.09 16.35 1.91 0.23 1.32 0.12 2.44 7.74 4.01 3.82 2.47
persistent (total number of families = 11)
mean = 72.26 1.87 1 15.23 2.60 2.26 37.14 0.69 26.68 95.42 27.69 38.75 32.01
sd = 7.58 0.12 1 1.86 3.25 0.23 1.87 0.07 3.08 6.90 3.77 3.12 3.87
membership, and to interpret the (biological) mean- bution offamily means, which facilitated the visual
ing of each dimension (Kendall and Stuart 1966; detection of skewness, and I tested the null hypoth-
Lachenbruch 1975; Gnanadesikan 1977; Karson esis that this distribution is normal using the Sha-
1982; Tabachnick and Fidell 1983; Reyment et al. piro-Wilk W-statistic. An assumption of normality
1984; Morrison 1990). In this study, successional was supported for eight of the attributes in every
status is the dependent variable with three groups class (a = 0.05). The distribution of one seedling
(serai, climax, and persistent). Morphological attri- feature, NSS (number of axillary short-shoots), ex-
butes act as independent variables. I performed all hibited severe positive skewness in the climax and
quantitative analyses for this study on a Macintosh persistent groups. Therefore, I did not probe A^^^S
Quadra 950 using JMP 3.1 application software or further using ANOVA. I performed four different
programs written by myself and executed with the statistical tests (O'Brien's, Brown-Forsythe, Levene
Microsoft QuickBASIC l.OOB Interpreter. F, and Bartlett's tests) to check the homogeneity of
variance among the three successional classes for
Results and Discussion each of eight morphological characteristics. None
=
I calculated the means and standard deviations of of the tests detected statistically significant (a
family means by successional class (serai, climax, 0.05) differences in variance among groups for any
and persistent) on each of 12 morphological traits of these traits.
ANOVAs
for a total of390 lodgepole pine seedlings (39 fam- Eight revealed that the differences
ilies). Examination ofthese statistics as well as his- among group means are statistically significant (a
tograms depicting the frequency distributions of = 0.05) for three morphological attributes: SHL
families for these features (not shown) indicated (shoot length), NLR (number of lateral roots), and
that five families (one serai, two climax, and two ROB (root biomass). In each case, t-tests identified
persistent) represent outliers. I made no further use the particular group differences responsible. The re-
of the data for these five families. Table 3 gives sults of the significant tests appear in Table 4.
summary statistics on each ofthe 12 morphological Year-old seedling shoots are, on average, longer
features by successional class for the remaining 34 in families derived from persistent populations
families. The mean offamily means on three ofthe (mean SHL = 72.26 mm) than in families belong-
attributes each exhibits very little variation among ing to either serai or climax populations (58.36 mm
successional classes: NLS (number ofaxillary long- and 61.00 mm, respectively). However, no real dif-
shoots), TPC (taproot caliper), and SHB (shoot bio- ference in seedling shoot length exists between ser-
mass). Consequently, I withdrew them from further ai and climax groups. The t-tests detected signifi-
statistical consideration. cant differences in the number of lateral roots be-
Analyses ofvariance. The validity of the results tween serai and climax and between climax and
of a series of ANOVAs depends upon the assump- persistent groups. In fact, the largest mean number
tion that the scores on a particular dependent vari- of lateral roots per seedling (mean NLR = 29.66)
able within each group are approximately normally characterize climax families. However, no evidence
distributed, and upon the assumption that each exists of a genuine difference in the number of lat-
group possesses a common variance on a given de- eral roots between serai and persistent families
pendent variable. I evaluated the normality assump- (27.20 and 26.68, respectively). The biomasses of
tion separately for the nine morphological traits seedling root systems are significantly smaller in
within each ofthe three successional classes in two serai families (mean ROB = 28.05 g), but the cli-
ways: I examined a histogram showing the distri- max and persistent groups are not statistically dis-
1 1
MADRONO
144 [Vol. 48^
Table 4. Results of Significant (a = 0.05) One-way Analyses of Variance and t-tests of the Differences
AMONG Successional CLASSES IN Mean Values ON EiGHT LoDGEPOLE PiNE SEEDLING Traits. df = degrees offreedom,
SS = sum of squares, MS = mean square. I rounded all numbers to two digits to the right of the decimal point for
display purposes.
SHL (shoot length)
Source df SS MS F P
successional status 2 1208.22 604.1 9.62 0.00
error 31 1946.33 62.79
'seraiipersistent — 4.14 Huif —— 70 P = 0.00
'climax:persistent = -3.46 df = 21 P = 0.00
NLR (number oflateral roots)
Source df SS MS F P
successional status 2 58.82 29.41 3.98 0.03
error 31 228.83 7.38
'seralxlimax = 2.33 df = 21 P < 0.03
'climax:persistent = 2.58 df = 21 P < 0.02
ROB (root biomass)
Source df SS MS F P
successional status 2 171.28 85.64 6.08 0.01
error 31 436.90 14.09
'seralxlimax = 3.43 df = 21 P = 0.00
'serahpersistent = 2.16 df = 20 P < 0.04
tinguishable on the basis of this trait (33.33 g and tions of them exhibit normality) is a prerequisite
32.01 g, respectively). for the use of discriminant function analysis. With
A series of ANOVAs yields the maximum small, unequal sample sizes (as in this study), val-
amount of information about the importance of idation of the multivariate normality assumption is
each attribute to the determination of group affili- largely a matter ofjudgment. By discarding the at-
ation only if none of those traits covary. A matrix tribute NSS (number of axillary short-shoots),
of Pearson product-moment correlation coefficients shown previously to possess a highly skewed dis-
(Table 5) indicates that, in fact, every pair of mor- tribution within each successional class, the validity
phological attributes is, to some degree, correlated. of the assumption ofmultivariate normality is like-
Thus, the claim that differences among successional ly.
classes on the mean values of SHL, NLR, and ROB A discriminant function analysis of34 lodgepole
are statistically significant incorrectly implies that pine family means on each of eight moiphological
successional status affects three independent phe- characteristics, which accounts for about 68.2% of
nomena. A closer examination of the relationship the total variation in seedling morphology, created
between seedling morphology and successional sta- two discriminant functions (Table 6). The first func-
tus demands a multivariate perspective. tion effectively ordinates all three successional
Discriminantfunction analysis. Justification of a classes by partitioning approximately 66.5% ofthe
multivariate normality assumption (within each variation among successional classes to achieve
group, the sampling distribution of the mean on group separation (Fig. 2). The second discriminant
each independent variable and all linear combina- function appropriates about 33.5% of the variation
Table 5. Correlation Matrix Obtained by Computing Pairwise Pearson Product-Moment Correlation Coeffi-
cients Across all Three Successional Classes for Eight Morphological Features.
SHL STC NPL BLL NLR LRL NTR ROB
SHL 1.0000
STC -0.0049 .0000
1
NPL 0.3757 0.1093 .0000
1
BLL 0.263 0.0554 0.2300 .0000
1
NLR -0.3149 0.4287 0.0759 -0.0122 1.0000
LRL 0.3808 0.2128 0.2941 0.4289 0.0791 1.0000
NRTORB 00..13615090 -00..02516916 00..31215859 00..44344276 00..32587714 00..64630216 01..40204060 1.0000
7
2001] BRADY: SEEDLING EVOLUTION IN LODGEPOLE PINE 145
Table 6. Discriminant Function Analysis of a Data Set Consisting of 34 Lodgepole Pine Family Means on
Eight Features Pertaining to Seedling Morphology.
!
standardized discriminant function coefficients
Discriminant
function SHL STC NPL BLL NLR LRL NTR ROB
1.0L38 0.1790 0.2367 -0.2789 0.0476 0.5094 -0.0759 0.2698
0.311 0.1089 0.2082 0.1081 0.2483 0.0600 -0.0327 0.5586
Comparison ofpredicted and actual group membership
Predicted Actual group membership
group
membership Serai Climax Persistent
Serai
Climax
Persistent
Loading matrix
Discriminant
function SHL STC NPL BLL NLR LRL NTR ROB
0.7301 0.3860 0.0434 -0.1769 0.5557 0.1818 0.1757 0.0939
0.5628 0.3780 0.5765 0.4998 0.3649 0.5492 0.4326 0.8950
among successional classes to distinguish seral port the assumption of homogeneous variance-co-
from the other groups (Fig. 2). variance matrices, which validate the use of dis-
A jackknife technique failed to expose any un- criminant function analysis in this study. Two re-
usually large Mahalanobis' distances (the distances sults confirm the statistical significance ofthis mul-
in multivariate space from family means to their tivariate inquiry: Firstly, an approximate F-ratio
group centroids); and plots of standardized family justified rejection of the null hypothesis that the
scores for the seral, climax, and persistent groups centroids associated with the three successional
revealed roughly equal dispersions. These findings classes are equal. Secondly, in a comparison of ac-
verify the absence of multivariate outliers and sup- tual group memberships and those based on pos-
terior probabilities obtained from Mahalanobis' dis-
tances, 82% of the 34 family predictions proved
2.5 correct (Table 6). The large value (0.73) of the as-
sociated Kappa statistic, which measures the agree-
ment between predicted and actual group affinity,
connotes that the results ofthe present discriminant
function analysis are, indeed, very reliable.
I referred to the loading matrix (Table 6) to ex-
plain the differences among seral, climax, and per-
sistent population types on each ofthe two discrim-
inant functions. Following statistical convention, I
deemed only those loadings of at least 0.50, which
implies an overlap in variance of about 25% be-
tween an independent variable and a discriininant
function, as eligible for interpretation.
A seedling's growth polarity. The first, and most
information-laden, discriminant function, or dimen-
sion of variation in seedling morphology, reflects a
-2.5 -1.5 -0.5 0.5 1.5 2.5 change in the pattern of asset allocation, not the
overall amount of growth. In morphometric terms,
Discriminant function 2
most of the variation among successional classes,
Fig. 2. The location ofeach ofthree lodgepole pine suc- as explained by the first dimension, is attributable
cessional classes in multivariate space as definedby adis- to the alteration of seedling "shape", not ''size".
criminant function analysis of eight attributes pertaining This finding differs markedly from the results of
tiozedseesdcloriensg omnortphheoliongdiyc.atEeadcdhisacxriismicnoannstisftusncotfiosnt.anAdadrodt- most multivariate morphometric studies, which
marks each group centroid (mean standardized discrimi- hopelessly confound the genetic determination of
nant function score). The area enclosed by a circle cor- form and phenotypic plasticity, and where the first
responds to the 95% confidence region around a given axis of variation corresponds to a generalized size
centroid. dimension (Reyment et al. 1984). All of the inde-
MADRONO
146 [Vol. 48
parative measure of the total stem photosynthetic
700- potential. The three points in Fig. 3 mark the av- i
persistent erage main stem surface areas for the three lodge-
"Ee 600- climax. \ pole pine successional classes (based on mean STC
ro 500- and mean SHL from Table 3). Due to their excep-
0()0 tional heights (mean SHL = 72.26 mm), seedlings
Q) 400- from persistent families possess the greatest stem
CO
1 300- photosynthetic potentials. Seedlings from serai and
E climax populations have comparable stem photo-
C0O) 200- synthetic potentials (mean SHL = 58.36 mm and
mm
lOOH 61.00 for serai and climax groups, respective-
ly)-
NLR (number of lateral roots) is the second mor-
Stemheight(mm) phological attribute making a large contribution to
the differentiation of successional classes along the
Foifg.st3e.m cSaulrifpearceanadreaheoigfhta.cTylhiendrtihcraele spotienmtsasinadifcuantcetitohne initial dimension of variation. As it provides an es-
average stem surface areas of lodgepole pine seedlings timate ofthe degree ofbranching, or subdivision of
derived from serai, climax, and persistent populations. the root system, NLR relates the thoroughness with
which a seedling can extract water and mineral nu-
trients from a given volume of soil (Fitter 1985,
pendent variables that contribute substantially to a 1991, 1994; Caldwell and Richards 1986). While
size dimension must exhibit discriminant function only its tip functions in uptake, a lateral root pro-
coefficients, correlations among themselves, and vides a ''platform" for numerous absorbing tertiary
loadings of like sign (Jolicoeur and Mosimann roots. Because the most efficient zone ofabsorption
1960; Reyment et al. 1984). In the present case, the occurs near the tip of any root, the total number of
two morphological attributes having the largest ab- root tips in a specified volume of soil acts as a
solute loadings on the first discriminant function measure ofthe absorptive capacity ofa root system
{SHL and NLR) possess discriminant function co- (Kramer and Kozlowski 1960; Caldwell and Rich-
efficients, correlations, and loadings of opposite ards 1986). Imagine a lodgepole pine seedling root
sign. The first dimension of variation clearly por- system that is completely embedded in acylindrical
trays an anisotropic pattern of change in seedling mass of soil. The height and vertical centerline of
growth among serai, climax, and persistent lodge- the cylinder correspond to the length and position
pole pine populations. of the taproot, and its radius equals the length of a
The trait SHL (shoot length) represents a measure lateral root. All lateral roots are identical in length,
of the total amount of seedling height growth dur- and each possesses the same number of attached
ing the first season of development, or a height tertiary roots. The following equation gives the to-
growth rate. Internode elongation, the primary tal number of root tips characteristic of the root
component of height growth, may lead to the pro- system per mm^ of soil:
dfluucxtidoennsiotfiense,witlmeaayvesmoivnereegxiiosntisngoflehaivgehserinptohostuonn- „ P ^ ^taproot^laterals ~^ ^taproot^Uiteralshateral^tertian-roots
light, and it tends to reduce self-shading by increas- '^(^llaatteerraall)' f'taproot
ing the distances among appendages attached to the (1)
stem (e.g., Horn 1971; Halle et al. 1978; Fisher
1986; Givnish 1986, 1988, 1995; Sakai 1986; Ko- where P is the number of taproot tips (usually, P
hyama 1987; Kuppers 1989). Since their epicotylar = 1); is taproot length; du,,,,„,, is a density, the
stems remained green during the first season ofde- number of lateral roots produced per unit length of
velopment, height growth via internode elongation taproot; lu,t,,,.ai is the length of a lateral root; and
may further contribute to the carbon economies of dtertiary roots t^c dcuslty of tcrtlary roots along a
lodgepole pine seedlings by promoting stem pho- lateral root. The product Itaprooidlaterals gives the total
tosynthesis (see Nilsen 1995). The total green stem number of lateral root tips. The product ^
surface area provides a means of comparing the aisllateral^tertiaryroots yicMs thc total numbcF of tcrtlary
stem photosynthetic potentials of different groups root tips. The denominator is the volume of the
of seedlings. As the epicotylar stems observed in reference cylinder of soil. Based on estimates ofP
this study lacked taper, I calculated the surface area (1), (160 mm) and d,„,„,, (0.29 per milli-
ofa stem by assuming that it is cylindrical in shape. meter of lateral root length) from this study (all are
The surface area of a stem increases with both essentially constant among the seedlings analyzed).
height and caliper (Fig. 3). Because long-shoots Fig. 4 presents Tas a function oflateral rootdensity
were relatively small and nearly constant in number for three different lateral root lengths. The number
across all lodgepole pine seedlings {NLS in Table of root tips composing a root system per unit vol-
3), the exclusive application of a simple cylindrical ume of soil increases in response to both an aug-
model to the main stem furnishes an adequate com- mentation in the number oflateral roots attached to
2001] BRADY: SEEDLING EVOLUTION IN LODGEPOLE PINE 147
10.0
7.5-
85 90 95 100 105
Lateral rootlength, ^^^)
//gfera/
Fig. 5. Total root length {R) as a function oflateral root
Fig. 4. The numberofroottips composing a lodgepole length dunerai)- The three labelled points mark the average
pine seedling root system permm"* ofsoil (T) as a func- total root lengths forthe three lodgepolepine successional
tion of lateral root density (<i/,,„,™/v) and lateral root classes.
length Thc three labelled points indicate the av-
Uiaierai)-
erage values of T for seedlings belonging to the serai,
climax, and persistent groups. lations among themselves, and loadings of similar
sign (Tables 5 and 6).
ROB (root biomass), the morphological charac-
the taproot and a decrease in the lengths of lateral teristic having the greatest loading on the second
roots (the radius of the cyhndrical mass of soil). discriminant function, grades the overall size ofthe
The three points in Fig. 4 indicate the average val- seedling root system. ROB does not convey infor-
ues of T for lodgepole pine seedlings belonging to mation about shape, i.e., the pattern of subdivision
the serai, climax, and persistent groups {d,^„^,^,,, is of the root system. Instead, it incorporates, and in-
twice the value of mean NLR from Table 3 divided evitably confounds, two aspects of development:
by a taproot length of 160 mm; lu„e,ai is mean LRL total root length (the sum ofthe lengths ofall roots)
from Table 3). Bearing in mind the simplifying as- and average root caliper. Consider a lodgepole pine
sumptions associated with the calculation of T, the seedling in which all lateral roots are identical in
absorptive capacities ofseedlings from climax pop- length, each lateral root possesses the same number
ulations only slightly exceed those of serai seed- of attached tertiary roots, and all tertiary roots are
lings, though, given their longer lateral roots, the equal in length. Under these conditions, the follow-
former may have access to water and mineral nu- ing equation expresses the total root length in mm:
trients from larger volumes of soil than either serai
or persistent seedlings. Seedlings derived from per- ^ ~
sistent populations possess distinctively lower ab- ^taproot ^~ lawra/.jlateral
sorptive capacities, due mainly to the production of
fewer lateral roots (mean NLR = 27.20, 29.66, and
26.68 for serai, climax, and persistent families, re- where l,„proot is taproot length; is the number
spectively). of lateral roots attached to the taproot; is the
length of a lateral root; d,^,,.,^^,,,, is a density, the
A seedling's resource acquisition potential. Dis- average number of tertiary roots that arise per unit
criminant function analysis identified a second, less length of lateral root; and mot is the length of
explanatory, dimension that highlights the variation a tertiary root. The quantity is the sum
'?/,,,,,,„/v/f„/,r,.<>r
in the ability of Blue Mountain lodgepole pine of the lengths of all lateral roots. The product
seedlings to obtain essential resources from their niateralsl,aprootd,ert,ar. ..Jtertiary root repreSCUtS thC SUm
surroundings. It utilizes information about overall of the lengths of all tertiary roots. The results of
seedling size and the sizes of individual organs in- this study supplied constant values for I,„j,,„„r (160
volved in the interception/uptake of light, water, mm), (55.8, twice the mean of 34 family
and mineral nutrients to separate the three succes- mmemans^2o/a,n,,,-rNt/,LR from Table 3), and d,^„i^,,,. (0.29
sional classes in multivariate space. In accord with '). The approximate median of the range of
the interpretation ofthe second dimension as a size mature tertiary root lengths of forest trees reported
vector, all five morphological attributes with high by Sutton and Tinus (1983) provided a reasonable
loadings (>0.50) on the second discriminant func- value for Z,^^,,,,,., root (2-0 mm). Figure 5 shows that
tion {SHL, NPL, BLL, LRL, and ROB) possess stan- the total length of a lodgepole pine seedling root
dardized discriminant function coefficients, corre- system {R) increases in direct proportion to lateral