Components of relative growth rate and their interrelations in 59 temperate plant species

New Phytol. (1997), 135, 395-417

Components of relative growth rate andtheir interrelations in 59 temperate plantspecies

BY RODERICK HUNT* AND J. H. C. CORNELISSEN

The NERC Unit of Comparative Plant Ecology, Department of Animal and PlantSciences, The University of Sheffield, Sheffield SIO 2TN, UK

{Received 22 April 1996; accepted 9 October 1996)

SUMMARY

Three groups of species (21 herbaceous monocotyledons, 22 herbaceous dicotyledons and 16 woody dicotyledons),including representatives of a wide range of natural habitats and life forms in inland Britain, were grown in theseedling phase in a resource-rich controlled environment and assessed over a 14-day period (21 d in the case ofwoody species). Mean values of relative growth rate (RGR), unit leaf rate (ULR), leaf area ratio (LAR), leaf weightfraction (LWT), specific leaf area (SLA), and the root-shoot allometric coefficient were derived.

In herbaceous species, the grand mean RGR was 0-20 d~^, comparable to values previously recorded. For woodyspecies, the mean was 0-09 d~ . An existing assumption linking high RGR to high allocation to photosyntheticbiomass was upheld by comparisons made between groups. Within groups, however, no pattern of this kind couldbe demonstrated.

When photosynthetically active radiation was increased from 125 to 250 /imol m~ s" , ULR was increased almostpro rata. The parallel response in RGR was only slight, being offset by considerable reductions in LAR. The apparentmean quantum yield for photosynthesis in herbaceous species (whole-plant d. wt basis) was 0-60 gmol"^.

There was no significant dependence of RGR on ULR in any of the three groups of species, although the absolutemagnitude of ULR declined in the order: herbaceous monocotyledons > herbaceous dicotyledons > woodydicotyledons. In all three groups, RGR was strongly dependent upon LAR but no differences emerged in absolutescale of LAR. The absolute scale of mean LWF decreased from herbaceous to woody species, but the dependenceof LAR on LWF strengthened. Groups showed no systematic differences in magnitude of SLA, but the correlation ofLAR with SLA was strong throughout.

Multiple regression showed that the leading determinants of RGR were ULR and SLA in herbaceous species andLWF in woody species. Principal components analyses (PCA) on each of the three groups explained at least 77%of variation and agreed closely with an optimal (non-hierarchical) classification. Only six cluster ' types' wererecognized out of the 16 theoretically possible combinations of 'high' or 'low' values of the four growthparameters. Strong evidence of evolutionary trade-offs emerged, most strikingly in that high RGR was never seenin combination with low SLA. The morphological/physiological types identified by an all-groups PCA separatedwoody from the herbaceous species, but dicotyledons were almost congruent with the monocotyledons.

The non-growth-analytical attributes most strongly correlated with mean RGR were percentage yield at a lowlevel of mineral nutrients, leaf nitrogen concentration, and seed weight. It was concluded that mean RGR plays acentral role in the identification of pathways of evolutionary specialization in herbaceous species.

Key words: Leaf area ratio, leaf weight fraction, relative growth rate, specific leaf area, unit leaf rate.

INTRODUCTION

The innate differences between plant species in theirmaximum rate of dry matter production underresource-rich, undisturbed conditions are funda-mentally important to plant ecology. Relative grow^thrate (RGR, the rate of dry matter production per unit

* To whom correspondence should be addressed.E-mail: [email protected]

of dry matter) is the most useful single comparator ofinnate growth potential because it is independent ofscale of organism (Evans, 1972; Hunt, 1990). Highand low maximum RGR have been shown to begeneral properties of species in resource-rich andresource-poor natural environments respectively(Grime, 1965; Parsons, 1968; Grime & Hunt, 1975)and maximum RGR is also one of the fundamentalaxes of plant specialization within the so-called C-S-

396 jR. Hunt andj. H. C. Cornelissen

R scheme of plant strategies or functional types screening studies by Blackman & Wilson (1951), Van(Grime, 1974, 1979). Andel & Biere (1990), Poorter & Remkes (1990),

Several screening programmes for measuring Poorter & Lambers (1991), Maranon & Grubbmaximum RGR in native plant species have been (1993) and Grubb et al. (1996).conducted. These may arbitrarily be defined as In addition, other screening has used subdivisionsstudies involving simple measurements (minimally of RGR which lie beyond these classical limits. These,shoot and root biomass), with at least two successive although 'growth-analytical' in nature (i.e. explana-harvests, on selections of at least 10 different species tory, holistic and integrative), address deeper levels(or populations, genotypes, genets or ramets), and of plant form and function than can be approachedwith plants grown in the juvenile phase without simply by gathering data on weights and areas,competition under resource-rich conditions. Among these are the quantitative treatments of theExamples include the studies by Blackman & Wilson nitrogen, or the carbon and nitrogen, economy of the(1951, 10 species). Grime & Hunt (1975, 132 plant by Hirose (1988), Poorter, Remkes & Lambersspecies), Fenner (1978, 12 species), Elias & (1990), Muller & Garnier (1990), Niemann et al.Chadwick (1979, 40 species and genotypes), Burdon (1992), Poorter & Bergkotte (1992), Poorter && Harper (1980, 48 genets). Hunt (1984, 18 ramets), Farquhar (1994), Vanaerendonk & Poorter (1994)Fenner (1983, 23 species), Shipley & Keddy (1988, and Garnier, Gobin & Poorter (1995).28 species), Dijkstra & Lambers (1989, 10 popu- Work involving daily, or even more frequent,lations and genotypes), Shipley et al. (1989, 23 harvesting (e.g. Hunt & Lloyd, 1987; Hunt, 1990)species), Shipley & Peters (1990, 68 species). Van and employing fitted, splined, growth curvesAndel & Biere (1989, 12 species), Poorter & (Parsons & Hunt, 1981; Hunt, 1982) has shown thatBergkotte (1992, 24 species), Maranon & Grubb the most valid developmental state in which to assess(1993, 27 species), Stockey& Hunt (1994, 14 species) maximum RGR is that of the fully independent,and Grubb etal. (1996, 11 species). juvenile seedling. Studies typically show a short

Such screening has shown that variation between period of hyper-exponential growth in the few daysspecies in maximum RGR is considerable. For following germination, followed by a short-livedexample, instantaneous RGR in seedlings of annual plateau in RGR, then a long decline in RGR towards theweeds can exceed 0-5 d ^ in the very short term plant's maturity. In large-seeded or very fast-(Hunt & Lloyd, 1987), implying a doubling time for growing species the hyper-exponential phase can bedry matter of not more than 1-4 d. By contrast, the reduced in duration.growth potential of young rosettes of the dune A practical feature of these deeper levels of growthwintergreen Pyrola rotundifolia ssp. maritima under analysis is that their scope is often relatively modesta very similar regime is only 0'007 d ^ (Hunt & in terms of number of taxa studied because it isHope-Simpson, 1990), with a doubling over 99 d. seldom, if ever, feasible to sustain high precision,Other studies of variations in RGR between species high realism and high generality simultaneously in a(e.g. Grime & Hunt, 1975) show that the frequency single study (Harper, 1982; Hunt & Doyle, 1984).distribution for mean RGR is approximately normal. Though the present study aimed to provide new orEven leaving aside the few exceptional cases, there is improved data on the growth of a wide range ofa 10-fold range of variation in mean RGR between British native species in the early seedling phase, it,species. At other levels of organization (e.g. Elias & too, had to strike the most advantageous compro-Cbadwick, 1979; Burdon & Harper, 1980; Dijkstra mise. The study was thus designed both to increase& Lambers, 1989) there is often a twofold range in the precision of the existing large-scale screeningmean RGR within the population, the genet or the programmes (such as that of Grime & Hunt, 1975)ramet (Hunt, 1984). by including a fuller selection than hitherto of the

As a gross measure of plant growth rate, RGR has more classical components of growth, and to increasebeen the single most useful index available to the realism of the screening (in the light of theecologists. However, following Briggs, Kidd & West information given above on the timing of maximum(1920), RGR has often been treated as the product of RGR in the seedling phase).a physiological component, the unit leaf rate (ULR. Because the period of 2 wk (herbaceous species) orthe rate of dry matter production per unit area of 3 wk (woody species) which immediately follows theleaf) and a morphological or allocational component, seedlings' first significant photosynthetic productionthe leaf area ratio (LAR, the quotient of total leaf area has been found to include the maximum instan-and total d. wt). Some of the screening for maximum taneous RGR (for any given environment) in virtuallyRGR has included information on these additional every instance studied, measurements of mean RGRparameters of growth, or on their further sub- which span this period are likely to be useful fordivisions, such as leaf weight fraction (LWF, the comparative purposes, even though they avoid thequotient of leaf dry weight and total d. wt) and frequent harvesting and subsequent curve-fittingspecific leaf area (SLA, the quotient of total leaf area which are necessary for the more precise derivationand leaf d. wt). Analyses of this nature occur in of a series of instantaneous rates. In consequence.

Components of RGR 397

this study was able to uphold the generality achieved C-S-R scheme defined by Grime (1974, 1977, 1979).by the larger screening programmes by including a Hunt et al. (1991 a) and Hunt et al. (19916) gavecomprehensive selection of herbaceous and woody concise introductions to the basis of this scheme. Inspecies representing a wide range of British habitats essence, three main functional types are defined as:and plant functional types. In particular, sufficient competitors (C), associated in the established phasenumbers of species were included to enable exam- of growth with low levels of environmental stressination of the possibly distinct roles of the various (pre-growth) and environmental disturbance (post-components of RGR in different plant groups. A growth); stress tolerators (S), associated with highrelated study by Cornelissen, Castro Diez & Hunt levels of stress and low levels of disturbance; and(1996) involves fewer growth parameters, but is ruderals (R), associated with low levels of stress andwholly devoted to 80 woody species. high levels of disturbance. There are four inter-

Obviously, the parameter values obtained under mediate, or secondary, types: CR, SR, SC and CSR.the resource-rich environment described here are Between each of the primary and secondary types,not necessarily the true maxima attainable across all there are a further 12 tertiary ty^pes. Figure \acombinations of conditions. Nor are they the values illustrates the location of all 19 plant types within C-that might be expected to obtain under any particular S-R space. For all 43 herbaceous species used in thisfield situation, or even the values that might be investigation, classifications at the level of tertiaryexpected at a later point in the same life cycle in the functional ty pes have been given by Grime et al.same environment (e.g. Swanborough & Westoby, (1988). The distribution of these species within C-S-1996), even though the ranking of perennial species R space is shown in Figure \b. For the purposes ofwith respect to RGR can largely be preserved into a this study, the partially^ w oody species Dryas octo-second growing season (Hunt & Lloyd, 1987). They petala, Helianthemum nummularium and Thymusare, however, useful single comparators of per- polytrichus w-ere treated as being 'herbaceous'. Theformance of a wide range of species. truly^ woody species chosen included several shrubs

The method of quantitative analysis was standard and trees, and one climber,throughout and a new, integrated package of analy seswas devised for this study. This is specified in an „ . , ,. ,. , . -,11 , ^ r 1 hxpemnental procedureAppendix and is available on demand irom theauthors. For each of the 43 herbaceous species, an amount of

Specifically, the present study set out to answer locally collected or commercial seed sufficient to givethe following questions, both for the selection of c. 40 germinated seedlings was placed onto poly-species as a whole and for the major taxonomic and propydene beads or filter paper on full 'Rorison'life-form types which it represents: (1) What is the nutrient solution (Booth et al., 1993) in closed,mean and range of mean RGR in a temperate fiora? (2) transparent containers, under the standard ISPHow is growth partitioned between root and shoot, growth-room environment. These conditions com-and what variations in water content are involved ? prised 250+ 15 fimol m" s~ of PAR at plant height,

(3) What is the interrelation of photosynthetically provided by 400 W metal-halide lamps and 100 Wactive radiation (PAR) and ULR in determining RGR ? tungsten-filament lamps, in the ratio 2:1. In a pilot(4) What are the relative contributions of the experiment, 29 of the species were also grown atmorphological and physiological components of 125 ±10/(mol m'^ s" (under similar but moremean RGR? (5) What is the combined influence of spaced illumination), but the measurements madethese components on mean RGR ? (6) Are there any were more limited and work at this level of par wasother important correlates of mean RGR ? subsequently discontinued. The red/far red quotient

(at 660/730 nm wavelength) was c. 1-4 throughout.Daylength was 14 h, day temperature was 20-22 °C

MATERIALS AND METHODS ^^^ ^^^^ temperature was 15-17 °C. Humidity w as

not controlled, but even outside the closed germ-Selection of species ination boxes it was high on account of the wateringThe investigation was conducted as a part of a large regime used elsewhere in the growth room,multidisciplinary project, the Unit of Comparative Twenty-four 200 ml pots were filled wdth water-Plant Ecology's Integrated Screening Programme washed silica sand and saturated with full-strength(known as the ISP, Grime, 1985; Hendry & Grime, Rorison solution (a balanced solution in which the1993). major nutrients N, P, K were respectively supplied

The 59 species (Table 1, with nomenclature at 56, 31 and 78 mg T^ at pH 4-5). Pots were placedfollowing Stace, 1991) were chosen to represent a in undrained seed trays filled to a depth of 5 mmwide range of natural habitats and life forms in with deionized water in the standard ISP environ-inland Britain (Grime, Hodgson & Hunt, 1988). At ment. As soon as possible after germination, twoleast for the herbaceous species, this also implies a seedlings were sown in each of eight of the 24 potswide representation of functional types within the and one seedling in each of the remaining 16 pots.

398 R. Hunt and J. H. C. Cornelissen

CR

R/CR

R

C

C/CR C/SC

C/CSR

CR/CSR SC/CSR

CSR

R/CSR S/CSR

SR/CSR

R/SR S/SR

SR

(a) Functional types

SC

S/SC

S

Can,Eh,Ud

Asy,Ga,Ha,Zm

Dp,Lp,

Ast,Cal Ac,

Cfo

Cc,Pa

-

(b) Location of

Ae,Dg Bp

Pt Be,Ov

Fr,HI,Km,PI Df,Ev

Cs,Hp,Lc,Po

Ao,Ra Bm,Cfl,Cro,Do

- Fo,Hn,Lh,Tp

At,Cri

43 herbaceous species

Figure 1. {a) The spatial interrelation of C-S-R functional types; the two environmental dimensions are:increasing stress, from top left to bottom right; increasing disturbance, from top right to bottom left, {b) TheC-S-R locations of the 43 herbaceous monocotyledons and dicotyledons used in the present study (for fullnames see Table 1).

On alternate days all pots were top-watered with fullnutrient solution at 50 ml per pot (on other days potswere bottom-watered as required with deionizedwater to maintain 5 mm depth of liquid in the seedtray).

At 7 d from planting out, all seedlings wereharvested from the eight-pot subset. Root, stem andleaves were separated for d. wt determinations andtotal leaf area per plant was measured. At 21 d fromplanting-out, the 16-pot subset was harvested simi-larly, with the additional measurement of totalsaturated f. wt per plant. Two independent, spacedsamples of 16 plants were thus available for growthanah^sis. Further information on the standardmethods used for seed collection, cleaning andstorage, for seed germination, the aerial and rootingenvironments, and harvesting methods can beobtained from relevant chapters in the volume byHendr>' & Grime (1993).

The procedure followed for the 16 woody specieswas the same as for the 43 herbaceous ones, exceptfor the following. To evaluate effects of seed reserveson seedling growth, oven-dry weights of seedsamples were made shortly after collecting. Largerpots, with correspondingly larger volumes of addedsolution were used for the woody species (70 ml ofsolution for 300 ml pots, 90 ml of solution for 400 mlpots). The PAR source was the same one used for the125/Amolm^^ treatment of the herbaceous species,but it supplied 135 + 10 /^mol m~^ s~ at plant height(on account of the sand surface in the larger potsbeing closer to the light source). Since we found thatwoody species can vary up to 10-fold in the timeinterval between germination and the onset of thephotosynthetically active phase (some hypogealspecies, for instance, extend a taproot long before theemergence of any shoot material), we adopted a strictontogenetic criterion (ratker than a temporal one)with regard to the timing of the first harvest. Thiswas performed when the modal group of seedlingswas that displaying one opened (but not necessarily

fully expanded) true leaf or leaf pair (not countingthe cotyledons). At this stage total leaf area per plantwas determined (including leafy cotyledons),together with d. wt of roots, stems (including anybranches and petioles), leaves, and seed remains (ifstill attached), and numbers of opened leaves andcotyledons; additionally, the shortest distance fromthe stem base to the top leaves (crown height) wasmeasured. The remaining seedlings were then grownon for a further 21 d, at which stage the sameparameters were again determined, but with sep-aration of leafy cotyledons and leaves, and with theinclusion of fresh weights of leaves and cotyledons.

Two of the herbaceous species, Poa trivialis andDigitalis purpurea, were grown under all threevariants of the experimental procedure, i.e. over 14 d

-2 „-!at 125 /^mol m ^ s \ over 21 d at 135 /^mol m " sand over 14 d at /tmol m"^ s~\ in order to makedirect comparisons.

Primary quantitative analysis

Theory. Growth analyses were conducted accordingto a purely ' classical' approach, involving harvest-interval calculations, rather than the 'functional' or'dynamic' approach, involving the use of fittedcurves (Evans, 1972; Causton & Venus, 1981, Hunt,1982, 1990), or a 'combined' approach involvingcurves fitted to classical values (Poorter, 1989). TheRGR in whole plant d. wt was derived across thesingle harvest interval, and its components ULR 'netassimilation rate' of some authors), (SLA), and (LWF)

were similarly derived from the same dataset, thoughthey were computed independently of one anotherand not obtained as mathematical subdivisions ofRGR. Instantaneously, these four terms are definedand related in the following way:

w


oog

(U

aaT3OO

o

aa

ooca

X,

o£

CO

Oi

03

CM

K

153

g

o£

LOCM

<

cn

ear

£-a

r-

CJ

'Clo

me

cCD

effi

cii

oo

>

0CO

imen

"•«—•

1C

c£

3 £

s(A

£

£

•c c

I

£

OK

p ^ p p p p p p ^ p ^ T ^ ^ ^ Ô O O C 5 O O O O O O O O O C D 6 O O O O C 6 O O O O 6 O C 5 6 O O O O O O

00

'—* O^ O 1^ H c*o O CO ON T H

oooc-oooooo^" O —' O "^ O ^^ O '-H O '"' O f O '" O '-H O " O ''"*p p o o p p p p p p p p p p p p p p p p p6 6 5 5 5 5 5

p

00 O (N

oo oop oO CO O O

00 p q pCO CO CO O

O 00 00r-H LTj 00 CNJ

00 O 00 O

CO CO 6 CO

VO tN • ^ T-H

o o o op p p po o o o

O O O O O O O O O O O O O O ' - Op o p p p p p p p p p p p o p p aoooooôo^cocôoôo^co^ -

,—T - H

(N

IIK

cnCo-a>.oCJoo

eoo

X5u<u

T-H 00( N COT-H T-H

ra po o

* T ^

1

ti

o

oroT - H

O

^^* P>i

<^

a

K

• ^

[-Ô<? CS

S C

a

o

S

1K

o c i > c o o o o c o c o o o c > c o o o _ 6 o _ g

r^ l o 00 T^

^ o ^ op p p p6 6 o 6

o o " rO - ^ T^ I^

CJ o r o^ p ^ pO CO O CO

vO (N T-< ro-* T- O O)O O '- Op p p pO CO O CO

fNj \ ^ 00 r ^ro ON ro LTj00 O ^ O

6 CO o CO

Ss

<3S

Ss*

Icq cq

•I

I

(3

!3

o

.ts

s

<3

C3

tse


H

aa

>1

-t3OO

o

o

D.a

O

o

oCNI

ai

ET3

S

Pi<P-,

C3

H

Ec^ S

< 83

.2- r j CO

C« S P

rT • "d

E

o o o o ( i > c ± > o o

OO

O O O O O O O O O O O pO O O O O

CS

d00 C^ '••O C^ **OCO C> O ci) O

ovOC3N

T-< 0 0t o CXJ

oCO

pON

CN

<z>

Ti" LOC^ CN

r o o 00

' CN

o

E

tJC

CN

, CN LT)

OCN ' C N ' C O " — ' ^-' ' C O " — ' • » . - r - i " — ' C N

p p ^ p ^ po o o o o o o o

oO0

CN ~—' -^H

p p p p ; p ^ p p p p p p ; p p p p p p p p pO O O O O O O O O C O O O O O O O O O O O O O O O O O O O O O O O O O

!

I

9 TJ

E "o

O O O Op p p p p p p po o o o o o o o

^ O ^ O O Op p p p p p

O O O O O

p p p o p p p p p p0 0 0 < D C J O C I C J C > O

O - ^ ^ O ^ - ' O - ' - ^ O — O ^ — ' O ^ ^ O ' - ' O ^ - ' O O O

p p p p p p p p p p p p o p o p o p oO O O O O O O O C ) O C > O O O C > 0 0 0

o O O O O O t0 0 0 0 0 0 0 0 0 ^ 0C N O ' T J O f N O ' ^ ^ OO O C J O t O O O O

n c

— <u . S< 83

0 0 • ^ "O r~• ^ r o OJ r^)r o ^ o J*c c o o •

c^ inLT) OLO r l -oo p<t> o

O CO

'7- pp ; po o (5

p6 o

' '^ CO ON LOO p CO p

o <o o CO

LOLO

0 0

A O

o00CO

CNpo^

ca'

d

00 ^CjN COO r-

ad

CS

d

C6 bp

D 3

t -^^ O LOri- ^ 00 ^O O O Oo o o oCO CO O O

CO O N CN T-<' ^ GO O 0000 O CN OT- p CN pO O O CO

00 O

o op pO CO

C3N o00 00

^ ^ CO ^ H0 0 CO CO i.-<0 0 0 0p p p po <o CO cii

O ON ON 00r-H o NO rC^ •^H CN O•r-< P T- pCO CO CO CO

CB

d

CB

d

O vC LO ^H t ^ COLO ^H t ^ CO O - ^O O O O - ^ OP p p p p pO O CO C5 O O

'— ON C lO O 00CN O CN LO •-— O• ^ O " ^ ^-" O T-H•7* P CN p CN pO O CO O O O

^ ONLO OO Op oCO CO

O Ol > CJNo o^ oo 6

CN r-<ON CNO Op pO CO

00 CMo op pCO CO

CO oO

r o<N OCO CO

opo^

ca]d

CS

dca

d

<s > m

O OCO CO

c ciCS caU 1)

o-0

§ I

•S'—i

Ia

t

a• 3 (3

CS

K

a

-CJ

o

Sa

O

s

^

a

4

aK


r o " ^ o CO 00• ^ li-i • * u-i O OC30 O OO O <> O

o o o dj o ot ^ p c»o cpO O O O o o o c 5 o o o c 5 o o o 6 o

o ^'-oc'ooocbcioooci

CB

dca

d Cca

dcad

cBd

cad

cad d

cad

cad

r^ CO

'O4

, 00

p p p p pO O O O O C J O O O O O O

p p p p p p p p p p ^ p p pO O O O O O O O O O O C i O O O C i O C i O o O O O

p p p p pO O O O O O O O O O

0000

_ o

O O O O O O Op p p p p p p p p p p pO O O O c O O O O O o O O

^ r c i ^ O t N < N T i o ^ r iO O O O O O O O ^ - O O O O O O O O O — ' OP p p p p p p p p o p p p p p p o p p p5 6

O O O O O O O O O O O Op p p p p p p p p p p p

5dO Opp

tooTp

ooooooooo

p p p p p p p 7 7 p p p pO C > C > C D O o O 0 C ) ( 0 C > O O o O O O O O o O o 6 o O O O C C 5 O O

cad

IS

dcad

cad

cad

cad

cad

cad

cad

cad

ca cad

CB

dcad

o o o o o o o o o o o op p p p p p p p p p p p ca

dca

dca

dca

dca

dca

dca

dca ca

dca

dCB CB

dca

dca

dca(—

ca

dca

d

C ! O O O O O O O C > O O Oca

dCB

dCB

dcB

dcB

dca

dca

dca

dca

dCB

dca

dCB

dca

dca

dCB

dCB

d

Sse

f lan

Si53

S

^ Thyi

QU

b

oT3

O O o ' ^

cB ca ' " • ^ ' ^

.5t/3

sK

s

. 3

I

.as

•2S

•a,

2!5

K

•ft,

SK

g

5sS

0;

a

^

3

o oC Era ca

>cd

CO

G

402 R. Hunt and jf. H. C. Cornelissen

where t is time, W is total d. wt per plant, L^ is totalleaf area per plant and L^ is total leaf weight perplant. The product of SLA and LWE, defined as LJWand known as leaf area ratio (LAR), was also derived,

Additionally, for the herbaceous species alone, theratio between total fresh weight per plant and totald. wt was calculated for the second (final) harvest,and the allometric coefficient for root/shoot de-velopment was determined. In woody^ species withseed storage, plant d. wt was first expressed withoutincluding seed remains and then, additionally (datanot show^n), with the mean loss in seed weight acrossthe harvest interval being added to all estimates ofplant d. wt at the first harvest.

Computation. For the principal analyses, three kindsof growth parameter were of interest. In genericterms these may be defined as l / F ( d F / d O ,1/Z (d Y/dt) and Z / Y, where Y and Z are differentplant variables and t is time. In the simplest case,with Y representing total d. wt per plant and Zrepresenting total leaf area per plant, the threeparameters RGR, ULR and LAR are obtainable. In othercases, the variables Z and Y can be given differentidentities in order to evaluate other ratios of thegeneral form Z / Y.

Fach growth parameter was estimated as a meanvalue, with variance, across the harvest interval7-21 d from germination (or the interval 0-21 dfrom leaf opening in the case of the 16 woody^species). The statistical methods followed those ofVenus & Causton (1979) and Causton & Venus(1981), whereby mean values of the various para-meters were obtained without the use of fittedfunctions and without the 'pairing' of plants acrossthe harvest interval. Statistical details on this methodare provided in an Appendix and a computerprogram written in GENSTAT 5 (Payne et al, 1987),which performs all of the necessary calculations, isavailable upon request from the authors.

The allometric coefficient for root/shoot devel-opment was obtained in the usual manner as thelinear regression coefficient of log^ root d. wt on log,shoot d. wt.

Secondary quantitative analysis

Further analyses, m the form of standard corre-lations, regressions, ordinations and classifications,were performed upon the results or the primaryanalysis m order to test subsidiary hypotheses. 1 hesemethods are specified where they are nrst mtro-A A --pu ^ ^- r 11 ^ *.- 4.- 1 1 f 11

duced. The presentation of all statistical work followsthe editorial code of the Royal Society (1995).

125 /imol m"^ s" and for seven parameters measuredat PAR 250/i^mol m"^ s~\ All values bear standarderrors of estimate. The 59 plant species are displayedin three approximately equal groups, comprising 21herbaceous monocotyledons, 22 herbaceousdicotydedons and 16 woody dicotyledons. Meanvalues of all parameters and their standard errors aredisplayed at the end of each group.

The herbaceous species included in this studyprovided for only a limited number of possiblephylogenetic contrasts at the level of the family.Nonetheless, for every growth parameter, paired t-tests were performed for grasses vs. non-grasses andfor Compositae vs. non-Comipositae. No comparisonproved to be significant at P < 0-05.

p^r Poa trivialis and Digitalis purpurea, the resultsobtained over a period of 14 d at 125 and250/^mol m"'s"^ appear in Table \a,h. In theadditional tests performed on these species over 21 d^t 135 [imol ra~"' s'^, the values of mean RGR for these^^^^ species were 0-201+0-005 d" and0-173+ 0-006 d ^ respectively

Within each group of species, each growth par-ameter has been correlated against all the others(Table 2). This has been done for exploratorypurposes only, without transformation and withoutregard to the statistical independence (or otherwise)

different variables.

D I S C U S S I O N

temperate flora.

Table 1 presents the primary results for the entirestudy. For herbaceous species, mean values are givenfor three parameters measured at PAR

of mean RGR in this

herbaceous species included in the presentstudy, the grand mean RGR across all species(measured at PAR 250 fimol m'^ s ) is 0-20 d"' and^j^^ g^ g around 0-009 d~\ The mean result may becompared with similar means derived from thelarger of the existing screening programmes, forexample with the mean value 0-17 d'^ for 38 of thegame herbaceous species assessed by Grime & Hunt(j 975) ^ ^ p ^ R of about 130 fimol m"' s~ over theperiod 14-35 d after germination, the mean value0-21 d~' reported by Shipley & Peters (1990) for 68herbaceous species of Canadian wetlands grown at390 pimolmT^ s~^ over the period 10—30 d aftergermination, and the mean value 0-22 d reportedby Maranon & Grubb (1993) for 27 Mediterraneanannuals grown at 200-400 fimol m s over theperiod 0—21 d after seedling emergence.

Clearly, for the herbaceous species it was ad-vantageous to growth to increase PAR from 125 to250 /imoi m s in the early stages of the presentstudy. This change of policy caused an increase inmean RGR from 0-17 d ^ to 0-20 d- \ All but twomonocotyledons and three dicotyledons individuallyregistered higher mean RGR at PAR 250 /^mol m"^ s" ^(Fig. 2a, c). With a mean SE of around 0-01 d"


o

**LO LO '—'

o rj "^\O O ^

d> o <

t^ toO ^-1 CN!

6 o 6

OLO

** ** *

^ o

00LOsO61

o"61

sO00LO

61

00

oCO61

OOCN

61

(N

*TV*

o* *TT ^

T-i LO p o vp pO O O O CD c5

O ^LO LOp CO6 6

fN Oo 00

0 0

**

-1 v- cs6 6

I IO . 2

I I

~S2

0)

CO

H

** * * *

00 ON O CO ON LO 00Th vO O O 00 Th COCO ^H t ^ LO O LO "+

L O

CO

O O O O O O OI I I I I

L O

o =+.;

O

* oTj- -H O LOO (M O CMT-< ^ ^ LO I

6 6 6 ft

***LOoON

Csl

6

ON O~_ LO^ "^6 K

t ^ T-H Th LO

CN ' ^ p ' ^

6 6 6 6

oLO

*** *CN ON 00 r^ OLO r o 00 rsO CO t o <N O

6 6 6 6 6 I O

* ** * * * *# * * * #

L O

ooooooo-Î i I I i <

*00

** L O

r " " " oT-HOÔNtÔONl^

tOp-^CNJCO^HT^tO I

6 6 6 6 6 6 6 6 «1 I I

CO

ON00CO

6

***CO

6

*00 CO O

^—' " ^

6 6

**so ON O lto •^H

6 6co

s3O<uo03

oLO

t o

^ o o o

J -:!

t o

-2

I ou

3O

**CO

LO O

6

*t o LO

o o o

oONCO

6

oLO

P6

LOCO

oLO

LOCO

ksto

OK

O OT3

LO

)

o< > :=i

L O

03 O <

d ^

sOCOCO

6 (

***o

B 6

oLO1

>

***

00

00

6

***vO00

r6

**ONCO\O

d 6

otoi

*-N|- CN ^CO ON sOCO LO p

6 6 6

COCOCO

oLO

***

CO

***00CO

**

6 6 6 6

LOCSl

O

L O

LOOJT 11

O

c/3

O

1)

wv ^v ^v ^v i-v 'o^ «v ^^ "

c c c c c c d c o

t o

C G C C C C C C C

toCO

o

LO

to

s_O

<

lOCO

1

K0a

LOCO

1

J

toCO

(

a<

to

CO

i

J

lOro—<

SU

4_j LO

JJ

oc

op6V***

p6V

LOp6V

1)uc

O

404 R. Hunt and Jf. H. C. Cornelissen

(a) Mean RGR-250 {r= 0-731*) {b) Mean ULR-250 (r= 0-700***)0-30

0-25

0-20

0-15

0-10

0-02

0-01

0-00

. * .

0-10 0-15 0-20 0-25 0-30

Mean RGR-125 (d"'')

(c) Mean RGR-250 (r= 0-477*)

0-00 0-01 0-02

Mean ULR-125 (mg ^

(cf) Mean ULR-250 {r= 0-404)

u-ou

0-25

0-20

0-15

0-10

•

«

• ^

*

Jf

•

•

• , - '

•

1 1 1

0-02

0-01

0-000-10 0-15 0-20 0-25 0-30

Mean RGR-125 (d"'')

0-00 0-01 0-02

Mean ULR-125 (mg mm^ d~ )

Figure 2. Scatter diagrams of growth parameters obtained at two diflferent levels of PAR (125 or250 fimol m"^ s"- ); parts (a) and (c) refer to mean relative growth rate and parts (b) and (d) refer to mean unitleaf rate; parts (a) and (b) refer to the herbaceous monocotyledons and parts (c) and (d) refer to the herbaceousdicotvledons.

across both levels of PAR, the increase in mean RGRwas significant at P < 0-05. Within the herbaceousmonocotydedons, the absolute range of the values ofmean RGR (the maximum minus the minimum) wasincreased from 0-10 to 0-17 d~ by increasing PARfrom 125 to 250/fmol m^" s^ . In herbaceousdicotyledons, the parallel increase was from 0-15 to0-21 d"^. Both increases in range were due to a rise inthe maximum of the distribution which was greaterthan the rise in its minimum. The results obtained atthe lower level of PAR achieved little advance on theposition reached by Grime & Hunt (1975). Work atthe higher of the two present levels of PAR thuspreserves the more effective contact with othermodern studies and provides the more discrimi-nating basis for uncovering inherent differencesbetween species with respect to mean RGR, thoughPAR at 250 fimol m"^ s"' is still not a high level ofphoton irradiance.

Strong correlations exist between results obtainedat the two levels of PAR used in the present study(Table 2 a, b); this is true both for the mono-cotyledonous and for the dicotyledonous herbaceousspecies. Results from the earlier study by Grime &Hunt (1975) are also highly correlated with thepresent data, the correlation coefficients with thepresent results for PAR 125 and 250/tmol m"^ s~

being, respectively, 0-570 and 0-645 in the mono-cotyledons, and 0-554 and 0-817 in the dicotydedons(all P < 0-001).

For Poa trivialis and Digitalis purpurea, eachgrown under three different regimes, the variation inRGR with regime was similar in both species. Relativeto the result obtained at 125 fimol m~^ s" over aperiod of 14 d, mean RGR was roughly 10% lower at135 jLimol m^^ s~ over a period of 21 d, the increasein period being more effective in reducing mean RGRthan the increase in PAR was in enhancing it, androughly 30 % higher over 14 d at 250 /imol m' ^ s~\ adirect enhancement by increased PAR.

The PAR level of 135/imol m~'s"^ is probablywell below the light saturation point of saplings ofsome of the woody species included in the presentstudy. However, since most British woody speciesregenerate under a woodland or grassland canopy,14 h of daily exposure to that level of PAR probablyprovides a daily radiation total which is higher thanthat commonly experienced as seedlings in the field.The grand mean RGR for woody species was littlemore than half that obtained for herbaceous speciesat 125/imol m~^ s"\ As the increase in period ofmeasurement from 14 to 21 d is less likely (than inthe case of herbaceous species) to have resulted in alowered estimate of mean RGR, this finding supports


earlier observations of generally lower RGR in woody shrubs and trees, they emerge similarly ranked inseedlingsasa whole (Jarvis&Jarvis, 1964; Grime & inverse order of mean RGR. High mean RGR, theHunt, 1975). reasoning went, could therefore be associated with

high allocation to photosynthetic tissue and low

How is growth partitioned between root and shoot "^^^"^ ^^^ ""''* ^^^h allocation to roots. A typicaland what variations in water content are involved? ^^^come of this process would be that fast-growing,

early successional species are nutrient-limited andThe simplest vehicle for a test of root-shoot that slow-growing, late-successional species arepartitioning is the allometric coefficient. This is an light-limited.index of the balance of growth between root and If this were a general phenomenon among plantshoot components of the plant integrated over a species, it could be expected to manifest itself, atperiod of time. Effectively, it is a ratio between root least in part, in a downwards drift in the allometricand shoot mean relative growth rates. As calculated coefficient with increasing mean RGR. Shipley &here, this dimensionless quotient takes a value of Peters (1990) went to some lengths to disprove thisunity during periods of equally balanced growth, assumption, drawing upon a dataset for mean RGRwhereas in 'shooty' growth its value is < 1 and in involving 68 Canadian wetland species. They es-' rooty' growth it is > 1. At the next level of analysis, tablished that a significant positive correlation (r =Korner (1994) justifiably held that biomass frac- 0-53, P < 0-001) existed between mean RGR and thetionation in plants should ideally be studied in at allometric coefficient (defined and calculated as here)least three or four compartments, allowing for the and therefore held that Tilman's (1988) assumptionseparation of both above-and below-ground material was disproved. However, the Shipley & Petersinto tissues of productive and supportive function. correlation (their Fig. 2) displays three importantHowever, the two-compartment analysis still outliers, without which the significance of theproduces interesting findings and retains contact correlation would be much reduced, though it iswith a very substantial literature (see Hunt & unlikely that its sign would be reversed. The currentNicholls, 1986). data contain no important outliers and, when

Grime & Hunt (1975) did not comment on separated into herbaceous dicotyledons, herbaceousroot:shoot allocation, even though the relevant data monocotyledons and woody dicotyledons, they showwere available. It was left to Hunt & Nicholls (1986) a mixed set of correlations between mean RGR andand Hunt & Lloyd (1987) to demonstrate a weak the allometric coefficient (Table 2a-c).positive correlation (in herbaceous species only) First, the coefficients obtained at 125 and atbetween the 1975 values of mean RGR and the 250/imol m"" s~ are themselves highly correlated,allometric coefficient. In this large and very het- demonstrating, as did the data obtained by Hunt eterogeneous collection of species, there was a al. (1987), that species differences in root:shootsignificant trend towards ' rootiness' in fast-growing allocation are robust to environmentally inducedspecies, but the scatter diagram was triangular, and perturbations, in this case an increase in PAR. Tableslow-growing species could tend either to 'rooty' or 1 shows, as expected, that the group means of theto ' shooty' growth. Much of the secondary variation allometric coefficient increase between 125 andwas doubtless due to the problems identified by 250 fivaol m~^ s , in both monocotyledonous andKorner (1994). However, within a smaller and much dicotyledonous herbaceous species, demonstrating amore homogeneous group of species. Hunt, Nicholls reduced allocation to shoot in the more productive of& Fathy (1987) then demonstrated a clear parallel the two aerial environments.drift upwards in mean RGR and the allometric Second, the only correlation with mean RGR whichcoefficient in a sample of 18 British grasses {P < approaches significance in the herbaceous species is0-001). The slope of this drift was stable to external the positive one for monocotyledons grown atperturbations in the form of shading or nutrient 125/tmol m"^ s~ (r = 0-389). This weakly supportsdeficiency, although, naturally, the absolute the observations made on 18 grasses by Hunt et al.magnitudes of the coefficients were affected by these (1987).factors. Third, the correlation with mean RGR in woody

It has long been realized that high maximum RGR dicotyledons is positive and highly significant (r =is a product of high activity both above and below 0-625, P < 0-01). In this group, which among theground (Donald 1958). Nevertheless, high maximum present species certainly include those with theRGR has been linked, assuming constant leaf per- lowest overall allocation to photosynthetic tissue (seeformance, to high biomass allocation to photo- the later discussion of leaf weight fraction), it is clearsynthetic structures (and vice versa) in the so-called that the faster-growing species allocate' resource ratio' model of grazed plant communities proportionally most towards root growth. This result(Huston & Smith, 1987; Tilman, 1988). This theory owes something to the special behaviour of threearose because, when species are ranked in order of slow-growing species, Aesculus hippocastanum,life form from annual herbs to perennial herbs, Juglans regia and Quercus robur, which all deployed


a substantial taproot before the initial harvest and available to conduct any trade-off was constrained ifthen engaged in highly shoot-orientated growth plant biomass had a high structural component or ifthereafter. However, omitting these three species the environment was one of low productivity,from the calculation still produced a correlation Swanborough & Westoby (1996) reached a similarcoefficient of 0-532 (P < 0-05). At the within-groups conclusion with respect to relative roles of biomass-level this apparently refutes Tilman's (1988) implicit based measures of production and allocation inassumption of a connection between low RGR and explaining RGR though they were not able to niatchhigh allocation to roots. Further, the woody species these measures directly against the equivalent area-as a whole engage in more shoot-orientated growth based parameters.than do the herbaceous species, displaying a mean In all, these findings support a more 'integrated'allometric coefficient of 0-89 (0-756 without the three view of partitioning in plants, in which there is aspecies mentioned) against the value of 0-828 highly interactive capture of resources by both shootsobtained across all herbaceous species grown at the and roots (Grime, 1994; Campbell, Grime &closely comparable PAR 125 /^mol m"" s~ . But when Mackey, 1991), as opposed to a general drift towardsthis comparison is performed on mean RGR, the a more biased allocation which is linked to mean RGRrespective values for woody and herbaceous species within any one plant group. This view accords withare 0-091 (0-099 without the three species mentioned) Davidson's (1969) 'functional equilibrium' hypoth-and 0-174 d^ . This therefore supports Shipley & esis in which whole-plant RGR is maximized (e.g.Peters's (1990) refutation of Tilman's assumption: Hilbert, Larigauderie & Reynolds, 1991) by main-slow-growing groups are not high root-allocators, or taining an approximately constant internal balancevice versa. between the products of total root and shoot activity.

However, Tilman (1991) pointed out that Shipley However, biomass allocation is only one of the& Peters (1990) had attempted a poor test of the components needed to secure this balance, the otherassumption linking RGR positively to photosynthetic being the specific activity (per unit mass) of root orallocation because the proportion of leaf material shoot. It is thus likely that the absence of systematicwithin shoot d. wt is variable, and that they had also trends in biomass partitioning among the groupsneglected the necessary assumption of constant leaf examined here is indicative that all groups also attachperformance. Between-groups, variability of allo- approximately equal importance to maximizingcation to leaf material is certainly present: in the specific activity both above-ground and below-present study, mean LWF was 0-702 for herbaceous ground. The existence of an evolutionary trade-offmonocotyledons, 0-711 for herbaceous dicotyledons which sacrifices either below-ground or above-and 0-497 for woody dicotyledons. Doubtless, such ground competitive ability (with respect to specificfindings elsewhere prompted Tilman's original as- activity) is thus called into question,sumption. However, at the within-groups level, no Finally, Shipley (1989) used his dataset on 68general patterns of growth involving differential Canadian wetland species to see whether mean RGRallocation to photosynthetic tissue can take place in the above-ground parts of the plant could be usedwithout there being a reflection of this process in the as a surrogate for mean RGR in the whole plant, withallometric coefficient. In the two groups in which a consequent saving of much experimental labour,shoot d. wt is most dominated by leaf d. wt (her- He found the highly significant correlationbaceous mono- and dicotyledons), no negative coefficient 0-98 for these two parameters of growth,association of the allometric coefficient with mean and a coefficient of 1-01 for mean whole-plant RGRRGR could be demonstrated, neither could any direct regressed onto mean shoot RGR, expressed in ident-association of mean LWF with mean RGR (Table 2a, ical units. Shipley recognized that what variationb). there was about this fitted line was due to differences

In a separate publication. Hunt & Cornelissen between species in above-ground and below-ground(1997) advised that all attempts to 'validate' the partitioning, concluding that the expedient ofTilman (1988) model by substituting-in independent measuring shoot growth alone would only be validexperimental data are ultimately irrelevant because when the variance between species with respect tothe model is a mathematical identity which is already partitioning was very small in comparison with thetrue by definition. It was useful, however, to see variance between species with respect to RGR. For thewhat values the physiological and allocational growth present selection of herbaceous species grown at aparameters such as ULR and LAR took among plants of PAR of 250 pimoX m"^ sT^, the coefficient of variationcontrastedtaxonomicgroupsandhabitataffinities.lt (standard deviation-:-mean) for mean RGR is 0-28,emerged that the trade-off occurring between physi- whereas that for the allometric coefficient is 0-19.ology and allocation in the struggle to optimize The latter coefficient, though smaller than that ofgrowth rate differed substantially between the leaf- RGR, is still appreciable in magnitude. It would not,area-based analyses (such as those reported here) and therefore, be advisable here to proceed with com-ones based upon leaf biomass (as calculated by Hunt parisons based upon shoot RGR alone.& Cornelissen, 1997). It was noted that the flexibility In respect of water content, contemporary theory


suggests that fast-growing species have thin leaveswith a high content (Garnier & Laurent, 1994).Fresh weight:dry weight ratio was measured inherbaceous species only, and at the final (21 d harvestonly. The data do show a relation along the linesdescribed, but only in the monocotyledons (Table2a, b). There, high f. wt:d. wt is associated withhigh values of each of RGR, LAR and SLA, but it is notcorrelated with ULR or with the allometric coefficient.It is likely that whole-plant water content is a crudemeasure which is correlated with RGR (or with theleaf parameters most closely tied to RGR) only whena large part of total d. wt of the plant resides in leafmaterial.

What is the interrelation of PAR and ULR indetermining RGR ?

measurements of small segments of fully grown,single-layered, uniformly illuminated, leaf material.Rather, they represent the net performance overmany days of whole, partly self-shaded, shootsystems growing in a natural display. Notsurprisingly, there is a significant negative cor-relation between this form of quantum yield and theLAR in these herbaceous species (at 250 fimol vnT^ s~\r = -0-359, P < 0-05). The most'leafy' species thushave the lowest quantum yields (cf. Poorter et al.1990).

The woody species included in the present studywere grown at only one level of PAR, so similarcalculations of PAR/ULR differences, and of quantumyield, were not possible.

What are the relative contributions of themorphological and physiological components of meanRGR?

The change in PAR from 125/tmolm ^s ^ to250 /imol m~ s"'- resulted in an increase in mean ULRfrom 0-0072 to 0-0136 mg mm~^ d~\ that is, a Mean ULR and mean LAR as determinants of meandoubling of PAR produced a ULR which was 188% RGR. The subdivision of RGR into a physiologicalof its former value, i.e. the value of this parameter component (ULR) and a morphological componentincreased almost pro rata. All monocotyledons and (LAR) has long been employed, both in the ecologicaldicotyledons individually registered higher mean and in the agricultural literature, as a means ofULR at 250/<mol m"" s ^ than at 125/^mol m"' s~ unravelling the causes of variations in growth rate(Fig. 2&, tf). However, the parallel response in mean (see reviews by Evans, 1972; Causton & Venus,RGR was a rise in value to only 117% of its former 1981; Hunt, 1982, 1990). The current understandingvalue. Although the relation RGR = ULR X LAR is far of this topic, in the case of European native species,from exact when dealing with harvest-interval means derives from work by Poorter (1990), Poorter &(Evans, 1972; Hunt, 1990), these relative changes in Remkes (1990) and Maranon & Grubb (1993).RGR and ULR imply very approximately that the The two publications involving Poorter bothhigher PAR has caused reductions to around two- employed similar forms of analysis: values of meanthirds of its former value in mean LAR, and that these RGR, ULR and LAR were obtained for a selection ofreductions partly offset the approximately pro-rata species and subjected to 'pathway anal^^sis'. In this,elevation of mean ULR. standardized data for the three variables (i.e. data

It is also possible to use the information on PAR converted to zero mean and unit standard deviation)and ULR to calculate a crude version of mean were regressed onto one another to discover thequantum yield for photosynthesis (whole-plant d. wt magnitude of the changes induced. In a sample ofproduction per quantum of PAR). Because, across eight grasses and herbaceous dicotyledons studiedall herbaceous species, 125 /imol m~ s" of by Poorter (1990), the respective coefficients of RGRadditional PAR produces 0-0064 mg mm"^ d" of on ULR and on LAR were 0-67 and 1-61, implying thatadditional ULR, this implies an apparent quantum the infiuence of LAR was at least twice as strong asyield of 0-60 g m o r \ Similar calculations, when that of ULR. In a larger sample of 24 grasses anddone on a species-by-species basis, show that within herbaceous dicotyledons, but including seven ofherbaceous monocotyledons the range of quantum those studied by Poorter (1990), Poorter & Remkesyields runs from 0-21 g mol"^ (in Agrostis capillaris) (1990) found that the coefficients of RGR on ULR andto 1-04 g mol"^ (in Catapodium rigidum), and within LAR were 0-51 and 0-96 respectively, implying thatherbaceous dicotyledons from 0-15 g mol"^ (in the infiuence of LAR was less strong than before, butEpilobium hirsutum) to 1-08 g mol"^ (in Plantago still considerably larger than that of ULR. But in bothlanceolata). Productive species thus appear to be instances Poorter found that any increases in ULRmore profiigate users of PAR. This observation were also associated with decreases in LAR and, as thewould seem to run counter to the commonly linkage between LAR and ULR was strongest of all, anyexpressed finding that species differences in RGR are increase in ULR led to a net decrease in RGR. Maranononly poorly related to innate differences in photo- & Grubb (1993) did not perform direct analyses ofsynthetic output per unit leaf area (e.g. Mooney, the efiFects of ULR on RGR and LAR, but as all threeFerrar & Slatyer, 1978; Nelson, 1988; Poorter e? a/, were regressed onto initial seedling weight, it is1990). However, these calculations of quantum yield possible to infer indirectly that, among their col-are not, as more usually, based upon direct lection of 27 Mediterranean annuals, RGR was

408 R. Hunt and jf. H. C. Cornelissen

(a) Mean RGR (d-'',r= 0-124) (b) Mean RGR (d"'', r=: 0-593**)

0-30

0-20

0-10

0-000-005 0-010 0-015 0-020 0-025

Mean ULR (mg ^ ^

10 20Mean LAR (mm^

30

30

20

10

0

(0

-

-

-

Mean U\R {mm^ mg"''. r=0-321)

•

• •

id) Mean LAR (mm^ nrig-i, r= 0-966***

0-2 0-4 0-6 0-8Mean LWF {dimensionless)

10 20 30Mean SLA(mm^

40

Figure 3. Relations between mean values of growth parameters in the herbaceous monocotyledons; parts (a)and (b) illustrate the subdivision of relative growth rate into its two components, unit leaf rate and leaf arearatio; parts (c) and (d) illustrate the subdivision of leaf area ratio into its two components, leaf weight fractionand specific leaf area.

inversely related to ULR and positively related to LAR(in the form of its two sub-components, LWF andSLA).

The present data exhibit no significant dependenceof RGR on ULR in any of the three groups of species(Table 2a-c), though the absolute magnitude of ULRdeclines systematically across the series, in themanner: herbaceous monocotyledons > herbaceousdicotyledons > woody dicotyledons. Figures 3 a,4 a, 5 a show this clearly. On the other hand, RGR isstrongly dependent upon LAR in all three groups(Table 2a-c and Figs 3b, 4b, 5b). This dependencealso increases across the series of plant groups,though no great differences between the groups existwith respect to the absolute scale of LAR.

Mean LWF and mean SLA as determinants of meanLAR. The component LAR may itself be subdividedinto a 'quality' sub-component (SLA) and a'quantity' sub-component (LWF) as a further meansof unravelling the causes of variations in RGR. Poorter(1990) and Poorter & Remkes (1990) did not extendtheir ' pathway analysis' to cover SLA and LWF, but itwas clear from their data that SLA was by far the moreinfluential parameter of the two (Poorter & Lambers,1991). Marafion & Grubb (1993) regressed SLA andLWF onto initial seedling weight of Mediterranean

annuals and found a weakly negative relation in thecase of LWF, and a strongly negative one in the caseof SLA. From this it can be inferred that RGR waspositively related both to LWF and to SLA, and morestrongly to the latter.

The present data exhibit two very interestingtrends across the groups of species. Though theabsolute scale of mean LWF decreases systematicallybetween herbaceous and woody species, the de-pendence of LAR on LWF increases strongly (Table2a-c and Figs 3 c, 4 c, 5 c). On the other hand,although there are no systematic changes in scaleacross the series in magnitude of SLA, the correlationof LAR upon SLA in the herbaceous monocotyledonsdisplays the highest coefficient in the entire dataset.Though this relation weakens across the seriesherbaceous monocotyledons to woody dicotyledons,it still remains strong in the latter (Table 2a-c andFigs 3d, 4d, 5d).

What is the combined influence of these componentson mean RGR?

Rationale. To help answer this question, three formsof multivariate analysis were performed on the meanvalues of RGR, ULR, LWF and SLA (using only thoseobtained at 250 /imol m"^ s"^ in the case of theherbaceous species), namely multiple regression.


(a) Mean RGR (d^'', r= 0-312)

0-30

0-20

010

0-000-005 0-010 0-015

Mean ULR (mnn

(b) Mean RGR (d"'', r= 0-652***)

0-020 0-025 0 10 20 30Mean LAR (mm^ nng"'')

[c) Mean LAR (mm^ mg"'', r= 0-494*)

30

20

10

0-2

(d) Mean LAR (nnm^ mg~\ r= 0-905^

W • • •

0-4 0-6 0-8Mean LWF(dimensionless)

10 20 30Mean SLA (mm^

40

Figure 4. Relations between mean values of growth parameters in the herbaceous dicotyledons; parts (a—d) asin Figure 3.

(a) Mean RGR (d"'', r=0-116)

0-30

0-20

0-10

0-00

(b) Mean RGR (d"'', r= 0-841***)

• #

0-005 0-010 0-015 0-020 0-025 0Mean ULR (mg

— I —

2010 20 30Mean LAR (mm^ mg"'')

(c) Mean LAR (mm^ mg-'', r= 0-818'

30

20

10

00-2

• •

(d) Mean LAR (mm^ nng"'', r= 0-786***)

20 30Mean SLA(mm^

400-4 0-6 0-8 10Mean LWF(dimensionless)

Figure 5. Relations between mean values of growth parameters in the woody dicotyledons; parts (a-d) as in

Figure 3.

410 R. Hunt and y. H. C. Cornelissen

Table 3. R-squared and P-values in the regression RGR =f(ULR, LWF, SLA) {at PAR level 250 only ; see Table1 for explanations and units)

R- ULR LWF SLA

Herbaceous monocotyledons {n = 21)Herbaceous dicotyledons (n = 22)Woody dicotyledons {n = 16)

0-0-0-

812782875

000

-000009-000015-000686

000

-020775-000571-000163

0-0000000-0000060-000448

principal components analysis (PCA) and non- ULR but relatively high in SLA. Axis 2 (the v-axis)hierarchical classification (optimal clustering). These pointed towards species of low RGR and high LWF.were done with the strictly limited objective of In herbaceous dicotyledons (Fig. 66), the positiveformalizing the underlying interrelatedness of these direction of Axis 1 pointed towards species of highvariables in different groups of species (cf. Reich, RGR and high SLA. Mean RGR appeared a second timeWalters & Ellsworth, 1992) and with a complete as a leading vector in Axis 2, which pointed towardsdisregard of the normal requirement for statistical species of low RGR and low ULR. In woodyindependence of the variables (cf. Petraitis, Dunham dicotyledons (Fig. 6 c), the positive direction of Axis& Niewiarowski, 1996). All analyses were performed 1 pointed towards species of low RGR and low SLA.by programs written in GENSTAT 5. Within each Axis 2 pointed towards species of low ULR and highgroup, data were analysed in species rank order (1 = LWF.high, n = low) to eliminate differences betweenvariables with respect to units (c. 100-fold) and Optimal clustering. An optimal (non-hierarchical)statistical distributions. clustering was performed separately on herbaceous

dicotyledons, herbaceous monocotyledons andMultiple regression. The model htt&d was of the form woody dicotyledons. The analysis used a sums ofRGR = / ( U L R , LWF, SLA) with linear terms in each of squares criterion to sort the species into clustersthe three 'independent' variables. The adjusted r^ which were as consistent as possible internally but asvalues for multiple regression, and the P-values for distinct as possible externally. The computationaleach of the coefficients (all were of positive sign) are approach was that followed by Grime & Hunt (1985)shown in Table 3. and Grime, Hunt & Krzanowski (1988). Briefiy, a

Clearly, these three variables jointly 'explained' a range of numbers of optimal clusters was calculatedgreat deal of the variation in mean RGR, and all of the for each set of species. A special ' stopping criterion'coefficients were highly significant (at at least P < (developed from work by Krzanowski & Lai, 1988)0-02, and mostly much less), but this was only to be was then used to identify the tightest possibleexpected. Of more interest is the finding that the clustering solution. Within the range one to ten, theleading variables (in terms of significance of optimal numbers of clusters were found to be threecoefficient) were different in the case of the her- in the case of herbaceous monocotyledons and woodybaceous and the woody groups. For both sets of dicotyledons, and four in the case of herbaceousherbaceous species the most significant variables dicotyledons.were ULR and SLA, (implying more special roles for Table 4 summarizes the typical attributes of eachleaf performance and quality than that of leaf of the clusters which were recognized. Only sixallocation). For woody dicotyledons, though with cluster ' types' were recognized out of the 16considerably less certainty, it was LWF which was theoretically possible combinations of 'high' meanmost specially linked to RGR, implying relatively values (within the upper half of the ranking) or ' low'lesser roles for leaf quality and leaf function. This mean values (within the lower half of the ranking) inresult agrees with the correlations performed on the the four plant variables. One of these six ('type 5')separate components. was represented within all three of the plant datasets;

two ('types 3 and 12') were represented only withinPrincipal components analysis. A simple PCA ^ê two herbaceous datasets; one ('type 2') waswas performed separately on the herbaceous represented only within herbaceous dicotyledons;dicotyledons, herbaceous monocotyledons and ^ê remaining two ('types 1 and 16') werewoody dicotyledons. The criterion was to maximize represented only within woody dicotyledons. Figurethe sums of squares and products withm linear 5 ^ôws how the six optimal clusters map onto PCAcombinations of the four variables. The first two axes ^^^^^ ^^^ -^ ^1^^ identifies the species involved.in every case included over three-quarters of thevariation (83%, 77% and 79% respectively). In Overview of multivariate analyses. There is excellentherbaceous monocotyledons (Fig. 6a), the positive agreement between the PCA and the clustering. Alldirection of Axis 1 (the x-axis) pointed towards clusters occupy distinct zones within PCA space, thespecies which, when defined in terms of the two only hint of any overlap being in the herbaceousleading vectors within the axis, were of relatively low dicotyledons, with the possible inclusion of smaller


(a) Herbaceous monocotyledons

Axis 1: ULR low, SLA high

(fa) Herbaceous dicotyledons

Axis 2: RGR low, ULR high

Axis V. RGR high, SLA high

(c) Woody dicotyledons

Axis 2: ULR low, LWF high

AXIS 1: RGR low, SLA low

Figure 6. Principal components analysis and non-hierarchical cluster analysis of three separate groups ofspecies, ordained or classified according to the four growth parameters relative growth rate, unit leaf rate, leafweight fraction and specific leaf area. The main contributors to the PCA axes are identified on the diagramsand the encircled groups of points represent the clusters defined in Table 4.

sub-cluster ('type 3') within a larger one ('type 5'),which was itself universal to all plant groups (seeFig. 66).

The clustering, in particular, reveals strong evi-dence of evolutionary trade offs (Table 4). Ten of thepossible 16 combinations of attribute are notrepresented at all, even in this diverse collection of

species and life forms. Most strikingly, high RGR isnever seen in combination with low SLA. When lowRGR is seen together with high SLA (e.g. in theherbaceous dicotyledons of 'type 2'), this differenceemerges on the second, less diagnostic PCA axis(Fig. 66). However, even within this apparentconstraint, different groups appear to reach different


Table 4. Results of cluster analyses

Type

123456789

10111213141516

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

RGR

lowlowlowlowlow-lowlowlowhighhighhighhighhighhighhighhigh

Parameter

ULR

ULR lowULR lowULR lowULR lowULR highULR highULR highULR highULR lowULR lowULR lowULR lowULR highULR highULR highULR high

combination

L W F

LWF lowLWF lowLWF highLWF highLWF lowLWF lowLWF highLWF highLWF lowLWF lowLWF highLWF highLWF lowLWF lowLWF highLWF high

SLA

SLA

SLA

SLA

SLA

SLA

SLA

SLA

SLR

SLA

SLA

SLA

SLA

SLA

SLA

SLA

SLA

lowhighlowhighlowhighlowhighlow-highlowhighlowhighlow-high

Groupsrepresented*

w2h2hl,h2—h i , h2, w2——————h i , h2———w2

* h i . Herbaceous monocotyledons; h2. Herbaceous dicotyledons; w2. Woodydicotyledons.

solutions: the route to high RGR in both sets ofherbaceous species is through high allocation toleaves of thin, limited-performance material, ratherthan to thicker, high-performing leaves. In the fast-growing woody species, however, the 'Utopian'combination comprising high values of all fourparameters is evident, though this occurs only withinthe local context of the woody group, the highestRGRs of which do not match those of the twoherbaceous groups. Similarly, the other extremecombination comprising low values of all parametersis also seen only among the woody species. Not manycommon solutions are reached by all groups, the onlyuniversal one being the 'moderate' combination ofefficient, thick leaves with low allocation to leafmaterial ('type 5'). There is, in general, a pooroverlap between the combinations displayed by theherbaceous and the woody species.

The morphological/physiological types identifiedby the multivariate analyses are not necessarily alsoecological types. It is true that the 'type 12' occurswithin both groups of herbaceous species andcomprises species found exclusively in resource-richhabitats. Some of them are annual and someperennial, and most are tall. On the other hand, someof the smaller clusters do represent coherent eco-logical types, e.g. 'type 3' almost entirely comprisesslow-growing species of resource-poor habitats.

The analyses of these species in three groups isclearly informative, but it also leads to difficultieswhere the local context (for example in absolutevalues of parameters) does not match the more globalone. A final perspective may be gained from an all-in analysis. A PC A involving all 59 species sim-ultaneously is presented in Figure 7.

The first two axes included 84 % of the variation.The positive direction of Axis 1 (the x-axis) pointed

towards species which, when defined in terms of thetwo leading vectors, were of relatively high RGR andrelatively high in ULR. Axis 2 (the j^-axis) containedonly one important vector and pointed towardsspecies of low SLA. Woody species occupied only the' low RGR' end of Axis 1 with very little overlap withherbaceous species. Axis 2 did not discriminatewithin woody species but, within herbaceous species,many dicotyledons occupied the end of the scale withhigh SLA, and many monocotyledons the end withlower SLA, though there was also much overlap.

Are there any other important correlates of meanRGR?

Data from screening programmes such as the ISPmay be analysed at three distinct levels. First, ashere, each plant attribute may be examined alone(although a subdivision into any components may beincluded) and the significance of the attributeassessed largely in isolation. Second, relations be-tween pairs of important attributes may beexamined. Sometimes these reveal a direct andsimple correlation but, more often, even the scatterdiagrams for significantly correlated attributes dis-play a triangular pattern, indicating that furtherattributes need to be considered in order to interpretthe trend satisfactorily (many examples of bothpatterns appear in Hendry & Grime, 1993). Third,more than two plant attributes may be examinedsimultaneously by means of multivariate statisticaltechniques. The third approach has been especiallyimportant in the search for primary ecologicalstrategies in both animals and plants. During thedevelopment of the ISP, this so-called 'rollingsynthesis' has been attempted on four previousoccasions (see Hendry & Grime, 1993, p. 223, for a


Axis 1: RGR high; ULR high

Figure 7. Principal components analysis of all 53 species taken together, ordinated according to the four growthparameters relative growth rate, unit leaf rate, leaf weight fraction and specific leaf area. The main contributorsto the PCA axes are identified on the diagrams and the encircled groups of points represent the three main plantgroups used in this study.

chronology) and a fifth synthesis is now the subjectof an extensive separate paper (Grime et al., 1997).

As a precursor to this fifth synthesis it wasnecessary to identify all of the attribute-to-attributecorrelations present within a matrix of 67 attributesby 43 species (the same herbaceous species reportedon here). As an up-to-date example of the secondlevel of ISP analysis, relations with the mostimportant three of the 20 attributes found to becorrelated with mean RGR (at P < 0-05 or better) maybe reviewed briefiy (some of the other relations havealready been published, for example, that involvingdegree of responsiveness to elevated COj, Hunt etal, 1993).

Responsiveness to low mineral nutrient avail-ability has been assessed as the percentage of thecontrol yield which is achieved at 1 % of ' full'nutrient level (Hendry & Grime, 1993, p.163). Thelow-nutrient response is negatively correlated withmean RGR in both monocotyledons and dicotyledons(Fig. 8 a, b), indicating very clearly that faster-growing species suffer proportionately greaterreductions in yield at low nutrient level. Thisconfirms much long-standing ecological theory andmeasurement (e.g. Grime, 1965; Parsons, 1968;Higgs & James, 1969; Shipley & Keddy, 1988).Fast-growing species may suffer such reductions

because they are innately more nutrient-demanding:this is borne out b}' the internal nitrogen con-centration of leaves of plants grown under fullnutrient conditions (Hendry & Grime, 1993, p. 156),which is positively correlated with mean RGR in bothmonocot^'ledons and dicotyledons (Fig. 8c, d) (cf.Hirose, 1988; Poorter et al, 1990; Niemann et al,1992; Poorter & Bergkotte, 1992; Vanaarendonk &Poorter, 1994; Garnier et al, 1995). Seed weight,however, is negatively correlated with mean RGR inboth monocotyledons and dicotyledons (Fig. Se,f).Although mean RGR, as measured in the currenttests, is diminished when large quantities of seedmaterial persist into the seedling phase (Grime &Hunt, 1975; Cornelissen, Castro Diez & Hunt,1996), there is also a genuine underlying relationbetween mean RGR and seed weight at all values ofthe latter, indicating that high seed weight clearly hasadaptive value in resource-poor habitats (e.g. Fenner,1978, 1983; Maranon & Grubb, 1993; Stockey &Hunt, 1994). However, in the Mediterranean systemsstudied by Lee et al (1996), the relation betweensmall-seededness and high RGR was not upheld,perhaps because the open state of such communitiescreates little selective pressure for large-seededness(which is so evidently advantageous to slow-growangspecies in more temperate, closed communities).

414 R. Hunt andjf. H. C. Cornelissen

(a) Rank in low-nutrient response(herbaceous monocots)

25

20

15

10

(fa) Rank in low-nutrient response(herbaceous dicots)

5 10 15 20Rank in mean RGR

25 5 10 15 20Rank in mean RGR

25

(c) Rank in leaf N-percentage(herbaceous monocots)

(d) Rank in leaf N-percentage(herbaceous dicots)

20

15

10

25 r

20

15

10

5


(e) Rank in seed weight(herbaceous monocots)


(f) Rank in seed weight(herbaceous monocots)

05 10 15 20Rank in mean RGR


25

Figure 8. Relations between mean relative growth rate and other important plant attributes (all data are plottedin ranked form). The correlation coefficients are (a) -0-623, (b) -0-559, (c) 0-540, (d) 0-433, (e) -0-447,if) -0-476. All are significant at P < 0-05.

Finally, mean RGR has played an important part inthe search for avenues of adaptive specializationwhich may conform to basic 'functional types'.Grime et al. (1997) used both DECORANA (or-dination by detrended correspondence analysis) andoptimal clustering (non-hierarchical classification) toestablish that a primary axis of specialization whichinvolves mean RGR exists within the 43 herbaceousspecies. The ordination and the classification analy-ses agreed very closely. The high-i-anked end of theprimary axis comprised species displaying all thefollowing attributes simultaneously: high rates oflitter decomposition, high precision in root andshoot foraging, high leaf nutrient concentration, lowpercentage yield at low nutrient levels, low leaf

longevity, high relative growth rate, low leaf tensilestrength, high palatability and low mycorrhizalstatus. Because similar patterns occurred in bothmonocotyledons and dicotyledons, this appeared toreflect a trade-off between attributes which confer anability for high rates of resource acquisition inresource-rich habitats and those responsible forretention of resource capital in resource-poor con-ditions. The relevance to this axis of mean RGRmeasured under resource-rich conditions in theseedling phase is fundamental.

ACKNOWLEDGEMENTS

The Integrated Screening Programme of the Unit ofComparative Plant Ecology is supported by the Natural


Environment Research Council at several UK researchsites. J.H.C.C. acknowledges support from the EC HumanCapital and Mobility Programme. Staff collaborating inthe present study included S. R. Band, P. L. Gupta, J. G.Hodgson, J. Laffarga, G. Montserrat-Marti, A. M. Neal,1. H. Rorison, R.E Spencer, A. Stockey and J.Whitehouse. Very helpful comments on the manuscriptwere provided by J. P. Grime and K. Thompson.

APPENDIX

For the principal statistical computations, threegeneric kinds of growth parameter were of interest.These may be defined as 1/ F (d Y/dt), l/Z (d Y/dt)and Z/ Y, where Y and Z are different plant variablesand t is time. Each growth parameter was estimatedas a mean value, with variance, across the harvestinterval 7 to 21 d from germination (or the interval 0to 21 d from leaf opening in the case of the 16 woodyspecies). The statistical methods followed those ofVenus & Causton (1979) and Causton & Venus(1981), whereby mean values of the variousparameters were obtained without the use of fittedfunctions and without the ' pairing' of plants acrossthe harvest interval.

In this method, with the symbols 'V and '^representing variance and covariance respectively,the expected value of mean RGR across the harvestinterval t-^ to t^, and its variance, are given by:

and

r (R) = [r (log, Y,) - r (log,

where F represents total d. wt per plant-as harvest-means in the case of R and as variate arrays in thecase of y/" (R).

For the expected value of mean ULR across thesame harvest interval, and its variance, the relevantexpressions are:

y j

and

r (E) = {dE/d +{dE/d

j) +

( y j

r (Z2)

2 {dE/d

2

where Y again represents arrays of data for totald. wt per plant and Z represents arrays of total leafarea per plant. The term E is an estimate of meanunit leaf rate between t^ and t^, as provided fromharvest-mean data by the conventional expression

Y,-

(see Evans, 1972; Causton & Venus, 1982, Hunt,1982, 1990).

For the value of the quotient Z/ Y at either t-^ or t^and for its variance-as for example in the case of leafarea ratio, i^-the relevant expressions are:

and

r (F) = exp[2 log, F+r (log, F)]

with arrays of quotients being used in the functiony^ (loggi^). From separate estimates at each harvestoccasion, the expected mean values across the harvestinterval t^ — t-^ can be obtained:

and

A computer program written in GENSTAT 5 (Payneet al., 1987), which performs all the above calcu-lations, is available upon request from the authors.

REFERENCES

Blackman GE, Wilson GL. 1951. Physiological and ecologicalstudies in the analj'sis of plant environment. VI. The constancyfor different species of a logarithmic relationship between netassimilation rate and light intensity and its ecologicalsignificance. Annals of Botany 15: 63-94.

Booth RE, Mackey JML, Rorison IH, Spencer RE, Gupta PL,Hunt R. 1993. ISP germination and rooting environments:sand, compost and solution culture. In: Hendry GAF, GrimeJP, eds. Methods in Comparative Plant Ecology. London:Chapman & Hall, 19-24.

Briggs GE, Kidd F, West C. 1920. A quantitative analysis ofplant growth. 11. Annals of Applied Biology 7: 202-223.

Burdon JJ, Harper JL. 1980. Relative grovi'th rates of individualmembers of a plant population. Journal of Ecology 68: 953—957.

Campbell BD, Grime JP, Mackey JML. 1991. A trade-ofFbetween scale and precision in resource foraging. Oecologia 87:532-538.

Cornelissen JHC, Castro Diez P, Hunt R. 1996. Seedlinggrowth, allocation and leaf attributes in a wide range of woodyplant species and types. Journal of Ecology 84: 755—765.

Causton DR, Venus JC. 1981. The biometry of plant groivth.London: Edward Arnold.

Davidson RL. 1969. Effect of root/leaf temperature differentialson root/shoot ratios in some pasture grasses and clover. Annalsof Botany 33: 561-69.

Dijkstra P, Lambers H. 1989. A physiological analysis of geneticvariation in relative growth rate within Plantago major L.Functional Ecology 3: 577-587.

Donald CM. 1958. The interaction of competition for light andfor nutrients. Australian Journal of Agricultural Research 9:421-432.

Elias CO, Chadwick MJ. 1979. Growth characteristics of grassand legume cultivars and their potential for land reclamation.Journal of Applied Ecology 16: 537-544.

Evans GC. 1972. The quantitative analysis of plant growth.Oxford: Blackwell Scientific Publications.

Fenner M. 1978. A comparison of the abilities of colonizers andclosed-turf species to establish from seed in artificial swards.Journal of Ecology 66: 953-963.

Fenner M. 1983. Relationships between seed weight, ash contentand seedling growth in twenty-four species of compositae. NewPhytologist 95: 697-706.

Garnier E, Gobin O, Poorter H. 1995. Nitrogen productivity

416 R. Hunt andj. H. C. Cornelissen

depends on photosynthetic nitrogen use efficiency and onnitrogen allocation within the plant. Annals of Botany 76:667^672.

Garnier E, Laurent G. 1994. Leaf anatomy, specific mass andwater content in congeneric annual and perennial grass species.New Phytologist 128: 725-736.

Grime JP. 1965. Comparative experiments as a key to the ecologyof flowering plants. Ecology 45: 513—515.

Grime JP. 1974. Vegetation classification by reference tostrategies. Nature 250: 26-31.

Grime JP. 1977. Evidence for the existence of three primarystrategies in plants and its relevance to ecological and evol-utionary theory. American Naturalist 111: 1169—1194.

Grime JP. 1979. Plant strategies and vegetation processes.Chichester: John Wiley & Sons Ltd.

Grime JP. 1985. Integrated screening programme. Annual Reportof the NERC Unit of Comparative Plant Ecology for 1985 : 6-10.

Grime JP. 1994. The role of plasticity in exploiting environmentalheterogeneity. In: Caldwell M, Pearcy R, eds. Exploitation ofEnvironmental Heterogeneity in Plants. New York: AcademicPress Inc., 1-19.

Grime JP, Hodgson JG, Hunt R. 1988. Comparative plantecology: a functional approach to common British species.London: Unwin Hyman.

Grime JP, Hunt R. 1975. Relative growth-rate: its range andadaptive significance in a local flora. Journal of Ecology 63:393-422.

Grime JP, Hunt R, Krzanowski WJ. 1985. Evolutionaryphysiological ecolog}^ of plants. In: Calow P, ed. EvolutionaryPhysiological Ecology. Cambridge: Cambridge University Press,105-125.

Grime JP, Thompson K, Hunt R, Hodgson JG, CornelissenJHC, Rorison IH, Hendry GAF, Ashenden TW, Askew AP,Band SR, Booth RE, Bossard CC, Campbell BD, CooperJEL, Davison AW, Gupta PL, Hall W, Hand DW, HannahMA, Hillier SH, Hodkinson, DJ, Jalili A, Liu Z, MackeyJML, Matthews N, Mowforth MA, Neal AM, Reader RJ,Reiling K, Ross-Fraser W, Spencer RE, Sutton F, TaskerDE, Thorpe PC, Whitehouse J. 1997. Integrated screeningvalidates primary' axes of specialisation in plants. Oikos (inpress).

Grubb PJ, Lee WG, Kollmann J, Wilson JB. 1996. Interactionof irradiance and soil nutrient supply on growth of seedlings often European tall-shrub species and Eagus sylvatica. Journal ofEcology M: 827-840.

Harper JL. 1982. After description. In: Newman El, ed. ThePlant Community as a Eunctioning Mechanism. Oxford:Blackwell Scientific Publications, 11—25.

Hendry GAF, Grime JP (eds). 1993. Methods in comparativeplant ecology : a laboratory manual. London: Chapman & Hall.

Higgs DEB, James DB. 1969. Comparative studies on thebiology of upland grasses. I. Rate of dry matter production andits control in four grass species. Journal of Ecology 57: 553-563.

mibert DW, Larigauderie A, Reynolds JF. 1991. Theinfiuence of carbon dioxide and daily photon-flux density onoptimal leaf nitrogen concentration and root: shoot ratio. Annalsof Botany 68: 365-376.

Hirose T. 1988. Modelling the relative growth rate as a functionof plant nitrogen concentration. Physiologia Plantarum 72:185-189.

Hunt R. 1982. Plant growth curves: the functional approach toplant growth analysis. London: Edward Arnold.

Hunt R. 1984. Relative growth rates of cohorts of ramets clonedfrom a single genet. Journal of Ecology 72: 299-305.

Hunt R. 1990. Basic growth analysis. London: Unwin Hyman.Hunt R, Cornelissen JHC. 1997. Physiology, allocation and

growth rate: a re-examination of the Tilman model. AmericanNaturalist (in press).

Hunt R, Doyle CJ. 1984. Modelling the partitioning of researcheffort in ecology. Journal of Theoretical Biology 111: 451-461.

Hunt R, Hand DW, Hannah MA, Neal AM. 1991a. Responseto CO., enrichment in 27 herbaceous species. Functional Ecology5: 410^21.

Hunt R, Hand DW, Hannah MA, Neal AM. 1993. Furtherresponses to COg enrichment in British herbaceous species.Functional Ecology 7: 661-668.

Hunt R, Hope-Simpson JF. 1990. Growth oi Pyrola rotundifolia

ssp. maritima in relation to shade. New Phytologist 114:129-137.

Hunt R, Lloyd PS. 1987. Growth and partitioning. NeivPhytologist 106 (Suppl.): 235-249.

Hunt R, Middleton DAJ, Grime JP, Hodgson JG. 19916.TRISTAR: an expert system for vegetation processes. ExpertSystems 8: 219-226.

Hunt R, Nicholls AO. 1986. Stress and the coarse control ofgrowth and root—shoot partitioning in herbaceous plants. Oikos47: 149-58.

Hunt R, Nicholls AO, Fathy SA. 1987. Growth and root-shootpartitioning in eighteen British grasses. Oikos 50: 53—59.

Huston MA, Smith TM. 1987. Plant succession: life history andcompetition. American Naturalist 130: 168-198.

Jarvis PG, Jarvis MS. 1964. Growth rates of woody plants.Physiologia Plantarum 17: 654-66.

Korner C. 1994. Biomass fractionation in plants: a recon-sideration of definitions based upon plant functions. In: Roy, JGamier, E eds. A Whole Plant Perspective on Carbon—NitrogenInteractions. The Hague: SPB Academic Publishing bv,173-185.

Krzanowski WJ, Lai YT. 1988. A criterion for determining thenumber of groups in a data set using sum-of-squares clustering.Biometrics 44: 23-34.

Maranon T, Grubb PJ. 1993. Physiological basis and ecologicalsignificance of the seed size and relative growth rate relationshipin Mediterranean annuals. Functional Ecology 7: 591-599.

Mooney HA, Ferrar PJ, Slatyer RO. 1978. Photosyntheticcapacity and allocation patterns in diverse growth forms ofEucalyptus. Oecologia 36: 103-111.

Muller B, Garnier E. 1990. Components of relative growth rateand sensitivity to nitrogen availability in annual and perennialspecies oi Bromus. Oecologia 84: 513-518.

Nelson CJ. 1988. Genetic associations between photosyntheticcharacteristics and yield: review of the evidence. Plant Physi-ology and Biochemistry 26: 543-554.

Niemann GJ, Pureveen JBM, Eijkel GB, Poorter H, Boon JJ.1992. Differences in relative growth rate in 11 grasses correlatewith differences in chemical composition as determined bypyrolysis mass-spectrometry. Oecologia 89: 567—573.

Parsons IT, Hunt R. 1981. Plant growth analysis: a program forthe fitting of lengthy series of data by the method of i?-splines.Annals of Botany 48: 341-352.

Parsons RF. 1968. The significance of growth-rate comparisonsfor plant ecology. American Naturalist 102: 595-597.

Payne RW, Lane PW, Ainsley AE, Bicknell, 10 others. 1987.Genstat 5 reference manual. Oxford: Clarendon Press.

Petraitis PS. Dunham AE, Niewiarowski PH. 1996. Inferringmultiple causality: the limitations of path analysis. FunctionalEcology 10: 421-4-31.

Poorter H. 1989. Plant growth analysis: towards a synthesis ofthe classical and the functional approach. Phvsiologia Plantarum75: 237-244.

Poorter H. 1990. Interspecific variation in relative growth rate:on ecological causes and physiological consequences. In:Lambers H, Cambridge ML, Konings H, Pons TL, eds. Causesand Consequences of Variation in Growth Rate and Productivityof Higher Plants. The Hague: SPB Academic Publishing bv,45-68.

Poorter H, Bergkotte M. 1992. Chemical composition of 24 wildspecies differing in relative growth rate. Plant, Cell andEnvironment 15: 221-229.

Poorter H, Farquhar GD. 1994. Transpiration, intercellularcarbon dioxide concentration and carbon isotope discriminationof 24 wild species differing in relative growth rate. AustralianJournal of Plant Physiology 21: 507-516.

Poorter H, Lambers H. 1991. Is interspecific variation inrelative growth rate positively correlated with biomass al-location to the leaves? American Naturalist 138: 1264-1268.

Poorter H, Remkes C. 1990. Leaf area ratio and net assimilationrate of 24 wild species differing in relative growth rate. Oecologia83: 553-559.

Poorter H, Remkes C, Lambers H. 1990. Carbon and nitrogeneconomy of 24 wild species differing in relative growth rate.Plant Physiology 94: 621-627.


Reich PB, Walters MB, Ellsworth DS. 1992. Leaf life-span inrelation to leaf, plant and stand characteristics among diverseecosystems. Ecological Monographs 62: 365-392.

Royal Society. 1995. Editorial code for the presentation ofstatistical analyses. Proceedings of the Royal Society Series B262: 379-380.

Shipley B. 1989. The use of above-ground maximum relativegrowth rate as an accurate predictor of whole-plant relativegrowth rate. Functional Ecology 3: 771-775.

Shipley B, Keddy PA. 1988. The relationship between relativegrowth-rate and sensitivity to nutrient stress in 28 species ofemergent macrophytes. Journal of Ecology 76: 1101-1110.

Shipley B, Keddy PA, Moore DRJ, Lemky K. 1989, Re-generation and establishment strategies of emergent macro-phytes. Journal of Ecology 11: 1093-1110.

Shipley B, Peters R. 1990. A test of the Tilman model of plantstrategies: relative growth rate and biomass partitioning.American Naturalist 136: 139-153.

Stace C. 1991. New flora of the British Isles. Cambridge:Cambridge University Press.

Stockey A, Hunt R. 1994. Predicting secondary succession inwetland mesocosms on the basis of autecological information onseeds and seedlings. Journal of Applied Ecology 31: 543-559.

Swanborough P, Westoby M. 1996. Seedling relative growthrate and its components in relation to seed size: phylogeneticallyindependent contrasts. Functional Ecology 10: 176-184.

Tilman D. 1988. Plant strategies and the dynamics and structure ofplant communities. Princeton, NJ, USA: Princeton UniversityPress.

Tilman D. 1991. Relative growth rates and plant allocationpatterns. American Naturalist 138: 1269-1275.

Van Andel J, Biere A. 1990. Ecological significance of variabilityin growth rate and plant productivity. In: Lambers H,Cambridge ML, Konings H, Pons TL, eds. Causes andConsequences of Variation in Growth Rate and Productivity ofHigher Plants. The Hague: SPB Academic Publishing bv,257-267.

Vanaerendonk JJCM, Poorter H. 1994. The chemical com-position and anatomical structure of leaves of grass speciesdiffering in relative growth-rate. Plant, Cell and Environment17: 963-970.

Venus JC, Causton DR. 1979. Plant growth anabasis: a re-examination of the methods of calculation of relative growthand net assimilation rates without using fitted functions. Annalsof Botany ^Z: 633-638.

Components of relative growth rate and their interrelations in 59 temperate plant species

Documents

Transcript of Components of relative growth rate and their interrelations in 59 temperate plant species