New Phytol. (1997), 136, 679-689
A comparative study of leaf nutrient concentrations in a regional herbaceous flora BY K E N THOMPSON^*, JOHN A. PARKINSON^ STUART R. AND R I T A E. S P E N C E R i ^ NERC Unit of Comparative Plant Ecology, Department of Anim.al and Plant Sciences, The University, Sheffield SIO 2TN, UK ^Institute of Terrestrial Ecology, Merlewood Research Station, Grange-over-Sands, Cumbria LAll SJU, UK {Received 5 December 1996; accepted 30 April 1997) SUMMARY Mineral nutrient concentrations were determined in leaves of 83 mostly herbaceous species collected from central England. Most samples were analysed for N, P, K, Ca, Mg, Na, Fe, Al, Mn, Cu and Zn. Concentrations of K, N and P showed similar levels of interspecific variability, with the highest concentrations being 6-9 times the lowest. Mg and (especially) Ca were much more variable, with the highest concentrations being 24 and 49 times the lowest respectively. Only in the case of P concentration was the niajority of the variance in the data found at or below the species level. Most of the variance in Ca and Mg concentrations was between monocots and dicots. Concentrations of N and P were strongly positively correlated with each other. Only Ca and Mn were consistently associated with soil pH, positively and negatively respectively. Dicots tended to accumulate more Ca and Mn from high soil concentrations than did monocots. Concentration of P was significantly positively correlated w-ith maximum potential relative grow th rate. Plants of woodland and arable habitats contained high concentrations of P, and those of pasture and skeletal habitats contained low concentrations of P. The P : N ratio was higher in plants of arable habitats. Species with P-rich leaves tended to be currently increasing in abundance. The results suggest that plants with nutrient-rich foliage grow quickly, dominate nutrient-rich ecosystems and are generally increasing as a result of the eutrophication and disturbance arising from human exploitation. Key words: Mineral nutrients, growth rate, pH, land use, phylogeny.
INTRODUCTION Previous field studies of nutrient concentrations in leaves and other plant parts have generally concentrated on comparatively few species from single communities (e.g. Woodwell, Whittaker & Houghton, 1975; Auclair, 1979; Abrahamson & Caswell, 1982; Grimshaw & Allen, 1987; Barhieri & Esteves, 1991; Shardendu & Ambasht, 1991; ElGhareeb, Ayyad & Gaballah, 1991; Henriques & Hay, 1992). They have also tended to focus on relatively narrow problems, such as the relationship between nutrient content of water and plant material in aquatic macrophytes, or the effect of grazing on the nutrient cotitent of pasture species. More general surveys of plant tissue nutrient concentrations have mostly abstracted data from the literature (e.g. Garten, 1976; Duarte, 1992), although large com* To whom correspondence should be addressed.
E-ma.1:
[email protected]
parafive surveys exist (Foulds, 1993). Previous comparative surveys have mostly confined themselves to remarking otT the relative constancy of the ratios between concentrations of particular nutrietits, and the similarities (and differences) between families or other broad groups, such as native and introduced species (Foulds, 1993). Recently, the ptvotal role of mineral tiutrients in the biology of individual plant species has become increasingly obvious (Lambers & Poorter, 1992; Reich, Walters & Ellsworth, 1992). Recent work has also demonstrated the role of eutrophication iti determining commontiess atid rarity in modern European floras (Hodgson, 19866; Thompson, 1994). Here, for the first time, we present data on leaf nutrient concentrations for a substantial fraction of the commoner species in a regional herbaceous flora, and attempt to relate these data to some important plant attributes and community processes, "^ .
,
r
IT
drawmg on data from the Integrated
JO
Screening
680
K. Tho77ipson and others
Table 1. Leaf nutrient concentrations {"„ d.wt), present status ( + , ijicreasifig; —, decreasing) and rank abundance in the herbaceous and dicarf shrub flora of the Sheffield region for the 83 study species Species Adoxa moschotellifia Agrostis capillaris AUiaria petioiata Allium ursinum Anemone nemorosa Angelica svlvestris Anisantha sterilis Anthoxanthum odoratum Anthriscus svlvestris Arabidopsis thaliana Arrhenatheruni elatius Arum maculatum Brachypodium pinnatum Briza media Bromopsis erecta Calhma vulgaris Caltha palustris Campanula rotundifolia Carex flacca Catapodium rigidinn Centaurea nigra Centaurea scabiosa Cerastium fontanum Chanierion angustifoiium Chenopodimir album Chrysosplenium oppositifolium Conopodiiim majus Conyza canadensis Dactylis glomeraia Descliampsia cespitosa Deschampsia flexuosa Digitalis pwpiirea Elytrigia repens Epilobium hirsutum Eriophoriim angustifoiium Eriophorum vaginatum Eestuca ovina Eestuca rubra Eilipendida ulmaria Galium aparine Galium saxatile Geranium roberttanum Hedera helix Helianthemum nummularium Helictotrichon pratense Heraclemn sphondylium Holciis lanatus Holciis mollis Hyacinthoides nonscripta Juncus effusus Juncus squarrosus Koeleria macrantha Lamiastrum galeobdolon Leontodon hispidus Lolium perenne Lotus corniculatus Luzida sylvatica Mercurialis perennis Minuartia verna Myrrhis odorata Nardus stricta Origanum vulgare Oxalis acetosella
Present Rank status abundance -1 f 1 0 0 0 2 1
0 1 4 0 -1 -4 -f -2 0 0 -1 -1 -2 0 1 0 0 0 2 1 1, 1 4 3 2 _2 1
-1 0 0 0 0 0 — -2 -3 1 0
0 0 0 -1 -2 -1 -1
D 1 0 0 -f 1 2
-2 0
251 8 196 148, 122 63 93 37 29 132 9 149 193 120 282 121 246 99 91 353 47 261 17 18 95 136 156 555 5 33 38 157 7 43 222 380 42 4 11 24 101 107 — 290 291 16 2 21 56 40 330 190 119 105 IS 41 436 70 460 419 138 343 114
n
pH
23 28 27 24 25 24 10 10 25 7 28 24 10 10 10 22 20 10 11 7 / 10 10 28 6 19 23 7 26 25 30 15 25 10 6 6 34 29 26 25 21 23 27 8 12 26 24 26 27 24 10 10 21 10 25 11 11 28 6 17 30 10 23
6-3! 4-79 7-03 6-30 5-54 5-75 6-53 4-95 6-58 6-17 6-46 6-41 7-01 6-35 6-67 4-22 6-33 6-15 621 7-24 6-77 6-64 5-37 5-87 6-48 6-22 5-23 6-96 6-07 5-69 3-87 4-25 6-67 6-25 3-57 338 5-33 6-77 6 53 6-52 4-21 6-29 6-30 6-93 6-68 6-80 5-41 5-05 4-93 4-88 3-61 6-63 5-85 6-26 6-89 6-81 5-35 6-25 7-07 6-66 4-34 6-41 5-16
Ca 2-56 0-39 1-83 1-23 2-09 2-06 0-77 0-38 2-25 2-41 0-76 1-08 0-31 042 0-34 0-53 1-43 2-09 0-48 0-68 1-84 3-30 Ml 1-34 1-75 1-42 2-15 2-80 0-38 0-37 0-17 0-97 0-74 2-64 0-16 0-15 5-33 0-47 0-94 2-12 0-91 2-20 1-47 1-76 0-45 1-94 0-57 0-57 0-96 0-15 0-08 0-55 1-07 2-59 0-53 2-45 0-32 2-72 1-75 1-34 0-19 1-84 0-70
K 5-05 2-58 3-79 3-86 2-70 2-96 2-24 1-99 4-23 2-24 2-94 3-23 1-58 2-05 2-16 0-77 3-01 2-46 1-53 076 3-02 2-3 3-51 2-02 6-03 2-51 4-98 2-93 3-29 2-00 1-91 3-74 2-31 1-65 1-28 0-83 0-26 2-00 1-94 5-31 2-76 2-71 1-66 1-21 1-93 3-93 3-45 3-88 4-23 147 1-84 1-81 4-55 3-73 3-55 1-28 2-84 3-33 1-92 3-34 1-25 2-50 2-37
Mg
0-54 0-18 0-39 042 048 046 0-21 0-12 0-32 0-20 019 034 0-13 0-10 0-18 0-17 040 041 0-13 0-20 045 047 0-57 039 1-85 044 0-34 0-51 0-20 0-16 0-11 0-35 0-16 0-35 0-14 0-15 146 0-15 0-57 044 0-22 0-22 0-26 0-34 0-15 042 0-21 0-25 0-25 0-10 0-12 0-10 041 044 0-21 044 0-14 049 0-13 045 0-08 0-30 0-24
N
P
n (Mn)
Mn
3-80 2-33 4-84 4-38 3-33 3-33 2-79 2-04 3-70 3-57 2-87 4-59 147 1-66 148 146 4-71 2-52 1-84 0-90 247 2-23 241 3-34 2-93 2-66 3-86 2-79 243 1-82 1-56 3-78 2-71 3-23 242 216 0-09 1-61 340 3-83 2-51 3-32 2-04 246 1-64 3-61 2-71 2-93 3-59 1-31 1-51 1-81 3-97 245 3-27 2-70 1'95 409 1-81 5-15 1-59 2-24 345
0-33 0-22 0-55 044 0-25 0-36 026 0-15 0-37 049 0-32 0-35 0-12 0-11 0-15 0-11 0-52 0-20 0-14 0-09 0-17 017 045 0-34 043 029 0-28 039 0-28 0-19 0-18 044 0-24 0-34 0-21 0-17 M7 0-22 0-28 043 0-29 0'31 0-19 0-23 0-12 0-38 0-33 0-27 0-29 0-13 0-19 0-16 0-27 0-21 044 0-13 0-20 0-27 0-22 043 0-17 0-17 0-26
23 28 27 24 25 24 — 4 25 — 28 24 5 — 6 22 20 — 5 — 7 — — 28 — 19 23 — 26 25 30 15 25 — 6 6 34 29 26 25 21 23 27 8 12 26 24 26 27 24 10 3 21 1 25 7 11 28 6 17 30 — 23
0-009 0-056 0-009 0-007 0-032 0-016 — 0-038 0-011 — 0-012 0-005 0-007 — 0-020 0-077 0-009 — 0-015 — 0-010 — — 0-021 — 0-017 0-037 — 0-016 0-008 0-049 0-116 0-003 — 0-015 0-023 0-024 0-009 0-016 0-016 0-137 0-012 0-028 0-008 0-015 0-010 0-035 0-058 0-012 0-028 0-019 0-008 0-016 0-010 0-007 0-012 0-027 0-004 0-000 0-008 0-052 — 0-034
Leaf nutrients in a regional flora
681
Table 1. [cont.) Species Pilosella officinarum Plantago lanceolata Poa annua Poa trivialis Potentilla erecta Pteridium aquiliniim Ranunculus ficaria Rumex acetosa Rumex acetosella Sanguisorba minor Scabiosa columbaria Silene dioica Stellaria holostea Stellaria media Thymus polytrichus Tussilago farfara
Present Rank status abundance n -2 0 0 2 -1
1 1 _ J
0 -1 -2 1
-1 1
-1 1
Typha latifolia
1
Urtica dioica Vaccinium myrtillus Viola riviniana
0 0 0
72 20 10 1 98 50 73 26 64 226 325 90 187 25 220 36 177 12 133 76
10 27 25 25 23 30 26 5 10 14 6 26 24 25 10 25 19 28 21 24
Mean Maximum Minimum Max/mm ratio
pH
Ca
K
Mg
N
P
6-18 6-43 5-89 6-61 4-69 5-43 6-05 5-96 4-05 7-12 7-33 5 99 6-09 6-66 6-56 6-66 6-52 6-68 4-06 5-73
0-92 1-87 0-58 0-74 1-47 0-57 1-64 0-70 0-80 2-12 2-23 1-33 0-80 1 20 2-44 3-45 1-13 3-87 0-99 1-04 1-28 3-87 0-08 49-24
2-18 2-09 4-19 4-86 1-97 2-17 4-89 3-11 2-39 1-27 2-31 6-08 4-92 6-59 1-58 3-63 2-24 2-61 M5 4-07 2-81 6-59 0-76 8-66
0-37 0-23 0-21 0-27 0-36 0-24 0-38 0-53 0-37 0-59 0-43 0-49 047 0-73 0-39 0-67 0-25 0-67 0-25 0-57 0-34 1-85 0-08 24-05
1-66 2-05 3-85 3-36 2-17 2-92 4-12 3-37 2-97 2-70 2-12 3-44 2-98 4-17 1-83 2-28 2-50 4-26 1-88 3-50 2-78 5-15 0-90 5-75
0-15 0-22 0-45 0-40 0-18 0-27 0-33 0-32 0-29 0-20 0-14 0-40 0-31 0-59 0-15 0-22 0-28 0-51 0-13 0-24 0-27 0-59 0-09 6-43
n (Mn)
Mn
27 25 25 23 30 26 5 1 14 6 26 24 25
0-005 0-013 0-006 0-107 0-022 0-018 0-033 0-160 0-015 0-006 0-033 0-102 0-040
25 19 28 21 24
0-008 0-101 0-010 0-199 0-027 0-031 0-199 0-000 —
pH refers to mean of surface soil of the sites from which leaves were collected, n = number of sites (note that for some species n for Mn is smaller). These data, plus those for Na, Fe, Zn, Cu and Al can also be found at http://www.shef.ac.uk/~nuocpe/. See text for fuller details.
Programme (Hendry & Grime, 1993; Thompson et al., 1996; Grime et al. 1997) and extensive vegetation surveys of the Sheffield region (Grime, Hodgson & Hunt, 1988). In particular we consider how far other plant traits and community properties can be predicted from a knowledge of leaf nutrient concentrations.
moschatellina), soil (e.g. Scabiosa columbaria, an
extreme calcicole, and Minuartia verna, practically confined to lead mine spoil), geographical distribution (e.g. Bromopsis erecta ~ a southern species) and life history (e.g. Catapodium rigidum - a winter annual). Conyza canadensis, apparently much the least frequent species, is a rapidly spreading alien and is now much commoner than when the surveys were conducted. MATERIALS AND METHODS The number of sites sampled for macronutrient Mature, non-senescent leaves were collected from 83 analysis varied widely between species, but was species, including 77 herbaceous monocots and never less than five. With very few exceptions, dicots, one fern {Pteridium aquilinum), one woody material was collected from the 3000 km' area climber {Hedera helix) and four dwarf shrubs surrounding Sheffield described in Grime et al. {Calluna tmlgaris, Helianthemum nummularium, Thv-(1988). Normally, three separate samples were mus polytrichus and Vaccinium myrtillus). Data are collected from each site, and these three samples presented for all 83 species, but Hedera is excluded were later analysed separately. In a minority of cases, from the analyses. Nomenclature follows Stace particularly where a species was uncommon or (1991). The species studied constitute a substantial produced only small amounts of leaf, a single bulk fraction of the commoner herbaceous and dwarf sample was collected from a site. In a very few cases, shrub species of the Sheffield region. If the her- two samples were collected and analysed separately. baceous and dwarf shruh species present in the At every site, a surface soil sample was also collected region are ranked in order of abundance, based on and its pH determined immediately on return to the extensive survey data (Grime et al., 1988; Hodgson, laboratory. unpublished), then the 82 include eight of the top ten Sample collection began in 1970, and 1388 of the and 29 of the top 50 (Table 1). Many of the less total number of 1555 samples were collected and abundant species were collected in order to include analysed between 1970 and 1979. The remaining 197 greater diversity in terms of taxonomy (e.g. Adoxa samples were collected in 1992 and 1993. During the
682
K. Thompson and others
earlier sampling period, a deliberate effort was made Table 2. Summarv of analysis of ztariance on to collect approx. equal numbers of samples from the concentration of five mineral nutrie7Jts in six species in major acidic and calcareous and upland and lowland ttvo sampling periods geological strata of the Sheffield region. No such Mean effort was made during the second sampling period. d.f. square F Most leaves were collected between June and Element September. Some, chiefly vernals, were collected as Mg early as April, and a very few, in every case evergreen 0-212 24-714 < 0-001 Species 0-033 3-825 0-056 species with long-lived leaves, as late as November. Sampling period 0-055 12-715 < 0-001 Interaction Fresh leaves were immediately dried to constant weight at 80 °C, then finely milled. Early (1970- Ca 7-064 47-596 < 0-001 Species 1979) samples were analysed for N, P, K, Ca, Mg, 0-377 2-541 0-117 Samphng period Na, Fe, Al, Mn, Cu and Zn, whereas later samples 0-025 0-165 0-974 Interaction were analysed only for N, P, K, Ca and Mg. Early K 9-749 < 0-001 1-055 Species samples were digested using a modified Kjeldahl 0-174 1-611 0-21 Sampling period method with hydrogen peroxide and sulphuric acid, 0-131 1-207 0-32 Interaction selenium catalyst and lithium sulphate (Parkinson & P Allen, 1975). N , P and Fe were determined colori0-004 Species 0-603 0-698 metrically using a Technicon autoanalyser, Na and 0 0-025 0-876 Sampling period 0-017 2-707 0-031 Interaction K by flame photometry using a Corning Model 1 50, and Mn, Cu, Zn, Ca, Mg and Al using a Pye Unicam N 2-226 Species 8-597 < 0-001 SP 1900 atomic absorption spectrophotometer. All 0-544 Sampling period 2-101 0-153 methods are documented in Allen et al. (1974). Later 0-611 Interaction 2-361 0-053 samples were analysed for N and P by digesting with a sulphuric acid/salicylic acid mix with a lithium sulphate / copper sulphate catalyst at 350 °C, Table 3. Percentage variance at each taxonomic level followed by N and P determination by flow injection for concentrations of five mineral nutrients analysis in a Tecator autoanalyser. For K, Mg and Ca Mg N K Ca, material was digested at 80 °C m 3 0 % HNO,,, and metals determined by atomic absorption spectro0 Subclass 0 42-7 35-0 7-4 photometry, using a Perkin Elmer 2100. 16Order 0 17-1 14-7 11-6 Family 2-5 35-0 13-4 0-9 18-1 Genus 5-8 15-8 0 0 20-0 RESULTS Species 12-1 15-0 31-1 26-1 Site 27-9 44-4 23-0 18-6 37-4 Table 1 summarizes leaf concentrations of N, P, K, Mg, Ca and M n for each species and the mean soil pH of the sites from which leaves of each species macrantha and Lotus corniculatus, with a total of 30 'old' samples and 32 ' n e w ' samples. Analysis of were collected. Of the major macronutrients, K and N were usually present in the greatest concen- variance of the elements common to the two trations, and P the least. Ca and Mg were in- sampling periods (N, P, K, Ca and Mg) reveals no termediate. Concentrations of K, N and P showed significant main effect of sampling period, although species x sampling period interaction is similar levels of interspecific variability, with the the highest concentrations being 6—9 times the lowest. significant for Mg and P (Table 2). We have therefore Mg and (especially) Ca were much more variable, amalgamated the two sets of data. with the highest concentrations being 24 and 49 times the lowest respectively. The taxonomic and Variation within and between taxa and sites environmental sources of this variability are explored later. Concentrations of Mn and Fe, Na, Cu, Al and Mineral nutrient data were analysed using nested in Zn (data not shown) were all highly variable; in analysis of variance ( N E S T E D function M I N I T A B 11), in order to determine the taxonomic every case one or more species had levels of these elements which v\ere either very low or below the level at which the variation occurs in each of the five major macronutrients. An extra 'site' level below threshold of detection. species was included to estimate variance between sites within species. Pteridium aquilinum was not Time of collection and analysis methods included in this analysis. In every case a substantial Most of the samples collected in the 1970s and the proportion of the variance is accounted for by 1990s concern different species, but there were differences between collection sites (Table 3). Otherapprox. equal numbers of samples from the two wise, however, the elements differ greatly. Only for periods for Anthoxanthum odoratum, Brachypodium P is the great majority of the variance ( > 7 5 % ) pinnatum, Bromopsis erecta, Carex fiacca, Koeleria found at the species level or below, although in N
Leaf nutrients in a regional flora
683
Table 4. Correlation coefficients between surface soil pH and element concentrations in leaves .411 species (w= 1528) pH Mn Zn Cu
Al Na Fe K N P Mg Ca
_04]*** -0-14 0-03 0-05 0-10 0-12 0-13 0-15 0-18 0-21* 0-35***
Monocots (n = 621)
Dicots [n = 907)
pH
pH
--0-30** -0-15 0-00 0-18 0-11 0-20* 0-10 0-14 0-21* 0-19 0-43***
hut monocots and dicots show different patterns of Ca and Mn concentration in relation to soil pH (Fig. 1). Both groups tend to accumulate more of both cations when they are available in excess in the soil, hut maximum concentrations in the leaves of dicots are much greater. Monocots appear to maintain relatively low concentrations of both elements over the whole pH range. The large proportion of variance in Mg and, particularly, Ca concentrations found at the subclass level is consistent with this very different hehaviour of monocots and dicots in relation to soil pH. The isolated, exceptionally high concentrations of Mn among the monocots in Figure 1 are both samples of Typha latifolia and, surprisingly, occur at high soil pH. This analysis means that only P concentration can legitimately he considered a species trait. Therefore, where we relate element concentrations to some other ecological variables, including growth rate and present status, we restrict our analysis to P. However, in e\'ery case where significant relationships exist with P concentration, similar relationships exist with N concentration. Variation hetween sites can be investigated on several different levels. Coefficient of \ariation for N and P was not related to number of sites sampled, i.e. mean element concentrations deri\^ed from sampling at 5—10 sites were usually not significantly less reliable than those derived from sampling at many more sites. A related question is whether element concentrations were much infiuenced by sampling date.
-0-51*** -0'20* 0-04 -0-05 0-10 0-04 0-11 0-11 0-13
0-20* 0-38***
***P < 0-001, **P < 0-01, *P < 0-05. and K close to half the variance is also found at this level. The pattern in Ca and Mg is quite different; here a substantial fraction of the variance is found at the subclass level, i.e. between monocots and dicots. An analysis of the relationship between leaf element concentrations and soil pH re\'eals the hasis of this difference hetween subclasses. There were few significant correlations between soil pH and leaf element concentrations across all species and sites in the study (Table 4). Only Ca and Mn were consistently associated with soil pH, positively and negatively respectively. These relationships hold for all species and for monocots and dicots separately. Dicots
Monocots
7 6 5 4 to CJ
o
3 2 1 0
6
1>
10
Dicots
Monocots
0-6
0-6
0-5
0-5 0-4
a 0.3
03
O O
0-2 0.1 00
0-1 0-0 10 2 4 6 pH of surface soil at collection site
Figure 1. Relationship between Ca and Mn concentrations and surface soil pH in individual collections of monocot and dicot leaves. For statistics see Table 4.
684
K. Thompson and others Table 5. Mean N, P and K coticentrations of leaves of three contrasted species sampled between May and October. Sig. is significance of a one-way ANOVA of difference between monthly means Species element
June
July
Aug.
Sept.
(3)
(12)
(H)
(2)
2-54 0-32
2-81 0-12
3-08 0-22
2-56 0-63
n.s
Mean
0-28 0-02
0-24 0-02
0-42 0-04
0-29 0-05
*#
Mean
2-87 0-13
2-66 0-14
3-34 0-25
2-50 1-00
n.s.
(1)
(4)
(7)
(8)
(2)
1-47 —
1-46 0-04
1-42 0-07
1-46 0-06
1-58 0-00
n.s
0-11 —
0-12 0-01
Oil 0-01
0-12 0-01
0-11 0-02
n.s.
0-76 —
0-73 0-03 (11)
0-79 0-05
0-99 0-25
n.s.
May
Arrhenatherum elatiiis (n)
Oct.
Sig.
K Mean SE T3
r SE
K SE
Calluna vulgaris (w) N Mean SE
p r
Mean SE
K Mean (1)
(3)
0-70 0-01 (11)
4-74 —
3-97 0-46
3-25 0-25
3-17 0-19
3-14 0-16
n.s
0-44 —
0-45 0-05
0-32 0-04
0-33 0-03
0-29 0-00
n.&.
2-60 —
2-14 0-30
179
2-14 0-19
2-20 0-00
n.s.
0-12
SE
Chamerioti angiistifolium (n) N Mean SE
(2)
P Mean SE
K Mean SE
***P< 0-001, **P < 0-01, * P < 0-005. Samples were not collected with this question in mind, and for many species were collected over a rather short period in spring or summer. However, in Tahle 5 we present data for N, P and K concentrations for three contrasted species collected over at least 4 months. In Calluna vulgaris and Chamerion angustifolium there was no significant difference in N, P or K concentrations in leaves collected at different times from May to October. In Arrhenatherum elatius P concentrations appeared to be higher in August than in other months, although N and K did not. There was a tendency for N and P concentrations to decline during the growing season in Cha7nerion, hut differences hetween July and August, when most samples were collected, were insignificant. On balance, it appears that variation between sampling dates contributed rather little to overall variation between species. Another approach to hetween-site variability is to compare element concentrations in samples from the five major geological strata represented in the region. Of these, Carhoniferous Limestone (CL) is basic and predominantly upland. Millstone Grit (MG) is acid and predominantly upland, Magnesian Limestone (ML) is basic and lowland, Bunter Sandstone (BS) is
acid and lowland, and the Coal Measures (CM) are acid and intermediate in altitude. These differences in base status and altitude are also associated with major difference in land use (Hodgson 1986a). For the 12 species for which we have at least five samples from each of these substrates, we examined the correlation between mean N and P concentrations across substrata (Table 6). Both elements are significantly correlated between all pairs of substrata, i.e. species with relatively nutrient-rich (or poor) leaves on one substrate also tend to be nutrient rich (or poor) on all other substrates, irrespective of soil pH, altitude and land use. This is moderately convincing evidence that leaf nutrient concentrations (with the notable exceptions of Ca and Mn) are predominantly species attributes, and that species rankings are not much affected by environmental variation. A caveat to this conclusion is that the 12 species analysed in Table 6 are among the relatively few species which could he sampled from several sites on all strata. The majority of species, for reasons of soil pH, climate or land use, are much more narrowlydistributed. This is merely one example of the way in which phylogeny and distribution are confounded at
Leaf nutrients in a regional flora
685
Table 6. Correlation coefficients between N and P concentrations of 12 species f sampled from five geological substrata Bunter sandstone Nitrogen Carboniferous limestone Coal measures Millstone grit Magnesian limiestone Phosphorus Carboniferous limestone Coal measures Millstone grit Magnesian limestone
0^66* 0-77** 0^66* 0-74** 0.77** 0-89*** 0.79** 0-82***
Carboniferous limestone
0-84*** 0.78** 0-82***
Coal measures
]\lillstone grit
0^58* 0^63*
0-80**
0-76** 0-83***
0-79**
0-85*** 0-87*** 0-79**
***P < 0-001, **P < O-Ol, * P < 0^05. t Agrostis capillaris, Dactylis glomerata, Lolium perenne, Chamerion angustifoiium. Anemone nemmosa, Holcus mollis, Hyacinthoides non-scripta, Stellaria medm, Anthriscus sylvestris, Poa annua, Galium aparine, Urtica dioica.
all levels. One way of illustrating this is to compare element concentrations in the ten congeneric pairs in Table 1 (Table 7). Mean ratios of concentrations of P, K and especially N are close to unity (compare ratios for whole dataset in Table 1). However, it is not possible to tell how much of this similarity is attributable to fundamental similarities in nutrient economy, and how much is due to congeners sharing similar habitats. The pairs where the congeners occupy the most dissimilar habitats (Galium and Stellaria) also have the most dissimilar nutrient concentrations. The sur\'ey was not designed to examine whether groups of species responded similarly to site variables, but we can identify small groups of species which were all sampled from the same pairs of sites. Data for seven common grassland and six commoti woodland species (Table 8) show that, at the level of individual sites, element concentrations in individual species are not predictable. For the grassland species, the direction of variation in N, P and K concentrations hetween sites appeared to be more or less arbitrary. In the six woodland species, most element
concentrations varied in the same direction, but even here one species (Hyacinthoides noti-scripta) varied in the opposite direction to the majority. Element ratios Strictly speaking, the pattern of phylogenetic variation of element concentrations does not permit us to comment on element ratios at the species level. However, the strong correlation between concentrations of all elements, and in particular between N and P (Table 9) is consistent with the findings of other surveys (Garten, 1976; Duarte, 1992). Since this correlation is also found in groups as diverse as phytoplankton and conifers, it seems fair to conclude that it is largely independent of phylogeny. Across a broad range of species, the P ; N ratio is close to 1 ; 10 Table 8. Ratios of N, P and K concetitrations between tivo pairs of sites, one from which seven grassland species were sampled, and one from which six ivoodland species ivere sampled Between-site ratios
Table 7. Ratios of mean leaf coticentrations of N', P and K between 10 congeners : species listed first have higher N concentration Genus
Species
Galium aparine / saxatile Stellaria media/holostea Deschampsia cespitosa/flexuosa Juncus squarrosus / effusus Poa annua / trivialis Rumex acetosa / acetosella Eriophorum angustifoiium/vaginatum Centaurea nigra/scabiosa Eestuca rubra/ovina Holcus mollis/lanatus Mean of congeneric pairs
N
I
]
1^52 1^40 M7 M5
1•50 1•93 1•07 1•47 1•13 1 •12 1 •21
1•92
M4 M3 M2 Ml
] •34
1^05 I •25
1 •16 1 •30 1 •54 •03 1•19 MO ] . 32 •37 1^08 1^ 20 1 •12 M9
! •
30
1
•32
N
P
K
0^82 \-23 146
0^51 0^90
1^24
1^05 1^26 0^98 0^86
0^96 0^97 1^36 0^85
Grassland spp. Agrostis capillaris Arrhenatherum elatius Dactylis glomerata Elytrigia repens Festuca rubra Holcus lanatus Plantago lanceolata Woodland spp. Allium ursinum Anemone nemorosa Arum maculatum Hyacinthoides non-scripta Lamiastrum galeobdolon Mercurialis perennis
\-3\ 2-\5
M5 113 121 MO 0^96 \-2(i
M2
ro7
b54
0^84 1^36 0^68 1^40 1^29
M7
1^28 0^73 \-Q9 0^56 r95 0^51 1^41 1^20
686
K. Thompson and others
Table 9. Correlation coefficients bettveen concentrations of individual elements across all species
r=0-55 P< 0-001
2-5
Ca K Mg X p
0-23* 0-51*** 0-38*** 0-27**
0-52*** 0-64*** 0-67***
Mg
N
0-39*** 0-41***
0-82***
1
1-5 X ro
K
**#p < 0-001, **P < 0-01, *P < 0-005, n = 1528. 1 0-5 0-1 0
0-2 0-3 0-4 Leaf P (% d. wt)
0-5
0-6
Figure 3. Relationship between maximum potential relative growth rate and leaf P concentration in 59 herbaceous and dwarf shrub species. point scale, where 5 is very common and characteristic of the particular habitat (°o frequency > 4 times that in the survey as a whole) and 1 is largely or completely absent from the habitat ( °o frequency > 0-25 times that in the survey as a whole); further details are in Grime et al. (1988). Plants of arable 1 2 3 4 5 6 habitats and, to a lesser extent, of woodland are Leaf N (% d. wt) generally rich in P, while plants of skeletal habitats Figure 2. Relationship between leaf N and P concen- and pasture are generally nutrient-poor (Table 10). trations in 82 herbaceous and dwarf shrub species {r = Since most pasture quadrats were found in upland, 0-82. P < 0-001). infertile parts of the survey area (the lowlands are principally arable), the latter correlation could well be an artefact of the survey area; different results (Fig. 2), hut this general relationship conceals much would almost certainly arise from a survey of interspecific variability. The ratio varies from 1 : 5 to lowland, improved pasture. 1:15 among non-legumes, with Lotus corniculatus, The arahle habitat shows much the strongest the only legume in the study, having an extreme ratio of 1:20. We consider variation in N : P ratio in relationship between habitat affinity and N : P ratio (Table 10). The relationship is negative; plants relation to habitat later. which are restricted to arable habitats, despite being rich in N, have low N : P ratios. Presumably the very Plant ecological attributes high relative concentrations of P in such plants are a For 59 of the 83 species studied here, estimates of consequence of the use of artificial fertilizers. This maximum potential relative growth rates (R^^j^,.) are pattern suggests that plant growth in other habitats available from Grime & Hunt (1975). Phosphorus might be limited mainly hy P availability, parconcentration is strongly correlated with R,^ax (Fig. ticularly in pasture, affinity for which is positively 3). Integrated Screening Programme data have correlated with N : P ratio (Table 10). revealed many other significant correlations hetween macronutrient concentrations (particularly P and N) Present status and, for example, leaf toughness and leaf longevity (negative) and palatability and rate of litter de- Fast-growing plants of fertile habitats are generally composition (positive). (Grime et al., 1996, 1997). increasing in the densely-populated countries of western Europe, while slow-growing plants of infertile habitats are generally declining in abundance Habitat preferences (Hodgson, 1986 a, 6; Thompson, 1994; Thompson, A measure of habitat preference can be obtained Hodgson & Rich, 1995). This polarization of the from extensive surveys of central England (Hodgson fiora into a declining, slow-growing component and et al., 1995). Affinities for skeletal, arable, pasture, an expanding, fast-growing component is very spoil, wasteland, woodland and wetland habitats for marked in densely-populated countries such as individual species are derived from a survey of over England and The Netherlands, but only slight in 10000 quadrats. Affinities are expressed on a five- sparsely-populated countries such as Scotland and
Leaf nutrients in a regional flora
687
Table 10. Spearman rank correlation coefficients betiveen element concentrations, N:P ratio and habitat affinity scores derived from extensive surveys of central England Ca
Arable 0-11 Wood 0-10 Waste 0-11 Wetland -0-24* Spoil 0-17 Skeletal 0-04 Pasture -0-29**
K
Mg
N
P
0-45*** 0-22* 0-40*** 0-59*** 0.47*** 0-24* 0-58*** 0-38*** 0-22* 0-13 0-11 0-06 -0-09 -0-12 0-03 0-08 -0-16 -0-01 -0-30** -0-09 -0-19 -0-15 -0-32** -0-32** -0-37*** -0-29** -0-44*** -0-49***
N:P -0-49*** 0-17 0-05 -0-11 -0-27* 0-12 0-24*
***P < 0-001, **P < 0-01, *P < 0-05. n = 82.
+ 5 for species which are increasing markedly everywhere to — 5 for species which are decreasing markedly everywhere (Tahle 1). Plants with P-rich lea\-es tend to he expanding and those with P-poor leaves to he declining (Fig. 4). Note that the 'present status' data in Tahle 1 record changing abundance on a very coarse scale indeed, and are particularly poor at detecting increases in species which are already very widespread. Thus scores of zero for Urtica dtoica, Cerastium fontanuni and Poa annua, for example, should not be taken as e\-idence that these species ha\'e not recently changed in abundance.
0-35 -
g
0-25 o o
a. 0-15 -
>2
Figure 4. Relationship between leaf P concentration and an index of current status based on recent patterns of increase or decrease in England, the Netherlands and western Germany. Bars are SE.
Ireland. We have therefore interpreted this pattern as a consequence of increasingly disrupti^'e human land use, which favours fast-growing species tolerant of eutrophication and disturbance (Thompson, 1994). Species which have recently increased or decreased in England can he identified from the Botanical Society of the British Isles Monitoring Scheme (Palmer & Bratton, 1995 ; Rich & WoodrufT, 1996). In the Netherlands and western Germany (Dutch Centraal Bureau Voor de Statistiek and Ellenberg et al. (1992) respectively), more detailed recording allows us not only to identify species which have recently increased or decreased, hut also to differentiate between species which have either slightly or markedly changed in abundance. We have scored species on the hasis of their current status in these three countries: species increasing in England score + 1 , those markedly increasing in the Netherlands score + 2 , those slightly increasing there score + 1, and similarly for western Germany. An identical system of negative scores is used for decreasing species, while species not changing in abundance score zero. The sum of these three scores provides a single species score, which can range from
DISCUSSION
These results confirm leaf nutrient concentration as one of the constellation of causally interrelated traits identified by Poorter (1989), Reich et al. (1992), Schlapfer & Ryser (1996), Nielsen et al. (1996), Grime et al. (1997) and others, comprising leaf longevity, SLA, photosynthetic rate and growth rate (among others), which are central to the ecology of plants, communities and ecosystems. Plants with nutrient-rich foliage grow quickly, dominate nutrient-rich ecosystems, and are generally increasing as a result of the eutrophication and disturhance arising from human exploitation. The pervasive infiuence of leaf nutrient concentrations extends right through ecosystems via their influence on palatahility and rates of litter decomposition (Aerts, 1995; Cornelissen, 1996; Grime et al., 1996; Cornelissen & Thompson, 1997). At the hroad geographic scale, the nutrient concentrations reported here are similar to those reported for herbs in many other sur\-eys (e.g. 1-4 °o N and 0-1-0-82 °o P in freshwater angiosperms: Garten, (1976); < l - 4 % N and < 0-1-0-6 °o P in \'ascular plants, lowest \-alues from wood}- species: Duarte (1992)). Nutrient concentrations in the leaves of trees and shrubs, however, are often lower (Woodwell et al., 1975; Specht & Moll, 1983). Some whole fioras seem to he relatively impoverished in nutrients. Concentrations of N and P in nati\'e Australian herhs (mean M °o N and 0-11 "o P) were
K. Thompson and others similiar to the lowest levels reported here, whereas concentrations in 'oligotrophic' families such as Proteaceae and Restionaceae (mean 0'7 °o N and 0-05'\, P) were below the minimum values we measured (Foulds, 1993). Australian species were also notably poor in Cu and Zn. Some of these low nutrient concentrations might represent genetic adaptation to chronically low nutrient availability, and as such are compensated for by high nutrient use efficiency through long leaf life-spans, but they are at least partly imposed directly by the environment. Introduced herbs had the highest nutrient levels (mean 1-3 "„ ^^^' and 0-16 "„ P), but even these were well below our mean levels. It seems clear that on a world scale, the leaves of British herbs contain moderate to high nutrient concentrations. It IS conventional to place great emphasis on the N status of plants and habitats, but the results of this study do not allow us to make much functional distinction between N and P. These two nutrients are closely correlated with each other, and with all the other variables described above. None of this should be surprising; the key role of both elements in metabolism means that they must maintain a relatively constant relationship, and published evidence suggests that this close relationship is universal throughout vascular plants, bryophytes and algae (Garten, 1976; Duarte, 1992). However, the fastgrowing weeds of arable fields were found to have unusually high P : N ratios. This is circumstantial evidence that in the modern British landscape, with its considerable anthropogenic N input direct from the atmosphere, plant growth might be increasingly limited by P availability. Leaf nutrient concentrations are strongly infiuenced by habitat and phylogeny, which are themselves highly correlated. Taxa (e.g. Chenopodium, Urticaceae) which are restricted to fertile, lowland habitats tend to have nutrient-rich foliage, whereas taxa (e.g. Cyperaceae, Ericaceae) which are mostly confined to upland, infertile habitats have nutrient-poor foliage. Similarities in leaf nutrient concentrations between related taxa (e.g. congeners) are thus a reflection parth' of similarities in biology and partly of restriction to similar habitats. Controlled experiments, or detailed measurements of nutrient availability of soils from which samples were collected, would be necessary in order to explore further the relative importance of habitat and phylogeny. Nevertheless, comparisons across geological strata confirm that rankings of nutrient concentrations in widespread species are remarkably consistent across very marked gradients of altitude and land use. Broadly speaking, therefore, leaf nutrient concentrations are essentially a species (or higher taxon) trait. There is, however, one instance where soil conditions exert an important and direct influence on mineral nutrient concentrations in leaves. Leaf Ca
and Mn concentrations are profoundly influenced by soil acidity, and this effect interacts strongly with phylogeny (Fig. 1; Table 4). All plants take up more multivalent cations when they are present in excess supply, but this effect is far more pronounced in dicots. This apparent inability to restrict uptake of multivalent cations is reflected in the generally lower tolerance of toxic levels of soil Fe in dicots (Snowden & Wheeler, 1993).
.'ACKNOWLEDGEMENTS Thanks are due to the numerous individuals who helped to collect plant material, and in particular to Phil Grime, Philip Lloyd and John Hodgson. .A.n earlier draft of this paper benefited from the connments of Alastair Fitter, Hans Cornelissen and two anonymous referees. The work was supported by the Natural Environment Research Council.
REFERENCES Abrahamson WG, Caswell H. 1982. On the comparative allocation of biomass, energy and nutrients in plants. Ecology 63: 982-991. Aerts R. 1995. The advantages of being evergreen. Trends in Ecology and Evolution 10: 402^07. Allen SE, Grimshaw HM, Parkinson JA, Quarmby C. 1974. The chemical analvsis of ecological materials. Oxford : Blackwell Scientific Publications. Auclair AND. 1979. Factors affecting nutrient concentrations in a Scirpus - Equisetum wetland. Ecology 60: 337-348. Barbieri R, Esteves FDA. 1991.The chemical composition of some aquatic macrophyte species and implications for the metabolism of a tropical lacustrine ecosystem: Lobo Reservoir, Sao Paulo, Brazil. Hydrobiologia 213: 133-140. Cornelissen JHC. 1996. .A.n experimental coinpanson of leaf decomposition rates in a wide range of temperate plant species and types. Journal cif Ecology 84: 573-582. Cornelissen JHC, Thompson K. 1997. Functional leaf attributes predict litter decomposition rate in herbaceous plants. Neiv Phytologist 135: 109-114. Duarte CM. 1992. Nutrient concentration of aquatic plants: patterns across species. Lim7wlogy and Oceanography 37: 882-889. El-Ghareeb R, Ayyad MA, Gaballah MS. 1991. Effect of protection on the nutrient concentration and uptake of some Mediterranean desert annuals. Vegetatio %: 113-126. Ellenberg H, Weber HE, Diill R, Wirth V, Werner W, PauliBen D. 1992. Zeigwerte von Pflanzen in Mitteleuropa. Scripta Geobotanica 18. Gottingen: Erich Goltze. Foulds W. 1993. Nutrient concentrations of foliage and soil m South-western Australia. New Phytologist 125: 529-546. Garten CT. 1976. Correlations between concentrations of elements in plants. Nature 261: 686-688. Grime JP, Cornelissen JHC, Thompson K, Hodgson JG. 1996. Evidence of a causal connection between anti-herbivore defence and the decomposition rate of leaves. Oikos 77: 489-494. Grime JP, Hodgson JG, Hunt R. 1988. Comparative plant ecology: A functional approach to common British plants. London: LJnwin Hyman. Grime JP, Hunt R. 1975. Relative growth rate: its range and adaptive significance in a local Bora. Journal of Ecologv 63: 393-422. Grime JP, Thompson K, Hunt R, Hodgson JG, Cornelissen JHC, Rorison IH, Hendry GAF, Ashenden TW, Aske-w AP, Band SR, Booth RE, Bossard CC, Campbell BD, Cooper JEL, Davison AW, Gupta PL, Hall W, Hand DW, Hannah MA, Hillier SH, Hodkinson DJ, Jalili A, Liu Z, Mackey JL, Matthe-ws N, Mowforth MA, Neal AM, Reader RJ, Reiling K, Ross-Fraser AM, Spencer RE, Sutton F, Tasker DE,
Leaf nutrients in a regional flora
689
Thorpe PC, Whitehouse J. 1997. Integrated screening valiCauses and Consequences of Variation in Grozcth Rate and dates primary axes of specialisation in plants Oikos 79 • Productivity in Higher Plants. The Hague: SPB Academic 259-281. Publishing, 49-68. Grimshaw HM, Allen SE. 1987. Aspects of the mineral nutrition Reich PB, Walters MB., Ellsworth DS. 1992. Leaf life-span in of some native British plants - inter..site variation. Vegetatw 70 • relation to leaf, plant, and stand characteristics among diverse 157-169. ecosystems. Ecological Monographs 62: 365-392. Hendry GAF, Grime JP. 1993. Methods in comparative plant Rich TCG, Woodruff ER. 1996. Changes in the vascular plant ecology. London: Chapman and Hall. floras of England and Scotland between 1930-1960 and Henriques RPB, Hay JD. 1992. Nutrient content and the 1987-1988: the BSBI Monitoring Scheme. Biological Consstructure of a plant community on a tropial beach-dune system ervation 75: 217-229. in Brazil. Acta Oecologica 13: 101-117. Schlapfer B, Ryser P. 1996. Leaf and root turnover of three Hodgson JG. 1986 a. Commonness and rarity in plants with ecologically contrasting grass species in relation to their special reference to the Sheffield flora. I. The identity, performance along a productivity gradient. Oikos 75: 398^06. distribution and habitat characteristics of the common and rare Shardendu, Ambasht RS. 1991. Relationship of nutrients in species. Biological Conservation 36: 199-252. water with biomass and nutrient accumulation of submerged Hodgson JG. 19866. Commonness and rarity in plants with macrophytes of a tropical wetland. Neu- Phvtologist 117: special reference to the Sheffield flora. II. The relative 493-500. importance of climate, soils and land use. Biological ConserSnowden RED, Wheeler BD. 1993. Iron toxicity to fen plant vation 36: 253-274. species. Journal of Ecology 81: 35—1-6. Hodgson JG, Grime JP, Hunt R, Thompson K. 1995. The Specht RL, Moll EJ. 1983. Mediterranean-type heathlands and electronic comparative plant ecologv. London: Chapman and sclerophyllous shrublands of the world: an over\-iew. In: Hall. Kruger FJ, Mitchell DT, Jar\is JL'M, eds. Mediterranean-type Lambers H, Poorter H. 1992. Inherent variation in growth rate ecosystems. Berlin: Springer ^'erlag, 41-65. between higher plants: a search for physiological causes and ecological consequences. Advances in Ecological Research 23 • Stace C. 1991. New flora of the British Isles Cambridge: Cambridge University Press. 188-242. Thompson K. 1994. Predicting the fate of temperate species in Nielsen SL, Enriquez S, Duarte CM, Sand-Jensen K. 1996. response to human disturbance and global change. In: Boyle Scaling maximum growth rates across photosynthetic organisms. Functional Ecology 10: 167-175 TJB, Boyle CEB, eds. Biodiversity, Temperate Ecosystems and Global Change. Berlin: Springer \'erlag, 61-76. Palmer MA, Bratton JH. 1995. A sample survey of the flora of Britain and Ireland. Peterborough: Joint Nature Conservation Thompson K, Hodgson JG, Rich TCG. 1995. Native and alien Committee. invasive plants: more of the same? Ecographv 18: 390-402. Parkinson JA, Allen SE. 1975. A wet oxidation procedure Thompson K, Hillier SH, Grime JP, Bossard CC, Band SR. suitable for the determination of N and mineral nutrients in 1996. A functional analysis of a limestone grassland community. biological material. Communications in Soil Science and Plant Journal of Vegetation Science 7: 371-380. Analysis 6: 1-11. Woodwell GM, Whittaker RH, Houghton RA. 1975. Nutrient Poorter H. 1989. Interspecific variation in relative growth rate. concentrations in plants in the Brookhaven oak-pine forest. In: Lambers H, Cambridge ML, Konings H, Pons T L , eds. Ecology 56: 318-332.
