Brett H. Robinson1, Robert R. Brooks1, John H. Kirkman1, Paul E.H. Gregg1 and Hugo Varela Alvarez2
1Department of Soil Science, Massey University, Palmerston North, New Zealand� and 2Computing Services Department, Massey University, Palmerston North, New Zealand
Received 18 December 1995. Accepted in revised form 16 September 1996
Key words: edaphic influences, principal components analysis, serpentine flora
关键词：土壤因素影响，主成分分析，蛇纹岩植物区系
1Abstract摘要
Ultramafic (“serpentine”) soils from and adjacent to the Dun Mountain Ophiolitic Belt, South Island, New Zealand were analysed for 11 elements in order to establish to what degree edaphic factors influenced the character of the overlying vegetation. Using principal components analysis with a mutual plot of the first two principal components, involving the total elemental concentrations in the soils, it was possible to divide the soils into 6 virtually nonoverlapping fields, each of which represented a specific vegetation community. Component 1 was essentially an “ultramafic plot” with heavy loadings from the elements chromium, cobalt, iron, magnesium, manganese, and nickel. Component 2 was a “non ultramafic” plot with heavy loadings from aluminium, copper and zinc. For elements extracted from the soils at pH 5.9, discrimination was somewhat poorer but confirmed the great importance of magnesium and nickel as controlling elements for the serpentine vegetation. It was concluded that the results indicated the overriding importance of edaphic factors in controlling the serpentine vegetation.
对新西兰南岛Dun Mountain蛇绿岩带及其邻近地区的超基性（“蛇纹岩”）土壤进行了11种元素的分析，以确定土壤因素在多大程度上影响其上覆植被的特征。基于土壤中元素全量，采用主成分分析并绘制前两个主成分的联合散点图，可将土壤划分为6个几乎互不重叠的区域，每个区域代表一个特定的植被群落。成分1实质上是一个“超基性”成分，铬、钴、铁、镁、锰和镍在其上具有较高载荷；成分2为“非超基性”成分，铝、铜和锌在其上具有较高载荷。对于在pH 5.9条件下从土壤中提取的元素，判别效果略差，但证实了镁和镍作为蛇纹岩植被控制元素的重要性。研究认为，结果表明土壤因素在控制蛇纹岩植被方面具有首要作用。
2Introduction引言
Ultramafic rocks cover a little less than 1% of the Earth and owe their origin to surface processes or metamorphism within the mantle far below the Earth’s surface (Coleman and Jove, 1992). Soils derived from these rocks are rich in iron and relatively so in chromium, nickel, and cobalt. They are also very rich in magnesium and relatively deficient in calcium. Nutrients such as potassium, phosphorus and nitrogen have very low concentrations.
超基性岩覆盖地球表面略少于1%，其成因可归于地表过程或地表以下深处地幔内的变质作用（Coleman and Jove, 1992）。由这类岩石发育的土壤富含铁，铬、镍和钴含量也相对较高；同时镁含量很高，而钙相对缺乏。钾、磷和氮等养分浓度极低。
There has been a long history of attempts to solve the “serpentine problem,” i.e. the infertility of serpentine soils (Brooks, 1987; Kinzel, 1982; Krause, 1958) but there appears to be no general agreement as to the causes of this infertility. Proctor and Nagy (1992) have however pointed out that whatever their variety, serpentine soils are always a major factor that controls the vegetation that they bear. Although physical and other factors may control serpentine floras, possible edaphic influences can be summarised as follows:
长期以来，人们一直试图解决“蛇纹岩问题”，即蛇纹岩土壤的贫瘠性（Brooks, 1987; Kinzel, 1982; Krause, 1958），但对于其贫瘠的原因似乎尚未形成共识。然而，Proctor and Nagy (1992)指出，无论蛇纹岩土壤属于何种类型，它始终是控制其上生长植被的主要因素。尽管物理因素及其他因素也可能控制蛇纹岩植物区系，但可能的土壤因素影响可归纳如下：
(1) Calcium and magnesium. Calcium deficiency per se was emphasized as an important factor in controlling a serpentine flora (Kruckeberg, 1954). The prevailing view is however that calcium levels in serpentine soils should be considered in relation to those of other elements such as magnesium. The role of magnesium is still unclear because of its complex inter-relationships with other elements such as its ability to reduce uptake of potassium by plants (Epstein, 1972), and to ameliorate nickel toxicity (Proctor and McGowan, 1976).
（1）钙和镁。钙缺乏本身曾被强调为控制蛇纹岩植物区系的重要因素（Kruckeberg, 1954）。但目前的主流观点认为，蛇纹岩土壤中的钙水平应结合镁等其他元素的水平加以考虑。由于镁与其他元素之间存在复杂的相互关系，例如镁能够降低植物对钾的吸收（Epstein, 1972）并缓解镍毒害（Proctor and McGowan, 1976），其作用至今尚不明确。
(2) Nickel. Nickel was once thought to have a dominant role in the control of serpentine floras but later evidencetends to negate this supposition. For example, Lee (1992) in a study of New Zealand serpentine floras concluded that only nickel in serpentine soils from the southwest of South Island was likely to reduce plant growth critically.
（2）镍。镍曾被认为在控制蛇纹岩植物区系方面起主导作用，但后来的证据倾向于否定这一假设。例如，Lee (1992)在对新西兰蛇纹岩植物区系的研究中得出结论：仅南岛西南部蛇纹岩土壤中的镍可能会严重抑制植物生长。
� FAX No: +6163505632. E-mail: R.Brooks@massey.ac.nz (3) Chromium. There is general agreement that chromium will have little effect on controlling a serpentine flora because of its lack of mobility. Elevated chromium levels in plant material are invariably associated with wind-borne surface contamination.
(4) Cobalt. Unlike chromium, cobalt is readily accumulated by plants. Indeed, the copper/cobalt deposits of Za¨ıre are host to nearly a score of species that are able to hyperaccumulate this element (Brooks et al., 1992). Despite this ability to accumulate cobalt, concentrations of this element in serpentine floras usually do not exceed 100 �g g �1 in dry tissue.
（4）钴。与铬不同，钴易被植物积累。事实上，扎伊尔的铜/钴矿床上生长着近20种能够超富集该元素的植物（Brooks et al., 1992）。尽管植物具有积累钴的能力，蛇纹岩植物区系中该元素的浓度通常不超过100 μg g−1（干重）。
(5) Low concentrations of nutrients. The concentrations of plant-available nutrients in serpentine soils are probably one of the main contributors to the “serpentine problem.” For example, in experiments on the Isle of Rhum (Scotland) and on the Keen of Hamar (Shetland), nutrient additions increased plant cover to a spectacular degree (Proctor and Nagy, 1992).
（5）养分浓度低。蛇纹岩土壤中植物有效养分的浓度可能是造成“蛇纹岩问题”的主要原因之一。例如，在Rhum岛（苏格兰）和Keen of Hamar（设得兰群岛）开展的试验中，添加养分使植被盖度显著增加（Proctor and Nagy, 1992）。
Some mention should also be made of physical factors such as water shortage due to the good drainage of serpentine soils. For example, Proctor and Craig (1978) found that drought was responsible for much of the lack of tree cover over The Great Dyke, Zimbabwe. They found well developed riverine forest over this body.
此外还应提及物理因素，例如蛇纹岩土壤排水良好所导致的缺水。例如，Proctor and Craig (1978)发现，津巴布韦大岩墙（The Great Dyke）上树木覆盖稀少在很大程度上是干旱所致。他们发现该岩体上发育有良好的河岸林。
We have recently carried out an extensive ecological study of the flora of the Dun Mountain Serpentine Belt in the north of South Island, New Zealand, in the course of which soils were analysed and their composition related to plant communities growing over them. The purpose of the present paper is to report these findings in which statistical analysis of the data has enabled us to propose a model for edaphic factors that control the serpentine flora.
我们近期对新西兰南岛北部Dun Mountain蛇纹岩带的植物区系开展了广泛的生态学研究，其间对土壤进行了分析，并将其组成与其上生长的植物群落相关联。本文旨在报道这些研究结果，通过对数据的统计分析，我们提出了一个控制蛇纹岩植物区系的土壤因素模型。
3Site description研究区概况
Geology地质
The Dun Mountain Ophiolitic Belt (Figure 1) extends southwards from d’Urville Island through Saddle Hill, and the Dun mountain, to Red Hill in the south. The Dun Mountain Massif where much of the work was carried out, is situated about 13 km east of Nelson and rises to an altitude of 1129 m. The ultramafic rocks intrude the older surrounding sediments and volcanic rocks. The main rock types are dunite and harzburgite with a little pyroxene. The ultramafic rocks of the Dun
Dun Mountain蛇绿岩带（图1）自d’Urville岛向南延伸，经Saddle Hill和Dun Mountain，直至南部的Red Hill。大部分研究工作所在的Dun Mountain岩体位于纳尔逊以东约13 km处，海拔达1129 m。超基性岩侵入周围较老的沉积岩和火山岩中。主要岩石类型为纯橄榄岩和方辉橄榄岩，含少量辉石。Dun
Mountain Massif have a concentric form with serpentinization at the contacts.
Mountain岩体的超基性岩呈同心状分布，接触带发生蛇纹石化。
Chrome and copper (mined at the turn of the century) are associated with a prominent fault near Wooded Peak. The surrounding rocks are predominantlyspilites interspersed in a Permian sedimentary sequence composed of greywackes, argillites and calcareous rocks (Lauder, 1965).
铬矿和铜矿（于世纪之交开采）与Wooded Peak附近的一条显著断层有关。周围岩石主要为细碧岩，夹于由杂砂岩、泥质岩和钙质岩组成的二叠纪沉积序列中（Lauder, 1965）。
The Cobb Valley asbestos mine is found in the Upper Takaka Valley. The deposits were mined from 1949 to 1963 (Williams,1965). All the ultramafic rocks are hydrothermally altered and the peridotitic differentiate has been completely serpentinised to chrysotile, bastite, and to an antigorite-like serpentine. Serpentinite from the asbestos workings averages only 650 �g g �1 nickel (low for ultramafics in this part of New Zealand), whereas the magnesium concentration averages 23%, and there is only a trace of calcium (Wellman, 1942). The magnesium/calciumquotient is therefore highly unfavourable for plant growth.
Cobb Valley石棉矿位于Upper Takaka Valley。该矿床于1949—1963年间开采（Williams, 1965）。所有超基性岩均经历了热液蚀变，橄榄岩质分异体已完全蛇纹石化为纤蛇纹石、绢石及类叶蛇纹石的蛇纹石。石棉采场蛇纹岩的镍平均含量仅为650 μg g−1（在新西兰该地区的超基性岩中属较低水平），而镁平均含量为23%，钙仅含微量（Wellman, 1942）。因此，其镁/钙比对植物生长极为不利。
The Dun Mountain Belt is floristically one of the most important of the ultramafic zones of New Zealand. The Dun Mountain itself was the type locality for dunite (FeMgSiO4) first described by Hochstetter in the first half of the 19th Century.
Dun Mountain带在植物区系上是新西兰最重要的超基性岩区之一。Dun Mountain本身是纯橄榄岩（FeMgSiO4）的模式产地，Hochstetter于19世纪上半叶首次对其进行了描述。
Soils土壤
The soils derived from the rocks of the Dun Mountain Belt have all of the characteristics that typify serpentine soils: (1) they contain predominantly ferromagnesian minerals, the silica and magnesium content decreasing as the soils are weathered; (2) they contain relatively large amounts of nickel, chromium, manganese and cobalt; (3) they have relatively low concentrations of nitrogen, phosphorus and potassium; (4) they have a high magnesium/calcium quotient; (5) they have a low content of organic matter; (6) they are sandy, relatively homogeneous, and generally shallow with typically 15 cm of soil on top of the parent material.
由Dun Mountain带岩石发育的土壤具备蛇纹岩土壤的全部典型特征：（1）以铁镁矿物为主，随着土壤风化，硅和镁含量降低；（2）含有相对大量的镍、铬、锰和钴；（3）氮、磷、钾浓度相对较低；（4）镁/钙比高；（5）有机质含量低；（6）质地偏砂、相对均质、一般较浅，母质上方的土层通常厚约15 cm。
The soils of the Dun Mountain Ophiolitic Belt are quite variable but a profile in the Lee Valley under stunted manuka (Leptospermum scoparium) and native grasses (15 � slope) has been described as “Dun Steepland” by Chittenden et al. (1966). Their description is as follows:
Dun Mountain蛇绿岩带的土壤变化较大，但Chittenden et al. (1966)将Lee Valley一处生长矮化麦卢卡（Leptospermum scoparium）和本土禾草（坡度15°）的剖面描述为“Dun陡坡地”（Dun Steepland），其描述如下：
(1) 0–2.5 cm – dark brown sandy loam, friable, moderately developed fine soft crumb structure.
（1）0–2.5 cm——暗棕色砂壤土，疏松，中度发育的细软团粒结构。
(2) 2.5–12 cm – dark brown silt loam with rock fragments, friable, strongly developed medium nutty structure.
（2）2.5–12 cm——暗棕色粉砂壤土，含岩石碎块，疏松，强度发育的中等核状结构。
Figure 1. Map of the Dun Mountain Ophiolitic Belt. Cg – Patuki Melange of a conglomerate of sedimentary, volcanic and ultramafic rocks. Ls – Wooded Peak Limestone consisting of poorly bedded grey limestone with lenses of grey siltstone. S – Greville Formation of interbedded laminated grey sandstone, siltstone and mudstone. Um – Layered dunite,harzburgite and pyroxenite with widespread serpentinisation. � – sampling sites represented by: 1 – Hackett Creek chromite mine, 2 – United Mine, 3 – site of sedimentary/ultramafic ecotone, 4 – Dun Saddle.图1. Dun Mountain蛇绿岩带地图。Cg——Patuki混杂岩，由沉积岩、火山岩和超基性岩组成的砾岩；Ls——Wooded Peak石灰岩，由层理不清的灰色石灰岩夹灰色粉砂岩透镜体组成；S——Greville组，由互层的纹层状灰色砂岩、粉砂岩和泥岩组成；Um——层状纯橄榄岩、方辉橄榄岩和辉石岩，普遍蛇纹石化。•——采样点：1——Hackett Creek铬铁矿，2——United矿，3——沉积岩/超基性岩交错带，4——Dun Saddle。
(3) 12–22 cm – dark olive-brown silt loam with many rock fragments, friable, moderately developed medium granular structure.
（3）12–22 cm——暗橄榄棕色粉砂壤土，含大量岩石碎块，疏松，中度发育的中等粒状结构。
(4) >22 cm – weathered serpentinite rock.
（4）>22 cm——风化蛇纹岩。
Elsewhere the soil cover consists of little more than weathered serpentine scree with little or no organic content. The modern nomenclature for this serpentine soil type is magnesic mafic brown (Anon, 1992) or typic argiudoll (vermiculitic) in the US taxonomy.
其他地方的土壤覆盖层几乎仅由风化的蛇纹岩碎屑构成，有机质含量很少或没有。该蛇纹岩土壤类型的现代命名为镁质铁镁质棕壤（magnesic mafic brown）（Anon, 1992），在美国土壤系统分类中为典型黏化湿润软土（蛭石质）（typic argiudoll, vermiculitic）。
Vegetation植被
Ultramafic areas nearly always support a characteristic vegetation, distinct from adjacent areas. The vegetational boundary between ultramafic and non-ultramafic soils is often very sharp, occurring over only a few metres. The Dun Mountain Belt is a particularly good example of this sharp differentiation.
超基性岩区几乎总是生长着有别于邻近地区的特征性植被。超基性土壤与非超基性土壤之间的植被界线往往非常清晰，仅在数米范围内即发生转变。Dun Mountain带是这种鲜明分异的一个特别典型的例子。
The vegetation of the Dun Mountain Belt has been described by Lee (1980; 1992) who recognises several distinct communities many of which are separated from the surroundingsedimentary forest by exceedingly sharp boundaries.We have carried out plant mapping at 6 of these communities (see below and in Table 1 for species lists).
Lee（1980; 1992）描述了Dun Mountain带的植被，识别出若干不同的群落，其中许多群落与周围沉积岩区森林之间以极为清晰的界线分隔。我们在其中6个群落开展了植物制图（见下文，物种名录见表1）。
ally no surface organic material. In the non-ultramafic forested areas, there was indeed a shallow organic layer, but it would have been inappropriate to compare this organic layer with the rest of the soils where such a layer was lacking. Moreover, in the forested areas, the plant roots were located deep into the lower horizons and had little contact with the surface material.
乎没有地表有机物质。在非超基性的森林区域确实存在一层较薄的有机层，但将此有机层与缺乏该层的其余土壤进行比较是不恰当的。此外，在森林区域，植物根系深入下部土层，与表层物质接触很少。
Vegetation mapping植被制图
4Materials and methods材料与方法
Soil sampling土壤采样
Visits were made to the Dun Mountain, Hackett chromite mine and Cobb Valley asbestos mine and to the ultramafic terrain surrounding these areas (Figure 1). Soils were collected from each site where samples were taken from the material immediately under the layer of humus and surface litter (which was usually sparse).
对Dun Mountain、Hackett铬铁矿、Cobb Valley石棉矿及其周围的超基性岩区进行了实地考察（图1）。在每个样点采集土壤，采样部位为腐殖质层和地表凋落物层（通常较稀薄）正下方的物质。
It might be argued that the surface organic layer should have been sampled rather than the lower inorganic fraction. However, in most cases there was virtu- We carried out plant mapping of the plant species occurring over the various geological units as shown below. The mapping procedure consisted of dividing the terrain into 10 m � 10 m quadrats and counting the number of plant species within each quadrat.
或许有人认为应采集地表有机层而非下部无机部分。然而，在大多数情况下几—— 我们对各地质单元上出现的植物物种进行了植物制图（如下所示）。制图方法为将研究区划分为10 m × 10 m的样方，并统计每个样方内的植物物种数。
(A) Gabbroic rocks adjacent to the Cobb Valley talc-magnesite workings carry a low forest dominated by species such as Nothofagusmenziesii and Weinmannia racemosa. There is a relatively sharp transition to the adjacent Leptospermum shrubland over serpentinite.
（A）Cobb Valley滑石-菱镁矿采场附近的辉长质岩石上生长着以Nothofagus menziesii和Weinmannia racemosa等物种为优势的低矮森林。它与相邻蛇纹岩上的Leptospermum灌丛之间过渡相对明显。
(B) The low transition forest between ultramafics and sedimentary/ultramafic melange consists of a low tree assemblage of variable composition depending on the predominance of oneormore parent materials inthe soil. The canopy is dominated by Nothofagus solandri var. cliffortioides.
（B）超基性岩与沉积岩/超基性岩混杂岩之间的低矮过渡林由组成多变的低矮乔木群落构成，其组成取决于土壤中一种或多种母质的优势程度。林冠以Nothofagus solandri var. cliffortioides为优势。
(C) Chionochloa tussock grassland is very typical of the open vegetation of the dunites and peridotites of the Dun Mountain Massif and increases in importance with altitude.
（C）Chionochloa丛生草地是Dun Mountain岩体纯橄榄岩和橄榄岩上开阔植被的典型类型，其重要性随海拔升高而增加。
(D) The vegetation of the Hackett Creek chrome mine closely resembles that of the Leptospermum scoparium shrubland (F) of the Dun Mt. Saddle except that the individual plants are more stunted. The only endemic plant found in this community is Pimelea suteri.
（D）Hackett Creek铬矿的植被与Dun Mt. Saddle的Leptospermum scoparium灌丛（F）十分相似，只是个体植株更为矮化。该群落中发现的唯一特有植物为Pimelea suteri。
(E) Leptospermum scoparium shrubland forms a large part of the serpentine vegetation and is host to endemics such as Pimelea suteri and Myosotis monroi. The partially endemic Notothlaspi australe is often found in damper places. The small gentian Gentiana corymbifera is also very common and is often the only coloniser of serpentine scree. Larger shrubs include Hebe odora and Cassinia vauvilliersii.
（E）Leptospermum scoparium灌丛构成蛇纹岩植被的很大一部分，其中分布有Pimelea suteri和Myosotis monroi等特有种。部分特有的Notothlaspi australe常见于较潮湿处。小型龙胆Gentiana corymbifera也十分常见，且往往是蛇纹岩碎屑上唯一的定居植物。较大的灌木包括Hebe odora和Cassinia vauvilliersii。
(F) The talc- and quartz-magnesite tailings of the Cobb asbestos mine are an environment extremely hostile to vegetation. On the slopes of these tailings the only species are Colobanthus strictus and Poa picta. In the more level parts of these tailings are to be found, Hebe odora and the ubiquitous Leptospermum scoparium.
（F）Cobb石棉矿的滑石-菱镁矿和石英-菱镁矿尾矿是对植被极为不利的环境。在这些尾矿的坡面上仅有Colobanthus strictus和Poa picta两种植物。在尾矿较平坦的部位可见Hebe odora和广布的Leptospermum scoparium。
Chemical analysis化学分析
All soils were sieved to 210 �m size and the samples (0.5 g) were weighed into polypropylene cups. A mixture (12 mL) of 1:1 HF/HNO3 was added to each cup and the solutions were evaporated to dryness on a water bath. Then 12 mL of 2 M HCl was added to each cup and the solutions warmed to redissolve the material. The chemical elements were determined in the solutions by flame atomic absorption spectrometry (FAAS). To measure the pH of the soils, 2.5 mL of water was added to 1 g of sieved soil and the mixture left overnight before measurement with a pH meter. A total elemental analysis by X-ray fluorescence spectrometry was also performed.
所有土壤过筛至210 μm，称取样品（0.5 g）置于聚丙烯杯中。向每个杯中加入1:1 HF/HNO3混合液（12 mL），在水浴上将溶液蒸发至干。随后向每个杯中加入12 mL 2 M HCl，加热使物质重新溶解。采用火焰原子吸收光谱法（FAAS）测定溶液中的化学元素。测定土壤pH时，向1 g过筛土壤中加入2.5 mL水，混合物静置过夜后用pH计测定。此外还采用X射线荧光光谱法进行了元素全量分析。
The extractable fraction of soil elements was determined by adding 10 mL of extractant (potassium hydrogen phthalate at pH 5.9) to 0.5 g of soil. The mixtures were shaken overnight and the supernatant decanted. The extracts were again analysed by FAAS.
土壤元素的可提取态通过向0.5 g土壤中加入10 mL提取剂（pH 5.9的邻苯二甲酸氢钾）进行测定。混合物振荡过夜后倾出上清液，提取液同样采用FAAS分析。
The question of what extractant should be used to simulate the plant-available fraction of metals in soils received a great deal of attention by us. The pH of the soil in the rhizosphere of plants is often more acidic than the bulk soil, two pH units lower being typical (Salisbury and Ross, 1978). The pH values of the soil samples taken in this survey ranged from 4.9 under forest litter to 7.7 under serpentine scrub (the majority of the samples). It was decided that extractionat pH 5.9 would simulate conditions at the rhizosphere of most of the plants in these soils; the pH of 5.9 being about 2 units below the pH of serpentine soils.
采用何种提取剂来模拟土壤中金属的植物有效态，是我们重点考虑的问题。植物根际土壤的pH通常比土体更酸，一般低约2个pH单位（Salisbury and Ross, 1978）。本次调查所采土样的pH值范围为4.9（森林凋落物下）至7.7（蛇纹岩灌丛下，占样品的大多数）。因此决定采用pH 5.9的提取来模拟这些土壤中大多数植物的根际条件；pH 5.9约比蛇纹岩土壤pH低2个单位。
There was a further rationale for selection of pH 5.9 as the one most likely to simulate uptake of elements from the soils. An experiment was carried out in which the serpentine-endemic Italian crucifer Alyssum bertolonii was grown for three months in serpentine soil from the Dun Mountain complex. This soil contained 6090 �g g �1 Ni. The plants had been sown in a tray containing 3.46 kg of soil and extracted 0.019 g of nickel. This represented an extraction of 5.5 �g g �1 of the original nickel content of the soils. The rhizosphere of plants is usually in contact with only 0.4–2.8% by volume) of the surrounding soil (Barber, 1984). If soil interception were the only factor, the total available nickel in the soil should therefore be in the range 196–1375 �g g �1. This figure is however far too high because it does not take into account mass flow and diffusion (Barber, 1984) that together can amount to ten times the root interception factor. If this combined figure is taken (i.e. 4–28%), the plant experiments indicate a probable 20–138 �g g �1 extractable nickel with a mean of 79 �g g �1. Our experimental value using KH phthalate at pH 5.9 was 84 �g g �1 Ni.
选择pH 5.9作为最可能模拟土壤元素吸收的条件还有另一依据。在一项试验中，将蛇纹岩特有的意大利十字花科植物Alyssum bertolonii种植于采自Dun Mountain岩体的蛇纹岩土壤中3个月。该土壤含Ni 6090 μg g−1。植物播种于盛有3.46 kg土壤的盘中，共吸取了0.019 g镍，相当于从土壤原有镍含量中提取了5.5 μg g−1。植物根际通常仅与周围土壤体积的0.4%–2.8%接触（Barber, 1984）。若根系截获是唯一因素，则土壤中有效镍总量应在196–1375 μg g−1范围内。然而该数值明显过高，因为未考虑质流和扩散（Barber, 1984），二者合计可达根系截获因子的10倍。若采用该综合数值（即4%–28%），植物试验表明可提取镍可能为20–138 μg g−1，平均值为79 μg g−1。而我们采用pH 5.9邻苯二甲酸氢钾测得的试验值为84 μg g−1 Ni。
Statistical treatment统计处理
Principal components analysis (PCA) (SAS Institute, 1987) was used to relate soil composition to the floristic composition of vegetation overlying the ultramafic substrate. The procedure involves finding one or more new variables (principal components) that condense as much as possible of the original data into a smaller set of uncorrelated variables. The variation in the first principal component accounts for as much as possible of the total variance in the original data set. The second principal component, at right angles to the first, expresses the next highest possible variance, and so forth. The main aim of PCA is to simplify the data. All data were transformed logarithmically because they represented lognormal, rather than normal distributions.
采用主成分分析（PCA）（SAS Institute, 1987）将土壤组成与超基性基质上覆植被的区系组成相关联。该方法旨在找出一个或多个新变量（主成分），将原始数据尽可能压缩为一组较少的互不相关的变量。第一主成分的变异尽可能多地解释原始数据集的总方差；第二主成分与第一主成分正交，表达次高的方差，依此类推。PCA的主要目的是简化数据。由于所有数据呈对数正态分布而非正态分布，故均进行了对数转换。
| Al | 1.53 | Ca | 0.262 | Co | 0.054 | Cl | 0.013 | Cr | 0.785 |
|---|---|---|---|---|---|---|---|---|---|
| Cu | 0.010 | Fe | 23.42 | K | 0.017 | Mg | 15.92 | Mn | 0.278 |
| Na | 0.009 | Ni | 0.609 | O | 41.5 | P | 0.034 | S | 0.052 |
| Si | 15.7 | Ti | 0.013 | V | 0.002 | Zn | 0.014 | Total | 100.2 |
Notes: (1) analyses performed by X-ray fluorescence spectrometry; (2) soil from Serpentine Rd between Rai Valley and Nelson.
注：（1）采用X射线荧光光谱法分析；（2）土壤采自Rai Valley与纳尔逊之间的Serpentine Rd。
5Results and discussion结果与讨论
Abundance data for heavy metals in soils and their extracts
土壤及其提取液中重金属的丰度数据
The elemental content of a typical serpentine soil from the Dun Mountain Ophiolitic Belt is given in Table 2. The mg/Ca quotient (61) is within the range of 16–845 for all other sites in the region.This is lower than values reported by Brooks (1987) for 230 New Caledonian serpentine soils (howeverboth elements had extremely low abundances in these soils) and higher than the quotient of 9.9 reported by the same source for Spanish ultramafics.
Dun Mountain蛇绿岩带典型蛇纹岩土壤的元素含量见表2。其Mg/Ca比（61）处于该地区所有其他样点16–845的范围内。该值低于Brooks (1987)报道的230个新喀里多尼亚蛇纹岩土壤的数值（但这些土壤中两种元素的丰度都极低），而高于同一来源报道的西班牙超基性岩的比值9.9。
Table 3 summarises abundance data for11 elements in the soils analysed in the course of this work. Table 4 lists abundances of 5 elements in soils extracts conducted at pH 5.9.表3汇总了本研究所分析土壤中11种元素的丰度数据。表4列出了pH 5.9条件下土壤提取液中5种元素的丰度。
From the magnesium, nickel and chromium values in Table 3, it is clear that the soils supportingvegetation communities D, E, and F are ultramafic in character.
由表3中的镁、镍和铬数值可知，支撑植被群落D、E和F的土壤具有超基性特征。
Correlation analysis相关分析
Correlation analysis of the total concentrations (transformedlogarithmicallybecause distributions were lognormal) of elements in the soils is shown in Table 5. Aluminium is negatively correlated with all other elements except calcium, copper and zinc. This reflects the low proportion of clays in the soils tested. Copper and zinc are typical chalcophile elements usually associated with sulphide mineralization which was virtually absent in all samples. Elements associated with serpentine soils (Co, Cr, Fe, mg, and Mn) were invariably mutually correlated.
土壤中元素全量（因呈对数正态分布而进行对数转换）的相关分析结果见表5。除钙、铜和锌外，铝与其他所有元素均呈负相关，这反映了所测土壤中黏土比例较低。铜和锌是典型的亲铜元素，通常与硫化物矿化有关，而所有样品中几乎都不存在硫化物矿化。与蛇纹岩土壤相关的元素（Co、Cr、Fe、Mg和Mn）之间均相互显著相关。
In the case of elements extracted from soils at pH 5.9, there were no significant correlations since the geochemical associations of the whole elemental concentrations were not reflected in the extracts. It will be noted that the extractabilities of only 5 of the original 11 elements has been reported because the concentrations of the other 6 (poorly extractable) were below the limit of detection (usually < 1 �g g �1) of the analytical method.
对于在pH 5.9条件下从土壤中提取的元素，未发现显著相关性，因为元素全量的地球化学组合关系未在提取液中得到体现。需要指出的是，原11种元素中仅报道了5种元素的可提取性，因为其余6种（难提取）元素的浓度低于分析方法的检出限（通常 < 1 μg g−1）。
Figure 2. Plot of component 2 vs. component 1 for logarithmically transformed total elemental concentrations in soils. Symbols as in Table 1.图2. 土壤中对数转换后元素全量的成分2对成分1散点图。符号同表1。
Principal components analysis (PCA)主成分分析（PCA）
Table 6 shows eigenvectors and cumulative eigenvalues for logarithmically transformed total elemental concentrations in the soils. The first three principal components altogether accounted for 90.1% of thetotal variance of the system. A plot of component2 vs. component 1 is given in Figure 2. From the plot it is clear that the soils of the 6 plant communities are almost completely separated from each other with very little表6给出了土壤中对数转换后元素全量的特征向量和累积特征值。前三个主成分共解释了系统总方差的90.1%。成分2对成分1的散点图见图2。由图可知，6个植物群落的土壤几乎完全相互分离，区域之间
| Ele | ment方法 | Ab (n= 10) | B (n= 10) | C (n= 5) | D (n= 5) | E (n= 12) | F (n= 5) |
|---|---|---|---|---|---|---|---|
| pH | 4.9 | 4.9 | 6.9 | 6.6 | 7.4 | 7.7 | |
| Al ( | %) | ||||||
| g.m.a | 0.93 | 4.10 | 0.53 | 0.10 | 0.22 | 0.05 | |
| s.d.r. | 0.45–1.94 | 2.51–6.69 | 0.30–0.95 | 0.06–0.16 | 0.06–0.78 | 0.04–0.08 | |
| Ca ( | %) | ||||||
| g.m. | 1.60 | 0.16 | 0.15 | 0.39 | 0.58 | 0.01 | |
| s.d.r. | 1.10–2.33 | 0.06–0.39 | 0.09–0.22 | 0.24–0.65 | 0.16–2.11 | 0.009–0.018 | |
| Co | |||||||
| g.m. | 35 | 33 | 217 | 214 | 176 | 121 | |
| s.d.r. | 29–43 | 23–48 | 173–272 | 180–254 | 112–278 | 105–138 | |
| Cr | |||||||
| g.m. | 373 | 146 | 887 | 936 | 1889 | 532 | |
| s.d.r. | 247–565 | 43–495 | 790–996 | 660–1328 | 1170–3048 | 470–603 | |
| Cu | |||||||
| g.m. | 10 | 35 | 57 | 10 | 34 | 10 | |
| s.d.r. | 5–20 | 19–65 | 45–73 | 6–16 | 22–53 | 6–17 | |
| Fe ( | %) | ||||||
| g.m. | 2.2 | 4.0 | 22.3 | 7.1 | 11.5 | 5.1 | |
| s.d.r. | 1.7–2.7 | 3.0–5.0 | 18.3–27.1 | 5.2–9.6 | 0.2–18.3 | 4.0–6.1 | |
| K | |||||||
| g.m. | 247 | 302 | 1681 | 694 | 1078 | 757 | |
| s.d.r. | 183–333 | 218–418 | 1614–1751 | 282–1707 | 608–1911 | 600–955 | |
| Mg | (%) | ||||||
| g.m. | 9.6 | 4.0 | 7.2 | 19.2 | 12.4 | 19.0 | |
| s.d.r. | 7.3–12.5 | 2.3–5.2 | 5.5–9.2 | 17.2–21.58 | 0.8–17.5 | 17.7–21.0 | |
| Mn | |||||||
| g.m. | 474 | 592 | 1801 | 1565 | 2318 | 382 | |
| s.d.r. | 368–612 | 426–823 | 1421–2282 | 1250–1960 | 1335–4014 | 287–509 | |
| Ni | |||||||
| g.m. | 221 | 164 | 57 | 2418 | 1478 | 2674 | |
| s.d.r. | 173–281 | 55–484 | 45–73 | 2210–2646 | 774–2823 | 2449–2919 | |
| Zn | |||||||
| g.m. | 14 | 46 | 106 | 25 | 67 | 14 | |
| s.d.r. | 10–20 | 30–71 | 88–129 | 16–37 | 53–84 | 11–17 |
ag.m. – geometric mean, s.d.r. – standard deviation range. bA – Cobb Valley beech forest, B – Dun Saddle beech forest, C – Cobb Valley Kanuka Scrub, D – Dun Mt grassland, E – Serpentine scrub over Hackett Creek chromite mine, F – Cobb asbestos mine.
a g.m.——几何平均值，s.d.r.——标准差范围。bA——Cobb Valley山毛榉林，B——Dun Saddle山毛榉林，C——Cobb Valley卡努卡灌丛，D——Dun Mt草地，E——Hackett Creek铬铁矿上的蛇纹岩灌丛，F——Cobb石棉矿。
overlap of the fields. Plots of other combinations of the first three components were not as successful at separating the fields and are not discussed further. Other statistical techniques exist (e.g. discriminant analysis) which would maximise separation of plant communities, but PCA is sufficient in this case.
重叠极少。前三个成分的其他组合散点图在区分各区域方面效果较差，不再进一步讨论。虽然存在可最大限度地区分植物群落的其他统计方法（如判别分析），但在本研究中PCA已足够。
From Table 6 it can be noted that the most important positive eigenvector coefficients in component 1 are those involving chromium, cobalt, iron, manganese, nickel and potassium in total soil analyses.
由表6可见，在土壤全量分析中，成分1中最重要的正特征向量系数涉及铬、钴、铁、锰、镍和钾。
It is appropriate at this stage to explain in greater detail the significance of these eigenvectors. The magnitude of the coefficientsin these vectors and their sign, determine where a given plant community will appear in the plot. In Table 6, there are negative coefficients in the first principal component. These are for aluminium and calcium. There are positive values for chromium to zinc. This implies that in a plot of Component 2 vs. Component 1 (Figure 2), groupings along the right-hand side of the x-axis will generally correspond to low concentrations of aluminium and calcium, and
此处有必要更详细地解释这些特征向量的意义。这些向量中系数的大小及其符号决定了某一植物群落在散点图中的位置。表6中第一主成分存在负系数，对应铝和钙；铬至锌则为正值。这意味着在成分2对成分1的散点图（图2）中，沿x轴右侧分布的类群通常对应于土壤中铝和钙浓度较低，而
| Ele | ment方法 | A (n= 10) | B (n= 7) | C (n= 4) | D (n= 5) | E (n= 7) | F (n= 3) |
|---|---|---|---|---|---|---|---|
| Ca | |||||||
| g.m. | 80 | 138 | 198 | 131 | 274 | 14.6 | |
| s.d.r. | 39–165 | 68–280 | 160–246 | 99–172 | 134–558 | 11.4–18.7 | |
| Fe | |||||||
| g.m. | 66 | 48 | 23 | 7.5 | 1.0 | 4.7 | |
| s.d.r. | 30–145 | 29–80 | 15–23 | 3.7–15 | 0.03–25 | 2.7–8.1 | |
| Mg | |||||||
| g.m. | 70 | 278 | 900 | 1622 | 1589 | 407 | |
| s.d.r. | 33–147 | 127–608 | 825–981 | 1304–2018 | 1110–2275 | 376–440 | |
| Mn | |||||||
| g.m. | 32 | 50 | 24 | 12 | 21 | 2.9 | |
| s.d.r. | 16–62 | 23–106 | 16–36 | 7.6–18 | 9.5–44 | 1.7–4.8 | |
| Ni | |||||||
| g.m. | 1.2 | 2.8 | 27 | 26 | 42 | 6.7 | |
| s.d.r. | 0.18–7.7 | 0.83–9.3 | 25–29 | 20–35 | 27–66 | 3.7–12 |
Symbols as in Table 3符号同表3
S�� = very highly significant (p <0.001), S� = highly significant (01> p >0.001), S = significant (0.05> p >0.01), NS = not significant (p >0.05). Negative sign implies inverse relationship.
S** = 极显著（p <0.001），S* = 高度显著（0.01> p >0.001），S = 显著（0.05> p >0.01），NS = 不显著（p >0.05）。负号表示负相关关系。
high concentrationsof the other elements (i.e. a strongly ultramafic grouping) in the soils. Thus community E (Hackett Creek) is a typical ultramafic grouping. To the left of the x axis we have communities A (Cobb Valley beech forest) and B (Dun Saddle beech forest) whose sedimentary substrates are high in aluminium and calcium but low in the other (ultramafic) elements.
其他元素浓度较高（即强超基性类群）。因此，群落E（Hackett Creek）为典型的超基性类群。在x轴左侧为群落A（Cobb Valley山毛榉林）和B（Dun Saddle山毛榉林），其沉积基质中铝和钙含量高，而其他（超基性）元素含量低。
Further discrimination of the vegetation communities is shown in Figure 2 where at the bottom of the yaxis (principal component 2) the Cobb Asbestos Mine community (F) grows over soils rich in the “ultramafic” elements chromium, cobalt, magnesium and nickel (large negative coefficients) and where at the top of the plot, the positive eigenvector coefficients indicate elevated levels of the other elements aluminium, copper, iron, manganese, potassium and zinc in the soils.
图2进一步显示了植被群落的区分：在y轴（主成分2）底部，Cobb石棉矿群落（F）生长于富含“超基性”元素铬、钴、镁和镍的土壤上（具有较大的负系数）；而在图的顶部，正特征向量系数表明土壤中其他元素铝、铜、铁、锰、钾和锌的含量较高。
The correlation coefficients of Table 5 are further evidence of the inverse relationship between aluminium and the typical “ultramafic” magnesium and heavy metals. The respective contributions of components 1 and 2 to the total variance of the system are 54.56% and 24.00% The third component, a further 11.56% of the variance, is weighted very heavily to “non-ultramafic” calcium (eigenvector coefficient of 0.85 in principal component 3).
表5中的相关系数进一步证明了铝与典型“超基性”元素镁及重金属之间的负相关关系。成分1和成分2对系统总方差的贡献分别为54.56%和24.00%。第三成分另解释了11.56%的方差，其权重高度集中于“非超基性”元素钙（钙在主成分3中的特征向量系数为0.85）。
When PCA was applied to a number of extractable metals (calcium, iron, magnesium, manganese and nickel), the cumulative percentage of the first three components was 89.99%. Considering ultramafic and non-ultramafic soils as separate populations,the spatial separationof the fields was better than in the case of the whole-element analyses (Figure 3) but there was some overlap of the individual fields. The major loadings of the eigenvector coefficients for component 1 were from nickel and magnesium (0.58 and 0.60 respectively) though there was a surprisingly high negative loading for iron (�0.45). Component 2 was dominated by calcium (0.70) and manganese (0.71). Even after making allowance for the smaller number of elements used for the extracts (poor detection limits rendered impracticable the determination of the other elements), it is clear that PCA using extractable fractions of the elemental content of the soils would also allow for a satisfactory discrimination of the different groups of soils and the vegetation communities that they supported.
对若干可提取金属（钙、铁、镁、锰和镍）进行PCA时，前三个成分的累积百分比为89.99%。将超基性土壤和非超基性土壤视为不同总体时，各区域的空间分离优于元素全量分析的结果（图3），但各区域之间存在一定重叠。成分1特征向量系数的主要载荷来自镍和镁（分别为0.58和0.60），但铁的负载荷高得出人意料（−0.45）。成分2以钙（0.70）和锰（0.71）为主。即使考虑到用于提取液分析的元素数量较少（检出限较差导致其他元素无法测定），显然基于土壤元素可提取态的PCA同样能够对不同土壤类群及其支撑的植被群落进行令人满意的区分。
| Variable变量 | Component 1 Compone成分1 成分 | nt 2 Component 32 成分3 |
|---|---|---|
| Total elemental content in soils | ||
| Eigenvectors | ||
| Aluminium | �0.30 0.37 | 0.11 |
| Calcium | �0.08 �0.02 | 0.85 |
| Chromium | 0.34 �0.12 | 0.36 |
| Cobalt | 0.39 �0.03 | �0.04 |
| Copper | 0.08 0.53 | �0.01 |
| Iron | 0.37 0.20 | �0.04 |
| Magnesium | 0.25 �0.45 | 0.11 |
| Manganese | 0.33 0.21 | 0.24 |
| Nickel | 0.37 �0.14 | �0.13 |
| Potassium | 0.36 0.09 | �0.17 |
| Zinc | 0.21 0.50 | 0.06 |
| Eigenvalues | 6.00 2.64 | 1.27 |
| Cumulative percentage | 54.56 78.56 | 90.12 |
| Elemental content of extracts at pH 5.9 | ||
| Eigenvectors | ||
| Calcium | 0.23 0.70 | �0.03 |
| Iron | �0.45 0.01 | 0.87 |
| Magnesium | 0.60 0.03 | 0.30 |
| Manganese | �0.21 0.71 | 0.02 |
| Nickel | 0.58 �0.03 | 0.38 |
| Eigenvalues | 2.22 1.56 | 0.71 |
| Cumulative percentage | 44.44 75.72 | 89.99 |
It was concluded that the vegetation groupings are strongly dependent on the chemical composition of the soils. Since both magnesium and nickel have a dominant role (high eigenvalues) in the major PCA component of both the total and extractable elemental concentrations in soils, these two elements are clearly important in defining the character of the overlying vegetation. This role is likely to be more important than in the case of elements of lower mobility such as chromium.
研究认为，植被类群强烈依赖于土壤的化学组成。由于镁和镍在土壤元素全量和可提取浓度的PCA主成分中均起主导作用（特征值高），这两种元素显然对决定上覆植被的特征十分重要。其作用可能比铬等迁移性较低的元素更为重要。
One problem that has to be addressed in making conclusions from this study, is the degree to which other “serpentine factors” such as lack of nutrients and drainage control the nature of the overlying flora. Such factors are however dependent on the chemical composition of the soils or the rocks from which they are derived. They are adjuncts to the edaphic factors rather than separate causative effects and cannot be considered in isolation.
从本研究得出结论时需要解决的一个问题是：养分缺乏和排水等其他“蛇纹岩因素”在多大程度上控制着上覆植物区系的性质。然而，这些因素取决于土壤或其母岩的化学组成。它们是土壤因素的附属因素，而非独立的致因，不能孤立地加以考虑。
6References参考文献
Figure 3. Plot of component 2 vs. component 1 for logarithmically transformed extractable concentrations of 5 elements in soils. Symbols as in Table 1.图3. 土壤中5种元素对数转换后可提取浓度的成分2对成分1散点图。符号同表1。
It is proposed that PCA applied to the elemental content of soils has proved to be a useful tool to establish perhaps the most important of the causes of the plant communities found over soils of the Dun Mountain Ophiolitic Belt and its immediate surroundings. In this example edaphic factors such as the elemental content of the soil have predominant influence on the development of vegetation. Our findings may not represent a final solution of the “serpentine problem” but we believe that they have highlighted the fact that it is likely that the solution to the problem will be found in these edaphic factors.
Anonymous 1992 New Zealand soil classification. DSIR Land Resources Sci. Rep. 19, 1–133.
Barber S A 1984 Soil Nutrient Bioavailability. Wiley and Sons, New York.
Brooks R R 1987 Serpentine and its Vegetation. Dioscorides Press, Portland.
Brooks R R, Baker A J M and Malaisse F 1992 Copper flowers. Nat. Geogr. Res. Explor. 8, 338–351.
Coleman R G and Jove C 1992 Geological origin of serpentinites. In The Vegetation of Ultramafic (Serpentine) Soils. Eds. A J M Baker, J Proctor and R D Reeves. pp 1–18. Intercept, Andover.
Chittenden E T, Hodgson L and Dodson K J 1966 Soils and agriculture of Waimea County, New Zealand. N.Z. Soil Bur. Bull. 30, 1–66.
Epstein E E 1972 Mineral Nutrition of Plants: Principles and Perspectives. John Wiley and Sons, New York.
Kinzel H 1982 Serpentinpflanzen. In Pflanzenphysiologie und Mineralstoffwechsel. pp 381–410. Ulmer, Stuttgart.
KrauseW 1958AndereBodenspezialisten. In Handuchder Pflanzenphysiologie. pp 755–806. Springer Verlag, Berlin.
Kruckeberg A R 1954 The ecology of serpentine soils. III. Plant species in relation to serpentine soils. Ecology 35, 267–274.
Lauder W R 1965 The geology of the Dun Mountain, Nelson, New Zealand. N.Z. J. Geol. Geophys. 8, 475–504.
Lee W G 1980 Ultramafic Plant Ecology of the South Island, New Zealand. PhD Thesis, Univ. Otago, Dunedin. 285 p.
Lee W G 1992 New Zealand ultramafics. The Ecology of Areas with Serpentinized Rocks. A World View. Eds. B A Roberts and J Proctor. pp 375–417. Kluwer Academic Publishers, Dordrecht.
Proctor J and Craig G C 1978 The occurrence of woodland and riverine forest on the serpentine of the Great Dyke. Kirkia 11, 129–132.
Proctor J and McGowan I 1976 Influence of magnesium on nickel toxicity. Nature 176, 234.
Proctor J and Nagy L 1992 Ultramafic rocks and their vegetation: an overview. In The Ecology of Areas with Serpentinized Rocks. A World View. Eds. B A Roberts and J Proctor. pp 469–494. Kluwer Academic Publishers, Dordrecht.
Salisbury F B and Ross C W 1978 Plant Physiology. Wadsworth Publishing Co., Belmont, CA.
SAS Institute Inc. 1988 SAS/STAT Guide for Personal Computers. SAS, Cary, NC. 705 p.
Wellman H W 1942 Talc-magnesite and quartz-magnesite rock, Cobb-Takaka District. N.Z. J. Sci. Technol. 24, 227–235.
7Acknowledgement致谢
We gratefully acknowledge Massey University for a doctoral scholarship and travel support for the senior author.
衷心感谢Massey University为第一作者提供博士奖学金和差旅资助。
Williams G J 1965 The Economic Geology of New Zealand. Australas. Inst. Mining Metall., Melbourne.
Williams G J 1965 The Economic Geology of New Zealand. Australas. Inst. Mining Metall., Melbourne.