DCA: analysis of interrelations within the species data 1 Introduction

Detrending by segments DCA was performed on the species matrix of 98 diatom taxa. Since the DCA is used to reveal ecological differences between species it is sensitive to rare species and therefore rare taxa were down-weighted (Eilersten et a l, 1990). The DCA axes were related by regression to external environmental variables which helps to interpret the resulting plot.

5.5.6.2 Results and discussion

The eigenvalues, cumulative percentage variance and the lengths of gradient accounted for the first 4 axes in DCA of 99 diatom species and 20 environmental variables regressed onto the axes are given in Table 5.3. (see section 5.5.4). The first two axes account for 23.4% of the variance in the species data. The low percentage of variance explained is typical for data with many taxa and many zero values (Stevenson et a l, 1991) and occurs in many ecological studies (e.g. Pienitz et a l, 1995, Korsman and Birks, 1996). The lengths of gradient for the first two axes are 2.55 and 1.80 SD respectively. This indicates that the sites placed on the opposite ends of the diagram have relatively different species composition (Hill and Gauch,

1980).

DCA revealed the major patterns in diatom distribution within the training set which are presented in Figure 5.5. The most acidified, dilute and clear lakes (e.g. KOLA17, KOLA 19) are grouped on the right side of the plot. Therefore, associated acidophilous species such as

Navicula hoefleri, N. seminulum, N. subtilissima, N. tenuicephala, Eunotia praerupta, E.

serra. Stauroneis anceps v. anceps, Cymbella perpusilla, C. naviculiformis, Pinnularia

stomatophora are also placed on the right part of the plot.

The left-hand side of the plot is occupied by the species occurring in the more alkaline waters with higher COND and cation concentrations, predominantly small benthic Achnanthes and

Fragilaria (i.e. Achnanthes didyma, A. suchlandtii, A. nodosa, A. pseudoswazi, A.curtissima,

A. levanderi, A. flexella, Fragilaria brevistriata and F. pseudoconstruens), and Cyclotella

pseudostelligera. Small benthic Achnanthes and Fragilaria, which prevail on the left part of

the diagram, are the typical species for arctic shallow lakes with higher alkalinity and conductivity (e.g. Smol, 1988, Jones and Juggins, 1995, Pienitz et a l, 1995b).

Figure 5.5 DCA with 24 sites and 98 diatom taxa. Diagram (1) illustrates the arrangement of sites, diagram (2) shows the variation in species data. Species centroides are shown as open triangles. Full names of diatoms are given in Tahle 6.2 (Chapter 6) 2.0 1 3 .5 — 1 7 ■ 1 8 2 3 1 5 21 20 0.0 0.0 1.0 2 .0 3 .0 D C A A X IS 1 (1) 5.0- 4.Q— 3 . 0 - C/5 2.0 —

§

1-OH 0.0 -1.0- -2.0- -3.0 ^ A A A C K R IE A A C L V U L A C C U R T A C Y R O S S A A U T E N E M C E S A .CMLAj JN A A P IS T O M A C D ID Y A C M C E S A .C M D E S C A A 1 AUNTVA Fuvmi E R G A A A n a s u b t A C M A R G A A A C M L U N A A M A A A , F U R H O M A A M A < A ' A A A C S U C H * A A ^ , A A A A N A T E N U N IC G R A A ^ An a s e m i E U S E R R -3 .0 -2 .0 -1.0 0.0 1.0 2.0 D C A A X IS 1 3 .0 4 .0 5 .0 6.0 (2 ) 7.0

The upper half of the plot is dominated by the complex of deep and clear water centric planktonic and meroplanktonic diatoms: Cyclotella rossii, C. pseudostelligera, Aulacoseira

lirata v. subarctica, A. distans v. nivalis, A. distans v. tenella, A. per glabra v. florinae.

These taxa are confined to the deep clear-water tundra and forest-tundra lakes with relatively high pH (6.4-7.1) and low TOC (e.g. KOLA5, KOLA24, K O LA ll, see Figure 5.4. (1)). DCA shows no clear separation between TOC and Max D gradients.

The lower half of the plot is occupied by the diatoms which prefer shallow and more productive lakes with higher TOC concentrations (e.g. Navicula vitiosa, N. minuscula,

Aulacoseira lirata v. lacustris). These are mainly the forest and forest-tundra lakes (e.g.

K0LA14, KOLA20, K0LA12).

The high scores and therefore the extreme positions of several species (which are not abundant in the whole training set) close to the edge of the diagram e.g. Stauroneis anceps v.

anceps, Navicula jaagii, indicate only the fact that these particular species occur in the lakes

with the extreme for this training set environmental parameters (i.e. pH and water depth), and such species can not be considered as indicators of acid conditions {S. anceps), alkaline conditions (A. suchlandtii) or low TOC environments (N jaagii). Although the rare species were down-weighted they still appeared to influence the analysis, and this is one of the shortcomings of DCA (ter Braak, 1987c).

Hence, the DCA effectively separates diatom species in the training set according to their water acidity (pH) and water TOC. The influence of these environmental variables on diatom distribution have been found in many ecological studies (e.g. Smol, 1988, Dixit et a l, 1991, Pienitz et al, 1995a, Korsman and Birks, 1996, Weckstrom et a l, 1997b).

Since the forest lakes have the longest pH gradient they are arranged along the first DCA axis according to their pH (Figure 5.5). The tundra lakes are predominantly grouped on the upper and central part of the plot because TOC gradient is much shorter in tundra lakes compared to the forest lakes. Generally, tundra lakes tend to have lower TOC than forest-tundra and forest lakes, which occupy the lower half of the plot. The forest lakes KOLA16 and KOLA13 are exceptions; they are situated in the upper part of the plot because of the extremely low TOC concentrations.

On the whole, the results of DCA of diatom distribution are in agreement with the findings in section 5.5 and with studies of diatom distribution in Fennoscandia and northern Canada (e.g. Pienitz and Smol, 1993, Pienitz et a l, 1995a, Weckstrom et a l, 1997b)

5.5.7 CCA: analysis of relationship between diatoms and environmental variables

In document A palaeoecological study of Holocene environmental change in a small upland lake from the Kola Peninsula, Russia (Page 152-155)