Hydrobiologia 502: 13–36, 2003. L. Naselli-Flores, J. Padis´ak, M. T. Dokulil (eds), Phytoplankton and Equilibrium Concept: The Ecology of Steady-State Assemblages. © 2003 Kluwer Academic Publishers. Printed in the Netherlands.
13
Life strategies, dominance patterns and mechanisms promoting species coexistence in phytoplankton communities along complex environmental gradients Nico Salmaso Dipartimento di Biologia, Universit`a di Padova, Via U. Bassi 58/B, I-35131 Padova, Italy E-mail:
[email protected] Key words: phytoplankton, ecological niches, community change rate, complex environmental gradients, multivariate analyses, deep lakes
Abstract This paper analyses the life strategies, the dominance patterns and the diversity in phytoplankton communities in large and deep lakes. The study was carried out on the largest Italian Lake (Lake Garda) from 1995 to 2000. Different statistical analyses were applied. For phytoplankton the time variable represents a complex environmental gradient driving annual succession; this gradient was made explicit by the application of PCA analyses to the environmental data. The use of Non Metric Multi Dimensional Scaling applied to Bray-Curtis dissimilarity matrices revealed an ordered and cyclic development of phytoplankton every year; the Bray-Curtis index, calculated between pairs of chronologically contiguous samples, was also used as a measure of the community change rate (β t ) over the temporal succession. A significant relationship between β t and the complex environmental gradient was assessed. Finally, for every phytoplankton species, the optimum conditions for growth and the realised niches were determined. The positioning of the species on the complex environmental gradient, and the contemporaneous application of cluster analysis based on the different specific environmental optima, highlighted primarily the existence of two groups at the extreme of the complex environmental gradient. The first group included the large late winter/spring diatoms, which developed during high water turbulence and strong physical control, high nutrient concentrations, low light conditions and reduced competition. The second group was composed by many heterogeneous summer species characterised by the ability to contrast losses by grazing and sinking in stratified and stable conditions, and the ability of tolerating nutrient deficiency. A third group of species developed during environmental conditions in the middle of the two previous extremes. These included the three master species Mougeotia sp., Fragilaria crotonensis and Planktothrix rubescens/agardhii. The endogenous and exogenous mechanisms promoting species coexistence are discussed, along with the applicability of competitive and equilibrium/non-equilibrium theories to phytoplankton dynamics.
Introduction Phytoplankton communities in lakes are composed of many different species. Sometimes dominants and subdominants may be evidenced, along with several rare species coexisting with the more abundant ones; a similar experience is provided by many areas covered by herbaceous or forest vegetation (Hutchinson, 1967). Most phytoplankton species are potential competitors for the same limited resources of their envir-
onment (principally nutrients and light). Under such conditions, the high number of coexisting species contrasts largely with the results that can be predicted from the application of competition theories. The Competitive Exclusion Principle (Gause’s Principle) states that two or more species cannot coexist if they make use of the same limiting resources, namely ‘complete competitors cannot coexist’ (Hardin, 1960) and each community should be dominated by very few species occupying different niches. Niche is intended in the sense of Hutchinson (1967), designating
14 the requirements of an organism abstracted from the spatially extended habitat. However, competitive dominance of the best adapted species is a process requiring a homogeneous habitat, environmental stability and equilibrium conditions; if communities develop in non-equilibrium conditions, then competitive exclusion is prevented. Hutchinson (1961) proposed that the coexistence of many species of phytoplankton could be due to environmental instability preventing equilibrium. Therefore, phytoplankton may be viewed as a non-equilibrium community of competing species and thus are not an exception to the principle of competitive exclusion (Krebs, 2001). Considering the high potential reproductive rates of phytoplankton organisms (with doubling times spanning from fractions of days to few days), environmental instability comprises factors characterised by very different temporal scales. Some factors act on short-medium time (i.e. days–weeks; e.g., meteorological and hydrological events, grazing, vertical and horizontal chemical gradients), while others evolve regularly on seasonal basis (e.g., changes in solar radiation, development and breaking of thermal stratification, replenishment and depletion of nutrients, seasonality of zooplankton grazers). The first group of factors – depending on the intensity and on the frequency of disturbances – is instrumental for supporting non-equilibrium dynamics and diversity (Connell, 1978; Padisák, 1994) or fast community reorganisation events (reversions or catastrophes). As for the second group of factors, the annual evolution of solar elevation in the medium and high latitude regions is crucial to determine the seasonal replacement of phytoplankton assemblages. In fact, for phytoplankton – and other organisms responding in a similar way to the same temporal scales – the time variable represents a complex environmental gradient driving annual successions. Different life history traits determine the success of different species at a particular region of any environmental gradient. When these traits (e.g., growth rates, nutrient requirements, shade tolerance etc.) are inversely correlated, successional replacement will result (McCook, 1994). On annual basis, the two temporally scaled groups of factors causing instability and environmental seasonal change are essential in supporting the diversity of phytoplankton (the ‘species richness’ of a waterbody). In a given habitat it is convenient to consider different components in species diversity (e.g., Whittaker, 1972). These include the richness in species of a given sample (α diversity) and the biological diversity along habitat
gradients (e.g., elevation or soil moisture: β diversity). Excluding particular cases (e.g., the world’s largest freshwater ecosystems, Bondarenko et al., 1996), the main differences in the composition of phytoplankton in the pelagic zone of the lakes evolve along temporal gradients. In this work, the extent of differentiation of the community along the temporal gradient will be indicated as β t . The relative importance of the two groups of factors determining environmental instability and change, as well as total phytoplankton diversity in a given waterbody, is mediated by the peculiar morphometric and hydrological characteristics of the different types of lakes. Large and deep lakes have the tendency to operate as large inertial systems, for they minimise the effect of external disturbances. Previous investigations carried out during the 1990s in the largest Italian lake (Lake Garda) revealed an ordered and coherent temporal succession of phytoplankton assemblages in the whole basin (Salmaso, 1996, 2002). On the contrary, the seasonal sequence of phytoplankton species in lakes located at the opposite morphometric and hydrological gradient (e.g., small and shallow mountain lakes) may be strongly influenced by meteorological (rains and storms) and hydrological events (snow melting, high water flushing or drought in the warm period), with different species being dominant in various years (Salmaso & Decet, 1997). The variations concerning phytoplankton tend to be more predictable in large inertial systems, because they are more dependent on the annual evolution of the environmental climatic forcing variables, and less dependent on stochastic events. Disturbances classified as ‘strong’ for small and shallow lakes may have little consequences for largest basins. In this work, I will compare the specific environmental requirements and life strategies of different algae, in order to explain their seasonal (successional) adaptation and dominance along temporal gradients in deep and large lakes. The specific objectives of this paper are: (i) to define the average, typical annual development and apparent optimum environmental conditions for growth of the most abundant phytoplankton species in the deep and large Lake Garda, based on 6 years of studying; (ii) to investigate the correlations between the temporal evolution of the main environmental factors and species turnover (β t ) and (iii) to discuss the seasonal changes in the dominant assemblages along complex environmental gradients, taking into account the specific competitive abilities defined for the most abundant taxa, and
15 evidencing the exogenous and endogenous factors promoting phytoplankton diversity and species turnover. The relevance of the above topics will be emphasised in relation to the applicability of equilibrium concepts to phytoplankton dynamics. Hereafter the term ‘community’ will be used to indicate the whole pool of species present in an annual cycle, i.e. the potential competitors and winners along the temporal gradient, whereas ‘assemblage’ will be used to indicate a generic seasonal phytoplankton group.
Study site Lake Garda is the largest Italian lake. Along with the lakes Iseo, Como, Lugano and Maggiore, it is one of the deep lakes (Insubrian lakes) located south of the Alps. Lake Garda has a volume of 49 km3, a surface of 368 km2 and a maximum depth of 350 m. The long theoretical water renewal time (27 years) is due to its low catchment (lake included)/lake surface ratio (around 6) and to low annual rainfall (790–1150 mm, IRSA, 1974). The main inflow is River Sarca, at the northern edge of the lake. The outflow, with an average discharge of 58 m3 s−1 , is River Mincio, at the southern edge of the lake. Details of the catchment and the lake are reported in IRSA (1974).
Materials and methods Methods in the field and laboratory The data refer to samples collected normally every four weeks between January 1995 and December 2000 in the deepest zone of the lake (west basin, Brenzone). The average values of the chemical variables and phytoplankton abundance in the upper 20 m were estimated from samples collected at discrete depths. From 1996 to 2000, water samples for phytoplankton analyses were collected at the integrated depths of 0–2 m, 9.5–11.5 m and 19–21 m with a 5 l, 0.5 m long Niskin bottle (for a total final volume of 20 l); in 1995 samplings were not carried out in the middle layer (9.5–11.5 m). Chlorophyll a was determined by spectrophotometry following the methods proposed by Lorenzen (1967). Subsamples of 200 ml were fixed with Lugol’s solution and stored in bottles of glass kept in the dark at 4 ◦ C for subsequent phytoplankton analyses. Algal cells were counted using inverted microscopes, following the criteria reported
by Lund et al. (1958). The counts include, besides the identified fraction, ultraplankton (naked or flagellate cells around 4 µm) and undetermined nanoflagellates (around 5–10 µm). A detailed description of the procedures used in the laboratory is reported by Salmaso (2002). Water samples for chemical analyses were collected at the surface and at 20 m depths, with the exception of December 2000. Soluble reactive (RP) and total phosphorus (TP), nitrate (NO3 -N) and ammonium nitrogen (NH4 -N), reactive silica and pH have been measured by the Veneto Region Environment Protection Agency (ARPAV, District of Belluno) following standard methods (APHA et al., 1989). The analytical procedures are described by Decet & Salmaso (1997). On each sampling, profiles of temperature were carried out with an underwater multiparameter probe. The differences of water density between 0–20 m (δ 0−20 ) and 0–150 m (δ 0−150) were taken as measures of the water column stability in the layer sampled for phytoplankton analyses and in the mixolimnion, respectively. Water density was computed from temperature measurements (Chen & Millero, 1986). Secchi disk transparency (zs ) was estimated using a bathiscope to reduce uncertainties in the measurements due to light reflections and wave motions. The euphotic depth (zeu ) was estimated from Secchi disc readings using the relationship: Zeu = 4.8 × ZS0.68 (Salmaso, 2002). The measurements of total solar radiation were obtained at the Agricultural Institute of S. Michele all’Adige (Section of Agrometeorology, publically available information). The measurement station was located at Arco, at the northern border of the lake, approximately 25 km away from the sampling station. As a measure of light availability for phytoplankton growth, the average values of solar radiation during the 3 days prior to sampling dates (I3d ) were considered (cf. Bleiker & Schanz, 1997). In only few cases, missing observations were supplied with the corresponding measurements made at S. Michele all’Adige (35 km from Arco). The 3-day averages were compared with the average annual solar radiation trend at Arco, obtained on the basis of fortnightly averages computed for the period 1983–2000.
16 Data analysis
optimum uk is:
Individual samples were ordered by principal components analysis (PCA) calculated from the correlation matrix of the physical and chemical untransformed variables. Biovolume based Shannon (H ) diversity (α diversity) was estimated using natural logarithms (Magurran, 1988); unidentified ultraplankton and nanoflagellates were not considered in the calculation. Phytoplankton data were analysed by Nonmetric Multidimensional Scaling (NMDS) (Kruskal & Wish, 1978; Salmaso, 1996); the ordinations were applied to Bray & Curtis’ dissimilarity matrices (Bray & Curtis, 1957) computed on biovolume values. Unidentified phytoplankton and rare species found in one occasion were neglected. Double square root transformation of the original data was applied to reduce the weight of the most abundant species (Field et al., 1982). The Bray-Curtis dissimilarity calculated between pairs of chronologically contiguous samples, was also used as an estimate of the differentiation of the community along the complex environmental gradient (community change rate, β t ) (cf. Whittaker, 1972; Salmaso, 1996). The single species were placed on the complex environmental gradient defined by the first two PCA axes (cf. Fabbro & Duivenvoorden, 2000) taking into account their biovolume variations (cf. Bray & Curtis, 1957) (Fig. 7). Each taxon k has been located computing the average values – weighted by the corresponding biovolume values – of the coordinates of the samples where it was determined. For the coordinates along the horizontal axis, xk is: xk =
n i=1
bik Xi
n
bik
i=1
where xk is the value of the coordinate of species k on the first PCA axis, bik the abundance of taxon k in sample i (bik ≥ 0), xi the value of the coordinate of sample i on the first PCA axis (cf. Fig. 2) and n the number of samples (n = 75). For every species, the observed optimum environmental conditions for growth were computed – analogously to the previous formula – by weighted averaging estimates (ter Braak & van Dam, 1989). A species k with a particular optimum for a variable, will be most abundant in samples where this variable is close to its optimum. The weighted averaging estimate of the
uk =
n
bik vi
i=1
n
bik
i=1
and the species’ tolerance, tk , or weighted standard deviation, is: 1/2 n n 2 bik (vi − uk ) bik tk = i=1
i=1
where bik is the biovolume of taxon k in sample i (bik ≥ 0) and vi the value of the variable of interest in sample i; the computations were carried out on the whole set of samples (n = 76 or, in the case of algal nutrients, n = 75). Finally, the optimum values (uk ) were utilised to identify the principal life strategies, classifying the phytoplankton species by cluster analysis (Pearson distance, Ward’s method). Phytoplankton diversity and dissimilarity indices were computed with SIMDISS 2.0 (http://www.bio.un ipd.it/limno/simdiss/), whereas multivariate analyses were carried out with SYSTAT packages.
Results Physical and chemical variables The principal environmental abiotic variables affecting the seasonal evolution of the phytoplankton community are reported in Figure 1. Solar radiation (I3d ) showed an annual oscillation, with values ranging from 2.5–5 MJ m−2 d−1 to 20– 25 MJ m−2 d−1 (Fig. 1a). Some differences emerged from the comparison of the I3d values with the average annual solar radiation trend (1983–2000). However, the 3-day averages were strongly correlated with the corresponding averages computed during the week before the sampling (7-day averages; r = 0.94; P < 0.01), so they seem to represent quite well the available income radiation for the growth of phytoplankton in a determined period of the annual cycle. From late spring to early autumn the lake displayed a marked thermal stratification. Maximum temperatures in the first metre reached 22–25 ◦ C, whereas the metalimnetic layer deepened down to 30– 40 m (Fig. 1b). The maximum winter euphotic depths ranged around 30–45 m (Fig. 1b). During summer, the lower limit of the euphotic layer was located between
17
Figure 1. (a) Temporal variations of the average values of solar radiation during the three days prior to sampling dates (I3d ), and average annual solar radiation trend estimated for the period 1983–2000. (b) Isopleths of temperatures (◦ C) with superimposed seasonal variations of the euphotic depth (m) and (c) temporal variations of the differences of water density between 0–20 m (δ0−20 ) and 0–150 m (δ0−150 ). (d) Temporal variations of pH, (e) nitrate nitrogen and reactive silica, (f) reactive (RP) and total (TP) phosphorus in the layer 0–20 m.
18 15 and 25 m from 1995 to 1998, and between 11 and 20 m in 1999 and 2000. The euphotic depth showed similar or greater values than the mixing depth (zeu /zmix ≥ 1) from May-June to September. In the layer 0–20 m complete mixing (δ 0−20 = 0 g dm−3 ) was generally observed in the period between October–November and March–April (Fig. 1c); the higher stability of the water column (δ0−20 > 1 g dm−3 ) was observed between June-July and August. The isopycnic layer extended down to 150 m (δ0−150 ∼ = 0 g dm−3 ) between December and March–April. This layer may be roughly considered the mixolimnion of the lake, because it undergoes thermal cooling and mixing during the late winter and early spring months. During the 1990s, complete vertical cooling (down to 350 m) and circulation was documented only in 1999 and 2000 (Salmaso et al., 2001). The seasonality of pH and nutrients is related to higher phytoplankton growth during the summer months and also to the vertical mixing of the water column occurring from late autumn to early spring. In the layer 0–20 m the pH values showed a regular seasonal evolution, with values ranging from 7.9 to 8.7 (Fig. 1d). Nitrate nitrogen and silica decreased together during the warmest months (Fig. 1e; r = 0.68; P < 0.01). The lower concentrations of NO3 -N in the 20-m layer were around 120–175 µg N l−1 from 1995 to 1998, and 60–70 µg N l−1 in 1999 and 2000. The minimum Si concentrations fluctuated from 0.15 to 0.25 mg Si l−1 in the whole case study. After the minimum summer values, the epilimnetic increase of NO3 -N and Si concentrations took place with the deepening of the mixing layer (Fig. 1c). NH4 -N concentrations were generally below 20 µg N l−1 and contributed only secondarily to the total amount of nitrogen. In 1999 and 2000, reactive and total phosphorus showed higher concentrations (up to 20 µg P l−1 of TP in 1999) in comparison to those measured in 1995– 1998 (Fig. 1f). These differences were caused by the different extent of the spring vertical mixing, which determined a major recycling of phosphorus from the deepest layers to the surface during the 2 years (1999– 2000) of complete overturn (Salmaso et al., 2001). A clear influence of the extent of the vertical water mixing was not evident for Si and NO3 -N.
Table 1. (a) Principal components analyses computed on the original physical and chemical variables and (b) on the same set of variables averaged over two consecutive samples. The two panels show the percentage of explained variance and the correlations between the first two components and the input variables. Significant correlations are reported in bold (P
