Palaeogeography, Palaeoclimatology, Palaeoecology
Quantifying species–environment relationships in non-marine Ostracoda for ecological and palaeoecological studies: Examples using Iberian data
Introduction
The term “indicator species” is sometimes used in ecology to describe a species which has a narrow ecological tolerance range with respect to one or several environmental factors and is consequently strongly indicative of such particular environmental conditions (Allaby, 1998). However, this expression is also used frequently in a broader sense, because all species can be indicative of a (more or less broad) set of environmental conditions. Furthermore, habitats are subjected to continuous change, and hence the absence of a particular species does not necessarily mean that the conditions that we measure prevent its survival at that precise moment. Consequently, the narrow concept of “indicator species” has been repeatedly regarded by ecologists as misleading (Wright et al., 1994, ter Braak, 1995). Therefore, it is suggested to use the adjective “indicator” in ecology in a broad sense, and thus we do throughout the present paper.
A strong interest in the value of Ostracoda as ecological indicators has long prevailed in the work of the international ostracod community, as illustrated by the topic of the first International Symposium on Ostracoda (ISO) (Puri, 1964): “Ostracods as ecological and palaeoecological indicators”. In this early meeting, Neale (1964) suggested that ostracod distribution was primarily controlled by salinity and temperature and, to a lesser extent, by other factors such as depth, pH, substrate, food, predation or parasitism. Further environmental factors have been since added to this list, including dissolved oxygen content, submerged vegetation, photoperiod, organic matter, water ionic composition, flow velocity or permanence of the water body (e.g. Delorme, 1969, Delorme, 1989, Carbonel et al., 1988, Griffiths and Holmes, 2000). We may distinguish between abiotic (physical and chemical) and biotic factors (predation, competence, parasitism). We will focus in this paper on the potential of non-marine ostracods as indicators of abiotic factors, based on the study of the relationships between species distributions and the physico-chemical habitat.
The range of major abiotic factors that may act directly on ostracod physiology, development and overall performance are water chemistry (solute composition, pH, oxygen content, calcite saturation index, concentration of toxic substances…), temperature, food, photoperiod, hydroperiod and flow energy. Water salinity has been regarded for a long time as one of the most important (e.g. De Deckker, 1981, Neale, 1988, Aladin, 1993). More recent research suggests that ionic composition, rather than salinity per se, is an equally if not more significant variable in determining the species composition of ostracod assemblages in continental water bodies (Carbonel and Peypouquet, 1983, Forester and Brouwers, 1985, Forester, 1986, De Deckker and Forester, 1988, Smith, 1993, Curry, 1999). From a physiological and evolutionary point of view, this may reflect the varied adaptations developed by different species to overcome problems and trade-offs in ionic regulation during development and, particularly, during calcification, as suggested by Mezquita et al. (1999a).
Researchers have frequently been interested also in the response of ostracods to other factors which have no direct influence on species' performances, usually because of the easy measurement of these variables or because of their important contribution for studying particular habitat characteristics, as in the case of organic pollution in water quality biomonitoring (e.g. Rosenfeld and Ortal, 1983, Bodergat et al., 1998, Mezquita et al., 1999b, Rosenfeld et al., 2000) or depth in water level reconstruction (e.g. Mourguiart and Carbonel, 1994) using ostracod assemblages. As Neale (1964) pointed out, “depth per se, is unimportant and its influence lies principally in its effect on other factors”, such as water temperature, light intensity or flow energy (Neale, 1964) (the effect of pressure on non-marine ostracods might be negligible in most water bodies, but needs to be checked). Therefore, any attempt to reconstruct lake-level history from present day distribution of ostracod species, should take into account all other variables (and their seasonal dynamics) that may covary with it and which directly affect ostracod performance (e.g. oxygen concentration, water temperature, water chemistry, currents, etc.) Similarly, when assessing the ecological status of a water body or analysing the relationship between certain ostracod species and water quality, we should measure those factors that may directly affect ostracod occurrence, such as the concentration of toxic elements (e.g. ammonia and heavy metals), oxygen content, pH, REDOX potential or salinity. In addition, when characterising species ecologies and relating habitat variables to the occurrence of these species, it is not admissible from an ecological point of view to take into account dead animals (valves or carapaces) in the same way as individuals collected alive; ostracod remains are not indicative of the environmental conditions that we measure when collecting them and, furthermore, these remains might have been transported from a different habitat, this being highly probable in streams but also common in standing waters with internal currents and waves.
Hutchinson (1957) introduced the concept of the ecological niche as a multidimensional hypervolume embracing the set of conditions that a species' population can withstand. The term has been used also to mean the functional role of a species in the ecosystem, but after the introduction of the “guild” concept by Root (1967), as a group of species all of which exploit similar resources in a similar fashion, it has been suggested to use “niche” only for “the set of conditions that permits a species to exist in a particular biotope” (Simberloff and Dayan, 1991). Furthermore, Margalef (1980) suggests that the concept of “niche” is most useful when the ecological hyperspace is defined in terms of the range of variables such as temperature or salinity, because these make the niche suitable for comparison and homogenisation.
When the niche concept is applied to a multispecies scene, it meets the continuum concept used in plant community ecology, which considers that changes in species composition of vegetation follow gradual responses to environmental gradients (Austin, 1985). In this context, it is now widely accepted that species abundance curves usually respond to environmental gradients approximating a normal (Gaussian, i.e. symmetric bell-shaped) curve, following Shelford's Law of Tolerance (Shelford, 1911) and that modes of species abundances are more or less randomly distributed along these gradients (Austin, 1985, ter Braak and Verdonschot, 1995). Consequently, when studying species assemblages distributed over wide gradients, unimodal ordination methods are preferred over linear methods for the separation of species niches (ter Braak and Prentice, 1988).
Ordination methods and other univariate and multivariate statistical techniques used in community ecology and palaeoecology are discussed in many different works, but two comprehensive and up-to-date reviews, which are understandable for non-statisticians, are those of Jongman et al. (1995) and Maddy and Brew (1995). In this paper we apply unimodal models to the study of species–environment relationships in non-marine ostracods.
One of the most suited methods for describing the unimodal response of species' distributions to environmental gradients is Gaussian Logistic Regression (GLR) but, under certain circumstances, the less computer-intensive weighted averaging (WA) approach, approaches GLR in its estimation of species' optima and tolerance ranges in relation to habitat variables (ter Braak and Looman, 1986, ter Braak and Looman, 1995). The method of WA, like GLR, can be used with species occurrence (presence/absence) data, and it has been the origin of powerful multivariate ordination methods widely used by community ecologists, e.g. Correspondence Analysis (CA) and Canonical Correspondence Analysis (CCA). The latter (CCA) is an extension of CA that allows direct gradient analysis, i.e. directly relates species data to habitat features. There has been an extensive array of papers published on CCA and related methods since ter Braak (1986) first described it (Birks et al., 1994), but only a few scattered works that apply this technique in community ecology of non-marine ostracods (e.g. Taylor, 1992, Malmqvist et al., 1997, Mezquita et al., 1999c, Mezquita et al., 2001).
In palaeoenvironmental research, CCA is used to study the response of modern species assemblages to environmental factors in calibration datasets. The species' response to these variables is then quantified usually using WA calibration; once species' optima and tolerances (and their significance) have been calculated, these are used to reconstruct past habitat conditions from fossil assemblages that share species with the modern proxy data (Birks, 1995).
Our aim is to show the utility and use of several unimodal statistical methods (GLR, CCA, WA) for the analysis and description of the ecological niches of non-marine ostracods. In addition, we will apply these techniques to derive an ostracod-based transfer function for ostracod-based palaeoecological reconstruction of Quaternary palaeoenvironments, using a heterogeneous ecological database of c. 500 ostracod samples collected from throughout the Iberian Peninsula. The performance of this transfer function is tested on sedimentary records from two Iberian Lakes.
Section snippets
Recent dataset
Our recent database encompasses 956 samples collected between 1979 and 1998 from 659 sites located on the eastern Iberian Peninsula (Fig. 1). These sites include different types of water bodies, such as coastal wetlands (56 stations), inland lakes (45 stations), springs (471 sites) and streams (87 stations). These data derive from several research projects and sampling campaigns, which are summarised below. Because of the heterogeneity of projects, not all sets of samples contain the same type
Gaussian logistic regression (GLR)
Logistic regression (LR) analysis is a particular form of Generalised Linear Modelling (GLM) that is usually applied to binary (presence/absence) data as the response variable and continuous or categorical data as the predictor variables (Norusis, 1985, Trexler and Travis, 1993). The univariate LR model can be written as:where p is the probability of an event (species occurrence in our case) as a function of the measured (environmental) variable x. The parameters b0 and b1
Ostracod occurrence and species' niches
In order to assess what set of environmental conditions non-marine Ostracoda, as a group on itself, withstand least well, we applied a MLR with stepwise forward selection of variables to a dataset including 720 samples with full environmental data (all variables shown in Table 2). Out of these, ostracods were found in 304 samples, and 416 samples yielded no ostracods. Because many environmental variables are highly correlated (Table 2), once one of these is selected to be included in the MLR
Conclusions and future work
Our results demonstrate the significance of a range of environmental variables on non-marine ostracod species distribution:
- •
Water ionic composition and concentration, which is principally controlled by lithology and hydrogeology, climate (precipitation / evaporation ratios) and marine influence. The reasons for this are still not well understood, but it seems that different species have different physiological abilities for ionic regulation and osmoregulation (e.g. Aladin, 1993, Mezquita et al.,
Acknowledgements
F.M. would like to thank A.M. Bodergat, N. Ikeya, A. Tsukagoshi, T. Kamiya and other members of the ISO 2001 organising committee for their invitation and support to attend the meeting. I. Corella is greatly thanked for preparing the map and for her help on other figures. Special thanks to the late Huw Griffiths for his support and suggestions. Many thanks also to all those who helped in field work, provided samples or water chemistry data or suggested improvements: A. Camacho, R. Hernández, S.
References (97)
- et al.
Paleoecology of limnic ostracodes: a review of some major topics
Palaeogeogr. Palaeoclimatol. Palaeoecol.
(1988) An environmental tolerance index for ostracodes as indicators of physical and chemical factors in aquatic habitats
Palaeogeogr. Palaeoclimatol. Palaeoecol.
(1999)- et al.
Diatom-based transfer functions for inferring past hydrochemical characteristics of African lakes
Palaeogeogr. Palaeoclimatol. Palaeoecol.
(1995) - et al.
An intercontinental comparison of chironomid palaeotemperature inference models: Europe vs. North America
Quat. Sci. Rev.
(1999) - et al.
Younger Dryas and Allerød summer temperatures at Gerzensee (Switzerland) inferred from fossil pollen and cladoceran assemblages
Palaeogeogr. Palaeoclimatol. Palaeoecol.
(2000) - et al.
Ecology and distribution of ostracods in a polluted Mediterranean river
Palaeogeogr. Palaeoclimatol. Palaeoecol.
(1999) - et al.
Ostracoda from springs on the eastern Iberian Peninsula: ecology, biogeography and palaeolimnological implications
Palaeogeogr. Palaeoclimatol. Palaeoecol.
(1999) - et al.
A theory of gradient analysis
Adv. Ecol. Res.
(1988) Quantification of paleotemperature changes during isotopic stage 2 in the La Draga continental sequence (NE Spain) based on the Mg / Ca ratio of freshwater ostracods
Quat. Sci. Rev.
(1996)Salinity tolerance, morphology and physiology of the osmoregulatory organ in Ostracoda with special reference to Ostracoda from the Aral Sea