Realized niche width of a brackish water submerged aquatic vegetation under current environmental conditions and projected influences of climate change
Introduction
Seagrasses and other plants of higher order are unique by occupying the subtidal photic zone of soft sediments. They form extensive habitats in sheltered near coastal zones (Reusch et al., 2005, Larkum et al., 2006) and are among the most productive habitats worldwide (Duarte, 2002). Furthermore, they provide a range of ecological functions such as coastline protection, sediment stabilization, wave attenuation, land-derived nutrient filtration, and carbon fixation, just to name a few; thereby providing some of the most valuable ecosystem services on Earth (Costanza et al., 1997, Short et al., 2011). The submerged aquatic species are also important as food, shelter and space for many invertebrates and fish, many of which are socioeconomically important (Hemminga and Duarte, 2000).
Human use of the coastal marine environment is increasing and diversifying worldwide. Given its multiple stresses, submerged aquatic species have gone through an unusually fast transition in terms of areal decline in habitat (Orth et al., 2006, Waycott et al., 2009). Although large scale processes on the formation of biotic patterns is not well known, it is plausible that contemporary climate change, interacting with other anthropogenic stressors, accounts for a large part of the observed decline. Ecologists have typically interpreted the composition of communities as the outcome of local-scale processes. However, in recent decades this view has been challenged emphasizing the importance of large-scale processes, including climate change, that may result in dramatic shifts in species distribution patterns and thereby affect community species composition, diversity, structure and productivity (Hawkins et al., 2013). Concurrent with recent climate change effects, large-scale fluctuations in water temperature is considered likely to control the distribution of submerged aquatic vegetation because with increasing temperature the photosynthesis-to-respiration ratio steadily decreases (Glemarec et al., 1997, Marsh et al., 1986, Zimmerman et al., 1989). In addition, heavy storms may create physical disturbance capable of reducing seagrass cover and increasing fragmentation of seagrass beds (Fonseca and Bell, 1998, Fonseca et al., 2000). At northern latitudes, elevated ice scouring likewise destroys submerged aquatic vegetation (Robertson and Mann, 1984, Schneider and Mann, 1991) but contemporary climate change may release vegetation from such a disturbance.
Non-independent effects are common in nature (Hoffman et al., 2003, Reynaud et al., 2003) and therefore it is expected that the combined effects of two or more variables cannot be predicted from the individual effect of each. However, anticipating the future consequences of the interacting pressures is a pre-requisite to sustainable management of coastal ecosystems under current environmental conditions and contemporary climate change. In recent years large investments have been made towards modelling of ecological systems and predicting their future behaviour (e.g. Müller et al., 2009). However, many of these models perform poorly because very little is known about how organisms might respond to multiple climate stressors (e.g. temperature and wave induced currents) and it is difficult to deal with complex and non-linear systems, such as those seen in the marine environment (see Byrne and Przeslaswki (2013) for an overview). Specifically, traditional statistical models tend to oversimplify the reality and/or statistical modelling itself may not be the most reliable way to disentangle the relationships between environmental variables and species distributional patterns because it begins by assuming an appropriate data model and the associated model parameters are estimated from the data. By contrast, machine learning avoids starting with a data model and rather uses an algorithm to discover the relationship between the response and its predictors (Hastie et al., 2009). The novel predictive modelling technique called Boosted Regression Trees (BRT) combines the strength of machine learning and statistical modelling. BRT has no need for prior data transformation or elimination of outliers and can fit complex non-linear relationships. The BRT method also avoids overfitting the data, thereby providing very robust estimates. What is most important from the ecological perspective is that it automatically handles interaction effects between predictors. Due to its strong predictive performance, the BRT method is increasingly being used in ecological studies (Elith et al., 2008).
The Baltic Sea hosts a mixture of submerged aquatic vegetation of marine, brackish or fresh water origin; each species characterized by its specific tolerance to environmental conditions (Snoeijs, 1999). Located at the margins of typical marine environments, the Baltic Sea is a vulnerable ecosystem and predicted dramatic climate change will challenge all the submerged aquatic species (Koch et al., 2013). Moreover, in the Northern Hemisphere high-latitude regions are expected to experience more severe warming compared to low-latitude regions (IPCC, 2013). In addition climate change is expected to prolong growing season, reduce ice cover as well as alter wind and precipitation patterns. Such changes are likely to have profound influences on water turbidity and salinity (Short and Neckles, 1999 and references therein). Due to the non-linear response of biota to the environment, even gradual changes in future climate may provoke sudden and perhaps unpredictable shifts in submerged aquatic plant communities as many are close to their physiological tolerance limit and different species have different response mechanisms. For example many submerged plant species are of fresh water origin but have a wide salinity tolerance and thus are often competitively superior over seagrasses under fluctuating salinity regimes (Stevenson, 1988, van den Berg et al., 1998).
Hydrodynamic conditions (Schanz and Asmus, 2003), nature of the substrate (Viaroli et al., 1997, De Boer, 2007), light (Peralta et al., 2002), temperature (Glemarec et al., 1997), salinity (Wortmann et al., 1997), water transparency (Krause-Jensen et al., 2008) and nutrient concentrations in the water column (Orth, 1977) are the key environmental variables affecting the distribution of submerged aquatic vegetation. In addition to this, ice conditions are also important in high-latitude regions (Robertson and Mann, 1984). Most of these variables are expected to change with changes in future climate; however, the strength of environment–biota relationships is likely a function of spatial scale. To date, the relationships between these environmental variables and the distributional patterns of aquatic vegetation have mostly been specified at one spatial scale but ignoring an infinite variety of other possibilities (Krause-Jensen et al., 2003, Appelgren and Mattila, 2005). In order to take into account these scale-specific effects, both small- and large-scale environmental variability should be incorporated into the models.
Seascape-scale (1–10 km) abiotic processes contribute to broadscale distributional patterns. Within these patterns smaller-scale processes (1–10 m) operate at a lower intensity to modify the distribution of species (Steele and Henderson, 1994). Among physical disturbances changes in water temperature and wave-induced current velocity are the key large-scale processes that are expected to alter species distributions. Specifically, elevated water temperatures favour plant growth and result in increased cover of submerged aquatic species (Xiao et al., 2010). Elevated wave stress affects sediment characteristics and water turbidity (Madsen et al., 2001), thereby favouring opportunistic species and disfavouring the pristine water species (Burkholder et al., 2007). The reduction of ice cover is also expected to have direct and indirect consequences. A direct consequence is the prolongation of the growth season and elevated macrophyte cover/biomass values. Currently, the co-existence of species is granted due to the presence of moderate ice disturbance that removes a significant amount of vegetation annually. In the absence of such disturbance, however, fast growing species are favoured over slow growing species. Finally, increased fresh water inputs favour higher order plants of fresh water origin over seagrasses (Touchette, 2007). But increased riverine inputs may also elevate sedimentation rates that impact negatively on the whole macrophyte community (Chambers et al., 1991). As species are shown to have strong individualistic responses to their environment we also expect large variability of responses among species (Bulleri et al., 2012).
The aims of this paper are to (1) identify the most important environmental variables defining the cover of submerged aquatic vegetation, (2) specify the spatial scales where such relationships are the strongest and (3) predict changes in the distributional pattern of the submerged aquatic vegetation from the current to future climate. The modelling approach aims to identify possible critical tipping points of all these variables where regime shifts in species distribution may occur, to provide a better understanding of the ecological frames in which outbreaks or local extinction are more likely to occur.
Section snippets
Study area
The study was carried out in the different sub-basins of the north-eastern Baltic Sea: the Baltic Proper, the Gulf of Riga, the West Estonian Archipelago Sea, and the Gulf of Finland (Fig. 1). The Baltic Sea is a geologically young semi-enclosed sea and one of the largest brackish water basins in the world. Due to short evolutionary history, low salinity and strong seasonality in temperature and light conditions, the number of submerged aquatic plant species is small, characterised by a mixture
Results
Submerged aquatic vegetation was found at 1891 stations out of 6516. Other sites were devoid of vegetation or dominated by perennial or ephemeral algae. Altogether, six submerged aquatic plant species were observed with the number of records indicated in brackets: M. spicatum (437), P. perfoliatus (381), R. maritima (333), S. pectinata (1274), Z. palustris (433) and Z. marina (293). Both single species and mixed stands of submerged aquatic vegetation were observed.
The distributional range of
Discussion
Our study suggests that (1) local and seascape-scale environmental variability affect the cover patterns of submerged aquatic species with local variability exceeding seascape-scale variability, (2) physical disturbance such as seawater warming, elevated wave-induced current velocity and reduced ice scour override the effects of salinity reduction, elevated turbidity and pelagic production and (3) finally, practically all of the studied submerged aquatic species benefit from the projected
Conclusions
Our study did not confirm the hypothesis that large scale abiotic processes define broad patterns of distribution and are the most significant factors in community variability. Instead, small and large-scale environmental variability both interactively contribute to the variability in the cover of submerged aquatic vegetation. Physical disturbance such as seawater warming, elevated wave-induced current velocity and reduced ice scour override the effects of salinity reduction, elevated turbidity
Acknowledgements
The authors acknowledge Dr. Kristjan Herkül for providing technical support at various stages of this study. Funding for this research was provided by Institutional research funding grant IUT02-20 of the Estonian Research Council. The project has received funding from BONUS project BIO-C3, the joint Baltic Sea research and development programme (Art 185), funded jointly from the European Union’s Seventh Programme for research, technological development and demonstration and from the Estonian
References (112)
- et al.
Turions of Myriophyllum exalbescens
Aquat. Bot.
(1979) - et al.
Variation in vegetation communities in shallow bays of the northern Baltic Sea
Aquat. Bot.
(2005) - et al.
Modelling macrophytes-nutrient-phytoplankton interactions in shallow eutrophic lakes and the evaluation of environmental impacts
Ecol. Eng.
(2001) - et al.
Ventilation of bottom water in the North Sea–Baltic Sea transition zone
J. Mar. Syst.
(2009) - et al.
Seagrasses and eutrophication
J. Exp. Mar. Biol. Ecol.
(2007) - et al.
Proximate composition, photosynthetic and respiratory responses of the seagrass Halophila engelmanni from Florida
Aquat. Bot.
(1987) Seagrass depth limits
Aquat. Bot.
(1991)- et al.
Effects of boating activities on aquatic vegetation in the Stockholm archipelago, Baltic Sea
Estuar. Coast. Shelf Sci.
(2004) - et al.
Effects of current on photosynthesis and distribution of seagrass
Aquat. Bot.
(1987) - et al.
The role of current velocity in structuring eelgrass Zostera marina meadows
Estuar. Coast. Shelf Sci.
(1983)