Oregon tidal wetlands and climate change (pt. 2)

Which factors affect the distribution of tidal wetland
plants? Will these gradients shift with climate change?

In the last post, I began to describe some of the work I conducted as an ecologist with the EPA. Our initial investigation was a regional survey to quantify plants, algae and environmental characteristics in wetlands scattered throughout four estuaries along the Oregon coast. After working through the algae, a second goal of the regional survey was to relate vascular plant species abundance, composition and diversity to the major environmental gradients present in these wetlands (link to the paper). Salinity and elevation relative to tides (this determines when and how long plants are flooded) were of particular interest. With future sea-level rise, both flooding and salinity exposure are expected to increase.

It is already well known that factors like elevation and salinity impact wetland vegetation in a general sense (e.g., Watson and Byrne 2009), but the specific relationships between plant communities and their local environment are less well known in the Pacific Northwest. Additionally, it is useful to know which environmental factors have the greatest effect on plant communities.

A first step was to look at different environmental factors as relative predictors of plant occurrence. From the plant surveys we had a simple dataset showing whether each species was present or absent at each location we sampled. We also had quantitative data at each location for five gradients of potential importance to plant distribution: tidal elevation, soil salinity, soil nitrogen content, soil grain size, and a hydrologic index that quantified the degree of marine versus river dominance for the estuary from which the data were collected. The plant and environmental data were put into logistic regression models for many of the common species. The exciting next step - which I learned about after encountering a study on butterfly habitat use - was to apply a technique called hierarchical partitioning. This statistical technique enables a researcher to assess the relative strengths of effects (independently and jointly with other factors) of different variables in a statistical model. Like all statistical methods, it has its limitations (for instance, it doesn’t perform well with non-linear relationships between dependent and independent variables), but it seemed like a promising technique to quantify the relative importance of selected environmental factors on plant distribution.

I ran the analyses for 20 of the more commonly-occurring species and obtained some interesting results. First, for quite a few species, soil salinity was the most important variable in explaining the presence or absence of the species in the wetlands. In the figure below, for instance, salinity was positively related to the presence of perennial pickleweed (Sarcocornia perennis). Pickleweed occurrence was also positively correlated with tidal elevation, estuarine river-dominance, and soil clay content, though less strongly. (The positive correlation with river flow seems somewhat counter-intuitive for this salt-loving species, but may be due to its high frequency of occurrence from low marsh at our most river-dominated site.)

Relative strength of abiotic factor effects on the occurrence of pickleweed (Sarcocornia perennis) in Oregon tidal wetlands. All factors had statistically significant effects, but soil salinity appeared to have the greatest effect in the statistical model.

Elevation turned out to be the most important variable predicting the presence or absence of some other species. And, more rarely, soil nitrogen stood out as a key environmental gradient. Grain size (percent clay) of the soils generally only weakly correlated with species presence and absence.

The logistic regression models enabled a species-by-species look at environmental correlates of plant occurrence, but wetland plant communities in the Pacific Northwest are very diverse and species associations occur in complex patterns. We used another exciting statistical technique, non-metric multidimensional scaling (NMDS), to investigate overall plant composition in our dataset. In a nutshell, NMDS is a computational technique that aims to represent all of the differences between pairs of samples in a simple 2 or 3 dimensional graphical display. Its value lies in its ability to take a complex multi-dimensional dataset and summarize that information in a visually-intuitive manner from which patterns can be deduced.

The plot below shows the results of our NMDS analysis based on the abundance of 20 common plant species. Points closer to each other are more similar in terms of species composition. In the figure, the samples are colored based on their tidal elevation. Brownish points are plots from lower wetlands (e.g., below mean higher high water, MHHW) and greenish points are plant assemblages from high tidal marsh that is less frequently flooded. The analysis shows that plant communities separate out on an elevation gradient, similar to the patterns of vertical zonation one would see with invertebrates and algae on a typical rocky shoreline.

Non-metric multidimensional scaling plot of vascular plant communities in Oregon tidal wetlands. Plots are colored according to their height above or below local mean higher high water (m). Plot stress = 0.11.

Below I’ve shown the same NMDS plot, but with points coded by summer-time soil salinity. Plant composition differs between more saline and fresher wetlands, but there is a gradual gradient as with elevation.

An observational study like this is valuable for generating hypotheses about which environmental factors affect the distribution of different wetland species. However, we know that species are affected my more than environment itself – interactions with other species matter too. To more clearly determine causality, and not just patterns in the data, controlled experiments are needed. With dozens of species in the tidal wetland flora of Oregon and many potential abiotic and biological factors of importance, comprehensive study of this question would be a massive undertaking! In the next two posts, I’ll discuss some limited experimentation we performed to assess the effects of a few abiotic factors on plant growth and germination.

References

Janousek CN and Folger CL. 2014. Variation in tidal wetland plant diversity and composition within and among coastal estuaries: assessing the relative importance of environmental gradients. J. Vegetation Science 25:534-545.

Watson EB and Byrne R. 2009. Abundance and diversity of tidal marsh plants along the salinity gradient of the San Francisco Estuary: implications for global change ecology. Plant Ecology 205:113-128.