One question we keep asking ourselves is “How does this matter?”
Sometimes it’s not clear how basic research will ultimately “matter”. There’s even an award: the Golden Goose given to science that seemed esoteric, but turned out to have far-reaching implications for humanity.
But here, we want the analyses we undertake to be useful and relevant to several different groups including: other scientists, people who make management decisions on natural resources, and the people whose tax dollars fund this work.
So, I asked myself how Kari’s work at the Goodwin Islands NERR site on relating the TX90p index to water temperature and changes in submerged aquatic vegetation (SAV) might be useful to researchers and managers working on bay restoration. I wanted to make a tool that would predict temperature stress. With some warning about heat stress events, managers and researchers could potentially monitor restoration projects more closely – with the goal of finding plant species or strains that continue to thrive under warm conditions.
The first question: is there evidence that heat stress affected submerged plants around the bay? We knew that the Goodwin Islands NERR site had experienced thermal stress, but we didn’t know whether freshwater or mesohaline regions were also affected. Fortunately, there is a terrific data set compiled by researchers at VIMS on how SAV has changed since 1978. I used the areal coverage of SAV by quadrat. These are small regions, not necessarily the same size, over which you might expect the same environmental conditions to apply. Figure 1 shows the yearly change in SAV area for each quadrat. There is a lot of variability! But when we average all the quadrat changes for a given year, we see some patterns that begin to look related to the temperature stress Kari examined on a more local basis. We can also see that both Maryland (North) and Virginia (South) quadrat means have similar time variations. This WAS a surprise, because the types of SAV can be very different in the different salinity regimes!
Kari had already found that the number of days when the highest daily temperature was above the 90th percentile (aka number of hot days) and the number of days when the daily minimum temperature was above the 90th percentile (aka number of hot nights) were both important in heat stress for SAV at Goodwin. So I combined these two indices to make a new index (Figure 1, bottom). The eyeball test suggested that we were onto something.
I also plotted the yearly precipitation, because we know that this is strongly related to streamflow, which affects water clarity. I wanted to see whether we could take these indices which are easy to measure, and for which we have future predictions, and use them to built a tool to predict SAV change. The first thing I did was to guess at some models using my intuition. The second thing I did was to write a brute-force checker that tested every possible model for every possible value and checked the model correctness using a quantitative measure of goodness of fit. I guess I shouldn’t be surprised, that my intuition turned out to be close to the computer, because the human brain is an excellent pattern recognizer.
The best model (Figure 2) uses thresholds in the temperature and precipitation indices to predict SAV stress as measured by decrease in areal coverage. It doesn’t have any predictive power for SAV increase – because its not meant to! Basically, if hot days occur more than 21% of the summer, then SAV gets stressed. If this increases to 30% of the summer, then the SAV gets very stressed. Including the impact of high precipitation (really high nutrient and sediment delivery) slightly improves the model and is statistically significant – so we include a threshold for precipitation too.
The model predicts the three major recent temperature and water clarity stress events (2002, 2005, 2010) pretty well, but we don’t have a very large number of events.
So, what did we learn? Most studies don’t seem to view thermal stress as a major contributor to SAV loss, but this analysis suggests otherwise. And now we have a tool that might warn us when we should be on the lookout for temperature stress rather than waiting until the next summer to discover it.
Because we used only our simple climate indices to develop the tool, we can use the climate models to predict how SAV thermal stress might change over time.
First we can look at how the temperature stress may have changed over the past century. When we run the model using observations, it suggests that thermal stress is a pretty new stressor for SAV in the Chesapeake, occurring mostly since 2000, and more strongly in the northern region of the bay (Figure 3).
Before we apply the climate models to the future, we should check how well they predict the past. We use 8 different climate models, but most of them show the same pattern over the past century. Warm events that would stress SAV were rare, followed by a small increase in frequency in the last decade (Figure 4).
More worrisome is the future. If the model future projections are correct, then temperature stress will become more frequent, increasing to ubiquitous over the next few decades (Figure 5).
You can see from Figure 1 that we only had SAV data up to 2012. When I saw that the climate models had such a strong increase in thermal stress, I asked Kari what the SAV did in 2013 and 2014, and how our indices had behaved. The answer: SAV at Goodwin Island thrived, and there were not a lot of hot days for 2013 or 2014.
So, should we all panic? There are three good reasons not to throw up our hands and pronounce doom. First, we don’t know whether this model, or the climate models, will predict into the future yet – but these results suggest we should be testing them. Second, it’s very likely that some species of SAV are more temperature tolerant and a gradual shift to these strains and species may occur. Third, temperature stress is not the only problem facing SAV. With improved water quality, temperature stress might cause less areal coverage loss because the plants are less stressed by other factors.
Hopefully this tool will help us to learn more about SAV and thermal stress so that we can improve restoration efforts.