publications > poster > lessons from the everglades: atypical isotope patterns in a complex ecosystem
Lessons from the Everglades: Atypical isotope patterns in a complex ecosystem
Carol Kendall, Bryan Bemis, Scott Wankel, Steve Silva, Cecily Chang, and Linda Campbell (USGS, Menlo Park CA)
We have been investigating foodweb structure in the entire Everglades ecosystem with isotopic techniques since 1995. This study was intended to aid in the understanding of spatial, temporal, and species-specific variations in methyl mercury in game fish.
![sampling sites location map](images/sitesmapth.jpg)
![green colored square indicating USGS Everglades site](images/grnsquare.gif) |
USGS Everglades Sites |
![pink colored circle indicating USEPA REMAP site for 1999](images/pinkcircle.gif) |
USEPA REMAP Sites (1999) |
![red colored circle indicating USEPA REMAP site for 1996](images/redcircle.gif) |
USEPA REMAP Sites (1996) |
|
To our immense surprise, almost none of our initial assumptions (which were based on literature reviews of previous isotope studies, most of which were in smaller and/or less complicated ecosystems in lakes and estuaries), turned out to be applicable to the Everglades. Hence, what started as a simple foodweb study at a few "typical" sites has turned out to encompass data from some 6000 samples collected from >300 sites, 15 of which were sampled intensively 1995-1998.
Our Everglades foodweb study is probably the largest isotope investigation of a wetlands ecosystem ever attempted. Thus far we have analyzed some 6000 samples for 13C and 15N, with another 1000 or so analyzed also for 34S. Sample locations are shown in this figure. The samples cover the entire Everglades marsh area, south of the agricultural areas south of Lake Okeechobee, and west of the urban coastal corridor.
Most of the work has been focused on some 15 USGS sites where we attempted to study the food chain up to gar by intensive sampling at each site, a few times a year from 1995-98.
We also collaborated with the USEPAs REMAP program and obtained samples of mosquitofish (Gambusia), periphyton (algal mats), and sediment from spatial synoptics in 1996 and 1999 (only the 1996 REMAP data will be discussed here).
Below are a set of expectations (hypotheses)
-- typical of most isotope studies elsewhere --
that proved untrue in the Everglades.
Expectation: "You are what you eat plus about 1 for C and 2-3 for N", as illustrated by the schematic below.
Reality: Wrong! Things arent so simple in the Everglades. |
![biomagnification schematic](images/magnifgrth.gif) |
Oligotrophic marsh sites (like U3, below and left) show the expected pattern (shown by the black arrow). However, all the nutrient-impacted (i.e., more eutrophic) marsh and canal sites (like F1, below right) show anomalous 13C values. These 13C values show a backwards trend that "appears" to indicate decreasing values up the food chain (as indicated by the black arrow).
U3
|
F1
|
![plot of delta15N and delta13C values for organisms on the food chain at U3](images/u3grth.gif) |
![plot of delta15N and delta13C values for organisms on the food chain at F1](images/f1grth.gif) |
The most likely explanation for the anomalous "backwards" 13C trend is that it is caused by seasonality in the 13C of the base of the foodweb. In particular, seasonal changes in the relative contributions of algae vs. macrophyte detritus to herbivores and omnivores would explain part of the pattern observed. It is NOT caused by lack of sampling of some critical plant with a very low 13C present at the sites -- we have tried and tried at many sites and many dates to find a "missing endmember", and looked and looked for plankton, but almost never found any present at the sites.
The mechanism by which seasonally variable mixing could cause the anomalous 13C pattern is illustrated by the set of small red foodweb arrows on the figure for F1.
Expectation: Trophic differences are the main control on 15N (and maybe 13C) values of organisms.
Reality: Wrong! The plots below show the spatial variations in the 34S, 15N, and 13C of Gambusia collected at REMAP sites (9/96).
The 6 range in 15N values and the 12 range in 13C values are NOT caused by spatial differences in trophic levels of the fish. Instead, the patterns are caused by spatial differences in biogeochemical reactions that affect the 34S, 15N, and 13C of the biota.
The color-contoured area is the marsh part of South Florida; the light blue denotes non-marsh areas. The dots show locations of Gambusia samples (5-10 per site).
Expectation: Disturbed sites (ones affected by high nutrients from agriculture) have shorter food chain lengths than more pristine sites.
The figures below (combined from Cabana and Rasmussen, 1994) show how 15N can be used to explain differences in Hg concentrations of the same species of fish in different ecosystems, by differences in food chain lengths.
The 15N values can distinguish between "Class 1" sites with short food chains (e.g., large fish eat zooplankton, so the fish are 1 trophic level above zooplankton) from "Class 3" sites with longer food chains (e.g., large fish eat smaller fish that preyed on insects that preyed on zooplankton, so that the large fish are 3 trophic levels above zooplankton). Shorter food chains result in less opportunity for biomagnification of contaminants.
Reality: Wrong! The diagram below (left) shows that there are NO consistent differences in food chain length (as indicated by the difference in 15N values between carnivorous fish and various possible diets) for nutrient-impacted sites and non-impacted (oligotrophic) marsh sites.
Trophic differences between carnivore fish and potential prey
|
|
The diagram above (right) summarizes the differences in food chain lengths for different diets for both 15N and 13C, and shows that there ARE consistent differences in the food chain lengths for C, but not for N, between impacted and unimpacted (relatively pristine marsh) sites.
Expectation: Temporal variations in 13C and 15N are minimal so that sampling organisms at each site 2-3 times a year will be sufficient to define foodweb relations.
Reality: Wrong! The plots below show that there is considerable temporal variation in the 13C and 15N of organisms at some sites (this is oligotrophic site 3A-15). There is a strong correlation between seasonal changes in marsh water levels and the values of algae.
selected organisms at site WCA3A -15
![plots of delta13C and delta15N of selected organisms over time](images/wcagrsth.gif)
Small seasonal changes in water level may change the balance of photosynthesis, respiration, and atmospheric exchange reactions in the water column that control the 13C of DIC. Such changes will probably also affect the 15N of dissolved inorganic N (DIN) because of corresponding changes in N uptake and redox conditions.
![plot showing the magnitudes of temporal changes in the delta13C and delta15N of biota](images/gambgrth.gif)
The magnitudes of temporal changes in the 13C and 15N of biota collected at USGS sites (above) are similar to the ranges of values seen at the spatial scale at REMAP sites (left). The values of organisms collected at the same site usually show an inverse relation between 13C and 15N over time, especially in more pristine environments.
Expectation: Temporal variations in 13C and 15N will be largest at the base of the foodweb and will decrease with increasing trophic level, as shown in the schematic of the "foodweb triangle" below (left).
Reality: Wrong! In the Everglades, the spatial variability in 13C and 15N (spatial variability data above) results in a "foodweb hourglass" instead of the expected triangle, as shown in the schematic above (right).
This hourglass shape reflects the differences in migration areas, ages, and integration times for organisms of different trophic levels.
Contrary to expectation, the bass at the top of the food chain do not show well-integrated 13C and 15N values.
Expectation: Algae (in the form of the ubiquitous periphyton mats) is the dominant base of Everglades foodwebs.
Reality: Wrong! The figure below suggests that bulk periphyton (or algae with a similar average 13C value) is generally NOT the base of the foodweb leading to Gambusia at most sites sampled by the REMAP program in 9/96 because the 13C of periphyton is almost always higher than the corresponding 13C of Gambusia, in contradiction to the expected pattern.
![graphs of delta15N and delta13C for gambusia and periphyton](images/gambperigrsth.gif)
![plot of delta15N gambusia vs. delta15N periphyton](images/ngambperigrth.gif) ![plot of delta13C gambusia vs. delta13C periphyton](images/cgamperigrth.gif)
Note that there is a much better correlation between the 15N values of Gambusia and periphyton, than the correlation between the 13C values. The average ![delta symbol](images/bigDelta.gif) 13C value is -1.8 .
(samples collected on plastic plates Oct.-Dec. 1997)
![plot showing how delta13C and delta15N vary over 3 months for 3 adjacent sites near U3](images/u3algaegrth.gif) |
Expectation: Algal mats will show little temporal variation in 13C and 15N, and so will the organisms that live in them and eat algae.
Reality: Wrong! The plot on the right shows how the 13C and 15N vary over 3 months for 3 adjacent sites near U3 (an oligotrophic marsh site). Some of the variability is due to species succession, but most is due to changes in the 13C and 15N of water column DIC and DIN due to changing water levels.
During the dry season, the marshes probably become more anoxic because of shallower water levels, less photosynthesis, and increased quantities of decaying vegetation. Note that the 13C and 15N of algae show strong positive correlations with water level fluctuations (figures above). These oscillations are substantially damped up the food chain, probably because of the longer integration times of animals vs. plants.
Take home message: There have been few isotope foodweb studies in wetlands and freshwater marshes.
Based on our experiences in the Everglades, use of stable isotopes for foodweb reconstructions in dynamic wetlands environments is likely to be MUCH more complicated than studies in lakes, rivers, and estuaries. |
Expectation: Non N-fixing aquatic plants will have similar 13C and 15N values since they all grow in the same environment.
Reality: Wrong! Look at the wide range of values for both algae and macrophytes on the plots below.
Isotopic compositions of selected marsh organisms
![plot of isotopic compositions of selected marsh organisms](images/isotopicth.gif)
The figure above shows a simplified version of the typical foodweb found at relatively pristine marsh sites (e.g., U3, 3A-15) where algae is the dominant base of the foodweb; data are normalized to 15N and 13C of mosquitofish. Only major species are shown (of the ~ 100 species that have been collected and analyzed multiple times).
In general, the data show the expected increase in the 15N and 13C of biota up the food chain, from algae and detritus (at the base of the foodweb) to Florida gar (at the top). Note that algae (e.g., periphyton and epiphyton have very different 15N and 13C values than macrophytes (e.g, sawgrass, lilypads, cattails, etc).
![plot of delta15N values of biota grouped into trophic categories based on gut contents data](images/trophcatgrth.gif) |
The figure on the left shows the 15N values of biota (normalized to mosquitofish), grouped into trophic categories based on the gut contents data. Note the very large range of 15N values for plants. |
The schematic on the right shows some of the processes that affect the isotopic compositions of DIC and DIN in aquatic systems, and as a result, the 13C and 15N of the plants and algae that grow there. |
Processes that affect plant values
(especially aquatic plants)
|
Expectation: Fish show increases in 15N and 13C with increasing size because of the increasing trophic level of the prey.
Reality: Wrong! Note the total lack of consistent 15N and 13C correlations with length in largemouth bass from different sites and dates (above, sites denoted by letters). This lack of correlation is caused by the spatial variability in 15N and 13C at the base of the food chain (see the spatial variability data above)
Expectation: Samples from a few sites and dates would be sufficient to characterize Everglades foodwebs.
Reality: Wrong! Over 6000 samples and 6 years later, we are now narrowing in on how the ecosystem works.
Morale: dont attempt to characterize foodwebs in large, dynamic, marsh ecosystems unless you have your own isotope lab (as we do) or have LOTS of money for analyses.
Click here for a printable version of this poster (note: document will open in a new browser window)
Related information:
SOFIA Project: Application of Stable Isotope Techniques to Identifying Foodweb Structure, Contaminant Sources, and Biogeochemical Reactions in the Everglades
|