We compared 13CDIC in 72 lakes from diverse regions using literature data, as well as new data for 3

1
Inter-Lake Variations in the Isotopic Signatures
of Dissolved Inorganic Carbon in Lakes
Within-Lake Processing Versus Watershed Loading
Darren L. Bade1, Stephen R. Carpenter1, Jonathan
J. Cole2, Paul C. Hanson1, and Raymond H.
Hesslein3 Corresponding author
DLBADE_at_WISC.EDU 1Center for Limnology, University
of Wisconsin Madison, USA 2Institute of
Ecosystem Studies, Millbrook, NY, USA 3Fisheries
and Oceans Canada, Freshwater Institute,
Winnipeg, MB, Canada
ABSTRACT ( B31C-0314)
PROCESS-BASED MODEL RESULTS/DISCUSSION

• The model predicts a similar range and pattern
  as observed ?13C-DIC (Fig. 5)
• The model does nearly as well at predicting
  ?13C-DIC as univariate regression models

We compared ?13C-DIC in 72 lakes from diverse
regions using literature data, as well as new
data for 32 lakes in the Northern Highland Lake
District of northern WI and the Upper Peninsula
of Michigan. We found that geochemical variables
(pH, DIC and alkalinity) account for a large
portion of the inter-lake variation in
statistical models. However a process-based model
including atmospheric gas exchange, inorganic
carbon speciation, and ecosystem metabolism was
evaluated for the Northern Highland Lakes. The
model provides a reasonable fit to the data
compared with the simplest of the statistical
models for lakes in which respiration exceeded
gross primary production (heterotrophic lakes
75 of lakes sampled). Lakes for which gross
primary production exceeded respiration
(autotrophic) were not fit well by the model. The
model demonstrates that external inputs of DIC
(e.g. groundwater) have relatively little
influence on ?13C-DIC and therefore, results from
the comparisons and models suggest that internal
lake processes are important for determining
?13C-DIC in lakes.
STATISTICAL MODELS RESULTS/DISCUSSION
Figure 5. Observed ?13C-DIC versus process-based
model predictions. Slope and intercept are not
significantly different from 1 and 0,
respectively.

• ?13C-DIC was significantly positively correlated
  with pH, DIC and alkalinity
• Suggests that external geochemical conditions
  (I.e. alkalinity) control ?13C-DIC
• Results generally similar between literature
  survey and the Northern Highland lakes
• ?13C-DIC was generally below the atmospheric
  isotopic equilibrium for a given pH (Fig. 3)
• Multiple regressions may suggest other factors
  such as morphometry (area) and biology (pCO2)

QUESTION
What processes are most responsible for the
inter-lake variation in d13C-DIC observed in
figure 1?
Figure 3. Relationship between pH and ?13C-DIC
for all data points in the literature survey. The
curved line represents atmospheric equilibrium at
a given pH.
PROCESS-BASED MODEL WITH EXTERNAL INPUTS
Figure 1. Boxplot of d13C-DIC values () obtained
from literature and other sources versus ranked
mean d13C-DIC values for each lake.

• The relative influence of external inputs of DIC
  and metabolism on ?13C-DIC were examined in 3
  lakes. (Table 2)
• Fractional turnover rates of DIC with respect to
  each process were calculated
• The largest fractional turnover rates are the
  most influential in determining the equilibrium
  ?13C-DIC
• Gradient in the dominance of different processes
• Heterotrophic respiration (RH) - Big Musky
• RH and Inflows - Allequash
• All inputs co-equal - Trout

Table 2. Characteristics of three lakes where
external inputs of DIC could be determined
Table 1. Multiple linear regression models. For
the literature survey, mean values are modeled.
All regression are significant plt0.01
Figure 6. Fractional turnover rates of processes
affecting DI13C
Methods

• Statistical models of data from a literature
  survey (see figure 1) and comparative study of 32
  lakes in the Northern Highland Lake District
  (NHLD figure 2)
• Process-based modeling of 25 lakes in NHLD

PROCESS-BASED MODEL

• Gross primary production (GPP) and respiration
  (R) were measured in 25 lakes in the NHLD and
  used as drivers in a process-based model
• The total DIC pool was modeled in addition to
  DI13C according to figure 4
• The model did not perform well in the 5 lakes
  with positive net ecosystem production
  (NEPGPP-R) and these lakes are omitted

Conclusions

• A large amount of variation exists in ?13C-DIC
  among lakes
• Statistical models suggests geochemical control
• Mechanistic models show the importance of
  biology
• A gradient of controls
• short water residence times or low ecosystem
  metabolism
• control by watershed inputs or atmospheric
  exchange
• long water residence times or high ecosystem
  metabolism
• control by biology or atmospheric exchange

Figure 2. Satellite imagery of the Northern
Highlands Lake District (Wisconsin and Michigan
www.lakesat.org)
ACKNOWLEDGMENTS

• University of Notre Dame Environmental
  Research Center (Dir. Dr. Gary Belovsky)
• University of Wisconsin-Madison Trout Lake
  Station (Dir. Dr. Tim Kratz)
• Andrew W. Mellon Foundation
• Anna Grant Birge Fellowship