1. Comparative Approaches to Ecosystem Analysis
Michael Pace
University of Virginia

7. Comparative Approaches to Ecosystem Analysis
Revisiting the ecosystem concept
Approaches to ecosystems
Anatomy of a comparative study
Some benefits and examples of comparative studies
Limitations of comparative studies

8. The Ecosystem as a Multidimensional Concept: Meaning, Model, and Metaphor
Meaning refers to the technical definition
Scale independent, any size
Flexible any abiotic/biotic complex can be defined as an ecosystem
Model required to translate into usable form
Example energy flow food web model
Domain of model must be specified
Components
Spatial/temporal scales
Physical boundaries
Connections
Constraints
Metaphor
Scientific to stimulate thought, creativity
Social
Pickett & Cadanaso 2002 Ecosystems

9. Pace, Unpublished Textbook Chapter
A) Study 1
Inputs
Outputs
B) Study 2
Inputs
Outputs
Sedimentation
Burial
Respiration
C) Study 3
Inputs
Outputs
Sedimentation
Burial
Pace, Unpublished Textbook Chapter

10. Approaches Ecosystem Science Long-term studies temporal context
trends and surprises
Comparisons
spatial context &
pattern
Ecosystem Science
Theory
integration
Experiments
measure responses to
perturbations
test mechanisms
Carpenter 1998 In: Successes, Limitations,
and Frontiers in Ecosystem Science

11. Anatomy of a Comparative Study
Nitrate export from rivers
Motivated by concerns about excess N-inputs to estuaries and coastal waters (N-budget for NY Bight)
Questions
What factors are related to NO3 concentrations and export from rivers (as inputs to coastal waters)?
Can we predict riverine NO3 concentrations?
Start with a simple conceptual model
Problem 1: measuring inputs and surrogate variables for inputs (e.g. watershed human population density)
Problem 2: considerable dedicated effort (time) needed to find and synthesize disparate measurements

12. Caraco web page at ecostudies.org

13. Peierls et al Nature

14. Alternate Models General Questions Depict Ideas Simple Models Pattern
Evaluation
Communication
Prediction
Mechanisms
Study
Design
Pace 2003 In:
Models in Ecosystem Science

15. Additional Research
Caraco, Carpenter, S., N.F. Caraco, D.L. Correll, R.W. Howarth, A.N. Sharpley, and V.H. Smith Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol. Appl. 8:
Caraco, N.F. and J.J. Cole Human impact on aquatic nitrogen loads: a regional scale study of large river basins. Ambio 28:
Caraco N. F. and Cole J. J Human influence on nitrogen export: a comparison of mesic and xeric catchments. Mar. Freshwater. Res. 52:
Bennett E. M., Carpenter S. R., and Caraco N. F Human impact on erodable phosphorus and eutrophication: A global perspective. Bioscience 51:
Seitzinger S.P., Kroeze C., Bouwman A.F., Caraco N., Dentener F., and Styles R.V Global patterns of dissolved inorganic and particulate nitrogen inputs to coastal systems: Recent conditions and future projections. Estuaries 25:
Caraco NF, Cole JJ, Likens GE, Lovett GM, Weathers KC Variation in NO3 export from flowing waters of vastly different sizes: does one model fit all? Ecosystems 6:

16. Types, Benefits and Examples of Comparative Studies
Test questions/hypotheses: Food chain length
Compare systems similar type: Watershed nitrogen
Compare systems of different type: Herbivory
Identify scale dependencies: Lake mixing depth
Identify exceptions: Hudson River
Study of studies – Invasive species
Synthesis: Lake respiration model
Prediction: Vertical flux of organic carbon
Generality: Some ‘rules of thumb’

17. Elton 1927 Animal Ecology

18. Food chain length Old questions Many hypotheses Few tests
How long are food chains?
How do lengths vary among ecosystems?
What controls length?
Many hypotheses
Primary production
Few tests
Critical problem of measurement

19. Background
Paper by Kaunzinger and Morin Nature1998 – “Here we show that experimental manipulations of productivity determine the length of microbial food chains in laboratory microcosms.”
Oceanographic perspective – what about the apparently long food chains of low productivity, oceanic ecosystems? For example, don’t tuna feed at the 4th to 6th trophic level?
Does ecosystem size matter?

20. Study of Food Chain Length In Lakes Using N-15 Isotope Trophic Fractionation
N-15 fractionates ~ 3 units for each trophic transfer; establish “trophic position”
Sampled 26 lakes ranging in size (> 106 km2) and productivity (> 102 mg TP m-3)
Established baseline of N-15 using a model to account for temporal and spatial variation in base isotope (by using long lived primary consumers from littoral and pelagic)
Established relative use of littoral and pelagic production from C-13
Measured isotope composition of fish predators and calculated maximum trophic position
Post et al. Nature 2000

22. Blue = Low TP
Red = Medium TP
Green = High TP

23. Conclusions and Implications
Food chain length not related to productivity
Food chain length longer in larger ecosystems
Why? In larger lakes:
New species and larger-bodied top predators
Additional species and trophic links in the middle of the food web
Possible reductions in omnivory

24. Compare Systems of Similar Type
Example: watershed nitrogen budgets
For a number of large eastern US watersheds that drain directly into the coastal zone
Papers from this work published in 2002 in a dedicated volume of Biogeochemistry (Vol. 57, No. 1)

25. Eastern Watersheds Study
Goal: Sources and Exports

26. New N Hudson: Atmospheric Deposition = 52 % Fertilizer = 10 %
N-Fixation Agriculture = 5%
N-Fixation Forest = 19%
N Food/Feed = 14%
Boyer et al Biogeochem.

28. Cyr & Pace Nature 1993

29. Conclusions and Questions from Herbivory Study
Aquatic herbivory > terrestrial
Herbivory does not decline across gradients of primary production (PP)
Herbivore biomass and herbivory increase at similar rates with PP
For a given level of PP aquatic and terrestrial herbivores reach similar biomass
But aquatic herbivores remove on average 3x more PP
Why the difference?
Nutrient quality of forage (see Cebrian papers)

30. Scaling Relationships
One of the most critical uncertainties in ecological/environmental work
Scaling up estimates often based simply on extrapolation from specific process studies (e.g. N-fixation at a few sites to N-fixation in the ocean)
Example: mixed layer depth in lakes

31. wow.nrri.umn.edu/urisa/

32. Fee et al L&O

33. Exceptions
Hudson River long-term study by scientists at the Cary Institute
Initial work focused on measuring key processes
Context for analysis came from comparative studies
Hudson often proved to be an exception, at least relative to most prior work

34. Cole et al. MEPS 1988
General pattern of
bacterial productivity
Findlay et al. L&O 1991
Hudson doesn’t fit

35. Pace et al FW Bio

36. Study of Studies – Review of Invasive Species Papers
Most studies of invasive species
do not specify time of invasion. Many
Done long after establishment.
Most studies of invasive species are
short. Few are long.
Strayer et al TREE

37. Synthesis Example – respiration in lakes
A tale of dealing with annoying book editors
Respiration is perhaps the best index of the flow
of organic matter in aquatic ecosystems, because
it integrates all the different sources of organic carbon
as well as the temporal variability in its rate of supply.”
Williams and del Giorgio 2005

38. Planktonic Data 70 lake mean values Dark bottle O2 consumption
North temperate lakes, primarily summer, primarily surface mixed layer
Range: to 6.73 mmol O2 m-3 h-1

39. log PR = log TP; R2 = 0.81

40. Sediment Respiration 30 lake mean values
Core incubations, O2 consumption
North temperate lakes
Sediment respiration normalized to standard temperature (10 deg. C)
Range: 1.6 to 33 mmol O2 m-2 d-1

41. log10 SR std 10 °C (mmol m-2 d-1)
0.2
0.4
0.6
0.8 1
1.2
1.4
1.6 .5
1.5 2
log10 Total Phosphorus (mg m-3)
Log SRstd = log TP; R2 = 0.69

42. Editors Right – but let’s have an estimate of global lake
respiration so we can put
it together with our other
numbers for the book’s
synthesis chapter (by us).
Editors

43. Global Estimate Use empirical equations for pelagic and sediment R
Global data on lake number and size distribution
Relationships of lake surface area with: lake abundance, depth, productivity

44. Area
(km2)
Global Total (1012 m2) PR
(Tmol C y-1) SR
0.01 – 0.1
0.38
1.3
0.7
0.1-1
0.37
2.8
0.6
1-10
0.35
6.2
10-102
0.34
12.5
0.5
0.33
11.4
0.4
0.30
7.9
0.3
0.29
6.1
Sum
2.69 58
3.8

45. Implications of Estimate
Global lake R (62 Tmol C y-1) is similar to (within 2x) of global R for rivers and estuaries
Global lake R is << global ocean and terrestrial R (each ~ 10,000 Tmol C y-1)
Global lake R > GPP (~ 54 Tmol C y-1)
Difference (8) ~ global CO2 evasion of 12 Tmol C y-1 (Cole et al. 1994)
Large lakes R especially in deep waters is poorly known
Pace and Prairie 2005 In:
Respiration in Aquatic Ecosystems

46. Prediction
Comparative studies often produce empirical equations that summarize the process y as a function of one or several independent variables:
Y = b0 + b1*x1 + b2*x bn*xn
These relationships can be applied to make predictions (but too seldom are)

47. Traditional approach to
measuring sedimentation
flux
Novel approach
Science 2007

48. Prediction from VERTEX program model:
Pace et al equation for vertical flux based on depth (z)
and primary production (PP):
POCf = 3.52 z0.73 PP1.00

49. VERTIGO POC Flux Predicted Versus Observed

50. Thoughts from Predictions
Buessler et al. make the case that the attenuation of flux at the two sites is different as a function of site-specific sources and processes.
Predictions from a model that includes site specific primary production partially accounts for the differences
Oceanographers should take a closer look at the scaling of flux with PP (is the coefficient =, >, or < 1?) as this might substantially affect variability of flux and sequestration

51. Generality from Comparative Studies
One goal of science is generality
Comparative studies often provide general or average expectations for an ensemble of ecosystems (e.g. vertical flux model just discussed)
These lead to “rules of thumb”

52. Example of “Rules of Thumb” from Comparative Studies
What is the ratio of bacterial production to primary production (in planktonic systems)?
From Cole et al – about 20%
What percent of recently fixed photosynthetic carbon is released by phytoplankton?
From Baines and Pace 1991 – about 13%
What is the relationship of Secchi depth to light extinction
From Koenings and Edmundson 1991 – about 2/zs

53. Limitations of Comparative Studies
Do not identify mechanism and hence may not advance understanding
May confound time or space scales
Heterogeneous data may compromise comparisons
Problem of extrapolation to novel conditions
Susceptible to false correlations (e.g. the number of alcoholics in cities can be predicted by the number of ministers)
Statistical and modeling problems

54. Comparative Studies Not Unique to Ecosystem Studies
Epidemiology
“Tobacco smoke was first shown to cause lung cancer in This study found that people who smoked cigarettes a day were 26 times more likely to die from lung cancer than lifelong non-smokers. Even people who smoked less than 15 cigarettes a day were still 8 times more likely to die from lung cancer than life-long non-smokers.”*
Allometry – relationships with body size in organisms
Comparative genomics – analysis of similarities and differences between species/strains (e.g. human vs. chimp genome) *

55. Think of Many Not One!

56. While ocean ecosystems are less discrete (than lakes), they are no less diverse – so compare relentlessly!
Image from NASA web site courtesy
of Norman Kuring SeaWiFS Project

57. Trophic Position:
2 + (d15Nfish – [d15Nmussel x a + d15Nsnail x (1-a)])/3.4
where:
a = (d13Cfish - d13Csnail)/(d13Cmussel – d13Csnail)
Mussel = long term integrator of pelagic production
Snail = long term integrator of littoral production