ECOTOXICOLOGY: A Comprehensive Treatment - Chapter 32 potx

Clements: “3357_c032” — 2007/11/9 — 12:39 — page 687 — #1
32
The Use of Microcosms,
Mesocosms, and Field
Experiments to Assess
Ecosystem Responses to
Contaminants and Other
Stressors
When factors are chosen for investigation, it is not because we anticipate that laws of nature can be
expressed with any particular simplicity in terms of these variables, but because they are variables that
can be controlled or measured with comparative ease.
(Fisher 1960)
32.1 INTRODUCTION
Results of field surveys and other descriptive approaches have provided a solid foundation by which
to evaluate the effects of contaminants on ecosystem processes. These studieshaveshownthatcertain
functional characteristics of ecosystems, especially productivity, nutrient flux, and decomposition,
are quite sensitive to anthropogenic disturbance. However, as we noted in the previous chapter,
descriptive studies are limited because of the inability to demonstrate cause-and-effect relationships
and because of difficulties identifying underlying mechanisms responsible for changes in these eco-
system processes. Complex interactions and indirect effects of chemicals are likely to be the rule
rather than the exception in many ecosystems. In addition, community inertia, defined as the tend-
ency for communities to persist under unfavorable conditions following disturbance (Milchunas and
Lauenroth 1995), complicates evaluation of ecosystem responses to perturbation. Isolating causal
mechanisms is particularly important in ecosystem studiesbecausetheseprocessesareoften complex
and controlled by an assortment of direct and indirect effects. For example, litter decomposition in
aquatic and terrestrial ecosystems is regulated by microbial processes and activity of invertebrates.
Because effects of contaminants on decomposition rate are dependent on the relative sensitivity
of microbial and macroinvertebrate communities, experimental approaches that isolate these dif-
ferent mechanisms are necessary to predict effects. This is an ideal application of microcosm and
mesocosm experiments, which are often designed to manipulate single or multiple environmental
variables, providing an opportunity to isolate specific factors and identify underlying mechanisms.

It is the ability to isolate and manipulate individual factors that makes application of microcosm and
mesocosm experiments particularly powerful in ecotoxicological research.
In this chapter we turn our attention to experimental approaches that have been employed to
demonstrate effects of contaminants and other stressors on ecosystem processes. We will examine the
use of both small-scale approaches such as microcosms, as well as larger, more ecologically realistic
687
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 688 — #2
688 Ecotoxicology: A Comprehensive Treatment
and field-based approaches such as mesocosms and whole ecosystem manipulations. Because of the
limited spatiotemporal scale, measuring responses of ecosystem processes to contaminants in micro-
cosms presents significant challenges. The duration of microcosm and mesocosm experiments is of
critical importance when assessing effects of contaminants. For example, a common phenomenon in
soil microcosms is the natural reduction in microbial biomass or activity over time. As a consequence,
effects of stressors on soil microbial processes become more difficult to quantify as experiments pro-
gress. Although microcosms are designed to simulate specific portions of a natural ecosystem, the
most valuable experiments investigating effects of contaminants on ecosystem processes have been
conducted in larger systems.
32.2 MICROCOSM AND MESOCOSM EXPERIMENTS
There is an increasing belief amongst risk assessors that model ecosystems do not possess ecological
advantages that were originally assumed, and that an instrumentalist approach to the prediction of toxic
effects in ecosystems will yield the most cost-effective results.
(Crane 1997)
Current approaches to ecological risk assessment of chemicals are ecologically naive and fail to include
current knowledge about effects of stressors on ecological communities.
(Pratt et al. 1997)
The genesis of microcosm and mesocosm research emerged from uncertainties regarding the use-
fulness of single species approaches for predicting effects of contaminants in nature (Cairns 1986).
While questions about ecological relevance of laboratory toxicity tests persist, microcosm and meso-
cosm experiments are now routinely employed in ecotoxicological research and to test ecological

principles (Fraser and Keddy 1997). The use of controlled experimental systems in aquatic ecology
has clearly increased over time (Figure 32.1), and these systems have been used to address both basic
and applied ecological questions. For example, if we consider contaminants as simply another form
of anthropogenic disturbance, model ecosystems can be used to characterize ecological resistance
and resilience. However, in his critique of model ecosystems used in ecotoxicological research,
Crane (1997) argues that research priorities should shift from understanding these ecological com-
plexities to questions regarding repeatability, precision, and the relationship between experimental
Ye a r
1989 1990 1991 1992 1993 1994 1995 1996 1997
Number of studies
0
10
20
30
40
50
60
Limnology
Toxicology
Microbiology
Terrestrial
FIGURE 32.1 The number of studies published between 1990 and 1996 that included the words microcosm
or mesocosm in the title or abstract. (Data from Table 1 in Fraser and Keddy (1997).)
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 689 — #3
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 689
and natural systems. He notes the “dangers of allowing model ecosystem studies to be driven by
ecological theory” and argues that model ecosystems are most appropriate as a tool to provide envir-
onmentally realistic exposure conditions. Because microcosm and mesocosm experiments often
provide complex responses across levels of biological organization, interpreting results can be chal-

lenging. Recall that problems with data interpretation were provided as a primary justification for the
decision by the U.S. EPA to drop mesocosm testing for pesticide registration (Touart 1988, Touart
and Maciorowski 1997, Chapter 23). Stay et al. (1988) argued that a lack of correspondence between
population, community, and ecosystem-level responses observed in their experiments indicated that
measurements at one hierarchical level may not be useful for predicting effects at other levels.
Sorting out these complex responses and relating alterations in community structure to ecological
processes should be a priority for microcosm and mesocosm research. We also believe that a critical
area of research is to determine the extent to which these experimental systems reflect natural con-
ditions. One of the challenges associated with the use of microcosms and mesocosms is the change
in functional measures over time, independent of the effects of contaminants. These changes, which
are often a result of container artifacts, compromise our ability to make comparisons among treat-
ments. Williams et al. (2002) compared structural characteristics of microcosms and natural ponds,
and recommended refinements to the design of model systems to improve their ecological realism.
Unfortunately, few studies have made this comparison based on functional measures or ecosystem
processes. Kurtz et al. (1998) measured reproducibility and stability of structural and functional
processes in estuarine microcosms. Both structural (relative abundance and density of sulfate redu-
cing bacteria) and functional (CO
2
assimilation, sulfate reduction) measures in microcosms were
similar to conditions in natural sediments after 7 days. However, the authors cautioned against longer
term experiments without modifying the system. Suderman and Thistle (2003) examined changes in
structural and functional measures in sediment microcosms derived from a shallow estuary. Chloro-
phyll a, primary production and most measures of meiofauna community composition remained
relatively stable over the 3-month period.
Another potential criticism of microcosm and mesocosm experiments is their relatively limited
temporal scale. Because long-term mesocosm experiments (e.g., >1 year) are rare, our understand-
ing of prolonged exposure to stressors is incomplete. This is an especially important issue when
considering ecosystem processes that often show delayed responses compared to structural altera-
tions. Bokn et al. (2003) exposed rocky intertidal communities to long-term nutrient enrichment in
marine mesocosms. Despite large inputs of N and P (maximum target concentrations were 32 and

2.0 µM, respectively) and significant increases in periphyton biomass, there were essentially no
effects on NPP, GPP, or respiration. These unexpected results were attributed to competition among
macroalgal species, grazing by herbivores, and physical disturbance. Although this study was con-
ducted for 2.5 years, a relatively long time period when compared with most mesocosm studies,
this was not a sufficient amount of time for opportunistic algal species to become established and
respond to nutrient enrichment (Bokn et al. 2003).
In addition to comparing processes in microcosms and mesocosms with those in natural eco-
systems and assessing changes in controls over time, additional research is necessary to optimize
experimental designs and to evaluate the statistical power of these systems (Kennedy et al. 1999). In
Chapter 23, we discussed strengths and weaknesses of different experimental designs (e.g.,ANOVA
versus regression; assignment of replicates to experimental units) for community-level assessments.
Consideration of statistical power is especially critical for assessments of ecosystem processes
because variability of these measures is often greater than for structural measures. Kraufvelin
(1998) estimated the number of replicates necessary to detect significant differences for 50 different
variables derived from land-based, brackish water mesocosms. Although calculations were based
on population and community-level variables, the results have important implications for meso-
cosm experiments designed to assess functional endpoints. Relatively few of the structural variables
examined had coefficients of variation (CV) less than 20%. Using an endpoint with a modest CV
of approximately 30%, 24 replicates were necessary to detect a statistically significant difference
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 690 — #4
690 Ecotoxicology: A Comprehensive Treatment
(α = 0.05) of 25% between control and treatment mesocosms. Kraufvelin (1998) also noted large
differences in the amount of variation among response variables. Assuming that ecosystem processes
will show a similar or greater level of variation, statistical power will obviously be an important
consideration when selecting functional endpoints.
32.2.1 MICROCOSMS AND MESOCOSMS IN AQUATIC RESEARCH
Microcosms and mesocosms have been employed extensively to assess the effects of contaminants
on processes in aquatic ecosystems. Although the majority of these investigations have focused
on changes in primary production, respiration and other aspects of ecosystem metabolism, end-

points related to nutrient processing and decomposition rates have also been considered. Most
experiments conducted in stream microcosms and mesocosms have focused on the response of
periphyton. Because of their small size, rapid rate of development, and diverse taxonomic composi-
tion, periphyton are sensitive indicators of water quality in natural and experimental streams (Lowe
et al. 1996). Changes in the structure and function of epilithic assemblages exposed to contamin-
ants can occur very rapidly. Colwell et al. (1989) attributed increased respiration in outdoor stream
microcosms treated with Zn to the establishment of Zn-tolerant bacteria and algae. We should also
remember that structural characteristics of ecosystems may directly or indirectly influence ecological
processes. For example, physicochemical characteristics in macrophyte-dominated systems are often
controlled by biological processes that are directly related to ecosystem metabolism (Brock et al.
1993). Kersting and van den Brink (1997) describe a dissolved oxygen–pH–alkalinity–conductivity
syndrome, in which each of these variables is expected to respond to toxic substances in parallel.
These interrelated responses may result in feedback between contaminants and ecosystem processes.
For example, it is likely that alterationsin community metabolism resulting from exposureto contam-
inants may affect pH and thereby modify contaminant bioavailability. To improve our understanding
of the complex responses frequently observed in microcosm and mesocosm experiments, sampling
protocols should be designed to quantify relationships between these physicochemical and biological
variables.
32.2.1.1 Separating Direct and Indirect Effects
One of the important applications of microcosm and mesocosm research has been to separate the
direct effects of contaminants from the indirect or secondary effects on ecosystem processes. Select-
ive application of contaminants that have specific effects on one group of organisms but relatively
limited effects on another group is a useful approach for quantifying these direct and indirect effects
(Pratt et al. 1997, Slijkerman et al. 2004). Pearson and Crossland (1996) measured photosynthesis
and respiration in outdoor experimental streams exposed to the herbicide atrazine and the insect-
icide lindane. Atrazine had a direct inhibitory effect on photosynthesis at 100 µg/L. In contrast,
photosynthesis increased in lindane-treated streams due to the elimination of grazing inverteb-
rates. Because the direct toxicological effect on invertebrates is limited, atrazine has been used
to identify bottom-up responses in model systems (Pratt et al. 1997) (Figure 32.2). In this example,
reduced food availability to higher trophic levels would be considered an indirect effect of herb-

icide exposure. Conversely, exposure of model systems to insecticides can have a direct effect
on invertebrate grazers, resulting in increased algal biomass and production. Predicting effects
of contaminants on intermediate trophic levels where elimination of one group may impact both
lower and higher trophic levels presents special challenges. Boyle et al. (1996) quantified indir-
ect effects of the insecticide diflubenzuron, a chitin inhibitor, on ecosystem processes in 0.1 ha
mesocosms. Significant declinesin abundanceof grazinginsects andzooplankton followingdifluben-
zuron treatment resulted in increased chlorophyll a biomass and GPP (a top-down response due to
reduced grazing) and reduced biomass of juvenile bluegill (a bottom-up response due to reduced
food supply). Brock et al. (1993) reported similar increases in periphyton and phytoplankton when
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 691 — #5
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 691
Atrazine level (µg/L)
Control 3.2 10 32 110 337
Dissolved oxygen (mg/L)
0
2
4
6
8
10
Number of species
20
30
40
50
60
70
Dissolved oxygen
Species richness

FIGURE 32.2 Changes in structural (species richness) and functional (dissolved oxygen concentration)
variables in microcosms treated with the herbicide atrazine. (Data from Table 1 in Pratt et al. (1997).)
mesocosms were treated with the insecticide chlorpyrifos, which reduced abundance of grazing
invertebrates.
The relationships between structural and functional components of model ecosystems are often
complex and may be dependent on contaminant concentration. Slijkerman et al. (2004) observed
that at intermediate concentrations of the fungicide carbendazim (17 µg/L), structural changes were
observed but there were no corresponding effects on ecosystem function. Functional impairment
occurred at higher exposure concentrations (219 µg/L), indicating that functional redundancy could
not compensate for changes in community structure. Similar results were reported by Carman et al.
(1995) for meiofauna exposed to PAHs in sediment microcosms. Despite significant changes in
meiofaunal community composition in high PAH treatments, there were no effects on bacterial or
microalgal activity.
Elimination of invertebrates by chlorpyrifos in experimental ditches had modest effects on eco-
system metabolism by decreasing respiration and increasing oxygen concentration; however, effects
on community structure and decomposition rates were much more dramatic (Kersting and van den
Brink 1997). Exposure of macroinvertebrates to chlorpyrifos reduced abundance of shredders and
resulted in decreased litter decomposition (Cuppen et al. 1995). These researchers also speculated
that elimination of grazing invertebrates by insecticides may enhance effects of eutrophication by
reducing top-down control of primary producers. Detenbeck et al. (1996) measured biomass, GPP,
and respiration in mesocosms treated with the herbicide atrazine in wetland mesocosms. GPP was
reduced at the lowest exposure level (15 µg/L), but respiration was either reduced (25 µg/L) or
enhanced (75 µg/L). An increase in ammonium, dissolved N, and dissolved P in treated mesocosms
was attributed to reduced nutrient uptake by periphyton. Bester et al. (1995) observed significant
reductions inprimary productionat low levels of atrazineexposure (0.12 µg/L) in marinemesocosms.
An increase in concentrations of dissolved organic N and P in treated microcosms was attributed to
release from damaged cell walls.
Mesocosm experiments also allow investigators to compare responses across levels of biological
organization, thereby providing opportunities to examine underlying mechanisms and relate struc-
tural changes to functional alterations. As noted in Chapter 31, reduced decomposition rate observed

in contaminated streams may result from either lower abundance of macroinvertebrate shredders or
changes in microbial activity. Stream mesocosm experiments have been used to assess the relative
importance of these two explanations. Newman et al. (1987) measured litter processing rates, shred-
der abundance, and microbial colonization in outdoor experimental streams dosed with chlorine.
Although no effects were measured at intermediate concentrations (64 µg/L total residual chlorine),
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 692 — #6
692 Ecotoxicology: A Comprehensive Treatment
lower rates of decomposition in streams receiving 230 µg/L were attributed primarily to reduced
abundance of amphipod shredders. The bacterial insecticide Bacillus thuringiensis significantly
increased microbial respiration and decreased decomposition in laboratory microcosms (Kreutz-
weiser et al. 1996). Although a similar trend was observed in outdoor stream channels, this trend
was not significant because of high variation among replicates.
32.2.1.2 Stressor Interactions
The key strengths of microcosm and mesocosm experiments are the opportunity to assess effects of
chemical mixtures and to quantify interactions among stressors under controlled experimental condi-
tions. Cuppen et al. (2002) observed significant effects of a mixture of insecticides (chlorpyrifos and
lindane) on decomposition rates of particulate organic matter in litterbags. Despite rapid dissipation
of both insecticides (t
1/2
= 9–22 days), elimination of shredders and reduced microbial activity res-
ulted in lower decomposition rates. Results of experiments measuring interactions between nutrients
and agricultural contaminants (e.g., herbicides, insecticides, sediments) are especially enlighten-
ing because these stressors frequently co-occur. More importantly, bioavailability of contaminants
may vary depending on the nutrient status and the amount of organic material in an ecosystem. For
example, sorption of contaminants is likely to be greater in more productive ecosystems. Barreiro
and Pratt (1994) used a factorial experimental design to measure the interactive effects of nutrient
enrichment and the herbicide diquat on primary productivity in microcosms. Although structural
variables responded to nutrients, there was no effect of diquat on algal biovolume, chlorophyll a,or
protein levels. In contrast, GPP was significantly reduced in treated microcosms. These researchers

also reported that recovery was greater in systems with higher nutrient levels, most likely due to
faster contaminant dissipation (Pratt and Barreiro 1998). The influence of nutrient concentration on
community resistance and resilience was also reported by Steinman et al. (1992), indicating that
functional responses to chemical perturbations are often context-dependent.
Comparatively few studies have examined effects of contaminants on N cycling and flux in
aquatic microcosms. Petersen et al. (2004) compared the effects of two antifouling biocides (zinc pyr-
ithione, ZPT; and copper pyrithione, CPT) on nitrification and denitrification processes in sediments.
Flux of nitrate from sediment increased significantly after additions of ZPT and CPT (Figure 32.3).
This increase was a result of increased nitrification (NH
4
→ NO
2
→ NO
3
) and/or a decrease in
denitrification (NO
3
→ N
2
). The greater sensitivity of nitrification observed in this experiment was
Control Low High
0
20
40
60
80
100
120
140
ZPT

CPT
(a)
NO
3
flux (µmol/m
2
/day)
TreatmentTreatment
Control Low High
0
100
200
300
400
500
(b)
NH
4
flux (µmol/m
2
/day)
FIGURE 32.3 Flux of NO
3
and NH
4
in microcosms exposed to zinc pyrithione (ZPT) and copper pyrithione
(CPT). Low and high treatments in the ZPT and CPT experiments were: 1.0 and 10.0 nmol ZPT/g and 0.1 and
1.0 nmol CPT/g, respectively. (Data from Table 1 in Petersen et al. (2004).)
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 693 — #7

Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 693
likely a result of greater functional redundancy of denitrification processes (Petersen et al. 2004).
Nitrification is a process performed by a limited number of bacteria, whereas denitrification is a
general process performed by many species.
Microcosm and mesocosm experiments can also be used to compare functional responses of
communities derived from different sources, thereby providing an opportunity to understand how
intrinsic features of an ecosystem may influence susceptibility to contaminants. Stay et al. (1988)
reported that effects of fluorene on respiration and rates of recovery differed among communities
depending on the source of these organisms. Fate of the insecticide chlorpyrifos in microcosms and
its effects on community metabolism, decomposition, and nutrient cycling was influenced by the
presence of macrophytes (Kersting and van den Brink 1997). Balczon and Pratt (1994) compared
effects of Cu on littoral and open water communities. Effects of Cu on oxygen production and
respiration were reduced in microcosms with an established littoral zone, most likely because of
greater adsorption and complexation by macrophytes and sediments. Although these results showing
variable responses in different ecosystems complicate our ability to make broad generalizations,
understanding the underlying mechanisms responsible for this variation may ultimately improve our
predictive ability.
Interactions between biotic and abiotic factors may also influence the response of primary pro-
ducers to contaminants. Steinman et al. (1992) observed that the physical structure and integrity
of periphyton mats influenced resistance and resilience of carbon fixation rates (a measure of
primary productivity) to chlorine exposure. Hill et al. (2000) measured bioaccumulation of Cd
by periphyton and subsequent effects on photosynthesis. Effects of Cd on photosynthesis were
regulated by periphyton biomass, with greater effects observed in treatments with less bio-
mass. Although there were differences in community composition among biomass treatments,
reduced effects in high biomass treatments were attributed to contaminant dilution and lower
Cd bioavailability.
32.2.1.3 Ecosystem Recovery
Although the short duration of many microcosm and mesocosm experiments precludes assessment
of recovery, some researchers have used these experimental systems to evaluate improvements in
ecosystem processes when contaminants are reduced or eliminated. Oviatt et al. (1984) measured

recovery of benthic respiration and nutrient flux for 21 months in mesocosms containing sediments
collected along a pollution gradient. Within 5 months, water quality characteristics (nutrients, chloro-
phyll a, and dissolved oxygen) and net system production were similar among treatments, indicating
that recovery may occur rapidly after pollutants are eliminated. Rapid recovery (4 weeks) of pho-
tosynthesis following exposure of marsh plants to crude oil was also reported by Pezeshki and
Deluane (1993). Similarly, periphyton productivity in outdoor experimental stream channels dosed
with the herbicide, hexazinone, was reduced by 80%, but recovered within 24 h following treatment
(Schneider et al. 1995). The estimated LC50 of hexazinone for periphyton production (3.6 µg/L)
was reported to be less than published values based on single species tests, demonstrating the greater
sensitivity of this functional measure.
32.2.1.4 Comparisons of Ecosystem Structure and Function
The majority of published microcosm and mesocosm experiments measure either structural or func-
tional characteristics. Because of concerns over sensitivity, variability, and the rate of response of
some functional indicators, we suggest that a practical application of these experimental systems
is to compare the efficacy of structural and functional endpoints. Questions such as the number of
replicates required to detect statistical differences between reference and treated microcosms and
the rate at which structural and functional variables respond to chemical stressors are of particular
importance. Rigorous control over exposureconditions and the ability to manipulate several variables
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 694 — #8
694 Ecotoxicology: A Comprehensive Treatment
simultaneously in microcosm and mesocosm experiments provide a unique opportunity to compare
effects of stressors on structural and functional characteristics (Culp et al. 2003).
The conventional wisdom is that, because of functional redundancy and greater variability of
functional measures, changes in community composition are likely to occur before alterations in eco-
system processes are observed (Schindler 1987, Schindler et al. 1985). However, like many examples
of conventional wisdom, there are exceptions to these generalizations in the literature. Some studies
have reported that functional measures are equally sensitive or even more sensitive than measures
of abundance, biomass, or community composition. Functional measures (periphyton productivity)
were considerably more sensitive than structural measures (periphyton biomass; macroinvertebrate

abundance and drift) to the herbicide hexazinone in outdoor stream mesocosms (Schneider et al.
1995). Concentration–response relationships between copper and several functional endpoints were
established by Hedtke (1984) in laboratory microcosms. GPP and respiration were reduced at 9.3 µg
Cu/L, but changes in community composition were observed only at higher concentrations of Cu
(30 µg/L), suggesting that ecosystem processes were more sensitive than structure in these experi-
ments. Clements (2004) reported that EC
10
values for heavy metals based on community respiration
and abundance of metal-sensitive species were similar. Jorgensen et al. (2000) calculated no effect
concentrations (NECs) for a variety of structural and functional measures in large pelagic meso-
cosms exposed to anionic surfactants (linear alkylbenzene sulfonates). Biomass (as chlorophyll a)
and biovolume of the dominant taxonomic groups were affected only at the highest concentrations
tested, whereas phytosynthetic activity was the most sensitive parameter for phytoplankton. After
4.5-day exposure, NECs for photosynthetic activity were similar to values for structural characterist-
ics (abundance of protozoans, crustaceans, and diatoms). Detenbeck et al. (1996) reported that gross
productivity of periphyton was significantly reduced in microcosms exposed to 15 µg/L of atrazine,
a concentration that significantly reduced survival of Daphnia but had no effect on other response
variables measured (biomass of cattails; growth of tadpoles and fathead minnows). Fairchild et al.
(1987) compared community composition, nutrient dynamics, leaf decomposition, and primary pro-
duction in experimental streams exposed to clean and contaminated (triphenyl phosphate) sediments.
Sediment exposures altered patterns of macroinvertebrate drift and increased nutrient retention, but
had no effects on leaf decomposition.
Some stream microcosm experiments have been conducted specifically to validate results of
laboratory toxicity tests and provide an opportunity to compare ecosystem functional measures
with more traditional toxicological endpoints. Exposure of stream mesocosms to relatively high
levels of Cd (143 µg/L) resulted in reduced abundance of grazing snails and increased periphyton
biomass, but had no effects on gross or net primary productivity (Brooks et al. 2004). Concurrent
single species toxicity tests with Ceriodaphnia dubia and Pimephales promelas showed that survival
was significantly reduced at this concentration. The lack of a response at lower Cd concentrations
(15 µg/L) was attributed to high concentrations of dissolved organic materials in these effluent-

dominated streams, which likely reduced metal bioavailability. Richardson and Kiffney (2000)
compared structural and functional measures in outdoor experimental streams dosed with mixtures
of metals. Significant concentration–response relationships were developed for several measures
related tomayfly abundance anddrift, but noeffects ofmetals on algal biomass or bacterial respiration
were observed. These researchers recommended that regulatory agencies should include estimates
of mayfly abundance and richness as indicators of metal impacts in streams.
Balczon and Pratt (1994) derived maximum allowable toxicant concentrations (MATCs) for
littoral and aquatic microbial microcosms exposed to Cu. The MATCs were generally greater for
process (photosynthesis, respiration) as compared to measures of community composition (species
richness, chlorophyll a biomass), indicating greater sensitivity of structural responses. Similar res-
ults were reported by Melendez et al. (1993) in which microbial communities were exposed to the
herbicide diquat. Changes in productivity and respiration were observed only at the two highest con-
centrations (10 and 30 mg/L), and these ecosystem-level responses recovered after 2 weeks exposure.
In contrast, the MATC for protozoan species richness and bacterial cell density was 0.32 mg/L, and
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 695 — #9
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 695
these responses showed little evidence of recovery. Barreiro and Pratt (1994) observed that gross
community productivity of periphyton was considerably more sensitive than chlorophyll a to diquat.
The lowest observable effect concentration (LOEL) for P/R in planktonic communities exposed to
fluorene, a polycyclic aromatic hydrocarbon (0.12 mg/L), was comparable to chronic toxicity values
based on single species tests with cladocerans, chironomids, and bluegill (Stay et al. 1988). How-
ever, the magnitude of change in ecosystem processes did not reflect the near complete elimination of
most zooplankton at concentrations exceeding 2 mg/L. Exposure of microbial communities derived
from natural sediments to a fungicide, herbicide, or insecticide reduced microbial biomass but had
no significant effects on respiration or denitrification (Widenfalk et al. 2004). These differences
among experiments suggest that not only is the relative sensitivity of structural and functional meas-
ures contaminant-specific, it may also vary with level of contamination and characteristics of the
exposure system.
In a comprehensive analysis of structural and functional responses of outdoor aquatic mesocosms

to the insecticide diflubenzuron, Boyle et al. (1996) observed relatively little effects on GPP, but
a significant increase in chlorophyll a, and reduced abundance and species richness of secondary
consumers (zooplankton, insects, and bluegill) in treated mesocosms (Figure 32.4). Although com-
munity metabolism and decomposition rates were affected in microcosms treated with chlorpyrifos,
these processes were generally less sensitive and occurred only after changes in structural measures,
suggesting functional redundancy of these systems (Brock et al. 1993). Cuppen et al. (2002) repor-
ted that no observable effects concentrations (NOECs) for decomposition rate of Populus leaves
and abundance of several macroinvertebrate shredders were similar. Interestingly, the structural and
Abundance
0
100
200
300
400
500
Richness
0
2
4
6
8
10
12
Zooplankton
Insects
Treatment
Control Monthly Biweekly
Biomass (µg/L) or production (mg O
2
/L /day)

0
10
20
30
40
50
Chlorophyll a
Production
Treatment
Control Monthly Biweekly
Biomass (kg/ha)
0
20
40
60
80
100
120
Bluegill (adults)
Bluegill (recriuts)
Zooplankton
Insects
FIGURE 32.4 Effects of the insecticide, diflubenzuron, on structural (abundance, biomass, richness) and
functional (primary production) measures in lentic mesocosms. Increased chlorophyll a biomass in mesocosms
treated monthly and biweekly compared with controls was attributed to reduced grazing pressure. (Data from
Boyle et al. (1996).)
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 696 — #10
696 Ecotoxicology: A Comprehensive Treatment
functional NOECs derived from this microcosm experiment were considerably less than the LC50

value derived from standard toxicity tests using known sensitive organisms. Because of the com-
plex and often unpredictable relationship between structural and functional measures observed in
some studies, we suggest that an appropriate strategy will be to include endpoints reflecting both
pattern and process when designing microcosm and mesocosm experiments. We agree with Brock
et al. (1993) that an understanding of contaminant effects on ecosystem function cannot be fully
appreciated without an understanding of community structure.
32.2.1.5 Effects of Contaminants on Other Functional
Measures
Although we traditionally consider changes in community composition to be a structural measure,
some researchers consider alteration in the abundance of groups that play an important functional role
(e.g., abundance of shredders in streams; abundance of grazing zooplankton in lakes) to be intimately
related to ecosystem processes and therefore an appropriate surrogate functional measure (Gruessner
and Watzin 1996, Wallaceet al. 1996). Field (Wallaceet al. 1982) and stream microcosm experiments
(Carlisle and Clements 1999) have assessed the effects of contaminants on functional feeding group
composition. The export or loss of materials from an ecosystem is an important functional process
that has received relatively little attention in the ecotoxicological literature. Similarly, emergence of
adult insects represents a net transfer of energy from aquatic to terrestrial habitats and therefore could
be considered a functional response. Gruessner and Watzin (1996) reported increased emergence of
insects in stream microcosms treated with atrazine. Culp et al. (2003) measured increased algal
biomass and changes in taxonomic composition in stream mesocosms dosed with 5% or 10% pulp
mill effluents. Although most measures of benthic macroinvertebrate community composition were
similar between treatments, emergence of mayflies was significantly reduced in treated streams.
Increased nutrient loading from nonpoint sources is expected to have significant impacts on aquatic
ecosystem structure and function. Elevated levels of nutrients are likely to produce excess organic
matter, which will result in greater biomass or increased export. An understanding of the ability
of an ecosystem to assimilate this excess production is necessary to predict the potential negative
effects of nutrient enrichment. Barron et al. (2003) observed no change in GPP, NPP, respiration,
or biomass following 27 months of nutrient addition in marine rocky intertidal mesocosms. Carbon
budgets calculated in this system showed that the lack of a response to nutrient enrichment resulted
from increased export of dissolved organic carbon. The ability of an ecosystem to export relatively

large amounts of excess carbon may offer some protection from nutrient enrichment in coastal areas.
32.2.2 MICROCOSMS AND MESOCOSMS IN TERRESTRIAL RESEARCH
While aquatic ecotoxicologists have long recognized the value of microcosms and mesocosms as
research tools for investigating effects of contaminants on ecosystem processes, these systems have
received considerably less attention in terrestrial ecotoxicology (Figure 32.1). Fraser and Keddy
(1997) reported that despite a general increase in the use of microcosms and mesocosms to address
basic and applied research questions during the mid-1990s, <5% of the studies were conducted in
terrestrial ecosystems. For practical reasons, much of the research using terrestrial microcosms and
mesocosms has focused on soil microbial systems. As described in previous chapters, alterations
in abundance and activity of soil microbes can have significant effects on decomposition rates and
nutrient processing. By examining both structure and function of soil communities, it is possible
to link direct and indirect effects of contaminants, and identify important regulating mechanisms
(Bogomolov et al. 1996). Although the vast majority (>95%) of soil respiration in terrestrial ecosys-
tems is a result of microbial activity, nematodes, arthropods, annelids, and other organisms contribute
significantly to decomposition. Experiments have been conducted to determine the relative contri-
butions of microbes and invertebrates to detrital food webs. Salminen et al. (2001) measured the
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 697 — #11
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 697
effects of heavy metals and detritivores (enchytraeid oligochaetes) on respiration in soil microcosms.
Invertebrate detritivores were eliminated at the highest Zn concentrations (>2500 mg/kg) and effects
of Zn on microbial respiration were dependent on detritivore density. Clear effects of Zn were only
observed in treatments with the greatest density of detritivores. These researchers reported some
evidence of functional redundancy, but noted that elimination of species that play an important role
in regulating soil microbial processes will have disproportionate impacts on ecosystems.
Unique properties of the soil environment may complicate our ability to assess bioavailability
and contaminant effects. For example, in aquatic ecosystems, the assumption that contaminants are
evenly distributed within the water column is generally valid. Although most experiments conducted
with soil microcosms attempt to achieve a relatively homogeneous distribution of contaminants,
chemicals in natural soils are often patchily distributed. In addition, small-scale spatial variation

in the physicochemical characteristics of soils may alter chemical bioavailability (Salminen and
Sulkava 1997). Some of the characteristics of soils that modify chemical bioavailability, such as
particle size and amount of organic material, are analogous to properties of aquatic sediments. We
will see that experimental designs that account for soil type and modify soil characteristics are a
common feature of many terrestrial microcosm experiments. In the following sections, we will
review some of the experiments conducted to assess the effects of heavy metals, organics, and other
stressors on soil processes.
32.2.2.1 Heavy Metals
Effects of heavy metals on ecosystem processes have been measured in soil microcosms containing
both natural and synthetic assemblages of microbes. In addition to measuring bacterial, fungal,
nematode, and arthropod abundance and biomass, typical functional endpoints reported in these
studies include soil respiration, nitrification, and N mineralization. Although some experimental
studies have directly measured leaf litter decay rates (Cotrufo et al. 1995, Kohler et al. 1995),
most of the research has focused on underlying microbial processes that regulate decomposition.
Bogomolov et al. (1996) measured a suite of structural and functional characteristics in microcosms
exposed to Cu. Increased pools of dissolved organic N and ammonium, reduced soil respiration, and
reduced litter decay were observed in Cu-treated microcosms. Soil respiration was the most sensitive
process examined, with effects observed at 50 mg Cu/L. These changes in ecosystem processes were
the result of direct toxic effects on structural measures (reduced microbial biomass and abundance
of nematodes).
As in aquatic ecosystems, one major advantage of microcosm and mesocosm experiments is
the ability to manipulate several independent variables or site characteristics to quantify factors
that determine contaminant effects and bioavailability. Khan and Scullion (2000) examined effects
of heavy metals on microbial biomass, respiration, and mineralization in soils with varying clay
and organic content. Metal bioavailability and effects were generally greater in sandy loams as
compared to soils with higher organic content. Ammonification and nitrification were found to be
more sensitive to Cd in calcareous soils than noncalcareous soils (Dusek 1995). Nitrate accumulated
in Cd-treated calcareous soils primarily as a result of the greater sensitivity of nitrite oxidizers.
Niklinska et al. (1998) established concentration–response relationships between heavy metals and
respiration in litter collected from beech-pine and oak forests. Despite differences in physical and

chemical characteristics of the two litter types, storage time was more important in controlling effects
of metals than litter type. Decreases in litter respiration rates with storage time were most likely a
result of rapid reduction in the amount of easily degraded material and/or increased respiration rate
immediately following litter collection (Niklinska et al. 1998). For litter respiration to be a useful
indicator of ecosystem responses to contaminants, a better understanding of the effects of storage
time is required. On the basis of the estimated EC50 values for respiration (Figure 32.5), Niklinska
et al. (1998) reported the following range of toxicity: Cu > Zn ≥ Cd  Pb. It is interesting to note
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 698 — #12
698 Ecotoxicology: A Comprehensive Treatment
Metal treatment
Cu Zn Cd Pb
EC
50
(mg/kg)
0
5,000
10,000
15,000
20,000
25,000
30,000
Beech-pine
Oak
FIGURE 32.5 Average EC50 values for soil respiration rates measured in microcosms exposed to heavy
metals. Forest litter was collected from beech-pine and oak-hornbeam forests. (Data from Niklinska et al.
(1998).)
Treatment
CC PC PP CP
Litter decay rate (k

)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
Control
litter
Polluted
litter
FIGURE 32.6 Litter decay rates of oak leaves (Quercus ilex) collected from control and metal polluted sites
after 2 months exposure to clean and metal contaminated soil. CC = soil and litter from control site; PC =
polluted soil and clean litter; PP polluted soil and polluted litter; CP = clean soil and polluted litter. (Data from
Table 3 in Cotrufo et al. (1995).)
that this order of toxicity is quite different from most aquatic studies in which Cd is significantly
more toxic to primary producers than either Cu or Zn.
Information concerning the relative toxicity of heavy metals in soil versus leaf litter is necessary
for remediation of metal contaminated sites. In microcosm experiments Cotrufo et al. (1995) com-
pared soil respiration, microbial and fungal biomass, and litter decomposition rates of oak leaves
collected from clean and metal-polluted sites. Lower respiration rates and reduced fungal abundance
were observed in litter contaminated by metals. Decomposition rates were significantly reduced for
metal-polluted litter, regardless of the soil source (Figure 32.6). These data indicate that heavy metals
in litter were directly responsible for reduced decomposition rates. The addition of organic material
through natural decomposition processes can reduce the effects of heavy metals in terrestrial eco-
systems. Boon et al. (1998) observed that the combined effects of low pH and Cu contamination on
soil ecosystem processes were significantly reduced in microcosms planted with Cu-tolerant grass.
The positive influence of Cu-tolerant plants on soil processes, which provided organic material and
© 2008 by Taylor & Francis Group, LLC

Clements: “3357_c032” — 2007/11/9 — 12:39 — page 699 — #13
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 699
reduced metal bioavailability, has important implications for restoring habitats impacted by heavy
metals.
32.2.2.2 Organic Contaminants and Other Stressors
Microcosms and mesocosms have been used to examine the effects of organic contaminants in
terrestrial ecosystems, with considerable effort devoted to assessing responses of soil communities.
Some papers have adopted a comparative approach and examined responses to a large number
of organic chemicals on a few ecosystem-level endpoints (Pell et al. 1998). Others have examined
effects of a single chemical or class of chemicals on several ecosystem processes. An excellent series
of papers describing the use of terrestrial microcosms in ecological risk assessment was published
in the journal Ecotoxicology in 2004 (Volume 13, Issue 4). This series of publications was the result
of a joint effort by university, private, and governmental partners to develop a standardized method
for conducting microcosm experiments in terrestrial ecosystems. The terrestrial model ecosystems
(TME) experiments were conducted with intact soil cores collected from several different field sites.
Initial experiments focused on ecosystem responses to carbendazim, a fungicide used extensively
for agricultural applications in Europe. Experiments conducted at sites in the United Kingdom,
Germany, Portugal, and the Netherlands examined effects of carbendazim on nutrient cycling and
organic matter processing. In general, nutrient dynamics were not affected by contaminant exposure
during the 16-week experiment, a result that was also supported by field experiments conducted
simultaneously (Van Gestel et al. 2004). In contrast to these results, Burrows and Edwards (2004)
reported significant effects of carbendazim on nutrient dynamics, soil dehydrogenase activity, and
several structural measures in soil microcosms treated with similar carbendazim concentrations.
Alteration in decomposition rate of organic matter is likely to affect nutrient dynamics and is
therefore considered an integrative functional endpoint in microcosm tests. Because decomposition
and nutrient dynamics in soils are closely coupled and regulated by microbial processes, structural
changes in microbial communities are likely to have important effects on ecosystem function. The
fungicide dithianon had relatively little influence on decomposition rates but significantly inhibited
microbial activity in soil microcosms (Liebich et al. 2003). These alterations in functional processes
corresponded to changes in microbial community composition and fungal biomass. Forster et al.

(2004) used cellulose paper as a standardized material to measure decomposition and assess inver-
tebrate feeding activity. A significant concentration–response relationship between carbendazim
concentration and decomposition rate was observed, with treated microcosms having 40%–80%
lower decomposition and showing a significant reduction in invertebrate feeding activity. Similar
LC50 values were calculated for both microcosm and field experiments (7.1–9.5 kg/Ha), lending
additional support for the application of TMEs to assess contaminant effects on soil processes.
Soil respiration, measured as evolution of CO
2
, is also a sensitive endpoint in microcosm experi-
ments. Salminen et al. (2002) compared the effects of several organic chemicals and heavy metals on
microbial respiration rate in soil microcosms. This functional measure was correlated with microbial
biomass and estimates of microarthropod and nematode abundance. Soil respiration was reduced in
all contaminant treatments, and effects increased with chemical concentration. Changes in respira-
tion were accompanied by decreases in microbial biomass and abundance of soil organisms resulting
from direct toxicity in most treatments. Because some of these responses were not observed until
late in the study, Salminen et al. (2002) recommend that soil microcosm experiments should be
conducted for a sufficient period of time to avoid false conclusions regarding responses.
Few investigators have conducted comparative studies of organic chemicals under different
physicochemical conditions in soil microcosm experiments. Chen andcolleagues (Chen and Edwards
2001, Chen et al. 2001) employed soil microcosms to examine the effects of fungicides on several
ecosystem processes. Soil properties, especially soil texture, regulated the effects of fungicides on
soil microbial activity and nutrient dynamics (Chen and Edwards 2001). Soil microcosms amended
with either alfalfa leaves or wheat straw, materials with very different C:N ratios (alfalfa leaves,
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 700 — #14
700 Ecotoxicology: A Comprehensive Treatment
C:N = 8.9; wheat straw C:N = 84), showed variable responses to the fungicides benomyl, captan,
and chlorothalonil (Chen et al. 2001). At recommended application rates, each of the fungicides
reduced soil respiration rates by 30%–50% in unamended soils; however, effects in soils amended
with alfalfa leaves or wheat straw varied among the three fungicides. Martinez-Toledo et al. (1996)

measured the effects of the herbicide simazine in soil microcosms with different physicochemical
characteristics. Simazine had relatively little effect on most structuraland functionalcharacteristics of
soil microflora. However, abundance of nitrifying bacteria was significantly reduced at application
rates normally used in agriculture, with greatest effects in soils with low organic content. These
researchers speculated that alterations in abundance of nitrifying bacteria are likely to have long-term
consequences for nutrient dynamics in soils and may disrupt the balance among nitrogen fixation,
denitrification, and nitrification.
Riparian habitats are located at the interface between terrestrial and aquatic ecosystems, and often
function as buffers to protect lakes and streams from anthropogenic disturbances. For example, pro-
cesses that occur in riparian soils, such as denitrification and microbial mineralization, can reduce
the effects of N enrichment in riparian ecosystems. The size of riparian buffers required to pro-
tect stream ecosystems from negative effects of agricultural runoff, urban discharges, and other
stressors has received considerable attention in the literature (Harding et al. 1998, Stewart et al.
2001, Wang et al. 1997). Ettema et al. (1999) compared the effect of N addition to soil micro-
cosms placed in two zones along a riparian corridor. Although microbial biomass and respiration
were not affected by N treatments in either zone, N added to microcosms located in the near-stream
zone was effectively removed by denitrification. In contrast, addition of N to an upslope zone did
not stimulate denitrification. These results suggest that denitrification can provide some protection
against increased N addition, but that responses vary among soil types and are relatively ephemeral
(Ettema et al. 1999).
Although top-down models of food web structure are frequently used to characterize aquatic eco-
systems, it is uncertain if these models are appropriate for soil communities. Many soil communities
are highly dependent on processes occurring in decomposer food webs and are therefore likely to
be controlled by bottom-up processes. Salminen and Sulkava (1997) measured microbial biomass,
abundance of soil invertebrates, and nutrient dynamics in soil microcosms treated with sodium penta-
chlorophenol (PCP), a wood preservative known to be toxic to many organisms. Reduced microbial
biomass resulted in the accumulation of nutrients in treated microcosms (Figure 32.7). Because of the
strong influence of PCP on the microbial food web and the similar sensitivity of predators and prey
to PCP, no evidence for top-down control was observed. On the basis of these results, it appears that
0

100
200
300
NH
4
+
-N (µg/g)
400
500
600
700
Control
Low
High
(a)
PO
4
3−
-P (µg/g)
Time
Week 2 Week 4 Week 8 Week 18
Time
Week 2 Week 4 Week 8 Week 18
0
20
40
60
80
100
(b)

FIGURE 32.7 Concentrations of NH
4
and PO
4
in soil microcosms exposed to sodium pentachlorophenate
(PCP) over an 18-week period. Low and high treatments correspond to 50 and 500 mg PCP/kg, respectively.
(Data from Table 3 in Salminen and Sulkava (1997).)
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 701 — #15
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 701
soil decomposer food webs are regulated by bottom-up processes and therefore traditional top-down
models employed in many aquatic systems are unsuitable for measuring effects of contaminants in
soils (Salminen and Sulkava 1997).
32.3 WHOLE ECOSYSTEM EXPERIMENTS
Ecosystem experiments are the most direct method available for improving predictions of environmental
response to management or inadvertent perturbation.
(Carpenter et al. 1995)
Whole ecosystem experiments have often been employed to assess the effects of contaminants on
functional processes. In fact, ecosystem-scale experiments designed to test the effectiveness of new
fertilizers, pesticides, and other crop treatments for agricultural applications played a prominent
role in the history of experimental design and statistical analyses. However, unlike argoecosystem
experiments that often have well-defined management objectives (e.g., increase crop production) and
can provide results in relatively short periods of time, conducting experiments in natural ecosystems
is considerably more challenging (Schindler 1990). Tests of Odum’s (1985) hypotheses concerning
how ecosystem processes should be affected by stress are generally not well supported in theliterature
(Rapport et al. 1985, 1998; Schindler 1990), highlighting the difficulty of predicting how ecosystems
will respond to perturbations.
Although the distinction between large mesocosm and whole ecosystem experiments is often
blurred, especially in terrestrial and marine systems, we will use the conventional definition of
Odum (1984) and consider mesocosms as partially enclosed experimental systems. Thus, whole

ecosystem experiments involve the planned application of contaminants or other stressors directly
to a system that is large enough to contain all of the important physical, chemical, and biological
processes of interest (Carpenter et al. 1998). In Chapter 23, we discussed the use of whole ecosystem
experiments for examining effects of contaminants on community structure and composition. Here
we will limit our discussion primarily to effects of stressors on ecosystem processes that regulate the
movement of materials and energy. As with experiments designed to assess changes in community
structure, there has been considerable discussion in the ecological literature concerning the trade-
offs between replication and ecological realism. High variability and the large number of replicates
necessary to obtain sufficient statistical power in ecosystem experiments complicate the use of
hypothesis testing approaches (Carpenter 1989). In fact, because most ecosystem experiments do
not involve true replication, traditional approaches that require hypothesis testing for assessing
statistical significanceare often inappropriate. Instead ofrelying on hypothesistests that often involve
insufficient statistical power (Carpenter 1989), approaches that evaluate alternative explanations
may be more useful. Instead of using additional ecosystems as replicates, investigators should
employ strong inference (Platt 1964), and experiments should be designed to evaluate multiple
alternative explanations (Carpenter et al. 1998). For example, comparing temporal trends in several
ecosystems that differed in important physical, chemical, or biological characteristics subjected to
the same stressor will allow researchers to evaluate alternative models. Note that this approach
differs in an important way from simple descriptive or comparative studies because the systems are
actually manipulated.Avarietyof statistical techniques, including interventionanalyses, multivariate
autoregressive models, dynamic linear models, and repeated measures, have been employed to
analyze changes in unreplicated ecosystem experiments.
32.3.1 AQUATIC ECOSYSTEMS
Experimental manipulation of ecosystems was originallyemployed in aquatic habitats when Chancey
Juday, a limnologist at the University of Wisconsin, conducted one of the first whole lake nutrient
enrichment experiments (Juday and Schloemer 1938). The tradition of whole ecosystem experiments
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 702 — #16
702 Ecotoxicology: A Comprehensive Treatment
continued at the University of Wisconsin when Arthur D. Hasler, a graduate student of Juday, added

lime to a brown water lake (Hasler et al. 1951). The primary objective of these early experiments was
to improve fisheries, but subsequent experiments were designed to examine effects of anthropogenic
stressors. Future students of Hasler’s, recognizing the usefulness of this experimental approach, went
on to establish two of the best-known whole ecosystem experimental facilities in North America:
The Hubbard Brook Ecosystem study area in New Hampshire and the Experimental Lakes Area in
Ontario, Canada.
As a consequence of these early experiments, a large amount of information concerning eco-
system responses of lakes and streams to anthropogenic stressors has been obtained (Carpenter
1989, Likens 1992, Likens et al. 1970, Schindler 1988, Wallace et al. 1986). The application of
valid experimental controls, analogous to experimental approaches used in agriculture, was a major
development in ecosystem sciences and allowed researchers to rigorously apply the scientific method
to test hypotheses regarding how ecosystems responded to perturbations (Likens 1985). Many of
these experiments have emphasized functional responses to anthropogenic perturbations and some
have provided a quantitative assessment of responses across levels of biological organization (Cot-
tingham and Carpenter 1998). The classic set of whole lake experiments conducted by Schindler and
colleagues at the Experimental Lakes Area (ELA) in Canada represented an important turning point
in the history of ecosystem research.
Some of the findings of this research seemed to question the usefulness of functional measures
for understanding how ecosystems respond to anthropogenic disturbances (Carpenter et al. 1995,
Howarth 1991, Odum 1985, Schindler 1987, 1990). In a review of several years of experiments
investigating whole-lake responses to eutrophication, acidification, and heavy metals, Schindler
(1987) concluded that measures of ecosystem processes, including primary production, nutrient
cycling, and respiration were relatively insensitive to perturbations and therefore “poor indicators of
early stress.” Table 32.1 summarizes major findings from ecosystem-level experiments conducted at
the ELA and illustrates Schindler’s perspectives on the potential limitations of ecosystem processes
as early warning indicators of perturbation. This somewhat pessimistic evaluation of the usefulness
of ecosystem processes for assessing stressor effects has led some researchers to abandon functional
endpoints altogether and rely entirely on structural measures. Issues such as functional redundancy,
feedback mechanisms, lower sensitivity, and high variability of ecosystem process variables were
discussed in Chapter 23 and are routinely cited in the literature as justification for focusing on

TABLE 32.1
General Responses of Aquatic Ecosystems and Mesocosms in the Experimental Lakes Area
to Chemical Perturbations
1. Phytoplankton biomass and production are limited primarily by the input of phosphorus and the rate of water renewal,
regardless of pollutant stress.
2. The rate of decomposition is governed by primary production and is not affected by pollutants.
3. Nutrient cycling is not affected by input of toxicants.
4. Abundance of species with short life cycles can respond quickly to perturbations and maintain ecosystem function.
Abundance of these species may serve as early indicators of ecosystem stress.
5. Diversity indices are less useful than changes in community composition for detecting stress in ecosystems.
6. Species with short life cycles and poor powers of dispersal are generally most sensitive to perturbation.
7. The most sensitive indicators of stress in most aquatic ecosystems include demographic changes in short-lived species
and changes in community structure.
8. Short-term toxicity tests and measures of ecosystem-level processes are not sufficiently sensitive to be used as early
indicators of ecosystem stress.
From Table 3 in Schindler (1987).
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 703 — #17
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 703
lower levels of biological organization. However, we feel that abandonment of process variables as
ecological endpoints in ecosystem experiments may be premature for some situations. For example,
Schindler (1987) notes that production is a very sensitive indicator of perturbation in some terrestrial
ecosystems. In contrast to the generalization that ecosystem variables are less sensitive, Cottingham
and Carpenter (1998) reported that primary production was a better indicator of nutrient enrichment
than abundance of individual species. Acomprehensive characterization of how ecosystems respond
to chemical perturbations requires that both patterns (structural measures) and processes (functional
measures) be considered. We agree with Cottingham and Carpenter (1998) that rather than asking
the question about which endpoints are better, a more appropriate approach is to develop general
guidelines for understanding why some indicators work better in some systems or for some classes of
stressors. It is quite likely that a list of sensitive indicators for toxic chemicals would be quite different

from a list of sensitive indicators for nutrient enrichment, climate change, or physical disturbance.
Two sets of experiments conducted in stream ecosystems deserve special attention because they
were designed specifically to assess the relationship between structural and functional measures.
Details of experiments conducted by Bruce Wallace and colleagues at Coweeta Hydrologic Labor-
atory (North Carolina, USA) were discussed in Chapter 23. Briefly, experimental introduction of
the pesticide, methoxychlor, into a small headwater stream significantly reduced leaf litter decom-
position (Figure 32.8) and downstream export of particulate organic material (Cuffney et al. 1984,
Wallace et al. 1982). Results showed significant variation among leaf species and that litter decom-
position was a sensitive indicator of contaminant effects. These alterations in detritus dynamics were
not associated with changes in microbial activity and were shown to be a direct result of mortality
to macroinvertebrate shredders and associated reductions in litter fragmentation. Higher production
to biomass ratios (P/B) in the treated stream resulted from a shift in community structure to smaller
organisms with shorter generation times and faster turnover.
A second set of whole ecosystem experiments conducted in the Kuparuk River, an Alaskan
(USA) tundra stream, was designed to assess the effects of P additions on community structure and
function (Peterson et al. 1993). Similar to the whole lake experiments conducted at the ELA, this
set of experiments relied on long-term assessments of unreplicated treatments. The experiments are
especially significant because of the long duration (16 years) of the manipulations and the large
diversity of response variables measured (Slavik et al. 2004). Summarizing the first 4 years of these
Dogwood
Red maple
White oak
Rhododendrum
Decay rate (k)
0.000
0.005
0.010
0.015
0.020
Reference stream

Treated stream
FIGURE 32.8 Exponential breakdown rates of leaves (mean ± 95% confidence limit) in reference streams
and streams treated with the pesticide methoxychlor at Coweeta Hydrologic Laboratory. (Data from Table 2 in
Wallace et al. (1982).)
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 704 — #18
704 Ecotoxicology: A Comprehensive Treatment
Phosphate
addition
Algal
production
Production
of Baetis and
Brachycentrus
Grayling
growth and
production
Bacterial
production
Decomposition
of allochthonous
POM and DOM
FIGURE 32.9 Observed and hypothesized top-down and bottom-up responses in an Arctic stream food web
following experimental addition of phosphorus. (Modified from Figure 15 in Peterson et al. (1993).)
experiments, Peterson et al. (1993) reported increased algal biomass and productivity in fertilized
reaches, but little change in diatom community composition. Decomposition rate of Carex litter was
also unaffected by P addition. Although total secondary productivity of insects was also unaffected
by nutrient addition, there was considerable variation among species, with some groups increasing
and others decreasing in the treated reach. Interestingly, Peterson et al. (1993) found evidence for
strong bottom-up effects of P additions on benthic food webs in the first 2 years of the experiment,

followed by strong top-down feedback in later years (Figure 32.9). Afollow-up paper published after
16 years of P enrichment reported a dramatic increase in abundance of the bryophyte Hygrohypnum,
an aquatic moss that replaced epilithic diatoms, resulting in significant changes in the benthic habitat
and a four times increase in NH
+
4
uptake (Slavik et al. 2004). The authors concluded that even
relatively long-term experimental manipulations (e.g., 4–8 years) were not adequate to predict these
striking responses to nutrient manipulations.
32.3.2 TERRESTRIAL ECOSYSTEMS
Changes in many different types of ecosystems are similar and predictable.
(Woodwell 1970)
Experiments conducted in natural terrestrial ecosystems (e.g., non-agroecosystems) to assess the
effects of anthropogenic stressors have been conducted for at least 50 years. Many of these
experiments emphasized structural responses to anthropogenic disturbance and were discussed in
Chapter 23. The pioneering work of George Woodwell in the 1960s at Brookhaven National Labor-
atory, New York (Woodwell 1970) exposed an oak-pine forest to chronic gamma radiation (
137
Cs).
Most of the focus of Woodwell’s experiments was on changes in vegetation structure and com-
munity composition. However, the mechanism proposed to account for these structural changes
invoked alterations in the relationship between GPP and respiration. Woodwell (1970) speculated
that larger plants such as trees are at their physiological limit in terms of the amount of surface area
available for respiration. Consequently, smaller plants are favored in disturbed environments because
they can withstand more damage to respiratory surfaces. Woodwell’s findings regarding responses to
anthropogenic disturbance were quite consistent with Odum’s “strategy of ecosystem development”
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 705 — #19
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 705
(Odum 1969) and stimulated considerable interest among terrestrial ecologists to seek broader gen-

eralizations. Enthusiasm for identifying consistent responses to disturbance persists (Rapport et al.
1998), although many terrestrial ecologists have beenfrustrated by the variability in responsesamong
ecosystems. A general consensus has emerged that greater understanding of terrestrial ecosystem
responses to disturbance can be achieved through controlled, long-term experiments.
Field experiments in terrestrial ecosystems have been designed to assess both direct toxicological
effects of contaminants and the indirect effects associated with the loss of food resources. Barrett
(1968) measured effects of the insecticide carbaryl on structural and functional characteristics in
a grassland ecosystem. Exposure to carbaryl had significant direct and indirect effects on insect
abundance, litter decomposition, and small mammal production. Not surprisingly, there was no
effect of the insecticide on NPP or plant community composition. However, biomass and abundance
of arthropods were reduced by >95% and litter decomposition was significantly lower in treated
plots. Reduced litter decomposition was attributed to the loss of microarthropods.
Field experiments conducted to assess effects of heavy metals on ecosystem processes are relat-
ively uncommon in the ecotoxicological literature. Korthals et al. (1996) examined long-term effects
(10-year exposure) of copper and low pH on microbial activity, respiration, and the abundance of
bacteria and nematodes in an agroecosystem (Figure 32.10). In addition to the relatively long dura-
tion of the study, this field experiment is significant because it examined interactions between two
stressors. The combination of high Cu and low pH reduced bacterial growth and abundance of nem-
atodes, but had little effect on other structural or functional endpoints. Significant reductions in the
abundance of herbivorous nematodes were attributed to changes in primary production and loss of
food resources. Laskowski et al. (1994) tested the hypothesis that mineral nutrients such as Ca, Mg,
and K would reduce the effects of heavy metals on forest litter respiration. Exposure of microbial
communities to Cu, Cd, Pb, and Zn reduced respiration rate as predicted, but mineral nutrients either
increased toxicity or had no effects on respiration rates.
Many of the recent experiments that emphasized functional processes in ecosystems examined
responses to nutrient additions, especially N. Inputs of N from agricultural, industrial, and domestic
sources to terrestrialecosystems have been recognizedas a serious environmentalproblem for several
decades. Deposition of N and the process of N saturation were considered potential threats to forest
ecosystems in Europe in the mid-1980s (Nihlgard 1985). Unlike toxic chemicals in which ecosystem
responses are often immediate and typically result in removal of sensitive species, subsidizing

Copper treatment
Control Low Medium High
Abundance (no. /100 g)
0
200
400
600
800
1000
1200
1400
1600
1800
6.1
5.4
4.7
4.1
Soil pH
FIGURE 32.10 Mean abundance of the plant-feeding nematode Pratylenchus in an agroecosystem after
10 years of exposure to copper and pH. Low, medium and high Cu treatments refer to 250, 500, and 750 kg
Cu per Ha, respectively. Reduced abundance of these herbivores was attributed to reduced primary production
and the loss of food resources. (Data from Table 2 in Korthals et al. (1996).)
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 706 — #20
706 Ecotoxicology: A Comprehensive Treatment
Treatment
Control Low N High N
Total N losses (g/m
2
)

0
5
10
15
20
25
Pine
Hardwood
FIGURE 32.11 Total N losses following 9 years of nitrogen addition (1988–1996) to experimental plots on
a pine plantation and hardwood forest at the Harvard Forest Long-Term Ecological Research site. (Data from
Table 3 in Magill et al. (2000).)
stressors such as N deposition can initially have positive effects. This “paradox of enrichment”
(Rosenzweig 1971) greatly complicates assessments of N deposition effects on ecosystems. For
example, while productivity of forest ecosystems is often related to N availability, long-term chronic
exposure to N deposition is detrimental and export of excess N may damage adjacent watersheds.
N saturationof anecosystem occurs when N outputapproximates orexceeds input (Argen andBosatta
1988). Although N saturation is an extreme and relatively unusual phenomenon, many disturbed
forest ecosystems either “leak” nutrients or have experienced damages because of N overloads. The
effects of N deposition on forest ecosystems were examined at Hubbard Brook Experimental Forest,
New Hampshire (Christ et al. 1995). These researchers reported that treated plots retained >95% of
the N added, and that effects of excess N on soil processes such as ammonification varied with soil
horizon. In contrast to expectations, excess N was not nitrified and appeared to be limited by some
other factor(s).
Although ecologists have long speculated about the potential ecosystem-level threats of N depos-
ition, there is relatively little information concerning how different forests will respond to N
saturation. The Chronic Nitrogen Amendment experiment conducted at the Harvard Forest Long-
Term Ecological Research site (Massachusetts, USA) compared responses of a pine plantation and
native deciduous broad-leaved (hardwood) forest to N enrichment (as ammonium nitrate, NH
4
NO

3
)
over a 9-year period (Magill et al. 1997, 2000). Total N inputs in treated plots were increased 6–19×
above background levels. These two forest types showed major differences in responses to N addi-
tion (Figure 32.11), and some responses could not be discerned until late in the experiment. Nitrate
losses were observed after 1 year of N additions and woody biomass production decreased in the
pine plantation. In contrast, the hardwood forest showed no indication of nitrate loss until year 8 of
treatment, and wood production increased under high N levels. Not only do these results demonstrate
variability in responses of two ecosystems to the same stressor, they support the contention of other
researchers that have questioned the adequacy of short-term ecosystem experiments for assessing
effects of N deposition (Milchunas and Lauenroth 1995). Despite significant responses to treatments,
one of the most interesting results of these experiments was the observation that forest ecosystems
were highly retentive of N, especially the hardwood forest that retained 96% of the N added (Magill
et al. 2000).
32.4 SUMMARY
Microcosm experiments, mesocosm experiments, and ecosystem-level manipulations are powerful
approaches for quantifying direct andindirect effects ofcontaminants onecosystem processes. Losses
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 707 — #21
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 707
of structural componentsof an ecosystem, such as reducedabundance of important functionalfeeding
groups, often have significant consequences for movement of materials and energy. Conversely,
alterations in some ecosystem processes, such as lower primary production, can result in structural
changes in communities dependent on these resources. Experimental approaches are often necessary
to characterize the complex relationships between patterns and processes. Despite the strength of
large scale manipulations for this purpose, clearly it will not be possible to conduct experiments for
all ecosystem and stressor combinations. One area where whole ecosystem experiments have clearly
lagged is in marine ecosystems. Some of the pioneering experiments in mesocosm research were
initially conductedin marine habitatsto investigate effects of oil spills (Chapter 23). However, whole-
ecosystem experiments in marine environments offer unique challenges. Because of the expense

and logistical difficulties associated with conducting ecosystem-level manipulations, as well as the
limited number of sites where these experiments can be conducted, a more efficient strategy is
necessary that will allow investigators to make generalizations from the current body of literature
(Schindler 1988). Attempts to generalize among ecosystem responses using conceptual models such
as the ecosystem distress syndrome (Chapter 25) will be useful for this purpose. However, large-scale
experimental manipulation of some classes of stressors, including elevated CO
2
, UV-B radiation and
atmospheric N, present significant logistical problems and will be examined in a later chapter. We
believe that a more efficient strategy for generalizing among ecosystem and stressor types is to
integrate what we already know about ecosystem responses to perturbations with a well-designed
program of observational and small-scale experimental approaches to identify critical questions
that can only be addressed using large-scale experiments. This strategy will allow researchers to
focus limited resources on those issues that are central to understanding how ecosystems respond to
anthropogenic perturbations.
32.4.1 SUMMARY OF FOUNDATION CONCEPTS AND PARADIGMS
• Isolating causal mechanisms is particularly important in ecosystem-level studies because
ecosystem processes are often complex and controlled by direct and indirect effects.
• Because of the limited spatiotemporal scale, measuring responses of ecosystem processes
to contaminants in microcosms presents significant challenges.
• The use of controlled experimental systems in aquatic ecology has clearly increased
over time.
• One of the more significant challenges associated with the use of microcosms and meso-
cosms is the change in functional measures over time, independent of the effects of
contaminants.
• Duration of experiments is especially important when considering ecosystem processes
that often show delayed responses compared with structural alterations.
• Most experiments conducted in stream microcosms and mesocosms have focused on
functional responses of periphyton.
• Mesocosm experiments also allow investigators to compare responses across levels of bio-

logical organization, thereby providing opportunities to examine underlying mechanisms
and correlate structural and functional alterations.
• Microcosm and mesocosm experiments can also be used to compare functional responses
of communities derived from different sources, thereby providing an opportunity to
understand how intrinsic features of an ecosystem may influence susceptibility to
contaminants.
• Because of questions concerning the sensitivity and variability of ecosystem processes,
one important practical application of mesocosm experiments is to compare the efficacy
of structural and functional endpoints.
• While aquatic ecotoxicologists have long recognized the value of microcosms and
mesocosms as research tools for investigating effects of contaminants on ecosystem
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 708 — #22
708 Ecotoxicology: A Comprehensive Treatment
processes, these systems have received considerably less attention in terrestrial
ecotoxicology.
• Typical functional endpoints reported in soil microcosm experiments include soil
respiration, nitrification, and N mineralization.
• Because decomposition and nutrient dynamics in soils are regulated by microbial pro-
cesses, structural changes in microbial communities are likely to have important effects
on ecosystem function.
• Alterations in abundance of nitrifying bacteria in soils are likely to have long-term
consequences for nutrient dynamics.
• Many soil communities are highly dependent on processes occurring in decomposer food
webs and are therefore more likely to be controlled by bottom-up processes than top-down
processes.
• Whole ecosystem experiments involve the planned application of contaminants or other
stressors directly to a system that is large enough to contain all of the important physical,
chemical, and biological processes of interest.
• A comprehensive characterization of how ecosystems respond to chemical perturbations

requires that both patterns (structural measures) and processes (functional measures) be
considered.
• It is quite likely that a list of sensitive indicators for toxic chemicals would be quite
different from a list of sensitive indicators for nutrient enrichment, climate change, or
physical disturbance.
• Improved understanding of terrestrial ecosystem responses to anthropogenic disturbance
can be achieved through controlled, long-term experiments.
• Unlike toxic chemicals in which ecosystem responses are often immediate and typically
result in removal of sensitive species, subsidizing stressors such as N deposition can
initially have positive effects on ecosystem processes.
• Alterations in structural components of an ecosystem, such as reduced abundance of
important functional feeding groups, often have significant consequences for movement
of materials and energy.
• Alterations in ecosystem processes, such as reduced primary production, can result in
structural changes in communities dependent on these resources.
REFERENCES
Argen, G.I. and Bosatta, E., Nitrogen saturation of terrestrial ecosystems, Environ. Pollut., 54, 185–197, 1988.
Balczon, J.M. and Pratt, J.R., A comparison of the responses of 2 microcosm designs to a toxic input of copper,
Hydrobiologia, 281, 101–114, 1994.
Barreiro, R. and Pratt, J.R., Interaction of toxicants and communities: The role of nutrients, Environ. Toxicol.
Chem., 13, 361–368, 1994.
Barrett, G.W., The effects of an acute insecticide stress on a semi-enclosed grassland ecosystem, Ecology, 49,
1019–1035, 1968.
Barron, C., Marba, N., Duarte, C.M., Pedersen, M.F., Lindblad, C., Kersting, K., Moy, F., and Bokn, T., High
organic carbon export precludes eutrophication responses in experimental rocky shore communities,
Ecosystems, 6, 144–153, 2003.
Bester, K., Huhnerfuss, H., Brockmann, U., andRick, H.J., Biological effectsof triazine herbicide contamination
on marine phytoplankton, Arch. Environ. Contam. Toxicol., 29, 277–283, 1995.
Bogomolov, D.M., Chen, S.K., Parmelee, R.W., Subler, S., and Edwards, C.A., An ecosystem approach to
soil toxicity testing: A study of copper contamination in laboratory soil microcosms, Appl. Soil Ecol.,

4, 95–105, 1996.
Bokn, T.L., Duarte, C.M., Pedersen, M.F., Marba, N., Moy, F.E., Barron, C., Bjerkeng, B., et al., The response
of experimental rocky shore communities to nutrient additions, Ecosystems, 6, 577–594, 2003.
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 709 — #23
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 709
Boon, G.T., Bouwman, L.A., Bloem, J., and Romkens, P.F.A.M., Effects of a copper-tolerant grass (Agrostis
capillaris) on the ecosystem of a copper-contaminated arable soil, Environ. Toxicol. Chem., 17,
1964–1971, 1998.
Boyle, T.P., Fairchild, J.F., Robinson-Wilson, E.F., Haverland, P.S., and Lebo, J.A., Ecological restructuring
in experimental aquatic mesocosms due to the application of diflubenzuron, Environ. Toxicol. Chem.,
15, 1806–1814, 1996.
Brock, T.C.M., Vet, J.J.R.M., Kerkhofs, M.J.J., Lijzen, J., Vanzuilekom, W.J., and Gijlstra, R., Fate and effects
of the insecticide Dursban(R) 4e in indoor Elodea-dominated and macrophyte-free fresh-water model-
ecosystems: III. Aspects of ecosystem functioning, Arch. Environ. Contam. Toxicol., 25, 160–169,
1993.
Brooks, B.W., Stanley, J.K., White, J.C., Turner, P.K., Wu, K.B., and La Point, T.W., Laboratory and field
responses to cadmium: An experimental study in effluent-dominated stream mesocosms, Environ.
Toxicol. Chem., 23, 1057–1064, 2004.
Burrows, L.A. and Edwards, C.A., The use of integrated soil microcosms to assess the impact of carbendazim
on soil ecosystems, Ecotoxicology, 13, 143–161, 2004.
Cairns, J., Jr., The myth of the most sensitive species, BioScience, 36, 670–672, 1986.
Carlisle, D.M. and Clements, W.H., Sensitivity and variability of metrics used in biological assessments of
running waters, Environ. Toxicol. Chem., 18, 285–291, 1999.
Carman, K.R., Fleeger, J.W., Means, J.C., Pomarico, S.M., and McMillin, D.J., Experimental investigation of
the effects of polynuclear aromatic hydrocarbons on an estuarine sediment food web, Mar. Environ.
Res., 40, 289–318, 1995.
Carpenter, S.R., Replication and treatment strength in whole-lake experiments, Ecology, 70, 453–463, 1989.
Carpenter, S.R., Chisholm, S.W., Krebs, C.J., Schindler, D.W., and Wright, R.F., Ecosystem experiments,
Science, 269, 324–327, 1995.

Carpenter, S.R., Cole, J.T., Essington, T.E., Hodgson, J.R., Houser, J.N., Kitchell, J.F., and Pace, M.L.,
Evaluating alternative explanations in ecosystem experiments, Ecosystems, 1, 335–344, 1998.
Chen, S.K. and Edwards, C.A., A microcosm approach to assess the effects of fungicides on soil ecological
processes and plant growth: Comparisons of two soil types, Soil Biol. Biochem., 33, 1981–1991,
2001.
Chen, S.K., Edwards, C.A., and Subler, S., Effects of the fungicides benomyl, captan and chlorothalonil on soil
microbial activity and nitrogen dynamics in laboratoryincubations, Soil Biol. Biochem., 33, 1971–1980,
2001.
Christ, M., Zhang, Y.M., Likens, G.E., and Driscoll, C.T., Nitrogen-retention capacity of a northern hardwood
forest soil under ammonium-sulfate additions, Ecol. Appl., 5, 802–812, 1995.
Clements, W.H., Small-scale experiments support causal relationships between metal contamination and
macroinvertebrate community responses, Ecol. Appl., 14, 954–967, 2004.
Colwell, F.S., Hornor, S.G., and Cherry, D.S., Evidence of structural and functional adaptation in epilithon
exposed to zinc, Hydrobiologia, 171, 79–90, 1989.
Cotrufo, M.F., Desanto,A.V.,Alfani,A., Bartoli, G., and Decristofaro,A., Effects of urbanheavy-metal pollution
on organic-matter decomposition in Quercus ilex L Woods, Environ. Pollut., 89, 81–87, 1995 .
Cottingham, K.L. and Carpenter, S.R., Population, community, and ecosystem variates as ecological indicators:
Phytoplankton responses to whole-lake enrichment, Ecol. Appl., 8, 508–530, 1998.
Crane, M., Research needs for predictive multispecies tests in aquatic toxicology, Hydrobiologia, 346, 149–155,
1997.
Cuffney, T.F., Wallace, J.B., and Webster, J.R., Pesticide manipulation of a headwater stream: Invertebrate
responses and their significance for ecosystem processes, Freshw. Invert. Biol., 3, 153–171, 1984.
Culp, J.M., Cash, K.J., Glozier, N.E., and Brua, R.B., Effects of pulp mill effluent on benthic assemblages in
mesocosms along the Saint John River, Canada, Environ. Toxicol. Chem., 22, 2916–2925, 2003.
Cuppen, J.G.M., Crum, S.J.H., van den Heuvel, H.H., Smidt, R.A., and van den Brink, P.J., Effects of a mixture
of two insecticides in freshwater microcosms: I. Fate of chlorpyrifos and lindane and responses of
macroinvertebrates, Ecotoxicology, 11, 165–180, 2002.
Cuppen, J.G.M., Gylstra, R., Vanbeusekom, S., Budde, B.J., and Brock, T.C.M., Effects of nutrient loading and
insecticide application on the ecology of Elodea-dominated fresh-water microcosms. 3. Responses of
macroinvertebrate detritivores, breakdown of plant litter, and final conclusions, Arch. für Hydrobiol.,

134, 157–177, 1995.
© 2008 by Taylor & Francis Group, LLC
Clements: “3357_c032” — 2007/11/9 — 12:39 — page 710 — #24
710 Ecotoxicology: A Comprehensive Treatment
Detenbeck, N.E., Hermanutz, R., Allen, K., and Swift, M.C., Fate and effects of the herbicide atrazine in
flow-through wetland mesocosms, Environ. Toxicol. Chem., 15, 937–946, 1996.
Dusek, L., The effect of cadmium on the activity of nitrifying populations in two different grassland soils, Plant
Soil, 177, 43–53, 1995.
Ettema, C.H., Lowrance, R., and Coleman, D.C., Riparian soil response to surface nitrogen input: Temporal
changes in denitrification, labile and microbial C and N pools, and bacterial and fungal respiration, Soil
Biol. Biochem., 31, 1609–1624, 1999.
Fairchild, J.F., Boyle, T., English, W.R., and Rabeni, C., Effects of sediment and contaminated sediment
on structural and functional components of experimental stream ecosystems, Water Air Soil Pollut.,
36, 271–293, 1987.
Fisher, R.A., The Design of Experiments, 7th ed., Oliver and Boyd, Edinburgh, 1960.
Forster, B., van Gestel, C.A.M., Koolhaas, J.E., Nentwig, G., Rodrigues, J.M.L., Sousa, J.P., Jones, S.E., and
Knacker, T., Ring-testing and field-validation of a terrestrial model ecosystem (TME)—An instrument
for testing potentially harmful substances: Effects of carbendazim on organic matter breakdown and
soil fauna feeding activity, Ecotoxicology, 13, 129–141 , 2004.
Fraser, L.H. and Keddy, P., The rate of experimental microcosms in ecological research, Trends Ecol. Evol.,
12, 478–481, 1997.
Gruessner, B. and Watzin, M.C., Response of aquatic communities from a Vermont stream to environmentally
realistic atrazine exposure in laboratory microcosms, Environ. Toxicol. Chem., 15, 410–419, 1996.
Harding, J.S., Benfield, E.F., Bolstad, P.V., Helfman, G.S., and Jones, E.B.D., Stream biodiversity: The ghost
of land use past, Proc. Natl. Acad. Sci. USA, 95, 14843–14847, 1998.
Hasler, A.D., Brynildson, O.M., and Helm, W.T., Improving conditions for fish in brown-water bog lakes by
alkalization, J. Wildl. Manage., 15, 347–352, 1951.
Hedtke, S.F., Structure and function of copper-stressed aquatic microcosms, Aquat. Toxicol., 5, 227–244, 1984.
Hill, W.R., Bednarek,A.T., and Larsen, I.L., Cadmium sorptionand toxicity inautotrophic biofilms, Can. J. Fish.
Aquat. Sci., 57, 530–537, 2000.

Howarth, R.W., Comparative responses of aquatic ecosystems to toxic chemical stress, In Comparative Ana-
lyses of Ecosystems: Patterns, Mechanisms, and Theories, Cole, J., Lovett, G., and Findlay, S. (eds.),
Springer-Verlag, New York, 1991, pp. 169–195.
Jorgensen, E. and Christoffersen, K., Short-term effects of linear alkylbenzene sulfonate on freshwater plankton
studied under field conditions, Environ. Toxicol. Chem., 19, 904–911, 2000.
Juday, C. and Schloemer, C.L., Effect of fertilizers on plankton production and on fish growth in a Wisconsin
lake, Prog. Fish Cult., 40, 24–27, 1938.
Kennedy, J.H., Ammann, L.P., Waller, W.T., Warren, J.E., Hosmer, A.J., Cairns, S.H., Johnson, P.C., and
Graney, R.L., Using statistical power to optimize sensitivity of analysis of variance designs for
microcosms and mesocosms, Environ. Toxicol. Chem., 18, 113–117, 1999.
Kersting, K. and van den Brink, P.J., Effects of the insecticide Dursban(R)4e (active ingredient chlorpyrifos)
in outdoor experimental ditches: Responses of ecosystem metabolism, Environ. Toxicol. Chem., 16,
251–259, 1997.
Khan, M. and Scullion, J., Effect of soil on microbial responses to metal contamination, Environ. Pollut., 110,
115–125, 2000.
Kohler, H.R., Wein, C., Reiss, S., Storch, V., and Alberti, G., Impact of heavy metals on mass and energy flux
within the decomposition process in deciduous forests, Ecotoxicology, 4, 114–137, 1995.
Korthals, G.W., Alexiev, A.D., Lexmond, T.M., Kammenga, J.E., and Bongers, T., Long-term effects of copper
and pH on the nematode community in an agroecosystem, Environ. Toxicol. Chem., 15, 979–985,
1996.
Kraufvelin, P., Model ecosystem replicability challenged by the “soft” reality of a hard bottom mesocosm,
J. Exp. Mar. Biol. Ecol., 222, 247–267, 1998.
Kreutzweiser, D.P., Gringorten, J.L., Thomas, D.R., and Butcher, J.T., Functional effects of the bacterial
insecticide Bacillus thuringiensis var Kurstaki on aquatic microbial communities, Ecotoxicol. Environ.
Saf., 33, 271–280, 1996.
Kurtz, J.C., Devereux, R., Barkay, T., and Jonas, R.B., Evaluation of sediment slurry microcosms for modeling
microbial communities in estuarine sediments, Environ. Toxicol. Chem., 17, 1274–1281, 1998.
Laskowski, R., Maryanski, M., and Niklinska, M., Effect of heavy-metals and mineral nutrients on forest litter
respiration rate, Environ. Pollut., 84, 97–102, 1994.
© 2008 by Taylor & Francis Group, LLC

Clements: “3357_c032” — 2007/11/9 — 12:39 — page 711 — #25
Microcosms, Mesocosms, and Field Experiments to Assess Ecosystem Responses 711
Liebich, J., Schaffer, A., and Burauel, P., Structural and functional approach to studying pesticide side-effects
on specific soil functions, Environ. Toxicol. Chem., 22, 784–790, 2003.
Likens, G.E., An experimental approach for the study of ecosystems—The 5th Tansley Lecture, J. Ecol., 73,
381–396, 1985.
Likens, G.E., The Ecosystem Approach: Its Use and Abuse, Oldendorf/Luhe, Germany, Ecology Institute, 1992.
Likens, G.E., Bormann, F.H., Johnson, N.M., Fisher, D.W., and Pierce, R.S., Effects of forest cutting and herbi-
cide treatment on nutrient budgets in Hubbard Brook Watershed-Ecosystem, Ecol. Monogr., 40, 23–47,
1970.
Lowe, R.L., Guckert, J.B., Belanger, S.E., Davidson, D.H., and Johnson, D.W., An evaluation of periphyton
community structure and function on tile and cobble substrata in experimental stream mesocosms,
Hydrobiologia, 328, 135–146, 1996.
Magill, A.H., Aber, J.D., Berntson, G.M., McDowell, W.H., Nadelhoffer, K.J., Melillo, J.M., and Steudler, P.,
Long-term nitrogen additions and nitrogen saturation in two temperate forests, Ecosystems, 3, 238–253,
2000.
Magill, A.H., Aber, J.D., Hendricks, J.J., Bowden, R.D., Melillo, J.M., and Steudler, P.A., Biogeochemical
response of forest ecosystems to simulated chronic nitrogen deposition, Ecol. Appl., 7, 402–415,
1997.
Martinez-Toledo, M.V., Salmeron, V., Rodelas, B., Pozo, C., and Gonzalezlopez, J., Studies on the effects of the
herbicide Simazine on microflora of four agricultural soils, Environ. Toxicol. Chem., 15, 1115–1118,
1996.
Melendez, A.L., Kepner, R.L., Balczon, J.M., and Pratt, J.R., Effects of diquat on fresh-water microbial
communities, Arch. Environ. Contam. Toxicol., 25, 95–101, 1993.
Milchunas, D.G. and Lauenroth, W.K., Inertia in plant community structure: State changes after cessation of
nutrient enrichment stress, Ecol. Appl., 5, 452–458, 1995.
Newman, R.M., Perry, J.A., Tam, E., and Crawford, R.L., Effects of chronic chlorine exposure on litter
processing in outdoor experimental streams, Freshw. Biol., 18, 415–428, 1987.
Nihlgard, B., The ammonium hypothesis—An additional explanation for forest dieback in Europe, Ambio, 14,
2–8, 1985.

Niklinska, M., Laskowski, R., and Maryanski, M., Effect of heavy metals and storage time on two types of
forest litter: Basal respiration rate and exchangeable metals, Ecotoxicol. Environ. Saf., 41, 8–18, 1998.
Odum, E.P., The strategy of ecosystem development, Science, 164, 262–270, 1969.
Odum, E.P., The mesocosm, BioScience, 34, 558–562, 1984.
Odum, E.P., Trends expected in stressed ecosystems, BioScience, 35, 419–422, 1985.
Oviatt, C.A., Pilson, M.E.Q., Nixon, S.W., Frithsen, J.B., Rudnick, D.T., Kelly, J.R., Grassle, J.F., and
Grassle, J.P., Recovery of a polluted estuarine system: A mesocosm experiment, Mar. Ecol. Prog. Ser.,
16, 203–217, 1984.
Pearson, N.and Crossland, N.O., Measurement of community photosynthesisand respiration in outdoorartificial
streams, Chemosphere, 32, 913–919, 1996.
Pell, M., Stenberg, B., and Torstensson, L., Potential denitrification and nitrification tests for evaluation of
pesticide effects in soil, Ambio, 27, 24–28, 1998.
Petersen, D.G., Dahllof, I., and Nielsen, L.P., Effects of zinc pyrithione and copper pyrithione on microbial
community function and structure in sediments, Environ. Toxicol. Chem., 23, 921–928, 2004.
Peterson, B.J., Deegan, L., Helfrich, J., Hobbie, J.E., Hullar, M., Moller, B., Ford, T.E., et al., Biological
responses of a tundra river to fertilization, Ecology, 74, 653–672, 1993.
Pezeshki, S.R. and Delaune, R.D., Effect of crude-oil on gas-exchange functions of Juncus roemerianus and
Spartina alterniflora, Water Air Soil Pollut., 68, 461–468, 1993.
Platt, J.R., Strong inference, Science, 146, 347–353, 1964.
Pratt, J.R. and Barreiro, R., Influence of trophic status on the toxic effects of a herbicide: A microcosm study,
Arch. Environ. Contam. Toxicol., 35, 404–411, 1998.
Pratt, J.R., Melendez, A.E., Barreiro, R., and Bowers, N.J., Predicting the ecological effects of herbicides, Ecol.
Appl., 7, 1117–1124, 1997.
Rapport, D.J., Regier, H.A., and Hutchinson, T.C., Ecosystem behavior under stress, Am. Nat., 125, 617–640,
1985.
Rapport, D.J., Whitford, W.G., and Hilden, M., Common patternsofecosystem breakdown understress, Environ.
Monit. Assess., 51, 171–178, 1998.
© 2008 by Taylor & Francis Group, LLC