Trajectories of zooplankton recovery in the Little Rock Lake whole-lake acidification experiment.

Abstract

Understanding the factors that affect biological recovery from environmental stressors such as acidification is an important challenge in ecology. Here we report on zooplankton community recovery following the experimental acidification of Little Rock Lake, Wisconsin, USA. One decade following cessation of acid additions to the northern basin of Little Rock Lake (LRL), recovery of the zooplankton community was complete. Approximately 40% of zooplankton species in the lake exhibited a recovery lag in which biological recovery to reference basin levels was delayed by 1-6 yr after pH recovered to the level at which the species originally responded. Delays in recovery such as those we observed in LRL may be attributable to "biological resistance" wherein establishment of viable populations of key acid-sensitive species following water quality improvements is prevented by other components of the community that thrived during acidification. Indeed, we observed that the recovery of species that thrived during acidification tended to precede recovery of species that declined during acidification. In addition, correspondence analysis indicated that the zooplankton community followed different pathways during acidification and recovery, suggesting that there is substantial hysteresis in zooplankton recovery from acidification. By providing an example of a relatively rapid recovery from short-term acidification, zooplankton community recovery from experimental acidification in LRL generally reinforces the positive outlook for recovery reported for other acidified lakes.

Full text

Page 1

353
Ecological Applications, 16(1), 2006, pp. 353–367
? 2006 by the Ecological Society of America
TRAJECTORIES OF ZOOPLANKTON RECOVERY IN THE LITTLE ROCK
LAKE WHOLE-LAKE ACIDIFICATION EXPERIMENT
THOMAS M. FROST,1,5JANET M. FISCHER,2,6JENNIFER L. KLUG,3SHELLEY E. ARNOTT,
4AND PAMELA K. MONTZ1
1Trout Lake Station, Center for Limnology, University of Wisconsin, 10810 County Highway N,
Boulder Junction, Wisconsin 54512 USA
2Department of Biology, Franklin and Marshall College, Lancaster, Pennsylvania 17604 USA
3Biology Department, Fairfield University, Fairfield, Connecticut 06824 USA
4Department of Biology, Queen’s University, Kingston, Ontario K7L 3N6 Canada
Abstract.
stressors such as acidification is an important challenge in ecology. Here we report on
zooplankton community recovery following the experimental acidification of Little Rock
Lake, Wisconsin, USA. One decade following cessation of acid additions to the northern
basin of Little Rock Lake (LRL), recovery of the zooplankton community was complete.
Approximately 40% of zooplankton species in the lake exhibited a recovery lag in which
biological recovery to reference basin levels was delayed by 1–6 yr after pH recovered to
the level at which the species originally responded. Delays in recovery such as those we
observed in LRL may be attributable to ‘‘biological resistance’’ wherein establishment of
viable populations of key acid-sensitive species following water quality improvements is
prevented by other components of the community that thrived during acidification. Indeed,
we observed that the recovery of species that thrived during acidification tended to precede
recovery of species that declined during acidification. In addition, correspondence analysis
indicated that the zooplankton community followed different pathways during acidification
and recovery, suggesting that there is substantial hysteresis in zooplankton recovery from
acidification. By providing an example of a relatively rapid recovery from short-term acid-
ification, zooplankton community recovery from experimental acidification in LRL gen-
erally reinforces the positive outlook for recovery reported for other acidified lakes.
Key words:acidification; biological resistance; pH stress; recovery; resilience; whole-lake ex-
periment; zooplankton.
Understanding the factors that affect biological recovery from environmental
INTRODUCTION
Acidic deposition has been recognized as a major
environmental stress in aquatic ecosystems over the last
25 years (Stoddard et al. 1999, Driscoll et al. 2001).
Legislative actions to reduce rates of sulfur dioxide
emissions in North America and Europe were moti-
vated by concern about the ecological impacts of acidic
deposition, and consequently, sulfur dioxide emissions
were significantly reduced in the 1980s and 1990s
(Stoddard et al. 1999, Driscoll et al. 2001). Chemical
recovery has been reported for some North American
aquatic ecosystems that had been anthropogenically
acidified (Stoddard et al. 1999, Driscoll et al. 2003,
Keller et al. 2003). Improvements in water quality have
also been noted in other locations, including Scandi-
navia (Evans et al. 2001, Skjelkva ˚le et al. 2001, Forsius
Manuscript received 2 December 2004; revised 5 May 2005;
accepted 17 May 2005. Corresponding Editor: J. J. Elser.
5Published posthumously. This paper is dedicated to the
memory of Thomas M. Frost (1950–2000): scientist, mentor,
and friend. We remember our conversations with Tom about
Little Rock Lake’s ‘‘acid champs’’ and ‘‘acid chumps’’ with
great fondness.
6Corresponding author. E-mail: jfischer@fandm.edu
et al. 2003) and eastern Europe (Evans and Monteith
2001, Evans et al. 2001). However, it is important to
note that chemical recovery has not occurred in some
locations (e.g., the Adirondacks of North America,
Germany, and parts of southeastern Canada and the
United Kingdom) and is not expected in some impacted
systems unless further reductions in emissions are leg-
islated (Evans et al. 2001, Henriksen et al. 2002, Dris-
coll et al. 2003, Jeffries et al. 2003a, b, Keller et al.
2003).
Because biological recovery is a major goal of leg-
islative action, monitoring programs focused on aquatic
biota have been undertaken in a variety of locations
(Locke and Sprules 1994, Arnott et al. 2001, Walseng
et al. 2001, Keller et al. 2002, Nilssen and Waeervagen
2002, Findlay 2003, Holt and Yan 2003, Snucins 2003,
Vrba et al. 2003, Waeervagen and Nilssen 2003). These
high-quality long-term data sets are critical for ad-
dressing questions about the rate and trajectory of re-
covery (Parr et al. 2003). Several studies have sug-
gested that biological recovery may be delayed com-
pared to chemical recovery (Driscoll et al. 2001, Dixit
et al. 2002, Yan et al. 2003). Furthermore, the length
of the lag for biological recovery may vary substan-

Page 2

354
THOMAS M. FROST ET AL.
Ecological Applications
Vol. 16, No. 1
tially among aquatic ecosystems. For example, Driscoll
et al. (2003) hypothesized that the lag between chem-
ical recovery and recovery of invertebrates in acidified
streams and lakes may range from three to 10 years.
Here, we report on variation in the recovery trajectories
of zooplankton following the experimental acidifica-
tion of Little Rock Lake (LRL), Wisconsin, USA.
Freshwater zooplankton are well-suited to investi-
gation of biological recovery from acidification be-
cause they are relatively easy to monitor and have been
studied in a variety of geographic locations that are
recovering from acidification (Keller and Yan 1998).
Although recovery can proceed rapidly in some lakes,
most studies indicate that zooplankton recovery re-
quires about a decade (Keller and Yan 1998). A wide
variety of metrics have been used to evaluate zooplank-
ton recovery from acidification, including univariate
measures of indicator species abundances, aggregate
measures of zooplankton species richness and diver-
sity, and multivariate metrics of community compo-
sition (Yan et al. 1996). Several studies indicate that
multivariate measures of community composition are
sensitive indicators of recovery (Yan et al. 1996, Holt
and Yan 2003). For example, a multivariate index of
zooplankton species composition provides evidence of
recovery in nine acidic lakes in Killarney Park, Ontario,
Canada, where pH had risen above 6, whereas zoo-
plankton species richness did not change during the
same time period (Holt and Yan 2003). These studies
underscore the value of quantifying changes in species
composition, in addition to coarser metrics of com-
munity response.
Detailed analyses of recovery of zooplankton species
composition following water quality improvements
have potential to reveal variation in recovery rates for
different components of the zooplankton community in
a given lake. For example, Arnott et al.’s (2001) anal-
ysis of recovery of plankton in Swan Lake, Ontario,
Canada, supports the hypothesis that recovery rates dif-
fer between taxonomic groups (e.g., rotifers vs. crus-
taceans). The rate of recovery to pre-acidified condi-
tions may also be influenced by the nature of the initial
response to acidification (e.g., whether the species de-
clined or thrived under acidified conditions). During
the first five years of recovery of Little Rock Lake from
experimental acid addition, several species that had
increased during acidification returned to pre-acidified
levels whereas the recovery of some species that de-
clined during acidification was delayed (Frost et al.
1998). Thus, patterns of biological recovery can be
complex and mechanisms driving these patterns remain
poorly understood.
The goal of our study was to summarize patterns of
zooplankton recovery following a whole-lake acidifi-
cation experiment conducted in Little Rock Lake, Wis-
consin. Because the choice of metric may affect our
perception of recovery, we compared recovery trajec-
tories for highly aggregated community metrics (e.g.,
total zooplankton biomass) and population level met-
rics (e.g., biomass of individual species). We also fo-
cused on the role of biological factors that may affect
recovery. For example, we compared recovery trajec-
tories for species that were negatively affected by acid-
ification (‘‘acid chumps’’) and species that thrived un-
der acidic conditions (‘‘acid champs’’). Our study rep-
resents an interesting comparison to other recent stud-
ies examining zooplankton recovery from acid rain
(Arnott et al. 2001, Holt and Yan 2003). The design of
the Little Rock Lake experiment included pre-acidifi-
cation data and parallel monitoring of a closely
matched reference system. These features are typically
lacking in studies of anthropogenic acidification due to
historical and/or logistic reasons. Overall, we believe
that the ability to follow zooplankton community tra-
jectories in Little Rock Lake for 16 years, including a
full cycle of response and recovery (see Mittelbach et
al. 1995), provided a unique opportunity to gain in-
sights about ecological resilience to acidification.
METHODS
Whole-lake experiment
In 1984, the two basins of Little Rock Lake, a bilobed
seepage lake located in northern Wisconsin, USA, were
separated using a vinyl curtain. Following a year of
baseline data collection, sulfuric acid was added to the
northern basin (hereafter, acidified basin). The pH of
the acidified basin was decreased sequentially to three
target pH levels each maintained for two years: 5.6
(1985–1986), 5.1 (1987–1988), and 4.7 (1989–1990).
The recovery phase began in 1991 when all acid ad-
ditions ceased and the acidified basin was allowed to
recover naturally. Data collection was completed in
2000. Throughout the entire period (1984–2000), the
southern basin (hereafter, reference basin) was unma-
nipulated and served as a reference system for the
changes that occurred in the acidified basin.
Sampling methods
Details of the limnological sampling are summarized
here and described in detail elsewhere (Frost and Montz
1988, Brezonik et al. 1993, 2003). From 1984 to 2000,
water samples for chemical analysis were collected
from a central station in each basin once every two
weeks during the ice-free season and approximately
monthly during winter. Here we present only the pH
data, as changes in chemistry during acidification and
recovery have been described elsewhere (Brezonik et
al. 1993, 2003, Frost et al. 1999, Sampson and Brezonik
2003).
Samples for enumeration of zooplankton were col-
lected with a 33-L Schindler-Patalas trap (53-?m mesh)
at the frequency described above for chemical samples.
Samples were collected from fixed depths (0, 4, and 8
m in the acidified basin and 0, 4, and 6 m in the ref-
erence basin) and preserved with 4% sucrose-buffered

Page 3

February 2006355
ZOOPLANKTON RECOVERY FROM ACIDIFICATION
formalin. With the exception of the surface sample, the
middle of the trap was positioned at the target depth.
Hypsometrically weighted mean abundances were cal-
culated for each basin. Zooplankton abundance was
converted to biomass using length–mass relationships
determined directly for Little Rock Lake species or
from the literature (Ruttner-Kolisko 1977, Pace and
Orcutt 1981, Culver et al. 1985, Yan and Mackie 1987).
Copepod nauplii were not identified to species but are
included in analyses of total zooplankton biomass, co-
pepod biomass, and the multivariate analyses of chang-
es in community composition. Additional details on
zooplankton analyses are provided in Frost and Montz
(1988).
Samples for enumeration of large predatory zoo-
plankton (Chaoborus punctipennis, water mites, Ep-
ischura lacustris, and Leptodora kindtii) were collected
with a 50 cm diameter, 253-?m mesh conical plankton
net at a fixed sampling station after dark. Tows were
collected approximately once every two weeks during
the ice-free season from 0.5 m above the deepest point
of the basin to the surface, except in 1995 when sam-
ples were not collected. Samples were preserved with
Lugol’s iodine. We used unpublished dry masses of
Chaoborus punctipennis (Fischer 1994) to transform
abundance to biomass. Biomass estimates for water
mites, Epischura lacustris, and Leptodora kindtii were
calculated using published values (Meyer 1989, Law-
rence et al. 1987, and Hawkins and Evans 1979, re-
spectively). Additional details on predatory zooplank-
ton sampling are provided in Sierszen and Frost (1993).
Data analysis
We evaluated zooplankton community recovery from
acidification using a variety of metrics ranging from
population level (e.g., biomass of an individual zoo-
plankton species) to taxonomic groups (e.g., rotifer bio-
mass) to highly aggregated community level (e.g., total
zooplankton biomass). For each metric, we compared
the time series of annual mean values in the acidified
basin to the long-term mean and variability for the same
metric (i.e., annual means) in the reference basin. Fol-
lowing the approach of Arnott et al. (2001), we used
the 10th and 90th percentiles for each metric across all
years (1984–2000) to represent variability in the ref-
erence basin. We concluded that a species or group of
species in the acidified basin responded to acidification
if its annual mean biomass was above or below the
range of variability in the reference basin during the
acidification years. In a few cases in which a species
was slightly outside the range for only one year early
in the acidification and returned to within the range for
several years (e.g., Bosminids in 1986), we did not
classify this species as responding to acidification dur-
ing the first excursion from the reference basin range.
We concluded that a species or group of species had
recovered from acidification once its annual mean bio-
mass was consistently within the range of variability
defined by the 10th and 90th percentiles of the reference
basin data. In the cases of Daphnia parvula and Tro-
pocyclops extensus, we concluded that recovery oc-
curred in 1997 despite subsequent departures from the
10th and 90th percentiles of the reference basin data
because similar dynamics were noted in the reference
basin. We recognize that this is a subjective approach
guided by statistical principle and intuition, but suggest
that it is a good way to assess general patterns in com-
plex long-term data sets like ours. We also compared
the timing of chemical and biological recovery for each
taxon. Recovery lag was defined as the time delay be-
tween chemical recovery (based on return to the mean
annual pH value in the year before the species respond-
ed to acidification) and biomass recovery of each spe-
cies.
To compare trajectories of change in community
composition in the acidified basin to community com-
position in the reference basin, we used correspondence
analysis (CA). Correspondence analysis is an ordina-
tion method widely used by community ecologists to
examine differences in species composition across en-
vironmental gradients (Jackson 1993, Gotelli and El-
lison 2004). We log10(x ? 1)-transformed the biomass
data, excluded rare species (?5% of biomass of their
taxonomic group) from the analysis, and calculated CA
using SAS (SAS Institute 1996). Large predatory zoo-
plankton (Chaoborus punctipennis, water mites, Ep-
ischura lacustris, and Leptodora kindtii) were not in-
cluded in the analysis due to one year of missing data.
We examined correlations between CA axis scores and
pH, secchi depth, and chlorophyll to explore relation-
ships between environmental factors and changes in
zooplankton community composition. It is important
to note that these analyses were constrained by the data
available and all possible environmental drivers were
not examined.
RESULTS
pH pattern
During the pre-manipulation year (1984), pH values
in the two basins of Little Rock Lake were very similar
(Fig. 1a). Sulfuric acid was added to the acidified basin
from 1985 to 1990, causing a stepwise decrease in pH.
Beginning in 1991, pH in the acidified basin was al-
lowed to recover. pH gradually rose until reaching pre-
manipulation levels in 1996 (Fig. 1). pH in the refer-
ence basin was variable over the 17-yr record but
showed no obvious directional trends (Fig. 1a). Com-
paring mean annual pH to the range of natural vari-
ability in pH in the reference basin provided a consis-
tent interpretation of recovery. Mean annual pH in the
acidified basin decreased below the range of variability
in the reference basin during the first year of acidifi-
cation (1985) and returned within the range of vari-
ability in the reference basin in 1996 (Fig. 1b).

Page 4

356
THOMAS M. FROST ET AL.
Ecological Applications
Vol. 16, No. 1
FIG. 1.
(REF) and treatment (TRT) basins of Little Rock Lake, Wisconsin, USA, from 1984 to 2000. Horizontal solid lines in (b)
indicate 10th and 90th percentiles in the reference basin for 1984–2000.
(a) Time series of pH values (samples were taken once every two weeks) and (b) annual mean pH in the reference
Response and recovery of aggregate groups
Despite a large decrease in pH, total zooplankton
biomass in the acidified basin stayed within the range
of variability in the reference basin until the final year
of acidification (Fig. 2a). Total zooplankton biomass
was lower in 1990 but recovered in 1991. During the
rest of the recovery period, total zooplankton biomass
was variable but generally stayed within the bounds of
the reference envelope (Fig. 2a). Although total rotifer
biomass increased during the later stages of acidifi-
cation, it did not increase above the reference basin
envelope until the second year of recovery (1992) and
returned to the range of reference basin variability in
1993 (Fig. 2b). Total cladoceran biomass responded to
acidification by increasing above the range of vari-
ability in the reference basin during the first year of
acidification (1985) and decreased to within the ref-
erence envelope early in the recovery period (1993)
(Fig. 2c). In contrast, total copepod biomass was rel-
atively insensitive to acidification (Fig. 2d). Total co-
pepod biomass stayed within the range of variability
in the reference basin until the final year of acidification
(1990) and recovered the following year (1991). Co-
pepods dominated the total zooplankton biomass
throughout the experiment (Fig. 2a, d).
Response and recovery in species composition
To assess changes in zooplankton species composi-
tion, we used correspondence analysis to compare the
community trajectory in the acidified basin to inter-
annual variation (i.e., the cloud of points) in the ref-
erence basin. When all species (except large predatory
zooplankton) were included in the analysis, dramatic
changes in the acidified basin zooplankton assemblage
were noted as early as 1986 when the community began
to move to the right along CA axis 1 (Fig. 3a, b). The
acidified basin community changed to a more acido-
philic composition during 1987–1991. However, a re-
covery trajectory began soon after acidification ended.
By 1994–1995, community composition in the acidi-
fied basin was similar to the reference basin. In 1996,
community composition had essentially returned to an
assemblage characteristic of the reference basin. Cor-
respondence analysis axis 1 was negatively correlated
with pH (n ? 34, r ? ?0.77, P ? 0.0001), whereas
CA axis 2 was negatively correlated with chlorophyll
(n ? 34, r ? ?0.40, P ? 0.02). It is interesting to note
that significant changes occurred in the reference basin
during the study period. The reference basin points with
the lowest scores on CA axis 2 correspond to 1996 and
1998–2000. The first two CA axes captured roughly

Page 5

February 2006357
ZOOPLANKTON RECOVERY FROM ACIDIFICATION
FIG. 2.
(REF) and treatment (TRT) basins of Little Rock Lake. Horizontal solid lines indicate 10th and 90th percentiles in the
reference basin for 1984–2000.
Annual mean total biomass of (a) zooplankton, (b) rotifers, (c) cladocerans, and (d) copepods in the reference
57% of the variance (38.61 and 17.98 for axes 1 and
2, respectively).
When only the rotifers were included in the CA, clear
changes in the acidified basin assemblage were noted
by 1987, the first year of the second stage of acidifi-
cation (Fig. 3c, d). Specifically, the acidified basin as-
semblage moved to the right on CA axis 1 towards a
composition characterized by Gastropus hyptopus,
Keratella taurocephala, and Synchaeta. Correspon-
dence analysis axis 1 was negatively correlated with
pH (n ? 34, r ? ?0.78, P ? 0.0001). During the
recovery period, rotifer community composition in the
acidified basin returned to the cloud of reference basin
points by 1994. The first two CA axes for the rotifer
community captured roughly 54% of the variance
(36.39 and 17.17 for axes 1 and 2, respectively).
The CA for cladocerans revealed substantial varia-
tion in the reference basin assemblage during the study
period across the first CA axis (Fig. 3e, f). It is im-
portant to note, however, that this pattern does not re-
flect simple directional change (e.g., from right to left)
in the reference basin assemblage. Instead, the trajec-
tory in the reference basin was more random. For ex-
ample, the five reference basin points with negative
scores on the first CA axis correspond to 1991, 1994,
and 1998–2000. Despite the variation in the reference
basin cladoceran assemblage, the acidified basin as-
semblage exhibited a clear response to acidification in
1986, reflecting a decrease in Daphnia dubia, Holo-
pedium gibberum, and Diaphanosoma and an increase
in Daphnia catawba. By 1991, however, the acidified
basin assemblage recovered to the broad region defined
by the reference basin points. CA axis 1 was positively
correlated with secchi depth (n ? 34, r ? ?0.41, P ?
0.02) and negatively correlated with chlorophyll (n ?
34, r ? ?0.35, P ? 0.04), whereas CA axis 2 was
negatively correlated with pH (n ? 34, r ? ?0.49, P
? 0.003). The first two CA axes for the cladoceran
community captured roughly 86% of the variance
(58.09 and 28.23 for axes 1 and 2, respectively).
Similar to cladocerans, the CA for copepods reveals
substantial variation in the reference basin assemblage
that complicates the interpretation of response and re-
covery in the acidified basin assemblage (Fig. 3g, h).