The Impact of Fish Exclusion on the Abundance and Species Composition of Larval Odonates:

Results of Short-Term Experiments in a North Carolina Farm Pond

Author(s): Peter J. Morin
Source: Ecology, Vol. 65, No. 1 (Feb., 1984), pp. 53-60
Published by: Ecological Society of America
Stable URL: http://www.jstor.org/stable/1939457
Accessed: 18-03-2015 21:37 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/1939457?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Ecological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Ecology.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
 colo(i %. 65(1). 1984. pp. 53--60
C 1984 by-the Ecological Society of America

THE IMPACT OF FISH EXCLUSION ON THE

ABUNDANCE AND SPECIES COMPOSITION OF
LARVAL ODONATES: RESULTS OF SHORT-TERM
EXPERIMENTS IN A NORTH CAROLINA
FARM POND'

PETER J. MORIN2
Department >t Zoology, Duke Ulnit'ersity, Durham, North Carolina 27706 USA

Abstract. Under natural conditions in a man-made North Carolina farm pond, a single small
species, Perithelnis tenera, numerically dominates a littoral assemblage of larval dragonflies. Censuses
of exuvia of larvae metamorphosing from the pond in three successive years demonstrated a consistent
negative correlation between numerical dominance in the assemblage and species-specific size in the
final larval instar. This pattern suggested that size-dependent processes, such as vertebrate predation,
might structure the odonate assemblage.
Exclusion of vertebrate predators from patches of emergent vegetation by screen enclosures for
3 mo increased total abundances of larval odonates by an order of magnitude. Dominance shifted
from small species to species of intermediate size, especially Pach'dipl/a lo/ngipenllis. where fish
were excluded. The largest species remained rare and were unaffected by fish exclusion. Four odonate
species were facultatively multivoltine, completing larval development within 1-2 mo after coloni-
zation of the fish enclosures. Fish exclusion also increased the abundance of large microcrustacea,
especially the cladoceran Sitnocephalus serrulatuis. Large microcrustacea increased in abundance
despite a corresponding increase in the abundance of their macroinvertebrate predators, where fish
were excluded.
The usual dominance of small Perithelnis in this assemblage was a probable consequence of
differentially heavy predation by fish on moderate-sized species, and the unexplained failure of large
species to recruit well under experimental or natural conditions. These results support the general
importance of predation in structuring freshwater communities and document an important mode of
population regulation operating during the aquatic larval phase of the odonate complex life cycle.
hey, ittords: Anisoptera: comninunit' structure; Iratgonflst; fish preda tion; Lepomis; Ilnacroen-
thlos; Micropterus; North Carolina; Perithemis: Pachydiplax; ponds; pred(ltion; zooplanktoni.

INTRODUCTION of community structure (e.g., Zaret 1980). Results from

The impact of vertebrate predation on the species a larger sample of communities are needed to correlate
composition of freshwater planktonic macroinverte- disparate responses to fish exclusion with other mea-
brates is well known (Brooks and Dodson 1965, Zaret sured differences among communities. Important fac-
1980) and provides a striking example of the efficacy tors responsible for these different responses may in-
of predation in structuring one component of fresh- clude variation in habitat complexity or natural levels
water communities. In contrast, the role of vertebrate of predator abundance among communities (Crowder
predation in structuring assemblages of freshwater and Cooper 1982).
macroinvertebrates remains controversial. Compara- In this paper I show that experimental exclusions of
tive studies suggest that fish predation may regulate fish and other aquatic vertebrates dramatically alter
the abundance and species richness of aquatic insects the abundance and dominance of species comprising
(e.g.. Macan 1977). One group of experiments indi- one important assemblage of littoral macroinverte-
cates that fish influence the biomass and abundance of brates, the aquatic larvae of dragonflies (Odonata: An-
benthic macroinvertebrates (Hall et al. 1970, Eriksson isoptera). These experiments tested a hypothesis in-
1979. Crowder and Cooper 1982). However, other ex- spired by the following observation: the odonate
periments suggest that exclusion of fish from the lit- assemblage in my study pond was dominated by a small
toral zone has little impact on the abundance or species late-breeding species, despite the fact that such species
richness of benthic macroinvertebrates (Thorp and are at a competitive or predatory disadvantage among
Bergey 1984I, h). Divergent outcomes of similar ma- larger, early breeding odonates in similar southeastern
nipulations of predators in different freshwater com- ponds (Benke 1978, Benke et al. 1982). I postulated
munities complicate the synthesis of emerging theories that dense populations of size-selective fish might foster
the predominance of small species by depressing the
I Manuscript received 28 July 1982; revised and accepted
abundance of larger odonates that might otherwise
27 December 1982.
2
Present address: Department of Biological Sciences, Nel- outcompete or prey on smaller species. These short-
son Labs, Busch Campus, Rutgers University, Piscataway, term exclusion experiments confirm the proposed
New Jersey 08854 USA. efficacy of vertebrate predators in controlling this im-

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
54 PETER J. MORIN Ecology, Vol. 65, No. 1

portant assemblage of freshwater invertebrate preda- long x 1.22 m wide x 0.61 m high) of wooden frames
tors (Benke 1978, Johnson and Crowley 1980), illus- completely covered by bottom and side panels of Fi-
trate one mode of population regulation in species with berglas screen (7 meshes/cm). I made four pens
complex life cycles (Wilbur 1980), and provide new permeable to large predators by cutting a 0.45 x 0.61
insight into the duration of larval development of sev- m window in the screen at the center of one side panel.
eral odonate species. Hereafter, these are termed open pens. The four re-
maining pens (hereafter termed enclosures) lacked such
MATERIALS AND METHODS
openings and excluded fish and turtles. The pens were
I studied the larval odonate assemblage of a 0.44-ha placed in a discrete 63 m long Paspal/um bed along the
man-made farm pond (Carroll's Pond) located in Or- north shore of Carroll's Pond on 22 June 1981. I added
ange County, North Carolina. Carroll's Pond is :28 natural densities of well-washed Paspalum to each pen
yr old, has a maximum depth of ;3 m, and provides to provide cover. Pen walls projected at least 0. 1-0.2
water for a small herd of cattle. The pond's narrow m above the waterline. Female dragonflies began ovi-
zone of emergent vegetation (1-2 m wide) is dominat- positing in the pens within several hours after pen
ed by an aquatic knotgrass, Paspalurn distichum. Beds placement. I augmented natural recruitment by adding
of emergent Paspalum are the principal oviposition first-instar hatchlings of several species to all pens,
sites patrolled by adult dragonflies, and most larval before it became obvious that larvae would naturally
dragonflies metamorphose in or near these beds. Elev- colonize the pens. On 16 and 20 July 1981, I added the
en species of dragonflies successfully metamorphose following numbers of first-instar odonates to each pen:
from the pond in most years. Fourteen additional 55 Tram ea lacerata, 55 Celithemis fasciata, 100 Li-
species patrol and oviposit, but their larvae fail to bellula incest, 100 Erythemnis sitpli/icollis, and 50
metamorphose from Carroll's Pond in most years (see Plathemis lydia. In addition, small instars of odonates
Appendix). Damselflies (Odonata: Zygoptera) are nei- in the Pa spa/lum bed could colonize the pens by pass-
ther conspicuous nor abundant in this community. The ing directly through the screen mesh. Of course, open
most abundant vertebrate predators in Carroll's Pond pens were permeable to odonates of all sizes.
are the fish Lepomnisinacrochirus (bluegill sunfish) and Paspaluin grew rapidly and provided dense cover
Micropteiuis sa/hnoides (large mouth bass). Lepomnisis and spatial heterogeneity for weed-dwelling odonates.
known to prey heavily on larval odonates (Sadzi- After 1-2 wk, habitat complexity within the pens re-
kowski and Wallace 1976). Other potential predators sembled that found in the unmanipulated Paspalum
include the turtles Chrysemnys scripta, Sternother-tis bed. Several millimetres of silt and detritus accumu-
odora tius, and Chelydra s e}p en tina. lated on the pen bottoms, providing ample habitat for
I measured natural patterns of larval odonate abun- burrowing and bottom-sprawling species. Grazing by
dance in the entire pond by collecting cast exoskele- tadpoles of Rana catesbeiana prevented fouling of pen
tons (exuvia) left around the margin of the pond by walls by periphyton. Tadpoles entered the open pens
larvae as they metamorphosed into teneral dragonflies. naturally, and each exclusion pen was stocked with 6-
Exuvial census is a standard technique for estimating 10 tadpoles.
the abundance of many aquatic insects (Benke and I assessed the effects of fish exclusion, by measuring
Benke 1975, Southwood 1978). Exuvia were collected the abundance of naiads either metamorphosing from
daily from the entire perimeter of the pond during each pens or remaining in the pens at the termination of the
annual period of metamorphosis (usually from early experiment in late September 1981. Larvae remaining
April through late September) in 1979, 1980, and 1981. in the pens in late September were diapausing individ-
The shoreline was bordered by closely cropped pas- uals that presumably would have continued develop-
ture and mown lawns, restricting suitable perches for ment during the following spring (Paulson and Jenner
metamorphosing odonates to the immediate margin of 1971). I collected exuvia daily within each pen, until
the pond, where exuvia were conspicuous and readily metamorphosis ceased in enclosures and the pond in
collected. Most exuvia were concentrated in a narrow mid-September. Direct observation of nocturnally
band within 1 m of the shoreline (Jacobs 1955). Al- eclosing naiads indicated that exuvia collected from
though exuvia were occasionally lost during periods the interior walls and vegetation of each pen originated
of heavy rain, total counts of exuvia collected from from larvae living in the pens. Overhanging wooden
the entire perimeter of the pond provided the best frames (;3 cm wide) on each pen's top prevented me-
available estimate of the abundance of each odonate tamorphosing naiads from wandering outside the pens
species metamorphosing from the pond in a given year. to eclose.
I experimentally examined the short-term impact of I made comparable daily collections of exuvia from
vertebrate predators on odonate species composition the 63-m Paspalurn bed surrounding the pens, and from
by measuring larval recruitment and metamorph pro- the remaining pond shoreline. I standardized counts
duction in screen pens that either excluded large ver- of exuvia from the pens and unmanipulated portions
tebrate predators or remained permeable to fish and of the pond to numbers of metamorphosing odonates
turtles. I constructed eight open-topped pens (1.22 m per metre of shoreline per census interval, to facilitate

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
February 1984 FISH PREDATION ON LARVAL ODONATES 55

comparison of emergence rates among natural and ex- I assumed that if open pens concentrated fish above
perimental habitats. I assumed that each pen corre- natural densities, more fish would be caught per min-
sponded to 1.22 m of shoreline. Emergence rates were now trap per day in the pens than in the weed bed.
calculated only over the census interval in which odo- Low visibility and thick vegetation precluded visual
nates metamorphosed from both the pens and the pond, methods of estimating fish densities.
23 July through 18 September 1981. I measured one other aspect of pen community
The experiment was terminated after =3 mo, on 22, structure to compare the resemblance of open pens to
23, and 24 September, when I carefully lifted the pens the natural weed bed. At the termination of the ex-
and their contents from the pond. Overwintering odo- periment, microcrustacea and other small arthropods
nate larvae retained by the screen mesh were sorted were sampled in each pen and at four haphazardly
from debris by hand and preserved. The contents of chosen sites in the weed bed. Samples were taken by
open pens provided a baseline estimate of the abun- plunging an inverted l-L polyethylene bottle into the
dance of odonates recruiting in the presence of pred- weeds to a depth of 0. 15 m and then allowing the
ators, via a sampling protocol that was directly com- bottle to fill with pond water and organisms. Samples
parable to the census of the contents of predator were filtered through an 88-,um mesh Nitex filter, and
exclosures. All sampling biases were assumed to be filtered organisms were preserved in a solution of su-
similar between treatments, and differences in the crose and formaldehyde (Haney and Hall 1973). All
contents of open and predator-exclusion pens were preserved microinvertebrates in each sample were
attributed to direct or indirect consequences of the identified and counted with the aid of a dissecting mi-
presence or absence of large vertebrate predators. Be- croscope. Counts of each taxon from the two kinds of
cause these experiments were conducted for less than pens and the unmanipulated weed bed were compared
a complete yearly cycle of odonate oviposition, de- (via ANOVA and Duncan's multiple range test) to de-
velopment, and metamorphosis, they cannot speak to tect differences in the availability of these odonate re-
entire patterns of odonate abundance described by sources (Johnson 1973, Benke 1978) between natural
collections of metamorphosing naiads over a full year. and experimental habitats. These samples also provid-
However, short-term predator exclusions effectively ed an indirect measure of the intensity of fish predation
document the consequences of predation in compa- in open pens and the natural weed bed. I assumed that
rable experimental communities, and can identify if the open pens trapped fish and concentrated verte-
species that are susceptible to predation. brate predators above natural densities, abundances
I used three statistics to describe differences in pop- of larger microcrustacea would be significantly lower
ulations of each odonate species between open pens in the open pens than in the weed bed, due to corre-
and fish enclosures: (1) the number of metamorphosing sponding differences in the intensity of predation by
naiads per pen, (2) the number of overwintering naiads size-selective fish (Zaret 1980).
per pen, and (3) the relative abundance of each species
RESULTS
in each pen. Relative abundance was defined by the
fraction of the total number of metamorphosing and Censuses of metamorphosing naiads revealed rela-
overwintering naiads in each pen belonging to each tively similar patterns of dominance and abundance
species. Relative abundance is a measure of domi- over three successive years of emergence (Table 1).
nance that is statistically independent of differences in The smallest species, Perithemis tenera, was an order
total odonate abundance. Species can conceivably dif- of magnitude more abundant than any other anisop-
fer in total abundance between treatments, while not teran in each year. In general, numerical dominance
differing in relative abundance, if dominance is inde- and species-specific body size were negatively corre-
pendent of variation in total odonate carrying capacity lated. Over the entire odonate assemblage, there was
among pens. a significant negative correlation between species-spe-
I also sampled fish densities at the termination of cific body length in the final larval instar and the log-
the experiment by rapidly tilting the open side of each arithm of the abundance of final-instar naiads meta-
open pen out of the water, trapping foraging fish. Pens morphosing from the pond within each year (Table 1).
were then completely lifted from the water, and trapped Species often varied considerably in abundance among
fish were collected, anesthetized, and preserved. years, but this variation had little impact on the broad
Densities of fish in open pens were calibrated against patterns of dominance described by the ranked abun-
natural densities during a similar experiment in 1982 dances of species within years.
using a catch-per-unit-effort technique. Plastic min- Exclusion of vertebrate predators by screen exclo-
now traps were placed in each of four open pens and sures dramatically altered both the absolute abun-
four haphazardly selected sites in the natural weed dance (number per pen) and relative abundance (per-
bed. Traps were left in place for 24-48 h, and then cent per pen) of metamorphosing and overwintering
trapped fish were censused. Large Lepounis could not odonates (Table 2, Fig. 1). Few odonates metamor-
enter the traps, but fish in the small size-classes pre- phosed from the open pens containing fish. However,
dominating in the pens in 1981 were readily captured. three species, Pachydiplax longipennis, Erythemis

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
56 PETER J. MORIN Ecology, Vol. 65, No. 1

TABLE 1. Numbers of odonate exuvia collected from the entire perimeter of CalToll s Pond in each of three successive
years of daily collections. Species-specific mean body lengths in the final instar are based on 5-10 individuals of each
species collected in 1979 and 1980, except for DronogioIg phus. Numbers in parentheses are the rank abundance of each
species within each year. Correlations at the bottom of the table describe the relation between the logarithm of abundance
and mean body length over all species in each year.

Body length Abundance (no./yr)

Species (mm) 1979 1980 1981
Perithelnis tenera 13. 1 2432 (1) 3766 ( 1) 5324 ( 1)
Ervthemlios simplicicollis 15.2 329 (3) 181 (3) 61 (4)
PachXvdiplax iongipenniis 16.9 491 (2) 589 (2) 234 (2)
Epitheca cVnosural 18.0) 57 (5) 137 (4) 21 (6)
Platheinis lvdia 19.7 68 (4) 99 (5) 104 (3)
Libellula luctuosa 21.7 17 (7) 25 (7) 32 (5)
Gniphus brisnle vi 22.2 11 (8) 26 (6) 6 (7)
Latdona deplanata 22.4 2 (10) 10 (8) 0 (10)
Epicordulia regina 23.6 4 (9) 0 (10) 1 (9)
Trainea lacerat(I 27.1 43 (6) 7 (9) 3 (8)
Dromogoinphus spinosus 30.9 1 (II) 0 (10) () (10)
Correlation between body length
and log, abundance -0.83 --0.90 -0.84
Correlation significance level .01 .01 .01

Sim/plic(ollis, and Perithemis teliera, metamorphosed entire perimeter of the pond (229.8 m) in most years
in abundance from fish exclusion pens. Small numbers (Tables I and 2).
of the large species Trameia l/(cer(at(a also metamor- Low metamorph production in the open pens was
phosed from the fish exlcusion pens. The first meta- similar to rates of metamorph production in the PucN-
morphs were collected from fish exclusion pens on 23 pa/lumi bed. Numbers of exuvia per metre of shoreline
July 1981, and metamorphs continued to appear through collected between 23 July and 18 September for the
18 September 1981. Because all odonates entered fish three most abundant species were: 6.5 Perithemiis/m,
exclusion pens as eggs or very small larvae, these four 0.2 Ervthemiislm, and 0.3 Pach vdipl/a/m in the Pas-
species clearly completed larval development in only p(/uclu bed; 1.8 Perithemnislm, 0.2 Er'theniislm, and
1-2 mo. 0.0 Pachx'dipl/x/m in the open pens; 18.4 Perithe-
Perithemis tencra, the smallest species, was the nu- inisIm, 13.7 Ervthemnislm,and 108.8 Pachyvdipllx/m in
merically dominant species metamorphosing from both the predator exclosures. Differences in average meta-
the open pens and the natural weed bed (Tables I and morph production between the Paspalunm bed and the
2). In contrast, predator exclusion generated a striking open pens were trivial compared to observed differ-
increase in the abundance of metamorphosing Pachy- ences between open and fish exclusion pens.
diplax longipeunis, a species of intermediate size. More Predator exclusion also greatly enhanced the abun-
Paochvtiplax metamorphosed from the four predator dance of overwintering naiads of moderate-sized
enclosures combined (the equivalent of 4.88 m or 2. 1% species (Table 2). Species with significantly enhanced
of the pond's shoreline) than were collected from the larval abundance in the absence of fish included Erx'-

TABLE 2. Absolute abundances (mean ? SD) of metamorphosing and overwintering larvae in open and fish exclusion pens.
Values in columns headed P indicate the level of significance (from ANOVA) of differences in the abundances of each
species attributed to the presence or absence of fish. Dashes indicate species that did not metamorphose from either
treatment.

Metamorphosing larvae Overwintering larvae

(no./pen) (no./pen)
Species Fish present Fish absent P Fish present Fish absent P
Perithemiis tenera 2.3 ? 2.6 22.5 ? 24.9 .150 22.3 ? 14.3 55.0 ? 64.2 .358
Er 'themiis simnplicicollis 0.3 ? 0.5 17.3 ? 4.6 .001 11.8 ? 5.5 184.0 ? 99.9 .014
Pachvdiplax longipennis 0.0 + 0.0 143.0 ? 62.3 .004 26.0 ? 15.3 423.3 + 113.6 .001
Epitheca c-ynosur- - 8.3 ? 8.5 53.0 ? 25.1 .015
Platheinis lvdia - 0.8 ? 0.5 20.0 ? 8.3 .004
Libell/lu ineesta - - - 0.8 ? 1.0 43.0 ? 12.8 .001
Libel/u/l luctuosa - - - 0.3 ? 0.5 48.0 ? 24. 5 .008
Epicorululia region - - - 0.5 ? 0.6 0.8 ? 1.0 .670
Triniea lacerata 0.0 ? 0.0 2.0 ? 2.2 .114 0.0 ? 0.0 0.3 ? 0.5 .355

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
February 1984 FISH PREDATION ON LARVAL ODONATES 57

60 fish
0 p~~~~~~~~~~~resent
50 LIZ......
-o
M
40 *t absent
~~~~~~~~~~~~fish

<30

20

Perithemis Erythemis PKchydipKax Epitheca Plathemis L.incests L. luctuosa

small * - Final Instar Body Size - N large

FIG. I. Mean relative abundance calculated from combined abundances of metamorphs and overwintering larvae of each
odonate species collected from the experimental pens. Relative abundances describe the percentage of the total number of
odonates in each treatment belonging to each species and describe changes in numerical dominance that are independent of
absolute total odonate density. Species of consistently low absolute and relative abundance are not shown. When a species
exhibited significant differences in relative abundance between treatments (via ANOVA), those differences are indicated by
P < .05 or ** P < .01 above the species bars.

tilheniis simplici(/5llis, P(l(hw(/ipl(ax longipeUnnis, Epi- vae of L. ilcesta were easily distinguished from L.
thieca cviosura, Pl(1the mis lNdia, Libell/lw infcest(a, and lucktuosa by counts of setae on the lateral lobes of the
Libell/lu /uctuos(a. In contrast, the abundance of over- labium (Needham and Westfall 1955).
wintering instars of small Perithe mis tenera did not Fish were abundant in all open pens at the termi-
increase significantly under fish exclusion, while its nation of the experiment, while only one small Le-
relative abundance (dominance) actually declined be- poinis was found in one of the four predator exclusion
cause of the increased abundance of other species (Fig. pens. Means and ranges of fish densities in the open
1). Relative abundances of Pach'(lipl/-x lonlipennis pens were 13.8 Lepomnis mal(Ilrofhirus/pen (range 12-
and Libellula spp. increased concomitantly with their 18) and 1.8 Micropterus s(/lhnoides/pen (range 0-4).
increased absolute abundance in fish exclosures. Most fish collected from the pens were small individ-
However, absolute abundances of Ervthetmis simpli- uals with mean standard lengths (snout to caudal pe-
collis, Plathemiis ltdia, and Epithecac(lonosur(l in- duncle; ranges in parentheses) of 43 mm (24-129 mm)
creased in proportion to the total abundance of the for Lepomiis and 50 mm (40-62 mm) for Mic ropterus.
entire odonate assemblage, resulting in unchanged rel- Comparison of the abundance of small Lepomiis
ative abundances for these species. The two largest captured in minnow traps placed in either open pens
species collected from the pens, Epicordulia regina or the weed bed in 1982 indicated that open pens did
and Traenca /acer(t(a, were rare in both treatments not concentrate fish above natural densities (Table 3).
and were apparently unaffected by fish exclusion. Ex- In fact, results from four successive trapping intervals
clusion of vertebrate predators produced a clear shift in July 1982 indicated that the number of fish per trap
in absolute and relative abundance from assemblages per sampling interval was slightly lower in the open
dominated by a small species to assemblages domi- pens than in the weed bed.
nated by species of intermediate size, especially Abundances of microinvertebrates sampled from the
Pac hx'diplax longipennis.
Three relatively uncommon species known to meta-
morphose from the pond (Ladiona deplanata, Gomn- TABLE 3. Means and ranges of the number of Lepoinis intic-
phuis brien/evi, and Dromnogoinphuis spinosus) failed to rochirus captured in plastic minnow traps placed in open
recruit in the pens. Ladtona (lepl/anat and Gomnphus pens or the unmanipulated weed bed on four sampling dates
in 1982. Open pens were identical to the pens used in 1981,
briieN/'1i are spring-breeding species that had com- and were located in similar positions in the weed bed. Sam-
pleted reproduction long before the pens were placed pling dates and sampling intervals indicate the timing and
in Carroll's Pond. Neither of these species was suffi- duration of each trapping effort. All means are for four
replicates in each habitat on each date.
ciently abundant under natural conditions to contrib-
ute conceivably to the fish exclusion effect. Droano-
Sampling Lepomis per minnow trap
gouiphus spinosus is quite rare at Carroll's Pond, and interval
no ovipositing adults were observed in 1981. Interest- Sampling date (h) Weed bed Open pens
ingly, Libellula incest, a species that failed to meta- 10 July 1982 24 1.8 (1-2) 1.3(0-3)
morphose from the pond despite the presence of large 11 July 1982 24 1.8 (1-3) 1.0 (0-3)
numbers of ovipositing adults, recruited in predator 12 July 1982 24 1.5 (0-4) 1.5 (1-2)
14 July 1982 48 2.8 (1-4) '.0(0-3)
exclusion pens at moderate levels of abundance. Lar-

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
58 PETER J. MORIN Ecology, Vol. 65, No. 1

TABLE4. Comparison of the abundance of various microinvertebrates in experimental pens and the unmanipulated weed
bed in 1981. Abundances are the mean (?SEM) number of animals in four replicate samples from each habitat. Overall
differences among habitats are indicated by significant values of P from an ANOVA for each species over all habitats.
Specific differences among habitats are indicated by results of Duncan's multiple range tests; means that do not differ
significantly are connected by underlining.

Microinvertebrate abundance (no./L)

Taxon Exclusion pens Open pens Weed bed ANOVA P
Sibnocephal/us serrulatus 320.3 ? 114.4 29.8 ? 8.6 17.8 + 3.9 .0004
Ostracods 22.5 ? 6.1 99.8 228.7 82.7 ? 9.3 .0080
Cyclopoid copepods 33.0 ? 4.6 62.5 + 22.9 34.3 ? 7.9 .53
Chvdorus sphaericus 16.3 ? 3.3 33.5 + 11.3 13.8 ? 3.5 .16
Scapholeberis kingi 0.8 + 0.8 4.5 + 1.7 1.0 ? 1.0 .08
Mites 1.0 ? 0.4 2.0 + 0.0 2.8 + 0.5 .03
Total 393.8 ? 118.7 232.0 + 17.3 152.3 ? 19.1 .04

open pens did not differ significantly from abundances in Carroll's Pond and elsewhere (Crowder and Cooper
in the Paspaluin bed (Table 4). In contrast, the aver- 1982) suggest that it may be more profitable to ask why
age abundance of Simocephalus serrulatus, a large lit- the impact of fish exclusion varies among communi-
toral cladoceran (adult body length 1-3 mm), was sig- ties, than to suggest that the species composition of
nificantly enhanced by fish exclusion. This increase in freshwater macroinvertebrates is generally unaffected
large microcrustacea occurred despite an order of by fish predation (Thorp and Bergey 198 la).
magnitude increase in the abundance of odonates which Differences in odonate abundance and species com-
also prey on Cladocera. These results further empha- position among caging treatments in Carroll's Pond
size the close resemblance of open-pen communities cannot be easily dismissed as cage artifacts. Several
to the natural weed bed and illustrate the pleiotropic lines of evidence suggest that the open pens faithfully
effects of fish exclusion on community structure. mimicked the natural community found in the Pas-
p(luln bed. Open pens did not concentrate fish above
DISCUSSION natural densities measured in the surrounding Paspa-
Striking differences in the abundance and domi- lum bed. Abundances of various microcrustacea were
nance of odonate species between open pens and fish similar in both the open pens and the weed bed. Emer-
exclosures clearly implied that fish can influence the gence rates of odonates in open pens and the weed
structure of odonate assemblages. These results sup- bed were similarly low compared to much higher
port the general importance of fish predation in struc- emergence rates in fish exclusion pens. No evidence
turing freshwater communities (e.g., Zaret 1980, suggests that open pens exaggerated the effects of fish.
Crowder and Cooper 1982) and experimentally con- Odonate responses to the presence of fish can be
firm previously untested hypotheses regarding poten- parsimoniously attributed to direct effects of verte-
tial relations between fish predation and odonate brate predation. Lepomlis Inacrochirus is known to prey
species composition (Benke 1978, Johnson and Crow- heavily on larval odonates (e.g., Sadzikowski and
ley 1980). Wallace 1976). Shifts of dominance from smaller to
It is unclear why previous fish exclusions in some larger odonate species were qualitatively similar to well-
communities containing many of the same odonates known responses of zooplankton species composition
failed to produce analogous results (Thorp and Bergey to fish predation; a small species predominated when
198 la). Disparate responses of communities to fish ex- fish were present, but larger species became dominant
clusion may simply reflect variation in the abundance when fish were excluded (e.g., Brooks and Dodson
of predators among ponds and lakes. Exclusions of 1965, Zaret 1980). This result appeared to be a simple
sparse populations of fish may have little apparent ef- consequence of the severe impact of fish on the abun-
fect on community structure when compared to results dance of odonates with moderate body sizes in the
of similar exclusions where fish occur at much greater, final larval instar. Fish did not enhance the abundance
but nonetheless natural, densities. Different responses of any odonate species, offering no evidence for shifts
to predator exclusion may also reflect variation among in dominance caused by predator-mediated competi-
communities in habitat complexity, given the impor- tion among prey (Paine 1966, Morin 1981).
tant role that spatial refuges play in moderating the The strong depression of microcrustacean abun-
intensity of predation in freshwater communities dance by fish suggested an additional mechanism that
(Benke 1978, Crowder and Cooper 1982). Striking re- may have contributed to the effect of fish exclusion on
sponses of the weed bed community to fish exclusion odonate abundance; fish and odonates may conceiv-

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
February 1984 FISH PREDATION ON LARVAL ODONATES 59

ably compete for microcrustacean prey. Fish exclu- habitat specificity, and competition with predator-re-
sion dramatically increased the abundance of the large sistant species in producing exclusions and patterns of
cladoceran Silnocephalus serraulatus, despite a corre- dominance require further study. For instance, ob-
sponding increase in the abundance of larval odonates served patterns of larval dominance may have simply
that also prey on microcrustacea. Similar increases in resulted from variation in the availability of critical
the abundance of Silnocephalus follow fish exclusion microhabitats (e.g., Crowley and Johnson 1982), and
in other communities (D. M. Johnson, personal coln- this pattern may have been only coincidentally cor-
minnication). Observed effects of fish on odonates may related with body size. A negative relation between
result both from direct predation on naiads and from final-instar size and dominance may not generalize to
competition between fish and surviving naiads for re- other communities (Thorp and Bergey 1981l, Crowley
duced stocks of microcrustacean prey. This latter pos- and Johnson 1982), but such comparisons are clouded
sibility is offset by the observation that benthic insect by differences in sampling techniques and potential
larvae may contribute more to odonate diets than do variation in predation intensity among studies.
microcrustacea (Benke 1978). In addition to documenting the effects of fish on
The consistent negative relation between body size odonate species composition, caging experiments can
of final-instar larvae and population sizes of various illuminate aspects of dragonfly natural history that are
species of metamorphosing odonates initially suggest- difficult to address with purely descriptive studies. For
ed that a simple relation might exist between body size instance, previous studies of dragonfly assemblages in
and susceptibility to predatory exclusion. Although all the southeastern United States have suggested that
odonates pass through a wide range of sizes during most odonates are univoltine (Paulson and Jenner 1971,
larval development, size in the final instar may be crit- Benke and Benke 1975). In contrast, the metamorpho-
ical in determining larval success in the face of size- sis of four species (Peritheinis tenera, Pachydiplax
selective vertebrate predation. If vulnerability to pre- longipen nis, Eryth enis simnplicicollis, and Tran'meala-
dation increases with body size, small species may cer(Ita) from fish exclosures in Carroll's Pond following
have an obvious advantage over larger species. Small- only 1-2 mo of development demonstrated these
er species can metamorphose at smaller, less vulner- species are facultatively multivoltine in this portion of
able body sizes than their larger counterparts. Larger their geographic range. The great rapidity of odonate
species are committed to metamorphosis at larger sizes, development observed in this study emphasized the
and may thereby incur greater intensities of size-de- importance of monitoring metamorphosing individu-
pendent vertebrate predation. Interspecific differences als, even during relatively short-term experiments.
in size among larvae metamorphosing from Carroll's Without such monitoring, significant differences among
Pond corresponded to well-known species-specific dif- treatments can effectively metamorphose and fly away,
ferences (Needham and Westfall 1955), indicating that while remaining undetected.
the negative relation was not simply a manifestation This experiment demonstrated that predation can be
of density-dependent growth. However, the large an important regulatory mechanism during the larval
species Epicordulia retina and Tralnea lalcera ta re- phase of the odonate complex life cycle. Fish greatly
mained equally rare, regardless of altered fish densi- reduce the number of larvae making the successful
ties. The failure of these large species to respond to transition between the aquatic larval phase and the
fish exclusion suggested that their low abundances in terrestrial/aerial adult phase of the life cycle. Such in-
Carroll's Pond were not explained by size-selective teractions suggest the potential for novel competitive
fish predation. interactions between fish and terrestrial species over
The negative relation between size and abundance aquatic and terrestrial phases of a shared odonate re-
in this odonate guild clearly has a complex basis that source. This admittedly speculative interaction de-
may be only partially related to predation. Further, serves further study and is supported in part by other
the consistent reproductive failure of some odonates observations. Certain birds can depend heavily on te-
listed in the Appendix cannot be generally attributed neral odonates for food during the breeding season
to fish predation. Only one of the species that regularly (Orians 1980). A negative relation between reproduc-
failed to metamorphose from Carroll's Pond, Libellula tive success in some waterfowl and the presence of
incest, increased in abundance in the absence of fish. fish is also correlated with reduced abundances of
The continued failure of larval recruitment of other shared macroinvertebrate prey where fish are present
abundant ovipositing species (e.g., Celitheinis faiscia- (Eriksson 1979). Predators with complete habitat sep-
ta, Libell/la cvanea, and others) suggested that their aration (aquatic vs. terrestrial) may conceivably com-
exclusion in the larval phase of the life cycle was not pete for different life history stages of the same prey
related to fish predation. Adults of these species pre- species, especially if prey abundance in both habitats
sumably maintain populations at Carroll's Pond only is tightly linked by metamorphosis and reproduction,
by migration from other ponds where they are repro- and if prey abundance in one habitat is limited by pre-
ductively successful. dation.
The potential roles of physiological constraints, The degree of coupling between the abundance of

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions
60 PETER J. MORIN Ecology. Vol. 65, No. 1

adult and larval odonates may be quite variable. Com- Johnson, D. M. 1973. Predation by damselfly naiads on
position of the aquatic larval phase of odonate assem- cladoceran populations: Fluctuating intensity. Ecology 54:
25 1-268.
blages probably reflects a differential "filtering" of eggs Johnson, D. M., and P. H. Crowley. 1980. Odonate "hide
and larvae of susceptible odonates by vertebrate pre- and seek"': Habitat-specific rules" Pages 569-579 in W.
dation (this study), predatory and competitive inter- Charles Kerfoot. editor. Evolution and ecology of zoo-
actions among larvae (Benke 1978, Benke et al. 1982, plankton communities. University Press of New England,
Crowley and Johnson 1982), and possible physiologi- Hanover, New Hampshire, USA.
Macan, T. T. 1977. The influence of predation on the com-
cal constraints (Corbet 1980). Composition of the adult position of freshwater animal communities. Biological Re-
phase of odonate assemblages probably reflects input views 52:45-70.
from the larval phase at that pond, migration of adults Morin. P. J. 1981. Predatory salamanders reverse the out-
from other breeding sites, and competitive and pred- come of competition among three species of anuran tad-
poles. Science 212:1284-1286.
atory interactions among adults. Integrated studies of Needham, J. G.. and M. J. Westfall, Jr. 1955. A manual of
both aquatic and larval phases of odonate assemblages the dragonflies of North America (Anisoptera). University
are needed to sort out the relative importance of these of California Press. Berkeley. California. USA.
potentially complex interactions in determining the Orians, G. H. 1980. Some adaptations of marsh-nesting
distribution and abundance of species with complex blackbirds. Princeton University Press, Princeton, New
Jersey, USA.
life cycles. Paine, R. T. 1966. Food web complexity and species di-
versity. American Naturalist 100:65-85.
ACKNOWLEDGMENTS
Paulson, D. R., and C. E. Jenner. 1971. Population struc-
This study would have been impossible without the co- ture in overwintering larval Odonata in North Carolina in
operation of John and Lucille Carroll, who own the pond. relation to adult flight season. Ecology 52:96-107.
Their patient tolerance of backyard ecology is greatly appre- Sadzikowski, M. R., and D. C. Wallace. 1976. A compar-
ciated. M. K. Morin, R. Harris, D. Campbell. A. Gotten, S. ison of the food habits of size classes of three sunfishes
Lewis, H. F. Nijhout, K. Clay. and H. M. Wilbur assisted (Lepomnis Inalrochirius Rafinesque, L. gibbosus (L innaeus)
in the field and/or commented on the manuscript. R. T. Paine and L. c'vaiuel//Is Rafinesque). American Midland Natural-
supplied an obscure reference, and comments by reviewers ist 95:220-225.
D. M. Johnson and A. C. Benke improved the manuscript. Southwood, T. R. E. 1978. Ecological methods. Second
Any errors of interpretation remain my own. edition. J. Wiley and Sons, New York, New York, USA.
During part of this study I was supported by National Sci- Thorp, J. H., and E. A. Bergey. 1981Ia. Field experiments
ence Foundation Grant DFB-791 1539 to H. M. Wilbur, and on responses of a freshwater. benthic macroinvertebrate
by a teaching assistantship in the Department of Zoology, community to vertebrate predators. Ecology 62:365-375.
Duke University. Computing funds were provided by the De- Thorp, J. H., and F. A. Bergey. 1981lb. Field experiments
partment of Zoology, Duke University. on interactions between vertebrate predators and larval
midges (Diptera: Chironomidae) in the littoral zone of a
LITERAtURE CITED)
reservoir. Oecologia (Berlin) 51:285-290.
Benke, A. C. 1978. Interactions among coexisting preda- Wilbur, H. M. 1980. Complex life cycles. Annual Review
tors: a field experiment with dragonfly larvae. Journal of of Ecology and Systematics 11:67-93.
Animal Ecology 47:335-350. Zaret, T. M. 1980. Predation and freshwater communities.
Benke. A. C., and S. S. Benke. 1975. Comparative dynam- Yale University Press, New Haven, Connecticut, USA.
ics and life histories of coexisting dragonfly populations.
Ecology 56:302-317. APPENDIX
Benke, A. C.. P. H. Crowley, and D. M. Johnson. 1982.
Interactions among coexisting larval Odonata: an in situ The dragonfly species pool at Carroll's Pond. Eleven species
experiment using small enclosures. Hydrobiologica 94:121- both patrol the pond as adults and successfully metamor-
130. phose in some years at this site. The remaining species patrol
and oviposit in the pond but do not successfully complete
Brooks, J. L., and S. 1. Dodson. 1965. Predation, body size
larval development. The species list is based on daily obser-
and composition of plankton. Science 150:28-35. vations in 1979 (2-5 h/d) and casual observations in 198()and
Corbet. P. S. 1980. Biology of Odonata. Annual Review of 1981. Species sighted on <5 d in 1979 are indicated with an
Entomology 25:189-217. asterisk.
Crowded. L. B., and W. E. Cooper. 1982. Habitat struc-
tural complexity and the interaction between bluegills and
their prey. Ecology 63: 180-18 13. Reproductively
successful species Reproductive failures
Crowley. P. H.. and D. M. Johnson. 1982. Habitat and
seasonality as niche axes in an odonate community. Ecol- Perith emi.s teiiera( Libel/lua iacesta
ogy 63: 1064-1077. Er-themni8si sipl(i(ollis Libel/lua (!'vanea
Eriksson. M. 0. G. 1979. Competition between freshwater Pa/ ch!dipl/ux /onlgipenuiis Celithemis fa's(iata
fish and Goldeneyes Buccphl/ot c/angula (L.) for common Epithe(a c aa.osra Tralnea arlin/al
prey. Oecologia (Berlin) 41:99-107. Pl/athemhislvdia Au ax~juius
Hall, D. J., W. E. Cooper, and E. E. Werner. 1970. An Libel/lla la/(tuosI A uti longipes*
experimental approach to the production dynamics and GoCInphus bridle 'vi Epicordulia princeeps B
structure of freshwater animal communities. Limnology and Ladonia (deplani(ltal Pantala Ihivienea *
Oceanography 15:839-928. Epicrdullia reginla Panitala Jlavesceas
Haney. J. F., and D. J. Hall. 1973. Sugar-coated Daplhnia: ira uiea laer (a(l tal Libelllda puli ell/t *
DIromiogoil plius Spillosits Ervtlirodiplax njiauscula*
a preservation technique for Cladocera. Limnology and
S'vuipetrium i'iiiliii *
Oceanography 18:331-333. Ce lit/ieiis e)loi(i* '
Jacobs. M. E. 1955. Studies on territorialism and sexual Celithemsszis e'lisa{g*
selection in dragonflies. Ecology 36:566-586.

This content downloaded from 128.235.251.160 on Wed, 18 Mar 2015 21:37:47 UTC
All use subject to JSTOR Terms and Conditions