species richness, species composition and population ... pubs...key-words: microcosms, protists,...

http://www.jstor.org

Species Richness, Species Composition and Population Dynamics of Protists in ExperimentalMicrocosmsAuthor(s): Sharon P. LawlerSource: The Journal of Animal Ecology, Vol. 62, No. 4 (Oct., 1993), pp. 711-719Published by: British Ecological SocietyStable URL: http://www.jstor.org/stable/5391Accessed: 27/08/2008 11:07

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you

may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

http://www.jstor.org/action/showPublisher?publisherCode=briteco.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed

page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the

scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that

promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

http://www.jstor.org/stable/5391?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/action/showPublisher?publisherCode=briteco

Journal of Animal Ecology 1993, 62, 711-719

Species richness, species composition and population dynamics of protists in experimental microcosms SHARON P. LAWLER Department of Biological Sciences, and Bureau of Biological Research Rutgers University Piscataway, NJ 08855-1059, USA; and *NERC Centre for Population Biology, Imperial College at Silwood Park, Ascot, Berks SL5 7PY, UK

Summary

1. Laboratory experiments using communities assembled from bacteria plus one, two or four protist predator-prey pairs tested whether species richness affects the persistence, abundance and temporal variability of the protist species. Comparisons among six different food webs of four protist species tested whether the effects of species composition rival the effects of species richness. 2. Most populations in species-rich food webs were less abundant than populations of the same species in food webs with fewer species. Increased species richness increased the variability of one species. Combining predator-prey pairs increased the number of extinctions. 3. The species composition of food webs also affected the persistence and mean abundances of many species, and changed the variability of two species. Strong effects of species composition may limit the ability of ecologists to explain population dynamics by examining general features of community structure.

Key-words: microcosms, protists, species composition, species richness, stability.

Journal of Animal Ecology (1993) 62, 711-719

Introduction

Community ecologists have long been interested in whether the structure of biological communities affects the stability of the populations that comprise the community (review: Pimm 1991). For example, species richness is thought to influence population stability, but there is surprisingly little data to bolster theoretical predictions about how species richness will influence populations. Another important question is whether community structure affects populations as strongly as more individualistic properties of the community (e.g. species composition). This study compares the relative effects of species richness and species composition on the stability of eight protist species. Below, I first discuss empirical measures of 'stability', and then consider how community structure might affect population dynamics.

In theoretical population biology, equations that model population dynamics are said to be stable when their parameters ensure that the populations return to previous, predictable behaviour after a perturbation. However, it is rarely feasible to demonstrate this type of stability in real populations,

* Present address.

so ecologists instead use empirical criteria for stability (reviews: Connell & Sousa 1983; Underwood 1989; Pimm 1991). Three measures are used in this study, the persistence of populations, variation in population size and mean abundance over time. These measures are not identical to mathematical stability, but may be correlated with it.

Persistence of the population is a common practical measure of stability (e.g. Hurd et al. 1971; May 1974; Abrams & Allison 1982). Although some mathematically unstable systems can persist for a long time, it seems likely that stable systems will persist for longer (on average) than unstable systems. Population variability is also frequently used as a measure of stability (e.g. Connell & Sousa 1983). Low population variability need not imply high stability in the mathematical sense, because population models that are mathematically stable may show stable cycles caused by density dependence (Leslie & Gower 1960; May 1974; Schoener 1985). However, models parameterized to be mathematically unstable show greater variability in abundances than the same models with stable parameter combinations (Taylor 1992). There is also empirical evidence that more variable populations are more likely to go extinct (all else being equal) (Pimm 1991). Therefore, low population variability may 711

712 Experimental communities of protists

be strongly correlated with persistence and reliably indicate population stability. In any case, high variability can predispose a population to extinction through stochasticity or Allee effects, if the cycle approaches zero (May 1974).

The abundances of the populations in these experiments were also monitored, because small populations are more vulnerable to extinction (review: Soule 1987).

Species richness or the number of species in a community may be a key element of community structure which influences population and community dynamics (MacArthur 1955; Elton 1958; May 1974; Pimm 1982, 1991). However, more experiments are needed to produce a general state- ment about the effects of species richness on community stability (Goodman 1975; Pimm 1984; McNaughton 1988).

Ecologists have also examined connectance (the proportion of actual to possible predator-prey links among species) and interaction -strength for their effects on the stability of community models (e.g. May 1972; Pimm 1982). Additionally, species composition could affect community stability as strongly as any of these factors.

The effects of species composition are often regarded as noise by ecologists searching for broad community patterns, because species composition tends to be unique to each community. However, it is important to compare the effects of general community properties, such as species richness, to the effects of species composition, to guard against oversimplification. For example, the consequences of adding species to an island may depend critically on the identity of the species rather than on the number of species, since island biotas are notoriously sensitive to particular introduced species (Atkinson 1989). If an effect of species richness on stability is detectable over many communities, but the stability of any given community is almost completely determined by the identity of its species, discussing only the former could mislead those who refer to the ecological literature when managing natural communities.

The effects of community structure on population dynamics remain largely unknown. This research addresses the dearth of empirical studies with an experiment designed to explore relations between species richness, species composition and population dynamics in laboratory microcosms of protozoans. 'Bottle experiments' such as this one have proven to be effective tests of ecological theory (e.g. Hairston et al. 1968; Luckinbill 1979; Dickerson & Robinson 1985; Drake 1991), and can provide the data necess- ary to advance theory (Kareiva 1989; Lawton 1989).

Materials and methods

EXPERIMENTAL DESIGN

To test whether species richness affects the persistence, variability and abundance of populations, several predator-prey pairs that could coexist independently were either grown separately or combined (May 1974; Luckinbill 1979). Pilot work identified predator-prey pairs that could coexist for at least 1 month (approximately 10-15 generations of the predators and over 60 generations of the prey). Three persistent pairs were used: Steinia spp. (Diesing) and Uronema spp. (Dujardin), Urostyla spp. (Ehrenberg) and Askenasia spp. (Blochmann), and Euplotes spp. (Ehrenberg) and Chilomonas spp. (Ehrenberg). Another pair which is sometimes unstable (Peter J. Morin, unpublished data), Ble- pharisma spp. (Perty) and Colpidium spp. (Stein), was also used and proved persistent in this experiment. The treatments are listed below. Treatment abbreviations are the first letter of the predator of each predator-prey pair. Predator-prey pairs:

Steinia and Uronema (S); Blepharisma and Colpidium (B); Euplotes and Chilomonas (E); Urostyla and Askenasia (U).

All possible four-species combinations of the predator-prey pairs:

Steinia and Uronema + Blepharisma and Colpidium (SB); Steinia and Uronema + Euplotes and Chilomonas (SE); Steinia and Uronema + Urostyla and Askenasia

(SU); Blepharisma and Colpidium + Euplotes and Chilomonas (BE); Blepharisma and Colpidium + Urostyla and Askenasia (BU); Euplotes and Chilomonas + Urostyla and Askena- sia (EU).

And the eight-species combination: Steinia and Uronema + Urostyla and Askenasia + Blepharisma and Colpidium + Euplotes and Chilo- onas (SBEU).

Each species was a member of two-, four- and eight-species food webs and each species was also a member of three different four-species food webs. This design allowed comparison of the effects of species richness to the effects of species composition.

Blepharisma, Colpidium, Euplotes and Chilo- monas were originally obtained from Carolina Bio- logical Supply Co, Burlington, NC, USA. Steinia were isolated from a Carolina Biological culture of Bursaria. Urostyla, Uronema and Askenasia were isolated from natural puddles in Middlesex Co., NJ.

713 S. P. Lawler

FOOD WEB ASSEMBLY AND MONITORING

Experimental microcosms were covered 240-ml glass jars containing 100 ml of medium (0-56 g of Carolina Biological protozoan pellets per litre of well water) plus two wheat seeds to provide additional nutrients. These materials were autoclaved before use. Medium was initially inoculated with four species of bacteria that are edible to many protozoans: Bacillus cereus and Bacillus subtilus var niger (Curds & Vandyke 1966), Serratia marcescens and Proteus vulgaris (Taylor & Berger 1976).

The medium used in the experiment was also inoculated with samples of the bacteria present in the stock cultures of all species used in the experiment. To -prepare this inoculum, 5 ml of medium from each stock culture was filtered through a 1-2-Rm millipore filter. The filter retained protozoans while allowing bacteria to pass through. All filtrates were mixed and the mixture was used to inoculate the medium used in the experiment. This procedure ensured that treatment effects were due to protozoans (directly or indirectly), rather than to initial differences in the composition of the bacterial trophic level.

Jars were randomized before bacterivorous protozoans (Uronema, Chilomonas, Askenasia, and Colpidium) were added. For each species, the stock culture was swirled to mix it and 10 drops were added to each jar where the species was required. After a 2-day time-lag to allow prey to become sufficiently abundant, predators were added. Pre- dators had been cultured on the prey that they were paired with in the design. 1 ml of each predator- prey culture was added where required. This contained approximately 20 Urostyla, 30 Blepharisma, 30 Euplotes, or 50 Steinia. Jars were placed in randomized locations on a laboratory shelf. During the remainder of the experiment, one sterile wheat seed was added to each jar weekly and jars were sampled every 2 days.

SAMPLING

To sample microcosms, each jar was swirled to mix its contents, and 10 drops of the medium and suspended organisms were placed on a tared Petri dish with a sterile pipette. The sample was weighed to 0 0001 g on a Sartorius balance and protists were counted with the aid of a stereoscopic microscope. Samples of bacterivores sometimes required dilution before counting. In these cases 4-10 drops were weighed on a tared Petri dish. Enough sterile medium was added to reduce the count to 5-20 protists per drop, then the dilution was sampled as above. The number of individuals in a standard sample was estimated by multiplying the count by the weight of the dilution divided by the initial weight of the sample. Ten drops were removed from every jar

for each sample. Counts were scaled to a 0.3159-g sample (the mean sample weight).

Experiments were sampled every 2 days for 38 days, beginning 4 days after the first protists were added to microcosms. Two weeks after the last sample, all jars were searched to determine which species were still present. Jars were mixed and approximately one-quarter of the contents was examined under the dissecting microscope to improve assessment of extinctions.

PREDATOR FEEDING TRIALS

Feeding trials tested which of the prey species each of the predators would consume. Information from these trials was used to estimate connectance (the proportion of realized trophic links to possible trophic links) in each food web type. Predators were placed in a Petri dish of protist-free medium for 1 hour, to standardize their hunger level. Meanwhile, stock cultures of prey were diluted with protist-free medium to approximately 50-100 individuals per 10-drop sample. 0-5 ml of this dilution was added to each of eight 1-ml wells of a sterile well plate. Four wells received 10 predators each and four were predator-free controls. These densities of predators and prey were within the range of densities of coexisting predator-prey pairs.

Two or three prey species were exposed to one predator per trial period. One species was always the prey that the predator was paired with in the two-species treatment. This served as a 'reference' species that the predator was known to eat, since the feeding rate of some predators changes over time and with hunger level (e.g. Salt 1961; Seshacher, Saxena & Girgla 1971). Feeding trial duration was less than 2 hours to ensure minimal prey division.

Prey were counted beginning 1 hour after predators were added. Wells were sampled as in 'Sampling' (above). ANOVAS on prey counts with predator pre- sence as the factor determined whether predators had reduced prey numbers.

To check whether predators could eat other predators, all possible two-way predator combinations were added to drops of medium on a Petri dish and observed for 15 min.

DATA TRANSFORMATION

Data from the first seven sampling dates were excluded to eliminate the initial low abundances and variance associated with species introductions from the data set (hereafter 'build-up dates').

In some cases, extinctions produced a series of zero abundances in the latter sampling dates. These zeros would reduce the mean abundances and their standard deviations for a trivial reason, so data from any replicates where the species was absent on


five or more dates and absent in the final sample (hereafter 'extinctions') were eliminated.

One case where a species went extinct in four out of five replicates of a treatment was excluded from analysis because it was not possible to determine within-treatment variation. Two replicates of the 'BU' treatment for Urostyla and one replicate of the 'SU' treatment for Askenasia were omitted, because only one individual was detected late in the experiment (after a 10-date absence).

All data were transformed with a log10 (abundance + 1) transformation, to eliminate any positive cor- relations between the mean and standard deviation of the abundances. After this transformation, mean abundances and mean standard deviations of each species were either uncorrelated or negatively correlated. The mean transformed abundances were used in ANOVAS with either species composition or species richness as the single factor. Where the ANOVA showed significant effects, significant differences between means were determined using a Ryan's Q test, which is powerful, yet controls the type 1 error rate well (Day & Quinn 1989).

MEASURING VARIABILITY

Mean standard deviations over time were calculated to estimate the variability of each population, using the transformed abundance data and omitting build- up dates, extremely low abundances and extinctions as above. The standard deviation of log10 (abundance + 1) can underestimate variability if there are many zeros in the data set (McArdle, Gaston & Lawton 1990). However, this data set contained few zeros because zero values due to extinctions were discarded, and inspection of the remaining data indicated that such bias was unlikely. The resulting standard deviations were used as the response values in non-parametric analyses of variance to determine whether species richness or species composition affected temporal variability. Differences were considered to be significant at P < 0 05, using a Kruskal-Wallis test. Analyses were repeated using the coefficient of variation of abundances. This measure has been recommended as a relatively unbiased measure of variability (McArdle et al. 1990). The two methods produced almost identical results, so only the former is presented.

Results

PATTERNS OF SPECIES EXTINCTIONS

Combining predator-prey pairs to increase species richness resulted in the extinctions of some of the species. There was only one extinction in the two- species webs; however, there were extinctions in at least some replicates of five out of the six four- species combinations, and in all replicates of the

Table 1. Extinctions. - Numbers give the number of replicates of the treatment (out of five) in which the species went extinct. 'Extinctions' were populations that were not detected in the last five samples and that were not present in the final search of the microcosms. Steinia, Euplotes and Blepharisma showed no extinctions, and there were no extinctions in the three two-species treatments not listed

Treatments

Species S SB SE SU BE BU EU SBEU

Askenasia - - - 4 - 5 3 5 Chilomonas - - 0 - 0 - 2 0 Colpidium - 0 - - 0 0 - 4 Uronema 1 1 3 0 - - - 2 Urostyla - - - 5 - 3 0 0

- Species not present in the treatment.

eight-species combination (Table 1). A chi-square test verified that extinctions were not random with respect to species richness (X2 = 7*7017, df = 2, P < 0.025).

The four- and eight-species treatments showed comparable numbers of extinctions (x2 = 0.4414, df = 1, P > 0 1). Twenty-two per cent of the populations (26 out of 120 populations) became extinct in the four-species combinations, and 28% of the populations (11 out of 40) became extinct in the eight-species treatment.

Extinctions seemed to be related to trophic level. All four of the prey species went extinct in some replicates of at least one treatment, whereas only one predator, Urostyla, went extinct in any replicate of any treatment (Table 1).

SPECIES RICHNESS AND TOTAL WEB

ABUN DANCE

Species richness did not affect the mean total abundance of all the species in the food web (ANOVA

df 2, 52, F= 0-32, P> 0-7245). Mean transformed

abundances for webs of each species richness were very similar (two spp. = 2*243, four spp. = 2.242, eight spp. = 2.103).

TROPHIC LEVEL, ABUNDANCE AND

VARIABILITY

Over all treatments, prey were more abundant than predators. Mean back-transformed abundance for prey was 32.2 individuals per sample, while there were 16*8 predators per sample (ANOVA df 1, 154, F=7-01, P<O.O1). Prey were more variable than predators (non-parametric ANOVA: prey mean SD =0 O4562, predator mean SD = O*2127, Kruskal- Wallis P<O.OOO1).

715

S.P. Lawler ABUNDANCE OF INDIVIDUAL SPECIES

Increased species richness decreased the mean abundance of five of the eight species, Askenasia, Blepharisma, Chilomonas, Steinia, and Uronema, and showed trends toward decreasing the abundance of Urostyla (Table 2).

Species composition had a strong effect on the abundances of at least four species (Table 3). Abun- dances of Blepharisma, Colpidium, Euplotes and Uronema all differed among the four-species treatments. This analysis was not possible for Askenasia and Urostyla, which persisted in some replicates when combined with Euplotes and Chilomonas, but went extinct in all replicates of the other four- species treatments.

VARIABILITY OF INDIVIDUAL SPECIES

Increased species richness did not affect the temporal variabilities of most of the species (Table 4). Only Steinia had a higher mean standard deviation over time in the food webs of greater species richness. For some of the other species, however, extinctions were more common in the four- and eight-species combinations (Table 1). If highly variable populations are more likely to go extinct, extinctions might have obscured some of the variability of more species-rich combinations.

Species composition affected the variabilities of at least two of the species, Blepharisma and Colpidium, and showed a trend toward affecting the variability of Euplotes (Table 5). This type of comparison was not possible for Askenasia and Urostyla, which persisted in only one four-species combination. However, the ability of species to persist in some four-species combinations, but not others is further evidence that species composition can determine whether a particular species will be stable in the community. There was no effect of species composition on the temporal variability of Chilomonas and Uronema, but both went extinct in one type of four-species community (Table 1).

FEEDING TRIALS AND FOOD WEB

CONNECTANCE

In the feeding trials, most of the predatory species ate most of the prey (Table 6). Euplotes, Steinia, and Urostyla each decreased the abundances of Chilomonas, Colpidium and Uronema. None of the predators decreased the abundance of active Askenasia. Strangely, Euplotes seemed to have a positive effect on Askenasia abundance. This out- come may have resulted from a poorly mixed dilution of Askenasia, because the duration of the trial was too short for significant prey division.

It seemed odd that none of the predators decreased

Table 2. ANOVA results for the effects of species richness on the abundance of each species. Tr = number of species in treatment. N = number of replicates. Ryan's Q = Ryan's Q multiple range test. Within species, means with the same letter were not different (a = 0-05). dfl and df2 are model and error degrees of freedom in the ANOVA, respectively. F and P values from the ANOVA are shown

Mean log Ryan's df Species Tr N (abund + 1) Q (1, 2) F P

Askenasia Two 5 1-65 A (1, 6) 69-88 0-0002** Four 3 0-45 B

Blepharisma Two 5 1-97 A (2, 22) 18-10 0-0001*** Four 15 1-80 A Eight 5 0-53 B

Chilomonas Two 5 2-38 A (2, 22) 5 90 0.0097* Four 13 1-58 B Eight 5 1-62 B

Colpidium Two 5 1-21 A (1, 18) 0.19 0-6675 Four 15 1*12 A

Euplotes Two 5 1-09 A (2, 22) 1.01 0-3822 Four 15 1-09 A Eight 5 0 94 A

Steinia Two 5 1-76 A (2, 22) 2-26 0-1281 Four 15 1-43 A Eight 5 1.22 A

Uronema Two 4 2-61 A (2, 15) 5-49 0-0163* Four 11 1.77 AB Eight 3 0-78 B

Urostyla Two 5 0-44 A (2, 8) 4-32 0-0533 Four 4 0-14 A Eight 2 0-09 A


Table 3. ANOVA results for the effects of species composition on the abundance of each species. Comparison of each species in three different four-species assemblages. Tr = treatment. N = number of replicates. Ryan's Q: results of Ryan's Q multiple range test. Within species, means with the same letter were not different (a = 0-05). dfl and df2 are ANOVA model and error degrees of freedom, respectively. F and P values from the ANOVA are shown

Mean log Ryan's df Species Tr N (abund + 1) Q (1, 2) F P

Blepharisma BE 5 2-13 A (2, 12) 18-43 0-0002** BU 5 2-09 A SB 5 1-21 B

Chilomonas SE 5 1-75 A (2, 10) 0-96 0-4152 EU 3 1-60 A BE 5 1-41 A

Colpidium SB 5 1-65 A (2, 12) 19-79 0.0002** BU 5 0-99 B BE 5 0-72 B

Euplotes EU 5 1-29 A (2, 12) 12-14 0.0013* SE 5 1-14 A

BE 5 0-84 B

Steinia SE 5 1-91 A (2, 12) 58-61 0.0001*** SU 5 1-52 B SB 5 0-86 C

Uronema SU 5 2-52 A (2, 8) 8-92 0.0092* SB 4 1-08 B SE 2 1-29 B

Table 4. Effects of species richness on population variability. Results of non-parametric ANOVAS on standard deviations (SD) of log-transformed abundances for each species in food webs of three different species richnesses. The three four-species combinations were lumped to give an overall mean standard deviation for food webs with four species. P values are from Kruskal-Wallis tests

Species Treatment N Mean SD P

Askenasia 2 spp. 5 0-32 0-4561 4spp. 3 050

Blepharisma 2 spp. 5 0-10 0.4594 4 spp. 15 0-15 8 spp. 5 0-15

Chilomonas 2 spp. 5 0-36 0-5306 4 spp. 13 0-42 8 spp. 5 0.53

Colpidium 2 spp. 5 0-45 0-3539 4spp. 15 0-37

Euplotes 2 spp. 5 0.20 0-5628 4 spp. 15 0-19 8spp. 5 0-25

Steinia 2 spp. 5 0-17 0.0024* 4 spp. 15 025 8 spp. 5 0-60

Uronema 2 spp. 4 0.47 0.7094 4 spp. 11 0.62 8 spp. 3 0.71

Urostyla 2 spp. 5 0-20 0-2340 4 spp. 4 0-19 8 spp. 2 0-13

Askenasia abundances, since Urostyla decreased Askenasia abundances in a previous experiment. A subsequent trial showed that Steinia and Urostyla ate the resting cysts of Askenasia, but Euplotes and Blepharisma did not.

Blepharisma seemed to feed on fewer species of

Table 5. Effects of species composition on population variability. Tr=treatment, N= number of replicates, Mean SD = mean standard deviation of logio (abundance + 1). P values are from Kruskal-Wallis tests

Species Tr N Mean SD P

Blepharisma BE 5 0 04 0-0092* BU 5 0 20 SB 5 0-22

Chilomonas BE 5 0-48 0 4061 EU 3 0-47 SE 5 034

Colpidium BE 5 0-32 0-0025* EU 5 0 57 SB 5 0-21

Euplotes BE 5 0-25 0-1023 EU 5 0-18 SE 5 0-13

Steinia SB 5 0-30 0-4724 SE 5 0 25 SU 5 0-21

Uronema SB 4 0-62 0-8319 SE 2 0-57 SU 5 0-63

717 S. P. Lawler

Table 6. Feeding trials. Results of ANOVAS comparing prey abundances in treatments with and without predators. Predator abbreviations are the first five letters of the genus. For all trials, dfl = 1, df2 = 6, N= 4. C = mean abundance of controls, p = mean abundance of predator treatments. The F statistic and probability are listed below means as (F, P)

Prey: Askenasia Chilomonas Colpidium Uronema

Pred. C p C p C p C p

Bleph. 50 49 65 65 94 76 61 48 (0-001, 0-947) (0-001, 0-982) (24-92, 0.002) (4-90, 0-068)

Euplo. 63 70 75 40 56 28 125 64 (11-08, 0.015) (21-02, 0-004) (32-96, 0.001) (32-42, 0.001)

Stein. 66 59 62 29 69 20 85 15 (0-91, 0.376) (23-28, 0-003) (86-8, 0-001) (108-0, 0-001)

Urost. 64 62 68 34 41 3 61 11 (0-01, 0.92) (22-46, 0.003) (41-31, 0.001) (447-1, 0-001)

protozoan prey in these systems than the other predators. Blepharisma decreased the abundance of Colpidium in one trial; however, it had no effect on Colpidium in two similar trials (ANOVA df 1,6; F= 0*45, P > 0*5227, F= 0*70, P > 0*4358), or in a trial where Blepharisma had been previously starved for 12 hours in an attempt to increase their hunger level (df 1,6; F= 0.25, P> 0.6330). These Blepharisma may not have been hungry, or may have switched to feeding on bacteria, since Blepharisma can feed either as a predator on protozoans or as a bacterivore (Giese 1973). Predators did not consume other predators.

Connectances were calculated based on the results of these feeding trials, to improve the description of the food webs. Although unobserved links may sometimes exist, they are probably rare in comparison with the links that are included. Connectance is the number of actual links divided by possible links. A feeding interaction between predator and prey equals two links (one from predator to prey and one from prey to predator). The number of possible links is n(n-1), where n = number of species. Connectances were: all two-species treatments = 1, SB, BE, BU and EU=0*5, SE and SU = 0*666, and SBEU = 0*428.

Discussion

EXTINCTIONS, ABUNDANCE AND

VA R1A BI LITY

Extinctions were associated with increased species richness, but it was unclear whether species richness per se caused the extinctions or whether introducing any other species to persistent combinations changes stability. There was only one extinction in the two- species combinations, most of which were preselected for persistence, but 22 and 28% of populations went extinct in the four- and eight-species webs, respectively. The latter percentages did not differ statistically.

Species richness did not affect the total abundance

of all species summed within a food web. Approx- imately the same number of individuals were dis- tributed among more species in food webs of greater species richness (Table 2). Such 'density compensation' (MacArthur, Diamond & Karr 1972), is expected in communities where species overlap broadly in resource use (MacArthur et al. 1972; Cody 1975; Faeth 1984). Similar total abundances among webs of different richnesses means that the resources in species-rich microcosms were not used more (or less) thoroughly or efficiently than resources in species-poor microcosms.

The increase in extinctions in the food webs with more than two protozoan species may result from the lower population sizes per species associated with additional species. MacArthur & Wilson (1967) predicted that species-rich communities would have lower abundances per species, and that smaller population sizes would increase the probability of extinction for each species because small populations fluctuate nearer to zero (see also Chesson 1978; Dickerson & Robinson 1985; Simberloff 1988). Increased species richness did not change the temporal variability in abundance of most of the species in this experiment (Table 4), but populations in species-rich webs did fluctuate closer to zero.

Species richness and species composition did not affect the temporal variability in abundance of most of the species. However, if population fluctuations do lead to extinctions, the extinctions may have removed the most intrinsically variable populations from the data set, weakening my ability to detect effects of species richness and composition on variability. It might be possible to test whether variable populations are more likely to go extinct, by comparing the pre-extinction variability of populations that go extinct to those that do not. This experiment did not provide enough data for such a test, because most of the populations that went extinct crashed early in the experiment when similar replicates were also decreasing in abundance. In most cases, there were no-obvious differences in variability between doomed populations and those that persisted.


CONN ECTANCE

Connectance decreased slightly with increasing species richness. In theory, decreased connectance should make it easier for species to coexist, while increased species richness should make it more difficult for species to persist (Robinson & Valentine 1979; Pimm 1991). Numerous extinctions in the species-rich food webs could mean that the negative effects of species richness outweighed any positive effect of lower connectance. Species-rich food webs contain more scope for competition and apparent competition, which may cause extinctions, but are not typically included in calculations of connectance (Holt 1977; Paine 1980; Pimm 1991). Extinctions caused by apparent competition and interference have been demonstrated in protist food webs similar to those in this study (Lawler, in press). Unless there is a correlation between the number of direct links in a food web and the number of strong indirect effects and non-trophic interactions, connectance may not be useful in predicting food web stability.

TROPHIC LEVEL AND VARIABILITY

Predator populations were less variable than prey populations. The lower variability of the predator populations may be a result of their ability to reduce metabolic rates and use stored nutrients when prey become scarce. Depending on the species, protozoans can starve for up to 27 days (Salt 1968; Jackson & Berger 1985).

The different generation times of the predators and prey could cause -differences in variability. Predators have longer generations than prey and longer-lived species tend to show lower variation in abundance (McArdle et al. 1990; Pimm 1991). However, other work in a similar system indicated that many protist bacterivores show low fluctuation in the absence of predation (Lawler & Morin, in press; Lawler, in press), so at least some of the difference between predator and prey variability may be driven by predation.

Prey went extinct more often than predators. This would be expected if predator-prey interactions were stronger than competitive interactions among the predators, which seems likely. Predator-prey interactions are immediately lethal to the prey, whereas competition among the predators may have less serious consequences. However, this experimental design cannot rule out competitive exclusion among the prey.

EFFECTS OF SPECIES RICHNESS AND

SPECiES COMPOSiTiON ON POPULATiON

S TA BiLi1T Y

Species richness and species composition were both important in determining the abundance, variability

and persistence of individual species. Each factor affected the abundances of at least half of the species, and changed the variabilities of one or two of the species. Since both of these factors had strong effects on approximately equal numbers of species, they can be regarded as comparable influences on community dynamics.

Most of the species in this experiment were affected by the species composition of the assemblages, but the community-specific nature of this factor will make it difficult to draw generalities about how species composition affects community stability. A closer look at the characteristics of individual species (e.g. reproductive rates, vulnerability to predation, feeding rates) might help ecologists to better understand how species composition will affect community stability.

Tilman (1989) advised ecologists to focus on the broad, general patterns among communities rather than on the particular characteristics of any one system, in order to abstract the complexity of the world into as few variables as possible. In this study, species richness (an easily measurable property of communities) affected the population dynamics of most of the species. Relatively species-rich communities had more extinctions and lower population sizes. However, the effects of each community's particular species composition were equally pro- nounced. No-one expects to completely explain the dynamics of any given population soly by examining general aspects of community structure. However, it would be useful to estimate how much we can hope to explain with general ecological principles, by performing additional studies that compare the effects of both universal and individualistic community properties on population dynamics.

Acknowledgements

I am very grateful to Dr Peter J. Morin for his excellent advice, generous help and enthusiastic support during all phases of this project. He and the other members of my graduate committee, Drs Nelson Hairston, Steven Handel, Henry John-Alder and Terry McGuire, provided valuable comments on the manuscript. I thank John Lawton and two referees for their helpful comments. There was also a third referee. This work was supported by the Arthur McCallum Fund and the Bureau of Biological Research of Rutgers University, and NSF grant BSR 90-06462 to P.J. Morin. Final preparation of the typescript was supported by the NERC Centre for Population Biology.

References

Abrams, P.A. & Allison, T.D. (1982) Complexity, stability, and functional response. American Naturalist, 119, 240-249.

Atkinson, I. (1989) Introduced animals and extinctions. Conservation for the Twenty-first Century (eds D. Western

719 S.P. Lawler

& M.C. Pearl). Oxford University Press, New York, USA.

Chesson, P. (1978) Predator-prey theory and variability. Annual Review of Ecology and Systematics, 9, 323-347.

Cody, M.L. (1975) Towards a theory of continental species diversity. Ecology and Evolution of Communities (eds M.L. Cody & J.M. Diamond), pp. 214-257. Belknap Press, Cambridge, USA.

Connell, J.H. & Sousa, W.P. (1983) On the evidence needed to judge ecological stability or persistence. American Naturalist, 121, 789-824.

Curds, C.R. & Vandyke, J.M. (1966) The feeding habits and growth rates of some fresh-water ciliates found in activated-sludge plants. Journal of Applied Ecology, 3, 127- 137.

Day, R.W. & Quinn, G.P. (1989) Comparisons of treatments after and analysis of variance in ecology. Ecological Monographs, 59, 433-463.

DeAngelis, D.L. (1975) Stability and connectance in food web models. Ecology, 56, 238-243.

Dickerson, J.E. & Robinson, J.V. (1985) Microcosms as islands: a test of the MacArthur-Wilson equilibrium theory. Ecology, 66, 966-980.

Drake, J.A. (1991) Community-assembly mechanics and the structure of an experimental species ensemble. American Naturalist, 137, 1-26.

Elton, C. (1958) The Ecology of Invasions by Animals and Plants. Chapman & Hall, London, England.

Faeth, S.H. (1984) Density compensation in vertebrates and invertebrates: a review and an experiment. Ecological communities: conceptual issues and the evidence (eds D.R. Strong, D. Simberloff, L.G. Abele & A.B. Thistle), pp. 491-509. Princeton University Press, Princeton, USA.

Giese, A.C. (1973) Blepharisma. Stanford University Press, Stanford, CA, USA.

Goodman, D. (1975) The theory of diversity-stability re- lationships in ecology. Quarterly Review of Biology, 50, 237-266.

Hairston, N.G., Allan, J.D., Colwell, R.K., Futuyma D.J., Howell, J., Lubin, M.D., Mathias, J. & Vander- meer, J.H. (1968) The relationship between species diversity and stability: an experimental approach with protozoa and bacteria. Ecology, 49, 1091-1101.

Holt, R.D. (1977) Predation, apparent competition, and the structure of prey communities. Theoretical Popu- lation Biology, 12, 197-229.

Hurd, L.E., Mellinger, M.V., Wolf, L.L. & McNaughton, S.J. (1971) Stability and diversity at three trophic levels in terrestrial successional ecosystems. Science, 173, 1134-1136.

Jackson, K.M. & Berger, J. (1985) Life history attributes of some ciliated protozoa. Transactions of the American Microscopy Society, 104, 52-63.

Kareiva, P. (1989) Renewing the dialogue between theory and experiments in population ecology. Perspectives in Ecological Theory (eds J. Roughgarden, R.M. May & S.A. Levin), pp. 68-88. Princeton University Press, Princeton, USA.

Lawler, S.P. (in press) Direct and indirect effects in microcosm communities of protists. Oecologia.

Lawler, S.P. & Morin P.J. (in press) Food web architecture and population dynamics in laboratory microcosms of protists. American Naturalist.

Lawlor, L.R. (1978) A comment on randomly constructed model ecosystems. American Naturalist, 112, 445-447.

Lawton, J.H. (1989) Food webs. Ecological Concepts: the contribution of ecology to an understanding of the natural world (ed. J.M. Cherrett), pp. 43-78. Blackwell Scientific, Oxford, UK.

Leslie, P.H. & Gower J.C. (1960) The properties of a stochastic model for the predator-prey type of inter-

action between two species. Biometrika, 47, 219-234. Luckinbill, L.S. (1979) Regulation, stability, and diversity

in a model experimental microcosm. Ecology, 60, 1098- 1102.

MacArthur, R. (1955) Fluctuations of animal populations, and a measure of community stability. Ecology, 36, 533-536.

MacArthur, R. & Wilson, E.O. (1967) The Theory of Island Biogeography. Princeton University Press, Princeton, USA.

MacArthur, R., Diamond, J.M. & Karr, J.R. (1972) Density compensation in island faunas. Ecology, 53, 330-342.

May, R.M. (1972) Will a large complex system be stable? Nature, 238, 413-414.

May, R.M. (1974) Stability and Complexity in Model Ecosystems. Princeton University Press, Princeton, USA.

McArdle, B.H., Gaston K.J. & Lawton, J.H. (1990) Variation in the size of animal populations: patterns, problems and artefacts. Journal of Animal Ecology, 59, 439-454.

McNaughton, S.J. (1988) Diversity and stability. Nature, 333, 204-205.

Paine, R.T. (1980) Food webs: linkage, interaction strength and community infrastructure. Journal of Animal Ecology, 49, 667-685.

Pimm, S.L. (1982) Food Webs. Chapman and Hall, London, UK.

Pimm, S.L. (1984) The complexity and stability of ecosystems. Nature, 307, 321-326.

Pimm, S.L. (1991) The Balance of Nature? Ecological Issues in the Conservation of Species and Communities. University of Chicago Press, Chicago.

Robinson, J.V. & Valentine, W.D. (1979) The concepts of elasticity, invulnerability and invadability. Journal of Theoretical Biology, 81, 91-104.

Salt, G. (1961) Feeding activity by Amoeba proteus. Exper- imental Cell Research, 24, 618-622.

Salt, G. (1968) The feeding of Amoeba proteus on Para- mecium aurelia. Journal of Protozoology, 15, 275-280.

Schoener, T.W. (1985) Are lizard population sizes unusually constant through time? American Naturalist, 126, 633-641.

Seshachar, B.R., Saxena, K.N. & Girgla, H. (1971) Some factors governing feeding behaviour of Homalozoon vermiculare (Ciliophora, Holotricha). Journal of Proto- zoology, 18, 90-95.

Simberloff, D. (1988) The contribution of population and community biology to conservation science. Annual Review of Ecology and Systematics, 19, 473-511.

Sokal, R.R. & Rohlf, F.J. (1981) Biometry. W.H. Freeman and Co., New York, USA.

Soule, M. (1987) Viable Populations for Conservation. Cambridge University Press, Cambridge.

Taylor, A.D. (1992) Deterministic stability analysis can predict the dynamics of some stochastic population models. Journal of Animal Ecology, 61, 241-248.

Taylor, W.D. and Berger, J. (1976) Growth responses of cohabiting ciliate protozoa to various prey bacteria. Canadian Journal of Zoology, 54, 1111-1114.

Tilman, D. (1989) Discussion: Population dynamics and species interactions. Perspectives in Ecological Theory (eds J. Roughgarden, R.M. May & S.A. Levin), pp. 89- 100. Princeton University Press, Princeton, USA.

Underwood, A.J. (1989) The analysis of stress in natural populations. Biological Journal of the Linnean Society, 37, 51-78.

Received 28 July 1992; revision received 8 December 1992

species richness, species composition and population ... pubs...key-words: microcosms, protists,...

Documents