UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The propagation of top-down and bottom-up signals in heterogeneous aquatic food webs Bell, Thomas 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2001-0318.pdf [ 3.05MB ]
Metadata
JSON: 831-1.0090066.json
JSON-LD: 831-1.0090066-ld.json
RDF/XML (Pretty): 831-1.0090066-rdf.xml
RDF/JSON: 831-1.0090066-rdf.json
Turtle: 831-1.0090066-turtle.txt
N-Triples: 831-1.0090066-rdf-ntriples.txt
Original Record: 831-1.0090066-source.json
Full Text
831-1.0090066-fulltext.txt
Citation
831-1.0090066.ris

Full Text

T h e propagation of top-down and bottom-up signals in heterogeneous aquatic f o o d webs by T H O M A S B E L L B . S c . (Honours), M c G i l l Universi ty , 1999 A T H E S I S S U B M I T T E D I N P A R T I A L F U L F U L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S Department of Z o o l o g y W e accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A September 2001 . © T h o m a s B e l l , 2001 In p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r reference and study. I f u r t h e r agree that permission f o r extensive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the head of my department or by h i s or her r e p r e s e n t a t i v e s . I t i s understood that copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l g ain s h a l l not be allowed without my w r i t t e n permission. Department of Z o q V o37 The U n i v e r s i t y of B r i t i s h Columbia Vancouver, Canada A B S T R A C T T h e two experiments described in this thesis attempt to identify the mechanisms that determine the biomass of trophic levels in freshwater communities. Ear ly work in this f ield suggested that trophic level biomass is determined b y resource supply and levels of predation. I present the results of two studies that investigate the impact of various other factors o n the biomass of trophic levels. In the first study, I describe the short-term impact of prey (phytoplankton) edibility on the phytoplankton response to nutrient enrichment and the addition o f a top (third) trophic level. T h e results suggest that the proportion o f inedible phytoplankton in the communi ty modifies the degree to which phytoplankton biomass is regulated by bottom-up (nutrient supply) or top-down (predation) forces. Compensation by inedible phytoplankton c o u l d therefore preclude the propagation of top-down signals in these communities. A s a result, experiments over longer time scales might show a progressive weakening of top-down signals. In the second study, I therefore describe the results o f a long-term (4-year) experiment in which zooplanktivorous fish are either present or absent. T h e results o f this experiment, combined with the results of an extensive literature analysis, provide evidence that there is in fact no obvious decline in the strength o f top-down signals with increasing experiment duration. T h e results do suggest, however, that the experimental results depend to some degree on the type of system in which the study is performed. i i T A B L E O F C O N T E N T S A B S T R A C T i i T A B L E O F C O N T E N T S i i i L I S T O F T A B L E S iv L I S T O F F I G U R E S v A C K N O W L E D G E M E N T S v i i CHAPTER I Introduction and summary 1 1.1 Trophic level biomass 2 1.2 T h e consequence of inedible phytoplankton 3 1.3 T h e effect of experiment duration. . . . . 4 CHAPTER II T h e ecological consequences of inedible phytoplankton 6 2.1 Introduction 7 2.2 Methods 10 2.3 Results 16 2.4 Discussion 19 CHAPTER III L o n g live the trophic cascade 37 3.1 Introduction 38 3.2 Methods 41 3.3 Results 47 3.4 Discuss ion 52 CHAPTER IV Conclusions 74 4.1 General conclusions 75 4.2 Further work 76 R E F E R E N C E S 77 A P P E N D I X I 86 iii L I S T O F T A B L E S T A B L E 2.1 Experiment design for Chapter I I 23 T A B L E 2.2 Effect of nutrient enrichment and phytoplankton communi ty structure on total phytoplankton biomass 24 T A B L E 3.1 F i s h dynamics in the two ponds to which fish were added 57 T A B L E 3.2 A m o n g - s t u d y mean effect o f fish on zooplankton and phytoplankton biomass for experiments that lasted for a single summer, for 4-12 months, and for longer than one year 58 T A B L E 3.3 A m o n g - s t u d y mean effect of fish on zooplankton and phytoplankton "in enclosure, mesocosm, pond, and whole-lake experiments 59 iv L I S T O F F I G U R E S F I G U R E 2.1 M e a n percent of the total phytoplankton biomass in the edible fraction for the two types of phytoplankton community 26 F I G U R E 2.2 M e a n phytoplankton biomass in the H i g h - and Low-nutrient enclosures for the two types of phytoplankton community 28 FIGURE 2.3 M e a n phytoplankton biomass averaged over the course o f the experiment in the two types o f phytoplankton community for enclosures with H i g h - and L o w -nutrient concentrations .30 FIGURE 2.4 M e a n zooplankton biomass on a single sampling date near the conclusion of the experiment (20 September 2000) in the two types o f phytoplankton community and under the two nutrient regimes 32 F I G U R E 2.5 M e a n phytoplankton biomass in enclosures to which zooplanktivorous fish have or have not been added for the two types o f phytoplankton community 34 FIGURE 2.6 M e a n effect of fish on phytoplankton biomass in the two types of phytoplankton community 36 F I G U R E 3.1 M e a n zooplankton biomass in control ponds and ponds with fish 61 F I G U R E 3.2 M e a n abundance of the invertebrate zooplanktivore Chaoborus in control ponds and ponds with fish 63 F I G U R E 3.3 M e a n phytoplankton bio volume in control ponds and ponds with fish 65 FIGURE 3.4 M e a n biomass of the inedible phytoplankton in control ponds and ponds with fish 67 FIGURE 3.5 T h e effect of fish on phytoplankton and zooplankton biomass for published experiments that lasted for up to 1 summer, 1 summer to 1 year, and longer than 1 year 69 FIGURE 3.6 Effect of fish on the biomass of phytoplankton and zooplankton over the course of the experiment 71 F I G U R E 3.7 Phytoplankton effect size as a function of zooplankton effect size f r o m published experiments 73 vi A C K N O W L E D G E M E N T S I a m particularly grateful to m y supervisor, D o l p h Schluter, for his invaluable advice and enthusiasm, and most especially, for giving me free reign to do as I wished. John Richardson gave freely o f his time and provided many useful comments on previous versions of this thesis. Diane Srivastava also gave many helpful comments, and as importantly, advised me on what to do next. I inherited m y enclosures f r o m B e a Beisner, who also instructed me on their use. B i l l N e i l l was thoughtful enough to organise the collection of m u c h of the data in the third chapter 7 years prior to m y c o m i n g to U . B . C . I a m thankful to Danusia D o l e c k i , who collected, preserved, and counted many o f the organisms that f o r m the basis o f the third chapter. Chapter III would not have been possible without the help of innumerable field assistants. T h e S O W D discussion group kept me honest, and helped to organise m y thoughts. T h e members o f the Schluter lab provided a stimulating atmosphere that I wi l l sorely miss. Steve V a m o s i always leant an ear, often both, and showed me the ropes. Jordana was patient with me. I was supported by a P G S - A scholarship f r o m the Natural Science and Engineer ing Research C o u n c i l ( N S E R C ) of Canada throughout m y degree. T h e cost of conducting the experiments was covered by N S E R C grants to D . Schluter and W . E . N e i l l . vii C H A P T E R I I n t r o d u c t i o n a n d s u m m a r y 1 1.1 TROPHIC L E V E L BIOMASS A m o n g the principal goals of ecology is to understand the mechanisms that determine the biomass of trophic levels. If the number of energy transfer steps determines the trophic position of an organism, then trophic levels are groups o f organisms that all have the same or similar trophic positions. A l t h o u g h omnivores obscure definitions o f trophic levels, some discrete trophic levels exist, including obligate primary producers and decomposers. T h e experiments presented in this thesis avoid some of the problems in defining trophic levels by concentrating on the primary producer trophic level. It is perhaps sufficient to note that simple food chain models perform surprisingly well , at least in aquatic systems, in predicting the response of f o o d webs to various perturbations. M u c h of the research on trophic levels has centred on whether trophic level biomass is principally determined by nutrient supply ("bottom-up") (Lindeman 1942), levels o f predation ("top-down") (Hairston et al. 1960), or an interaction between the two (Oksanen et al. 1981). Theoretical and experimental work has shown that manipulating either the concentration of nutrients or the biomass of the topmost trophic level can alter the trophic level biomass of any trophic level. A l t h o u g h the top-down (Hairston et al. 1960, Fretwell 1977, Carpenter et al. 1985) and bottom-up (Lindeman 1942, White 1993) views were originally presented in opposition to each other, it is n o w widely recognized that such a dichotomy does not exist. If all else is equal, nutrients determine the potential biomass, and predation determines the actual biomass of any particular trophic level (Oksanen et al. 1981, Hunter and Price 1992). Trophic level biomass is therefore determined not only by resources and predation but also by the interaction between the 2 two. Recent reviews of the subject, however, have shown m ixed results (Brett and G o l d m a n 1997). Al though it is evident that both resources and predation levels are important, it is also clear that there must be a number of other factors that modify the effect o f these to determine the eventual trophic level biomass. S u c h factors include the structure of the physical environment (spatial heterogeneity), temporal heterogeneity in resource supply and predation levels, behavioural responses to alterations o f the biotic and physical environment, the architecture of the f o o d web in w h i c h interactions between organisms occur, as well as historical processes, chance, and several others (Polis et al. 2000). M u c h of this thesis deals with one of these factors, the architecture o f the f o o d web, and more particularly, the role of unpalatable prey in determining the biomass of primary producers in freshwater systems. 1.2 T H E CONSEQUENCE OF INEDIBLE P H Y T O P L A N K T O N Recent theoretical and laboratory studies have suggested that inedible or unpalatable prey m o d i f y the degree to which trophic levels are regulated by bottom-up or top-down forces such that primarily edible communities are regulated by predation and primarily inedible communities are regulated by resource supply. Despite the hypothesized importance of prey edibility, experiments have only rarely examined the response o f primary producers to nutrient enrichment or the addition of a top trophic level fo l lowing the reduction or elimination of inedible species. I performed two experiments in aquatic enclosures in w h i c h the prey (phytoplankton) had been manipulated to create treatments composed either nearly exclusively of only edible phytoplankton or with both 3 edible and inedible phytoplankton. In the first, the two types o f phytoplankton community were subjected to either high or low nutrient concentrations. In the second, the two types of phytoplankton community were present in enclosures with either 2 or 3 trophic levels. I found that the impact of both the nutrients and the fish on total phytoplankton biomass was modif ied by the edibility o f the phytoplankton community. A l t h o u g h enclosures with only edible phytoplankton were able to increase with enrichment, there was a greater overall phytoplankton biomass in enclosures with both edible and inedible phytoplankton. T h e addition of a third trophic level had a positive effect o n phytoplankton biomass when only edible phytoplankton were present, but had no effect on phytoplankton biomass when both edible and inedible phytoplankton were present. These results therefore provide support for the hypothesis that the proportion of inedible phytoplankton determines the degree to which communities are regulated by top-d o w n or bottom-up forces. 1.3 T H E EFFECT OF E X P E R I M E N T D U R A T I O N T h e results of the previous experiment demonstrate that inedible phytoplankton have a substantial impact on the propagation of top-down signals in aquatic systems. A number of long-term, usually unreplicated, experiments have further demonstrated that the removal o f the third trophic level (zooplanktivorous fish) results in a net negative effect on phytoplankton biomass, but that the magnitude o f the effect is dampened by a presumably concomitant increase in inedible phytoplankton. If inedible phytoplankton are able to completely compensate for losses in the edible phytoplankton, it is possible 4 that the "strength of the trophic cascade" (i.e. the degree to which the addition of a top trophic level alters the biomass of trophic levels below) w i l l decline with increasing experiment duration. Despite attracting considerable interest, there have been few replicated studies o f trophic cascades for longer than a summer f ield season, and none for the time required to estimate the long-term result o f press perturbations. I present the results of a 4-year study of trophic cascades in experimental ponds in w h i c h the top predator, a zooplanktivorous fish, was either present or absent. I tested two predictions of the trophic cascade hypothesis: (1) that the addition o f the top predator (fish) results in a sustained increase in the primary producer (phytoplankton) biomass (2) that there is a relationship between the effect of fish on zooplankton and the effect of fish on phytoplankton. I tested the same two predictions on the results o f 91 published trophic cascade experiments. I found that trophic cascades are important in determining trophic level biomass, but that there was no relationship between the effect of fish on zooplankton and the effect o f fish on phytoplankton, both in m y long-term experiment and m y between-studies comparison. I present evidence that there are quantitative differences in the strength o f trophic cascades among enclosure, mesocosm, pond, and whole-lake studies, but that the strength of trophic cascades does not diminish with increasing experiment duration. 5 CHAPTER II T h e ecological consequences of inedible phytoplankton 6 2.1 INTRODUCTION A number of studies have recently suggested that the impact of nutrients or predators on trophic level biomass is determined, at least in part, by the architecture of the f o o d web (Hunter and Price 1992, Strong 1992, Persson et al. 2001). F o o d web architecture generally refers to the pattern and strength of feeding relations between organisms in a community . It includes such features as the edibility of prey species, the average connectance between species, and the prevalence of omnivory , shared predators, and intraguild predation (Holt and L a w t o n 1994, M o r i n and L a w l e r 1995, Hol t and Polis 1997, D i e h l and Fe ibel 2000). There has been particular recent interest in the role of prey edibility in determining the biomass of trophic levels. T h e "edibil i ty hypothesis" ( L e i b o l d 1989) focuses on how the presence o f less-edible prey alter the predictions o f f o o d web theory. T h e model of L e i b o l d (1989) and subsequent extensions explore the consequences of inedible species on the distribution o f biomass (Grover 1995, Bohannan and L e n s k i 1999) and the stability (Abrams and Walters 1996, G e n k a i - K a t o and Y a m a m u r a 1999, H u x e l 1999) o f trophic levels and o f predator-prey systems. If a tradeoff between competitive ability and edibility is assumed, as appears to be the case for freshwater algae (Agrawal 1998) and terrestrial plants (Agren and Schemske 1993, M u t i k a i n e n and Walls 1995), then the models suggest that both the standing crop biomass and the stability or variability o f trophic levels are altered by the presence of inedible species. In general, the models make the fol lowing predictions. First, that the inedible species (but not the edible prey) w i l l increase with increasing resource levels in systems with 2 trophic levels. 7 Second, that edible species (but not inedible species) w i l l increase with the addition of a third trophic level. E d i b l e and inedible functional groups therefore play a critical role in communities because the balance between these two functional groups determines the degree to which the prey are regulated by bottom-up or top-down forces. Comparative studies provide some support for these predictions. S u c h studies suggest that the communities composed of homogeneously edible prey (phytoplankton) are regulated by resource levels, and that prey communities heterogeneous in their edibility to predators are regulated by predation ( M c C a u l e y et al. 1988). A s predicted, lakes with greater nutrient concentrations also tend to have a greater proportion o f inedible phytoplankton (Watson et al. 1988, 1992, M a s s o n et al. 2000). Despite the obvious appeal o f the theory in resolving the debate between proponents o f the top-down and the bottom-up views, experimental tests of these hypotheses are rare. However , some examples exist f rom terrestrial (Schmitz 1994), aquatic ( L e i b o l d 1989, Hansson et al. 1998, Persson et al. 2001), and microbial systems (Bohannan and L e n s k i 1999). These studies manipulated resource levels and the presence of the top (third) trophic level and measured the response of the edible and inedible species. In general, these studies have provided only m ixed support for the qualitative predictions of the models (above). A l t h o u g h enrichment and the addition o f a third trophic level often results in increases in inedible and edible prey respectively ( L e i b o l d 1989, Hansson et al. 1998, Bohannan and L e n s k i 1999, Schmitz et al. 2000), a number o f experiments report contrary patterns ( L e i b o l d 1989). These correlational studies effectively document the consequences of nutrient and f o o d web manipulations on the proportion of edible to inedible biomass. However , 8 notably absent are studies that investigate how the response to these manipulations is altered by also controlling prey edibility. Apart f r o m studies using highly s implif ied microbial f o o d webs (Bohannan and L e n s k i 1999, 2000), the experiments that have been performed to date therefore did not explicitly examine the role o f prey edibility because it was not manipulated in the experiments. T h e present study is to m y knowledge the first that investigates the consequences o f inedibility by manipulating the edibility of phytoplankton in natural food webs of phytoplankton, zooplankton, and zooplanktivorous fish in aquatic enclosures. In particular, I examine the response of total phytoplankton biomass to nutrient and secondary consumer (zooplanktivorous fish) additions to f o o d webs in w h i c h the proportion o f inedible phytoplankton has been manipulated. T h e theory predicts (above) that top-down forces should regulate communities with a homogeneous prey trophic level (all prey edible), and bottom-up forces should regulate communities with a heterogeneous prey trophic level (edible and inedible species). T h i s study therefore tests the fo l lowing hypotheses: (1) There is a positive response to enrichment in heterogeneous but not homogeneous prey communities in systems with 2 trophic levels. (2) There is a positive response to the addition o f a third trophic level in homogeneous but not heterogeneous communities. 9 2.2 M E T H O D S Study site and enclosures T h e study was conducted in enclosures in an experimental p o n d at the University of Brit ish C o l u m b i a , Canada. T h e p o n d is 23 x 23 m and slopes to a m a x i m u m depth of 3.5 m. T h i s p o n d has never contained fish since the ponds were built in 1991. Phytoplankton biomass in previous years was dominated by highly edible small flagellates, especially Chlamydomonas and Cryptomonas, although blooms of Tetraedron and small (<50 um) dinoflagellates occasionally occurred. T h e larger phytoplankton, thought to be inedible to most zooplankton, are dominated by Ceratium and by filamentous blue-green (Anabaena) and green algae. Zooplankton biomass was dominated by Daphnia pulex and.calanoid and c y c l o p o i d copepods. Smaller cladocerans, such as Chydorus, Diaphanosoma, Bosmina longirostrus, and rotifers were also c o m m o n , but most often contributed little to the total biomass. T h e study was conducted in enclosures in the limnetic zone o f an experimental p o n d f r o m 13 July to 5 October 2000. T h e enclosures were constructed f rom U V -protected plastic bags suspended f r o m a floating w o o d e n frame. T h e bags were 1 m z by 2.5 m deep, and contained approximately 1000 L o f water. T h e y were closed to the sediment because the conditions were intended to mimic those o f the pelagic zone of a small lake. Experimental design I created two types o f phytoplankton community . O n l y edible phytoplankton were present in the first (Homogeneous prey), and both edible and inedible phytoplankton were 10 present in the second (Heterogeneous prey). T o estimate the response of the phytoplankton community to nutrient enrichment, I added nutrients to half of the replicates {High nutrients) or left nutrients at ambient levels {Low nutrients). In a second experiment, zooplanktivorous fish were added to half o f the High nutrient replicates (5 trophic levels) while the rest remained without fish (2 trophic levels). A l t h o u g h the study is treated as two separate experiments, the results of the High nutrient+2 trophic level treatment is used in both experiments (Table 2.1). F ish were also added to half of the Low nutrient replicates, but these were not used in the analysis o f the results because of unexpected fish mortality. There were 2 replicates for each treatment combination. Manipulation of the phytoplankton community Previous studies have shown that phytoplankton larger than approximately 30 jum are for the most part inedible, or at the least highly unpalatable, to c o m m o n zooplankton grazers (Burns 1968, Vanderploeg 1981, L e h m a n and Sandgren 1985, M c C a u l e y and D o w n i n g 1985). I therefore attempted to eliminate the inedible phytoplankton f r o m half of the enclosures by filtering the water through fine netting as it was pumped into the enclosures. It is impossible in practice to eliminate all large (inedible) phytoplankton f r o m the enclosures. T h e purpose of the manipulation was rather to create treatments with ambient and very low concentrations of inedible phytoplankton. Water was added f r o m the p o n d to the enclosures f r o m 3 July to 12 July 2000. A l l water was first filtered through 202 |im Nitex netting to remove the macrozooplankton. In half o f the enclosures {Homogeneous prey), the water was also filtered through 20 \xm Nitex netting as it was added to the enclosures to remove the large inedible phytoplankton. In addition, approximately 2000 L o f water in each enclosure containing 11 Homogeneous prey was passed through 20 u m netting subsequent to the initial filtering. I therefore estimate that the water in each enclosure was filtered 3 times before initiating the experiments. T h e macrozooplankton trapped on the 202 u m netting were subsequently introduced into the enclosures. A l t h o u g h the filtering procedure eliminates the smaller zooplankton (20-202 um) f rom the Homogeneous prey enclosures, previous work in the same ponds (see Chapter III) suggests that small zooplankton have little effect o n phytoplankton dynamics when large cladocerans are present in the system, as was the case in this study. Size-fractionated phytoplankton samples were taken once every 10 days for the first 31 days of the experiment to check the efficacy o f the phytoplankton manipulation. Water was withdrawn f r o m the center of each enclosure and deposited in a jar using a hollow glass tube (1.5 m long, 18 m m diameter) fitted with a removable stopper. T w o samples of 120 m l each were withdrawn f r o m the jar. T o estimate total phytoplankton biomass, one of the 120-ml samples was passed through a 25 m m diameter Whatman G F / F glass-fiber filter in situ. T o estimate the contribution of the inedible phytoplankton (>30 |ira) : the same procedure was fol lowed but the 120-ml sample was first passed through 30 u m Nitex netting. B o t h samples were incubated in 95% acetone overnight (>18 h) at 4 C . Phytoplankton C h i a was estimated using the fluorometric technique ( L i n d 1979). T h e contribution of the edible phytoplankton (<30 um) was obtained by subtracting the >30 p m C h i a f rom the total phytoplankton C h i a. A n a l y s i s of these samples indicated that the filtering procedure described above was successful in creating treatments with a higher biomass o f edible phytoplankton, and that the manipulation was sustained throughout the experiment. S m a l l edible 12 phytoplankton contributed on average 93.1% of the total phytoplankton biomass in Homogeneous prey enclosures, and only 51.7% in Heterogeneous prey enclosures (Figure 2.1). Repeated measures analysis o f variance ( A N O V A ) confirmed that this difference was statistically significant (FJJO = 84.2, P < 0.0005). There was furthermore no difference in the total phytoplankton biomass between Homogeneous and Heterogeneous prey enclosures immediately f o l l o w i n g the manipulation of the phytoplankton community ( F j ^ = 0.51, P = 0.69). Experiment 1: The effect of enrichment T h e experiment was initiated on 24 July 2000 when 0.175 u g L ' 1 K H 2 P 0 4 and 3.883 ug L 1 N a N 0 3 were added to the High nutrient enclosures. T h e nutrients were first dissolved in 5 L o f water f rom the enclosure into which the nutrients were being added. T h e water in the enclosures was mixed briefly immediately f o l l o w i n g the nutrient addition to ensure an initial even distribution of the nutrients within the enclosures. N o nutrients were added to the Low nutrient enclosures, but these enclosures were also briefly mixed. Phytoplankton were usually sampled twice weekly beginning 20 July 2000. T o t a l phytoplankton biomass was measured as described in the previous section. Zooplankton were sampled near the beginning (7 August 2000) and conclusion (20 September 2000) of the experiment. I used the same glass tube to collect the zooplankton as was used to collect the phytoplankton. Fifteen liters o f water were obtained f r o m each enclosure and was sieved through 100 urn Nitex netting. T h e zooplankton trapped o n the netting were then placed in scintillation vials in 95% ethanol. Zooplankton were enumerated and measured under a dissecting microscope after the 13 conclusion of the experiment. Zooplankton biovolume was estimated f rom the m a x i m u m length and width of each individual . Because the 100 u m netting is unreliable for sampling the smallest zooplankton (e.g. rotifers and c o p e p o d nauplii) , total zooplankton biomass was estimated as the biomass of the macrozooplankton (post-naupliar cladocerans and copepods). A l t h o u g h additional zooplankton samples were also obtained over the course of the experiment f r o m smaller samples o f water (2 L ) , insufficient zooplankton were obtained per sample to accurately estimate zooplankton density. These samples were therefore not included in the analysis. Experiment 2: The effect of zooplanktivorous fish additions I used the Limnet ic species o f the threespine stickleback (Gasterosteus sp.), a zooplanktivorous fish (Schluter 1993), as the top predator trophic level. Parental fish were captured f r o m Paxton Lake , British C o l u m b i a , Canada ( 4 9 ° 4 3 ' N , 1 2 4 ° 3 1 ' W ) during M a y 2000. E g g s f r o m a single female were fertilized with the sperm f r o m a single male in the laboratory. T h e progeny were raised in the laboratory for 6 weeks before being added to the appropriate enclosures at the initiation o f the experiment (4 September 2000). F i v e individuals (mean weight = 88.9 m g , mean standard length = 21.1 mm) were added to each o f the appropriate enclosures. T h e fish added to each enclosure were drawn haphazardly f r o m the available stock. There was no difference in the mean weight of the fish added to each enclosure ( A N O V A : F3,i6 = 0.83, P = 0.50). A l t h o u g h the addition of predators was 46 days after the initial phytoplankton manipulation, there was little change in the proportion of edible phytoplankton in the Homogeneous prey enclosures (Figure 2.1), indicating that the manipulation o f the phytoplankton community persisted throughout the experiment. 14 T h e fish were transported to the enclosures and al lowed to acclimatize to pond temperatures overnight in plastic bags half f i l led with aquarium water before being added to the enclosures. T h e y were removed f r o m the enclosures at the conclusion of the experiment first using minnow traps overnight for 3 consecutive nights, and then by adding rotenone (C23H22O6). Phytoplankton and zooplankton sampling methods were the same as for Experiment 1. Phytoplankton were sampled every 10 days beginning 9 days prior to the addition o f the fish (26 August 2000). Analyses F o r Experiment 1, I tested for differences in the phytoplankton biomass among the treatments using repeated measures analysis o f variance ( A N O V A R ) , where the phytoplankton community (Homogeneous and Heterogeneous prey) and nutrient concentration (High and Low) were the dependent variables. T o investigate more general patterns in the data, I calculated the grand mean across the time series for each treatment, and compared treatments using a factorial A N O V A . F o r Experiment 2, I wanted to test the hypothesis that zooplanktivorous fish had a greater positive effect on phytoplankton biomass in the Homogeneous prey treatment then in the Heterogeneous prey treatment. Phytoplankton biomass was therefore converted to effect sizes by dividing the phytoplankton biomass in each replicate enclosure with 3 trophic levels by the phytoplankton biomass in the appropriate controls (2 trophic levels). T h e data were then standardized for differences between enclosures prior to the start of the experiment by subtracting the difference between the treatment (3 trophic levels) and the appropriate control (2 trophic levels) for each data points in the time series. T h e resulting time series 15 represents the effect of fish on phytoplankton biomass. T h e time series were compared using repeated measures A N O V A . A l l statistical analyses were performed on log-transformed data to homogenize the variances. T h e statistics were computed using S Y S T A T 5.05. 2.3 RESULTS Experiment 1: The effect of enrichment In this experiment, I examined the effect o f enrichment o n the two types of phytoplankton communities. I predicted (see Introduction) that phytoplankton biomass w o u l d increase in the Heterogeneous prey enclosures but not in the Homogeneous prey enclosures. T h e prediction was confirmed to some degree because there was a marginally non-significant increase in the phytoplankton biomass with enrichment in the Homogeneous prey treatment ( A N O V A R : Fj,2 = 13.6, P = 0.065), and a significant increase in the Heterogeneous prey treatment ( A N O V A R : F ; ,2 =47.6, P = 0.020) (Figure 2.2). F o r the latter, there was also a significant T i m e x Nutrient interaction ( A N O V A R : F 18,36= 2.3, P = 0.018). However , when the status o f both the phytoplankton community {Homogeneous and Heterogeneous prey) and the nutrient concentration (High and Low) were included in a repeated measures analysis o f variance, the nutrient concentration but not the edibility o f the phytoplankton community had a significant effect o n the total phytoplankton biomass (Table 2.2). T h e significant T i m e x Nutrient x Edibi l i ty interaction suggests that the edibility o f the phytoplankton community does play some role in the response of the total phytoplankton biomass to enrichment. 16 I averaged the phytoplankton biomass across the time series o f each enclosure to look at coarser trends in the data. T h e data show (Figure 2.3) that as predicted (above), the biomass o f the Heterogeneous prey phytoplankton communities was able to increase substantially with enrichment to 3.0-times the biomass in enclosures that did not receive any nutrients. Contrary to the prediction, however, the Homogeneous prey phytoplankton communities were also able to increase with enrichment to 3.6-times the biomass without nutrient inputs. W h e n both nutrients and phytoplankton edibility are included in a fully factorial A N O V A , both nutrients ( F ; , 5 =28.71, P = 0.006) and phytoplankton edibility (Fy,5 = 8.35, P = 0.045) had a significant effect on total phytoplankton biomass, but there was no interaction between the two (Fi,s = 1.07, P = 0.40). H o w e v e r , there was no difference in the phytoplankton biomass between High nutrients+Homogeneous prey and Low nutrients+Heterogeneous prey, suggesting that the growth o f the phytoplankton in the Homogeneous prey might be constrained by higher zooplankton grazing rates. However , the absence of a difference between the two High nutrient treatments (Homogeneous and Heterogeneous prey) indicates that zooplankton could not have greatly affected the phytoplankton biomass in Homogeneous prey enclosures. If nutrient enrichment results in a higher density o f edible phytoplankton, then zooplankton density should also increase with enrichment. There was no difference in the total zooplankton biomass among the treatments near the beginning (7 August 2000) of the experiment ( A N O V A : F?,4 = 1-22, P = 0.41). A s predicted, manipulation of the phytoplankton community and nutrient concentrations resulted in a difference in the zooplankton biomass among the treatments near the conclusion of the study ( A N O V A : F?,4 = 13.73, P = 0.014) because o f a significantly higher zooplankton biomass in the 17 enriched Homogeneous prey enclosures (Figure 2.4). There is therefore a significant interaction between nutrient levels and phytoplankton community structure when the data are included in a factorial A N O V A (FlA = 12.0, P = 0.026). Experiment 2: The effect of zooplanktivorous fish additions A s expected, there was a lower zooplankton biomass in enclosures with 3 trophic levels (t6 = 2.87, P = 0.028), presumably because of grazing by zooplanktivorous fish. I predicted (see Introduction) that the addition o f zooplanktivorous fish w o u l d have a positive effect on phytoplankton biomass in both Homogeneous and Heterogeneous prey treatments, but that this effect would be more pronounced in the Homogeneous prey enclosures. There was no significant effect of fish on phytoplankton biomass in the Homogeneous prey enclosures ( A N O V A R : F/,2 = 3.57, P = 0.20) but there was a significant T i m e effect (FJ,<J = 28.3, P = 0.001) and T i m e x Treatment interaction ( F j g = 8.9, P,= 0.013) (Figure 2.5). T h e biomass of phytoplankton was higher (marginally non-significant) in enclosures with fish for the Heterogeneous prey treatment ( A N O V A R : F/,2 = 15.02, P = 0.061), although this appeared to be due to initial differences between treatment (3 trophic levels) and control (2 trophic levels) enclosures (Figure 2.5). After calculating phytoplankton effect sizes and standardizing for initial differences between treatments and controls (Figure 2.6), there was no significant difference between phytoplankton effect sizes in the Homogeneous and Heterogeneous prey treatments ( A N O V A R : F ; ) 2 = 0.52, P = 0.54). However , the significant T i m e x Treatment interaction (FJ,<J = 50.0, P < 0.0005) suggests that the qualitative divergence between the Homogeneous and Heterogeneous prey effect sizes as the experiment 18 progressed is real. Pairwise comparisons at each sampling date indicated that there was a significantly higher effect o f zooplanktivorous fish on phytoplankton biomass in the Homogeneous prey enclosures on the final sampling date (Tukey test: qw,s = 2.25, P = 0.032), but there was no significant difference on any o f the other sampling dates. I therefore conclude that the fish had a positive effect on phytoplankton biomass, but perhaps only in the Homogeneous prey enclosures. Further, the hypothesis of a greater effect o f fish in Homogeneous prey enclosures is upheld, at least by the conclusion of the experiment. 2.4 D ISCUSSION T h e experiments in this study c o n f i r m the suggestions of previous authors ( L e i b o l d 1989, A g r a w a l 1998) that prey (phytoplankton) edibility, here represented as phytoplankton size, can have considerable effects on the dynamics o f the phytoplankton trophic level. T h e results of Experiment 1 demonstrate that the phytoplankton respond differently to nutrient enrichment in the two types o f phytoplankton community (Homogeneous and Heterogeneous prey). T h e o r y predicts that prey biomass should increase with enrichment in 2-trophic level systems only if some of the prey are inedible (Le ibold 1989, Kretzschmar et al. 1993, Bohannan and L e n s k i 2000) or unpalatable (Graver 1995). If there are only edible prey, any increases in primary production w o u l d simply result in increases in primary consumer biomass as they offset increases in their prey. If inedible phytoplankton are present, not only are edible algae susceptible to grazing losses, but the inedible algae are at a competitive advantage because they act as 19 nutrient "sponges" by sequestering the available nutrients (Watson et al. 1988, Bohannan and L e n s k i 1999), which might therefore lead to further reductions in the biomass o f the edible phytoplankton. A s predicted, there was an increase in the phytoplankton biomass with enrichment in the Heterogeneous prey enclosures. Contrary to the predictions, there was also an increase in the phytoplankton with enrichment in the Heterogeneous prey enclosures, and as a result there was no interaction between nutrients and prey c o m m u n i t y categories. Despite the absence of an interaction, there was a differential effect o f enrichment on the two types o f phytoplankton community that was manifest in an overall higher total phytoplankton biomass in enclosures with heterogeneous prey. T h i s might occur if the two size fractions not only differ in their edibili ty to zooplankton grazers but also in their resource requirements. T h e models that have investigated the effects o f prey edibility assume that phytoplankton differing in their edibili ty have identical resource requirements. Such an assumption may be unrealistic, especially i f phytoplankton with anti-predator traits, such as thicker cell walls, require different ratios o f the c o m m o n micronutrients, as is the case for many inedible blue-green algae ( M a c K a y and Elser 1998). T h e mechanism by which phytoplankton were able to increase in the Homogeneous prey enclosures despite concomitant increases in the zooplankton also remains unclear. It is possible that the zooplankton were not given sufficient time to respond to increases in edible phytoplankton biomass, but experiments show that zooplankton are able to approach equil ibr ium densities over comparable time scales (Walters et al. 1987, Attayde and Hansson 2001). Alternatively, increased interference 20 among zooplankton with increased zooplankton densities ( M c C a n n et al. 1998), including interference competition and intraguild predation, could also lead to the observed increase in edible phytoplankton with enrichment. F inal ly , it is also possible that small but inedible algae became abundant in the Homogeneous prey enclosures. Unfortunately, the methods used during the experiment were insufficient to account for this possibility. Further work w o u l d be required to distinguish among these hypotheses. There is now considerable evidence that top-down forces are important in regulating trophic level biomass in aquatic systems (Brett and G o l d m a n 1996). There is a great deal of variability, however, in the degree to which phytoplankton biomass is affected by the trophic levels above. Several authors have suggested that the strength of trophic cascades are contingent on the degree o f heterogeneity, or functional complexity, in the lower trophic levels (Hunter and Price 1992) such that trophic cascades are c o m m o n in " s i m p l e " aquatic f o o d webs, but rare in more complex terrestrial systems (Polis 1994). It is evident that the addition of a third trophic level should have little effect on lower trophic levels i f the primary producer biomass is dominated by organisms that are inedible to the primary consumers. Primary producer communities that are homogeneous in their edibility to primary consumers should therefore be more tightly regulated by top-down forces than heterogeneous communities. F o r Experiment 2, I therefore predicted (see Introduction) that there w o u l d be a greater response to the addition o f a third trophic level in the Homogeneous prey treatment. T h e results f r o m this study uphold this prediction. I demonstrated that, by the conclusion of the study, the magnitude of the effect of the secondary carnivore (zooplanktivorous fish) on primary producer biomass depends on the edibility of the 21 primary producer trophic level. Results f r o m previous enclosure experiments have been restricted to the observation that inedible phytoplankton c o m m o n l y , but not always, increase with increases in zooplankton biomass ( L e i b o l d 1989) and therefore dampen the trophic cascade. Similar ly , several hypotheses based o n comparative data have suggested that the degree to which top-down forces regulate trophic levels depends on nutrient concentrations (Coley et al. 1985, M c Q u e e n et al. 1986, Elser and G o l d m a n 1991). A l t h o u g h there is some debate as to whether trophic cascades are important in oligotrophic (nutrient-poor) lakes, there is general agreement that trophic cascades are weak in nutrient-rich aquatic systems because they are dominated by inedible phytoplankton (Watson et al. 1988, 1992). Interestingly, terrestrial plants appear to exhibit the opposite pattern, with decreased plant defense in nutrient-rich environments (Coley et al. 1985). Comparisons between terrestrial and aquatic systems are complicated, however, for example by differences in plant and herbivore life-history strategies between terrestrial plants and phytoplankton (e.g. perenniality, size differences between primary producers and herbivores) as well as by differences in the ratio and absolute concentration of available nutrients, which might influence the relative cost of defense. Unfortunately, comparative studies cannot separate the effects o f nutrient concentration f r o m phytoplankton edibility and other covariates of lake nutrient concentrations (such as lake size) o n the strength o f the trophic cascade. T o m y knowledge, Experiment 2 is the first manipulative experiment to demonstrate that phytoplankton edibility alone can have a strong impact on the strength of trophic cascades. 22 Table 2.1 S u m m a r y o f the experimental design for the 2 experiments in this study. E a c h cell is a replicate (2 replicates for each treatment combination). L o w nutrients H i g h nutrients 2 trophic levels 2 trophic levels 3 trophic levels Homogeneous prey Heterogeneous prey 1 7 2 8 10 11 12 Experiment 1 Experiment 2 23 Table 2.2 Results of a repeated measures A N O V A to determine the effect of nutrient enrichment (High and Low nutrients) and of the phytoplankton community (Homogeneous and Heterogeneous prey) o n total phytoplankton biomass. Source of Variat ion F N u m . df D e n . d f P Between subjects Nutrients 14.55 1 4 0.019 E d i b i l i t y 1.01 1 4 0.372 Nutrient x edibility 0.49 1 4 0.523 W i t h i n subjects T i m e 5.16 18 72 <0.0005 T i m e x nutrients 2.22 18 72 0.009 T i m e x edibility 0.76 18 72 0.740 T i m e x nutrient x edibility 1.94 18 72 0.026 24 Figure 2J: M e a n percent of the total phytoplankton biomass ( C h i a) in the edible (<30 urn) fraction for the Homogeneous and Heterogeneous prey enclosures. Error bars are the standard error of 4 enclosures for each sampling date. 25 Homogeneous prey Heterogeneous prey T i m e (d) Figure 2.1 26 Figure 2.2: M e a n phytoplankton biomass ( ± S E ) in the High and Low nutrient enclosures for Homogeneous and Heterogeneous prey phytoplankton communities . 27 28 Figure 2.3: G r a n d mean of the phytoplankton biomass ( ± S E ) averaged over the course of the experiment in the two types of phytoplankton communi ty (Homogeneous and Heterogeneous prey) and under the two nutrient regimes (High and Low nutrients). T h e lines connecting the columns represent pairwise comparisons between treatments using the T u k e y test on log-transformed data (* P < 0.05). Non-significant comparisons are not indicated. 29 70 oo oo ca 6 o 1 « o U C 60 50 40 30 20 10 0 • L o w nutrient • H i g h nutrients Homogeneous prey Heterogeneous prey Figure 2.3 30 Figure 2.4: M e a n ( ± S E ) zooplankton biomass o n a single sampling date near the conclusion of the experiment (20 September 2000) in the two types of phytoplankton community (Homogeneous and Heterogeneous prey) and under the two nutrient regimes (High and Low nutrients). T h e lines connecting the columns represent pairwise comparisons between treatments using the T u k e y test on log-transformed data (* P < 0.05). Non-significant comparisons are not indicated. 31 800 700 600 H 500 H 400 A 300 200 100 "H • Low nutrients • High nutrients Homogeneous prey Heterogeneous prey Figure 2.4 32 Figure 2.5: M e a n ( ± S E ) phytoplankton biomass in enclosures to which zooplanktivorous fish have (3 trophic levels) or have not (2 trophic levels) been added for the two types of phytoplankton community (Homogeneous and Heterogeneous prey). Negative values on the x-axis are data f r o m before the start of the experiment. 33 Figure 2.5 34 Figure 2.6: M e a n ( ± S E ) effect of fish on phytoplankton biomass in Homogeneous and Heterogeneous prey enclosures. Effect sizes are calculated by d i v i d i n g the phytoplankton biomass in the 3 trophic level treatments with the average value o f the control (2 trophic levels) for the same date. Negative values on the y-axis are data f r o m before the start of the experiment. Pairwise comparisons were performed at each sampling date using the T u k e y test on log-transformed data (* P < 0.05). 35 Figure 2.6 36 CHAPTER III Long live the trophic cascade 37 3.1 INTRODUCTION A l t h o u g h the effect of predators on their prey has been wel l documented, the importance of the indirect effect of the top predator in determining the biomass of lower trophic levels via their direct effects on their prey remains controversial (Carpenter and K i t c h e l l 1992, D e M e l o et al. 1992, Schmitz et al. 2000, Halaj and W i s e 2001). T h i s indirect effect of the top trophic level on lower trophic levels is called the trophic cascade (Carpenter et al. 1985). There has been considerable interest in trophic cascades over the last two decades, particularly in freshwater systems (Persson 1999). T h e interest has arisen both f rom the apparent generality o f the response o f freshwater communities to adding or deleting the top predator (Brett and G o l d m a n 1996, L a w t o n 1999), and the applicability o f this knowledge to controlling eutrophication through "biomanipulat ion" (Gophen 1990, Reynolds 1994, Drenner and Hambright 1999). T h e hypothesis that trophic cascades are important and c o m m o n in aquatic ecosystems is most often tested by manipulating the top trophic level, usually zooplanktivorous fish. T r o p h i c cascades are implicated if the elimination of zooplanktivorous fish results in increased zooplankton biomass and decreased phytoplankton biomass. A meta-analysis o f top-predator addition experiments in aquatic systems has shown that a great majority o f such studies conform to the predictions o f the trophic cascade hypothesis (Brett and G o l d m a n 1996). Despite that success, this meta-analysis showed a great deal o f variability in the strength of the trophic cascade, that is, in the degree to which the top-predator (zooplanktivorous fish) affects herbivore (zooplankton) and primary producer (phytoplankton) biomass. 38 M a n y hypotheses have been advanced to explain the observed variability in the strength of trophic cascades. Principal among these is that top-down effects interact with bottom-up effects to produce results that are not predictable unless nutrient dynamics are also taken into account (Oksanen et al. 1981, Polis 1994). A l t h o u g h there have been some experimental results that support this view ( L e i b o l d and W i l b u r 1992, Balciunas and L a w l e r 1995), the hypothesis was rejected by a recent meta-analysis o f 11 independent experiments in freshwater enclosures in which both the presence of a third trophic level and the concentration of nutrients were manipulated (Brett and G o l d m a n 1997). T h e meta-analysis showed that both nutrients and predators had a significant effect on the response of primary producer to these manipulations, but that there was no significant interaction between the two. Several other factors, especially spatial heterogeneity (Scheffer 1998), and f o o d web heterogeneity, such as the prevalence of o m n i v o r y (Strong 1992, D i e h l 1995), interference among predators ( M c C a n n et al. 1998) and heterogeneity in prey edibility (Le ibold 1989), have been suggested to modify the outcome of f o o d web manipulations under certain conditions (Polis et al. 2000). However , it is not clear whether any of these factors is consistently important in determining the strength of trophic cascades. A more troubling suggestion is that the strength o f trophic cascades cannot be predicted because of inadequacies in the methodology. Replicated trophic cascade experiments typically run over very short time scales (one summer or less) and therefore last for only a fraction of a single generation o f the top predator, which is insufficient to estimate the eventual outcome of the manipulation (Yodzis 1988). T h e consistent top-d o w n effects that occur with the addition of a trophic level (Brett and G o l d m a n 1996) 39 might disappear once top-predator density decreases as a result of declining prey density or because of delayed compensation by species that are adapted to avoid predators. Unreplicated whole-lake experiments typically endure for m u c h longer than enclosure and p o n d experiments, but qualitative reviews of whole-lake experiments have suggested that the initial strong effect of removing or adding a top predator can decline in subsequent years (Shapiro 1990, M c Q u e e n 1998). It is therefore unclear whether the strong trophic cascades often observed in aquatic systems are weakened or even persist over relevant time scales. There is also debate as to whether the enclosures and mesocosms c o m m o n l y used in such experiments are appropriate models o f whole lakes. W h i l e it is relatively easy to obtain the necessary replicates in enclosure and mesocosm studies, these systems do not imitate many o f the processes that may be important determinants of community dynamics in lakes (Bloesch et al. 1988, Frost et al. 1988, Carpenter 1996). Conversely, the interpretation of unreplicated perturbation experiments, such as whole-lake studies, remains controversial (Stewart-Oates et al. 1992). M a n y o f the conclusions regarding trophic cascades are derived f r o m enclosure and m e s o c o s m studies, but there has been little attempt to investigate whether these results are comparable to those o f whole-lake manipulations. A l t h o u g h some of the large-scale processes that occur in lakes but not in mesocosms, such as benthic-pelagic coupling and long range biotic nutrient transport, have been shown to affect lake communities, no one has yet shown how they affect the strength of trophic cascades. W h i l e ecologists continue to debate whether large spatial scale is more important than replicability o f results, the m i n i m u m spatial scale necessary to approximate whole-lakes remains unknown. 40 T h i s study presents experimental and comparative data to address these concerns. I summarize the results o f a 4-year (4 top-predator generations) replicated trophic cascade experiment. T h i s experiment was more than 18 times longer than the average duration of replicated trophic cascade experiments, and almost twice as long as the longest published experiment. It is the only replicated experiment in freshwater of w h i c h I a m aware that satisfies the requirement of Y o d z i s (1988) that experiments last for at least 2 generations of the top predator. T h e experiment was conducted in ponds, w h i c h are a compromise between lakes and mesocosms. T h e purpose of the experiment was to test two predictions o f trophic cascade theory. First, that the strength o f trophic cascades is undiminished over prolonged periods. Second, that the magnitude of the effect o f fish on zooplankton predicted the magnitude of their indirect effect o n phytoplankton. I furthermore perform a quantitative literature analysis to compare m y results to those of other trophic cascade studies and to investigate whether the preceding hypotheses h o l d for trophic cascade experiments in general. T h e data f r o m the literature al lowed me to test the additional hypothesis that trophic cascade experiments in different types o f study system (lakes, ponds, mesocosms, enclosures) produce quantitatively different results when the topmost trophic level is manipulated. 3.2 METHODS Study site and Experimental Design T h e study was conducted in 4 experimental ponds at the Universi ty o f Brit ish C o l u m b i a , Canada from 6 July 1993 to 21 August 1997. T h e ponds are 23 x 23 m 2 , and 41 slope to a m a x i m u m depth of 3 m. T h e y were built in 1991 and have not since been drained. T h e experiment consisted of two treatments, each twice replicated. Zooplanktivorous fish (see below) were added to 2 of the 4 ponds, and the 2 other ponds were designated as controls. T h e control ponds used in this experiment have never contained fish. O n e of the ponds with fish had contained fish during 1992 using the same species as that used in the present study. F i s h had been removed f r o m this p o n d 4 months prior to the beginning of this study using minnow traps. Fish F i s h were introduced to 2 of the ponds during M a y 1993. I used a Limnet ic species o f the threespine stickleback (Gasterosteus sp.), one o f several populations endemic to 5 lakes (formerly 6) in Bri t ish C o l u m b i a , Canada. Sticklebacks are ideal to use in such an experiment because they are cosmopolitan and c o m m o n in most temperate countries, and because they have a relatively short generation time (usually 1 year). Furthermore, the Limnet ic species feeds primarily on open-water zooplankton in the wild , although adults, especially males, also consume littoral invertebrates during the spring breeding season (Schluter 1993). Parental fish were caught in Paxton L a k e , Brit ish C o l u m b i a ( 4 9 ° 4 3 ' N , 1 2 4 ° 3 1 ' W ) . E g g s were obtained and fertilised in the laboratory (Hatfield and Schluter 1999, V a m o s i et al. 2001). One thousand 8-week o ld juvenile sticklebacks were introduced into one p o n d on 31 M a y 1993, and 689 into the other pond on 7 July 1993. T h e latter pond was supplemented with 161 2-week o l d juveniles o n 9 A u g u s t 1993 for a total o f 850 42 individuals (unexpected mortality in the laboratory precluded introducing the full 1000 individuals) . It is unlikely that differences in the initial populations could alter the outcome of the experiment because a short generation time w o u l d allow fish numbers to quickly adjust to available resources. F i s h were trapped using minnow traps over a 24-h period on 15 M a r c h 1994. T r a p p e d fish were marked by c l ipping the first dorsal spine and released immediately after being marked. T h e same number of traps was used to recapture fish over the same amount of time 10 d fol lowing the marking. Because 76.3% of fish captured in the second trapping session were marked, the total number captured during the first session was a relatively accurate index of the total population size in the spring. F o r this reason, fish captured using the same number o f traps in subsequent years were counted but not marked nor recaptured. F o r those years, I estimated the fish population size in each p o n d by assuming that the number caught in 24 h was equal to the same proportion of the total p o n d population that was captured in 24 h during M a r c h 1994. A total of 150 fish were removed f r o m the ponds over the course of the experiment to enumerate gut contents. Zooplankton Zooplankton were sampled periodically during the summer and early fall by pumping 15 to 75 L of water through 62 u m Nitex netting. T h e zooplankton trapped on the netting were preserved in 5% formalin. Zooplankton were sampled haphazardly in areas devoid of macrophytes or mats of filamentous algae at a depth of 1 m. Zooplankton were enumerated and identified under a dissecting microscope. A l l o f the organisms in the sample were counted unless zooplankton were very abundant in which case 43 zooplankton were counted until 100 large (post-naupliar) crustacean zooplankton had been enumerated. Zooplankton biomass was calculated f r o m abundance and length data using length-weight relationships f r o m the literature (Bottrell et al. 1976, D o w n i n g and Rigler 1984). Phytoplankton Phytoplankton were sampled using a bilge p u m p f r o m haphazard locations at 1 m and 2 m depths in areas clear of macrophytes and mats o f filamentous algae during the summer and early autumn (May-October) . Total phytoplankton biovolume did not differ between the two depths (paired t-test: t43 = 1-2, P = 0.12). A l l data were therefore averaged over depth prior to statistical analyses. Phytoplankton were placed in 250 m l glass jars, stained with approximately 3 m l L u g o l s solution, and stored until they were analyzed in the laboratory starting M a y 2000. Phytoplankton were settled in 10 m l chambers overnight (> 18 h), and were then counted using an inverted microscope. Phytoplankton with a greatest linear dimension <10, <60, and >60 p,m were counted at magnifications of 600, 400, and lOOx respectively. Forty fields of view were counted at each magnification, which was equal to 0.011, 0.19, and 0.73 m l o f water respectively. C e l l size and shape were recorded to calculate biovolume. Literature Analysis I searched the literature for experiments for experiments in w h i c h zooplanktivorous fish were present or absent. I searched the Science Citat ion Index for articles containing the keywords: trophic cascade, biomanipulation, pond, enclosure, 44 mesocosm, fish, and combinations thereof. Enclosure studies prior to 1996 were obtained f r o m Brett and G o l d m a n (1996), and supplemented with database searches. Whole- lake manipulation data were obtained primarily f r o m the reviews o f L e i b o l d et al. (1997) and Hansson et al. (1998). Systems were divided into ponds, enclosures (bags and lake enclosures and exclosures), mesocosms (cattle tanks, plastic pools, etc.), and lakes. T o distinguish lake f r o m p o n d studies, a p o n d study was defined as one in w h i c h there were replicated man-made water bodies o f similar size and shape each with an area <1 ha, and which contained an evident benthic and limnetic invertebrate community . Unreplicated studies o f lakes <1 ha were excluded f rom the analysis. Ef fec t sizes were calculated by comparing the treatments with and without zooplanktivorous fish in replicated studies, and by comparing high and low zooplanktivore abundance in whole-lake studies unless data f r o m appropriate control lakes were available. I used some whole-lake manipulations that reduced planktivore abundance by adding a piscivorous fish, but only when this method was accompanied by some other method of zooplanktivorous fish removal (e.g. netting, rotenone). T h e data were further divided into 3 duration categories: studies that lasted for a single summer (Summer), for longer than a summer but less than a year (4 to 12 months) (1 year), or for longer than a year (>1 year). D a t a were taken f rom published figures using a digit izing tablet. A full list of the publications f r o m which data were taken is included in the A p p e n d i x . Phytoplankton chlorophyl l -a was used in preference to other measures o f phytoplankton biomass. B i o v o l u m e and fluorescence were used when chlorophyl l a data were not available. Zooplankton dry weight was used to estimate zooplankton biomass. Zooplankton 45 abundance was converted to dry weight using length-weight relationships, but typical weights of c o m m o n zooplankton species (Hal l et al. 1970, Wetze l and L i k e n s 2000) were used when zooplankton size data were not available, as was most often the case. In some cases, crustacean or cladoceran biomass were used instead of total zooplankton biomass when data o n the whole zooplankton community were lacking. Statistical analyses Repeated measures analysis of variance ( A N O V A R ) was used to estimate treatment effects over the course of m y experiment using S Y S T A T 5.05. D a t a were log-transformed prior to statistical analyses to homogenize the variances. I calculated "effect sizes" to assess the impact of zooplanktivorous fish on zooplankton and phytoplankton biomass. T h e zooplankton effect size was the zooplankton biomass when the zooplanktivore was absent or at low density divided by the zooplankton biomass when the zooplanktivore was at high density. T h e phytoplankton effect size was the phytoplankton biomass at high zooplanktivore density divided by the phytoplankton biomass when the zooplanktivore was absent or at low density. F o r comparisons of the zooplankton and phytoplankton effect sizes between system types (mesocosm, enclosure, pond, lake) and duration categories (Summer, 1 year, >1 year) I used analysis of variance ( A N O V A ) on log-transformed data to estimate differences among categories. I also tested for differences in the coefficient of variation among the duration categories and types o f system (Zar 1996). 46 3.3 RESULTS Fish F i s h numbers fluctuated dramatically both between and within years (Table 3.1). Observations f r o m w i l d populations ( D . Schluter, personal communication) indicate that this may have been largely due to within-year differences in trappability. Unfortunately, there are insufficient data to test this hypothesis. T h e gut contents of the fish were also highly variable. N o single species consistently dominated stickleback guts. T h e small herbivorous cladoceran Chydorus was often numerically the most c o m m o n item (mean percent o f total gut organisms per fish for both ponds combined = 33.4%), but may have had little importance in supporting stickleback populations because of their low weight compared to larger cladoceran and copepod species. C a l a n o i d and c y c l o p o i d copepods, Bosmina longirostrus, amphipods, Diaphanosoma, and ostracods were all numerically important during different years or in different ponds, but were never consistently important. Zooplankton Zooplankton biomass was considerably higher in control ponds ( A N O V A R : F/,2 = 258.9, P = 0.004) even though the initial average zooplankton biomass was almost identical in control ponds and ponds with fish for the 4 sampling dates at the outset of the experiment in 1993 (Figure 3.1). There were both a significant T i m e effect (F 10,20 - 7.2, P < 0.0001) and a significant Treatment x T i m e interaction (F 10,20 = 3.9, P = 0.005). T h e significant interaction term most l ikely results f r o m the delayed effect of fish on 47 zooplankton biomass over the first year of the experiment. T h e lower zooplankton biomass in ponds with fish was accompanied by a shift in zooplankton community structure. Ponds with no fish were dominated by larger zooplankton, especially large Diaphanosoma, calanoid copepods, and Daphnia pulex. Ponds with fish were dominated by smaller zooplankton species, especially Bosmina and Chydorus. T h e invertebrate predator Chaoborus was present at low densities at the outset o f the experiment in both control and treatment ponds (Figure 3.2), but became more abundant in control ponds after the first two years of the experiment ( A N O V A R : F ; , 2 = 52.4, P = 0.019). Phytoplankton There was a higher total biovolume o f phytoplankton in ponds with fish present compared to ponds without fish ( A N O V A R ; F1>2 = 69.3, P = 0.014) (Figure 3.3). There was no significant time effect (F 12,24 = 1.0, ns), nor was there any interaction between time and treatment effects (Fi2,24 = 0.6, ns). Phytoplankton were divided in 2 categories based on their edibility to c o m m o n zooplankton. Phytoplankton greater than approximately 30-60 u m are in general m u c h less edible to even the largest zooplankton grazers than are smaller phytoplankton (Burns 1968, Vanderploeg 1981, L e h m a n and Sandgren 1985, M c C a u l e y and D o w n i n g 1985). W h e n these size categories were analyzed separately, further analysis indicated that the higher total phytoplankton biovolume in ponds with fish is the result of a higher contribution o f small edible phytoplankton (<60 urn diameter) to the total phytoplankton biovolume ( A N O V A R : Fy,2 =123.3, P = 0.008). In contrast, there was a higher b i o v o l u m e of large inedible 48 phytoplankton (>60 p m diameter) in the control ponds (Figure 3.4; A N O V A R : F/,2 = 85.65, P = 0.011). Effect of experiment duration T h e results f r o m m y experiment indicate that the zooplankton were consistently greater in ponds without fish for all data points subsequent to the first year o f the study (Figure 3.1). T h e phytoplankton response to zooplanktivorous fish additions was more variable. Overal l , there was a greater phytoplankton b i o v o l u m e in ponds with fish even though there was no difference in the total phytoplankton biovolume during the first year of the experiment ( A N O V A R : F/,2 = 0.11, ns). T h e phytoplankton biomass was considerably higher in the p o n d with fish on all sampling dates except during the 4 t h year o f the study (1996) when there was little difference in the phytoplankton biomass between the fish and control ponds. B y 1997, a higher phytoplankton biomass in ponds with fish had resumed. There was therefore little evidence for a damping o f the trophic cascade at later dates (Figure 3.3). T o assess whether the strength o f trophic cascades was generally affected by experiment duration, I divided published studies into 3 duration categories {Summer, 1 year, >1 year) (Figure 3.5). Effect sizes (the magnitude o f change resulting f r o m the addition of fish) were calculated for each independent experiment (Table 3.2). There was no significant difference among categories in the log-transformed phytoplankton effect sizes ( A N O V A : F2,ss= 1-1, ns). There was, however, a significant difference among the log-transformed zooplankton effect sizes ( A N O V A : F2,75 = 4.96 , P = 0.009). Pairwise comparisons indicated that this difference was due to significantly smaller zooplankton 49 effect in the >1 year category than the Summer category ( T u k e y test: 975,3 = 3.81, P = 0.024) and the 1 year category (Tukey test: 975,5 = 3.96, P = 0.018) categories. There was no difference between the Summer and 1 year effect sizes ( T u k e y test: 975,3 = 1.14, P = 0.701). I also tested to see whether the variability (coefficient of variation) of the effect sizes differed among the duration categories. I found that the coefficient of variation of the phytoplankton effect sizes was significantly higher in experiments that lasted for one summer than for experiments that lasted for longer than a year (U= 2.24, p = 0.025) but that there was no significant differences for any o f the other pairwise comparisons. T h e coefficient of variation of the zooplankton effect sizes were significantly higher in the Summer category compared to <1 year (t<x>= 2.31, p = 0.022) and >1 year (t*>= 2.21, p = 0.028) experiments, and was also higher for <1 year compared to >1 year experiments (t_= 4.24, p < 0.0005). Whole- lakes were the only study system that had sufficient data f rom all 3 duration categories to warrant statistical tests that could control for the system of study. There was a significant difference in the phytoplankton effect sizes among the 3 duration categories when only the whole-lake data were used (Table 3.2; F226 = 7.5, P = 0.003) owing to a significantly larger phytoplankton effect in the Summer category compared to the <1 year category (Tukey test: 95,26 = 4.03, P = 0.022) and >1 year studies (Tukey test: 95,26 = 5.45, P = 0.002). There was no significant difference among the 3 categories for the zooplankton effect sizes {Fi.n - 1.5, ns), although there were only 2 data points in the Summer category. 50 Trophic coupling Tight coupl ing between trophic levels is impl ied i f the magnitude o f the effect o f a disturbance on one trophic level is transmitted at the same magnitude to the trophic level above or below. In m y long-term experiment, the addition of zooplanktivorous fish resulted in an average 18.24-fold decrease in zooplankton, and a 5.69-fold increase in phytoplankton. However , there was no difference between these effect sizes ( fe = 1.22, P = 0.12). In m y across-studies comparison, the addition o f zooplanktivorous fish in general resulted in a 2.55-fold decrease in zooplankton and a 2 .18-fold increase in phytoplankton, but there was similarly no difference between these effect sizes (tj67 - 0.97, ns). Tight coupling is also implied if there is a significant relationship between zooplankton and phytoplankton effect sizes, either over time within an experiment, or across studies. A l t h o u g h the pattern o f effect sizes was somewhat similar in m y experiment between zooplankton and phytoplankton (Figure 3.6), there was no relationship between zooplankton and phytoplankton effect sizes for the dates where data were available for both zooplankton and phytoplankton (Fij = 0.66, ns, R 2 = 0.09). T h e absence of a relationship was due to a more rapid response o f the zooplankton to zooplanktivore additions at the outset o f the experiment, and a dip in the phytoplankton effect size during 1996 that did not occur in the zooplankton. In m y across-studies comparison, I similarly found that there was no relationship between zooplankton and phytoplankton effect sizes (Figure 3.7; Fjj6 = 0.83, ns). W h e n data points f r o m particular system types (ponds, mesocosms, enclosures, whole-lakes) were analyzed separately (Table 3.3), there was a significant relationship between zooplankton and phytoplankton effect size for p o n d studies (Fi,9 = 10.9, P = 51 0.009, = 0.55), but not for mesocosms, enclosures or whole-lakes (Figure 3.7). There was no difference in the effect o f fish on phytoplankton biomass a m o n g studies that were conducted in different system types = 0.4, ns), but there was a significant difference among zooplankton effects (F3J4 = 7.0, P = 0.0003). T h i s was due to a significantly larger zooplankton effect size in mesocosm studies compared to enclosure (Tukey test: q75,4 = 4.49, P = 0.012) and whole-lake ( T u k e y test: q75,4 = 5.44, P = 0.002) studies (Table 3.3). T h e zooplankton effect was also significantly higher in ponds compared to lakes ( T u k e y test: q75,4 = 3.90, P = 0.016) but was heavily influenced by 3 outliers with very high zooplankton effects in ponds (Figure 3.7). T h e variability (coefficient o f variation) among studies in the zooplankton effect sizes was significantly lower in lake studies compared to p o n d ( U o = 3.67, p = 0.008) and enclosure ( L o = 2.58, p = 0.01) studies, and was also significantly lower in mesocosm compared to ponds (k*= 2.08, p = 0.039). N o other comparison was significant. There was no significant difference between any pair o f system types in the coefficients o f variation of phytoplankton effect size. 3.4 DISCUSSION Several authors have argued that trophic cascades are unlikely in complex f o o d webs, where inedible species, omnivores, and interference among competing predator species are likely to preclude the propagation of top-down signals (Strong 1992, M o r i n and L a w l e r 1995). T h e results o f m y across-studies comparison, supported by other similar meta-analyses (Brett and G o l d m a n 1996, L e i b o l d et al. 1997) indicate that the 52 trophic cascade is an important and indeed ubiquitous feature of lentic freshwater systems, f r o m small-scale enclosure and m i c r o c o s m experiments to whole-lake manipulations. Previous experiments have shown, however, that a number of factors can m o d i f y the outcome of trophic cascade experiments, inc luding spatial and temporal heterogeneity, heterogeneity in prey edibility, and self regulation of trophic levels through omnivory , intraguild predation, and territoriality (Polis et al. 2000). G i v e n the multiplicity of these secondary factors, it is surprising that even the qualitative hypotheses of the trophic cascade are supported in the great majority of studies. A l t h o u g h these factors appear to be important in preventing trophic cascades in many terrestrial systems (Schmitz et al. 2000) and might play a role in some of the instances of failed trophic cascades in freshwater systems ( D e M e l o et al. 1992, Badgery et al. 1994), m y across-studies comparison shows that they appear to be insufficient to block trophic cascades in the great majority of cases in freshwater. Compensat ion within trophic levels is often cited as being responsible for the absence of a trophic cascade. Compensat ion occurs when higher levels of predation do not result in reduced biomass of the prey, but instead in an increased abundance o f less edible prey. T h e long-term experiment provides some of the clearest documented evidence for compensation by large phytoplankton when zooplankton are abundant, but this compensation still fails to prevent a net decline in the phytoplankton biomass in the absence of zooplanktivorous fish. Similarly , compensation by the invertebrate zooplanktivore Chaoborus in the absence o f zooplanktivorous fish has the potential to stabilize zooplankton densities and therefore also phytoplankton densities. M y long-term 53 experiment shows that even clear compensation by Chaoborus is insufficient to preclude the trophic cascade. However , m y across-studies comparison indicates that zooplanktivorous fish can have a wide range o f effects o n zooplankton without subsequent effects o f similar magnitude o n phytoplankton (Figure 3.7). Therefore, although several authors have presented clear evidence that zooplankton refuges (e.g. macrophytes) play a key role in preventing declines in zooplankton biomass and therefore increases in phytoplankton biomass (Scheffer 1998), m y comparative data show that the zooplankton biomass alone is a poor predictor of phytoplankton biomass. T h e communi ty composit ion of the zooplankton and phytoplankton rather than only biomass m a y therefore be more important determinants of their response to manipulations of higher trophic levels. T h e absence o f a relationship between zooplankton and phytoplankton effect sizes unfortunately makes it difficult to predict the outcome o f programs to control eutrophication by biomanipulation. M y experimental results clearly demonstrate that the addition of a zooplanktivorous fish to replicate ponds results in a decrease in total zooplankton biomass and an increase in phytoplankton biovolume over prolonged periods. T h e results reported here are f rom the longest replicated experiment o f aquatic trophic interactions o f which I a m aware. These results show that trophic cascades are maintained for 4 generations o f the top predator despite a lower phytoplankton biomass in ponds with fish during the fourth year (1996) of the experiment. S u c h a prolonged experiment provides a g o o d estimate o f the long-term consequences o f this press perturbation ( Y o d z i s 1988). T h e great majority of previous reports of replicated mesocosm, enclosure, and p o n d 54 experiments have lasted for considerably less than a single generation of the top predator. A l t h o u g h the duration of whole lake manipulations is typically longer than for whole-lake manipulations, few have run for the amount of time required to estimate the eventual outcome o f the manipulation (Carpenter 1988, M c Q u e e n 1998). I hypothesised that the strength of the trophic cascade w o u l d decline fol lowing the original reduction of zooplankton, either because of delayed compensation by inedible phytoplankton and zooplankton or because of declines in zooplanktivore densities as they adjust to the lowered zooplankton biomass. T h e results indicate, however, that inedible phytoplankton compensate rapidly fol lowing declines in the edible phytoplankton, making it unlikely that there w o u l d be any long-term declines in the strength of the trophic cascade. T h e resolution o f m y estimates o f stickleback densities are probably marred by intra-annual variation in trappability, which makes it difficult to see how they adjust to the initial declines in the zooplankton standing crop. I therefore conclude f r o m m y experimental results that there is no evidence for a decline in the strength of the trophic cascade over prolonged periods. M y analysis of published reports is in general agreement with the results of the long-term experiment. T h e results indicate that the great majority o f studies also show a decrease in the zooplankton and an increase in the phytoplankton biomass, and that these results do not diminish over longer time scales. Rather, m y analysis shows that trophic cascades are not only ubiquitous, but are also persistent. T h e higher variability o f phytoplankton and zooplankton effect sizes for very short experiments (one summer or less) suggests that researchers should remain cautious when extrapolating short-term results to longer time scales. A recent experiment (Attayde and Hansson 2001) 5 5 demonstrated that even modest increases in the duration of their experiments ( from 14 to 28 d) can considerably decrease the variability o f the phytoplankton response to manipulations of higher trophic levels. T h e comparisons are unfortunately confounded by differences between the type of system in w h i c h the experiments are performed, with the duration of small-scale mesocosm and enclosure experiments typically much shorter than that o f whole-lake manipulations. There is some evidence to suggest that that phytoplankton effect size is lower for long-term manipulations when only whole-lake data are used, but the comparison might be biased by a low sample size o f shorter experiments. T h e results of m y literature analysis also demonstrate that there are systematic differences in both the mean and variance of the results o f trophic cascades that depend on the type of system in which the study is conducted. A l t h o u g h such differences between small scale and whole-lake manipulations have been suggested previously, especially by proponents of whole-lake manipulations (Carpenter 1996), there has been little effort to verify the accuracy of this assumption. T h e criteria necessary for a successful biomanipulation have, o f necessity, relied o n comparative data. Unfortunately, this makes it difficult to understand the variables that actually determine the strength of the trophic cascade because many of the morphometric and biological characteristics of lakes covary. F o r example, because of recent biomanipulation failures in some large lakes, it has been suggested that strong trophic cascades are unusual in large deep lakes, perhaps because of compensation by inedible algal species and by invertebrate zooplanktivores ( M c Q u e e n 1998). T h e results suggest that researchers should be cautious in extrapolating the results of small-scale studies to larger systems. 56 Table 3.1 F i s h dynamics in the two ponds to which fish were added. Density estimates are the number of fish caught in minnow traps over 24 h. A mark recapture experiment in 1994 indicated that o n average 76.3% o f the population was caught in 24 h. M e a n population sizes (mean number of fish per pond) were therefore estimated as the mean Density multiplied by 1.237. T h e first row of data is the number o f fish introduced into the ponds. Date Density in P o n d 1 (fish caught 24 h"1) Densi ty in P o n d 2 (fish caught 24 h"1) M e a n population size (fish pond" 1 ) 6 July 1993 925 26 M a r c h 1994 182 164 244.9 23 June 1994 102 253 93.4 6 June 1995 768 793 965.5 11 A p r i l 1996 625 486 443.5 1 A p r i l 1997 486 629 554.2 57 Table 3.2 A m o n g - s t u d y mean effect of fish on zooplankton (control/fish) and phytoplankton (fish/control) biomass for experiments that lasted for a single summer, for 4-12 months, and for longer than one year. Means , standard errors (SE) , and coefficients of variation ( C V ) are calculated f r o m the untransformed data. Samples sizes (n) of phytoplankton and zooplankton effect sizes are not equal for a given duration category because some studies reported phytoplankton but not zooplankton biomass. Phytoplankton effect size Zooplankton effect size Durat ion n M e a n S E C V n M e a n S E C V A l l data S u m m e r 52 3.41 0.57 1.21 43 7.03 1.76 1.64 4-12 months 16 3.41 0.78 0.91 14 56.98 47.14 3.15 >1 year 23 2.18 0.25 0.54 21 1.25 1.59 0.26 Whole- lakes S u m m e r 4 10.98 5.06 1.08 2 2.81 2.20 0.90 4-12 months 4 2.17 0.54 2.00 4 3.15 1.30 1.21 >1 year 21 2.13 0.27 1.74 19 1.56 0.27 1.33 58 Table 3.3 A m o n g - s t u d y mean effect of fish on zooplankton (control/fish) and phytoplankton (fish/control) in enclosure, mesocosm, pond, and whole-lake experiments. Means , standard errors (SE) , and coefficients of variation ( C V ) are calculated f rom the untransformed data. Phytoplankton effect size Zooplankton effect size n M e a n S E C V M e a n S E C V Enclosure 25 2.68 0.43 0.92 3.32 1.27 1.91 M e s o c o s m 17 3.36 0.84 1.03 10.77 3.40 1.30 P o n d 11 3.26 0.72 0.77 73.23 59.59 2.70 L a k e 25 3.36 0.87 1.39 1.91 0.33 0.86 59 Figure 3.F. M e a n ( ± S E ) total zooplankton biomass in control ponds and ponds with fish. Error bars are asymmetric on a logarithmic scale. T h e shaded areas are fall sampling dates (1 September to 1 November) . 60 61 Figure 3.2: M e a n ( ± S E ) abundance of the invertebrate zooplanktivore Chaoborus in control ponds and ponds with fish. Error bars are asymmetric on a logarithmic scale. T h e shaded areas are fall sampling dates (1 September to 1 N o v ember ) . 62 63 Figure 3.3: M e a n ( ± S E ) phytoplankton bio volume in control ponds and ponds with fish. Error bars are asymmetric on a logarithmic scale. T h e shaded areas are fall sampling dates (1 September to 1 November) . 64 Figure 3.3 65 Figure 3.4: M e a n ( ± S E ) abundance of inedible phytoplankton in control ponds and ponds with fish. Er ror bars are asymmetric on a logarithmic scale. T h e shaded areas are fal l sampling dates (1 September to 1 November) . 66 67 Figure 3.5: T h e effect of fish on (A) zooplankton and (B) phytoplankton for published enclosure, mesocosm, pond, and whole-lake experiments that endured for up to 1 summer, 1 summer to 1 year, and longer than 1 year. References f r o m which data were obtained are given in the Appendix . 68 1000 100 o 4-1 D a o C ' "a. o o N o ,v c o -4—* M C _S "a, O PH 10 d 1 i 0.1 + 10 0.1 0 o o B 8 o o o B • • • • A A O D 1 A 5 A A t O £ Summer 4-12 months o Enclosure • M e s o c o s m • P o n d A L a k e A A >1 year Figure 3.5 69 Figure 3.6: Effect of fish on the biomass of phytoplankton (fish/control) and zooplankton (control/fish) over the course of the experiment. Large symbols indicate data points used to test for a relationship between zooplankton and phytoplankton effect sizes. T h e shaded areas are fall sampling dates (1 September to 1 November) . 70 Figure 3.6 71 Figure 3.7: Phytoplankton effect size (treatment/control) as a function of zooplankton effect size (control/treatment) f r o m published experiments in ponds, mesocosms and enclosures and whole-lake manipulations. T h e data above and to the right of the dashed line (1s t quadrant) are in agreement with the predictions of the trophic cascade hypothesis (zooplankton and phytoplankton effect sizes >1). References f r o m w h i c h data were obtained are given in the Appendix . 72 100 10 1 i 0.1 : -• o - • A • o o • • V • • • • D 0 A O _ * • * A O o O i D 1 •0 --1 1 1 1 o 1 1 1 1 M i l l 1 1 1 — 1 1 1 1 1 | 1 1 1 1 1 1 1 1 0.1 • Pond • Lake n M e s o c o s m o Enclosure 1 10 Zooplankton effect 100 1000 Figure 3.7 73 C H A P T E R I V C o n c l u s i o n s 74 4.1 GENERAL CONCLUSIONS T h e combined results o f the two experiments presented in Chapter II demonstrate the importance of prey heterogeneity in determining the balance between top-down and bottom-up regulation. I show that the presence o f inedible phytoplankton can alter the response of the phytoplankton to nutrient additions and the addition of a trophic level. Theoretical ( G e n k a i - K a t o and Y a m a m u r a 1999, H u x e l 1999) and experimental (Bohannan and L e n s k i 1999) work has further shown that not only the standing crop biomass but also the variability and stability of the biomass is affected by the presence of inedible species. Unfortunately, there are few experiments that explicitly manipulate prey edibility despite the hypothesised importance of inedible species in determining the dynamics of trophic levels. I hope m y work wil l stimulate further efforts. Despite the importance of inedible species, they were unable to compensate for the losses incurred by the edible phytoplankton in the long-term experiment presented in Chapter III. There was therefore little evidence for a diminution of the trophic cascade over the course of the experiment. Similarly , the across-studies comparison demonstrated that there was no strong effect of experiment duration o n the strength o f the trophic cascade. T h e decline in the strength of the trophic cascade with increasing experiment duration when the analysis is restricted to whole-lake manipulations warns that such analyses are not conclusive. T h e analysis demonstrates that short-term and small-scale experiments produce results that are highly variable, and consequently, difficult to extrapolate larger systems. Naturally, this conclusion also extends to the results of the experiments presented in Chapter II. 75 4.2 FURTHER WORK T h e experiments presented in Chapter II underline the importance of understanding the dynamics o f functional groups within a trophic level. R e m o v a l o f other types o f functional groups, such as omnivores (Diehl 1995), dominant herbivores (Persson et al. 2 0 0 1 ) o r intraguild predators (Polis and H o l t 1992, R o s e n h e i m et al. 1993, M o r i n 1999) has similarly proved to be a fruitful method o f understanding the dynamics o f trophic levels and communities. S u c h studies, and including the present one, demonstrate that the debate on whether the abundance o f organisms is principally determined by nutrient supply or predation levels w i l l o n l y be resolved in the context o f the f o o d web in w h i c h the organisms o f interest are embedded. M i c r o c o s m s and mesocosms are the only systems in w h i c h m a n y o f the ideas in ecology can be tested because of the difficulty and expense o f manipulating particular variables while holding all others constant in whole-lake manipulations. A s the results o f Chapter III are among the first to show, the results of small-scale experiments often produce quantitatively different results f r o m those o f large-scale manipulations. I believe that, rather than dismissing the results of such experiments as inadequate at the spatial scale o f a whole lake, it is necessary to identify the way in w h i c h trophic cascades are altered because of the small spatial alone. S u c h an understanding is also crucial for understanding the importance of trophic cascades in terrestrial and marine ecosystems, where small-scale enclosure and exclosure experiments are often the only recourse. 76 REFERENCES A b r a m s , P . A . , and C J . Walters. 1996. Invulnerable prey and the paradox of enrichment. E c o l o g y 77: 1125-1133. A g r a w a l , A . A . 1998. A l g a l defense, grazers, and their interactions in aquatic trophic cascades. A c t a Oecologica 19: 331-337. A g r e n , J. , and D . W . Schemske. 1993. T h e cost of defense against herbivores: A n experimental study of trichome production in Brassica rapa. A m e r i c a n Naturalist 141: 338-350. Attayde, J . L . , and L . - A . Hansson. 2001. Press perturbation experiments and the indeterminacy o f ecological interactions: Effects o f taxonomic resolution and experimental duration. O i k o s 92: 235-244. Badgery, J., D . J . M c Q u e e n , K . H . Nichol ls , and P . R . Schaap. 1994. Biomanipulat ion at Rice L a k e , Ontario, Canada. L a k e and Reservoir Management 10: 163-173. Balciunas , D . , and S.P. Lawler . 1995. Effects o f basal resources, predation, and alternative prey in m i c r o c o s m f o o d chains. E c o l o g y 76: 1327-1336. Bloesch , J., P. Bossard, H . Buehrer, H . R . Buergi , and U . Uehlinger . 1988. C a n results f r o m l imnocorral experiments be transferred to in-situ conditions: Biomanipulat ion in limnocorrals V I . H y d r o b i o l o g i a 159: 297-310. Bohannan, B . J . M . , and R . E . L e n s k i . 1999. Effect o f prey heterogeneity o n the response o f a model food chain to resource enrichment. A m e r i c a n Naturalist 153: 73-82. Bohannan, B . J . M . , and R . E . L e n s k i . 2000. T h e relative importance o f competition and predation varies with productivity in a model community . A m e r i c a n Naturalist 156: 329-340. 77 Bottrell, H . H . , A . D u n c a n , Z . M . G l i w i c z , E . Grygierek, A . H e r z i g , A . Hil lbr ight -Illowska, P. Larsson, and T . Weglenska. 1976. A review of some problems in zooplankton production studies. Norwegian Journal o f Z o o l o g y 24: 419-456. Brett, M . T . , and C R . G o l d m a n . 1996. A meta-analysis of the freshwater trophic cascade. Proceedings of the National A c a d e m y o f Sciences U S A 93: 7723-7726. Brett, M . T . , and C . R . G o l d m a n . 1997. Consumer versus resource control in freshwater pelagic f o o d webs. Science 275: 384-386. Burns , C W . 1968. T h e relationship between b o d y size o f filter-feeding Cladocera and the m a x i m u m size of particle ingested. L i m n o l o g y and Oceanography 13: 675-678. Carpenter, S .R. 1988. Transmission of variance through lake f o o d webs. pp. 119-135. in: S. Carpenter (ed.). C o m p l e x interactions in lake communities , Springer-Verlag , N e w Y o r k . Carpenter, S .R. 1996. M i c r o c o s m experiments have l imited relevance for community and ecosystem ecology. E c o l o g y 77: 677-680. Carpenter, S .R. , and J .F. Ki tchel l . 1992. Trophic cascades and biomanipulation: Interface o f research and management- A reply to the comment by D e M e l o et al. L i m n o l o g y and Oceanography 37: 208-213. Carpenter, S .R. , J .F . Ki tchel l , and J .R. H o d g s o n . 1985. Cascading interactions and lake productivity. B i o s c i e n c e 35: 634-639. C o l e y , P . D . , J .P. Bryant, and I. F . S . Chapin . 1985. Resource availability and plant antiherbivore defense. Science 230: 895-899. D e M e l o , R. , R. France, and D . J . M c Q u e e n . 1992. Biomanipulat ion: H i t or myth? L i m n o l o g y and Oceanography 37: 192-207. 78 D i e h l , S. 1995. Direct and indirect effects of o m n i v o r y in a littoral lake community . E c o l o g y 76: 1727-1740. D i e h l , S., and M . Feibel . 2000. Effects of .enrichment on three-level f o o d chains with omnivory . A m e r i c a n Naturalist 155: 200-218. D o w n i n g , J . A . , and F . H . Rigler (ed.). 1984. A manual o n methods for the assessment of secondary productivity in fresh waters. B l a c k w e l l Scientific , O x f o r d . Drenner, R . W . , and K . D . Hambright . 1999. Biomanipulat ion of fish assemblages as a lake restoration technique. A r c h i v fur H y d r o b i o l o g i e 146: 129-165. Elser , J.J., and C R . G o l d m a n . 1991. Zooplankton effect o n phytoplankton in lakes of contrasting trophic status. L i m n o l o g y and Oceanography 36: 64-90. Fretwell , S . D . 1977. T h e regulation of plant communities by f o o d chains exploiting them. Perspectives in B i o l o g y and M e d i c i n e 20: 169-185. Frost, T . M . , D . L . D e A n g e l i s , S . M . Bartell, D . J . H a l l , and S . H . Hurlbert . 1988. Scale in the design and interpretation of aquatic community research, pp. 229-258. in: S. Carpenter (ed.). C o m p l e x interactions in lake communities, Springer-Verlag , N e w Y o r k . G e n k a i - K a t o , M . , and N . Yamamura . 1999. Unpalatable prey resolves the paradox of enrichment. Proceedings of the R o y a l Society of L o n d o n B 266: 1215-1219. G o p h e n , M . 1990. Biomanipulat ion: Retrospective and future development. H y d r o b i o l o g i a 200/201: 1-11. Grover , J.P. 1995. Competit ion, herbivory, and enrichment: Nutrient-based models for edible and inedible plants. A m e r i c a n Naturalist 145: 746-774. Hairston, N . G . , F . E . Smith, and L . B . Slobodkin . 1960. C o m m u n i t y structure, population control, and competition. A m e r i c a n Naturalist 44: 421-425. Halaj , J . , and D . H . Wise . 2001. Terrestrial trophic cascades: H o w m u c h do they trickle? A m e r i c a n Naturalist 157: 262-281. H a l l , D . J . , W . E . Cooper , and E . E . Werner. 1970. A n experimental approach to the production dynamics and structure of freshwater animal communities . L i m n o l o g y and Oceanography 15: 839-928. Hansson, L . - A . , C . Bronmark, P. Nyst rom, L . Greenberg, P. Lundberg , P . A . Ni lsson , A . Persson, L . B . Pettersson, P. Romare, and L . J . T r a n v i k . 1998. Consumption patterns, complexity and enrichment in aquatic f o o d chains. Proceedings of the R o y a l Society L o n d o n B 265: 901-906. Hatfield, T . , and D . Schluter. 1999. E c o l o g i c a l speciation in sticklebacks: Environment-dependent hybrid fitness. E v o l u t i o n 53: 866-873. H o l t , R . D . , and J . H . L a w t o n . 1994. T h e ecological consequences of shared natural enemies. A n n u a l Review of E c o l o g y and Systematics 25: 495-520. H o l t , R . D . , and G . A . Polis. 1997. A theoretical framework for intraguild predation. A m e r i c a n Naturalist 149: 745-764. Hunter, M . D . , and P . W . Price. 1992. Playing chutes and ladders: Heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. E c o l o g y 73: 724-732. H u x e l , G . R . 1999. O n the influence of f o o d quality in consumer-resource interactions. E c o l o g y Letters 2: 256-261. 80 Kretzschmar, M . , R . M . Nisbet, and E . M c C a u l e y . 1993. A predator-prey model for zooplankton grazing on competing algal populations. Theoretical Population B i o l o g y 44: 32-66. L a w t o n , J . H . 1999. A r e there general laws in ecology? O i k o s 84: 177-192. L e h m a n , J . T . , and C D . Sandgren. 1985. Species-specific rates of growth and grazing loss among freshwater algae. L i m n o l o g y and Oceanography 30: 34-46. L e i b o l d , M . A . 1989. Resource edibility and the effects of predators and productivity on the outcome of trophic.interactions. A m e r i c a n Naturalist 134: 922-949. L e i b o l d , M . A . , J . M . Chase, J . B . Shurin, and A . L . D o w n i n g . 1997. Species turnover and the regulation of trophic structure. A n n u a l R e v i e w of E c o l o g y and Systematics 28: 467-494. L e i b o l d , M . A . , and H . W . W i l b u r . 1992. Interactions between food-web structure and nutrients on p o n d organisms. Nature 360: 341-343. L i n d , O . T . 1979. H a n d b o o k of c o m m o n methods in l imnology . M o s b y , St. L o u i s . L i n d e m a n , R . L . 1942. T h e trophic-dynamic aspect of ecology. E c o l o g y 23: 399-418. M a c K a y , N . A . , and J.J. Elser. 1998. Factors potentially preventing trophic cascades: f o o d quality, invertebrate predation, and their interaction. L i m n o l o g y and Oceanography 43: 339-347. M a s s o n , S., B . P i n e l - A l l o u l , and V . H . Smith. 2000. T o t a l phosphorus-chlorophyll a relationships in southern Quebec lakes. L i m n o l o g y and Oceanography 45: 732-740. 81 M c C a n n , K . S . , A . Hastings, and D . R . Strong. 1998. Trophic cascades and trophic trickles in pelagic f o o d webs. Proceedings o f the R o y a l Society of L o n d o n B 265: 205-209. M c C a u l e y , E . , W . W . M u r d o c h , and S. Watson. 1988. Simple models and variation in plankton densities among lakes. A m e r i c a n Naturalist 132: 383-403. M c C a u l e y , E . , and J. D o w n i n g . 1985. T h e prediction of cladoceran grazing rate spectra. L i m n o l o g y and Oceanography 30: 202-212. M c Q u e e n , D . J . 1998. Freshwater food web biomanipulation: A powerful tool for water quality improvement, but maintenance is required. Lakes & Reservoirs: Research and Management 3: 83-94. M c Q u e e n , D . J . , J .R. Post, and E . L . M i l l s . 1986. Trophic relationships in freshwater pelagic ecosystems. Canadian Journal o f Fisheries and A q u a t i c Science 43: 1571-1581. M o r i n , P. 1999. Productivity, intraguild predation, and population dynamics in experimental f o o d webs. E c o l o g y 80: 752-760. M o r i n , P.J . , and S.P. Lawler . 1995. F o o d web architecture and population dynamics: T h e o r y and empirical evidence. A n n u a l R e v i e w of E c o l o g y and Systematics 26: 505-529. Mut ikainen , P. , and M . Walls . 1995. G r o w t h , reproduction and defence in nettles: Responses to herbivory modif ied by competition and fertilization. O e c o l o g i a 104: 487-495. Oksanen, L . , S . D . Fretwell , J. A r r u d a , and P. N i e m e l a . 1981. Exploitation ecosystems in gradients of primary productivity. A m e r i c a n Naturalist 118: 240-261. 82 Persson, A . , L . - A . Hansson, C . Bronmark, P. Lundberg , L . B . Pettersson, L . Greenberg, P . A . Ni lsson , P. Nystrom, P. Romare, and L . Tranvik . 2001. Effects of enrichment on simple aquatic f o o d webs. A m e r i c a n Naturalist 157: 654-669. Persson, L . 1999. Trophic cascades: A b i d i n g heterogeneity and the trophic level concept at the end of the road. O i k o s 85: 385-397. Polis , G . A . 1994. F o o d webs, trophic cascades and c o m m u n i t y structure. Australian Journal of E c o l o g y 19: 121-136. Polis , G . A . , and R . D . Holt . 1992. Intraguild predation: T h e dynamics of complex trophic interactions. Trends in E c o l o g y and E v o l u t i o n 7: 151-154. Polis , G . A . , A . L . W . Sears, G . R . H u x e l , D . R . Strong, and J. M a r o n . 2000. W h e n is a trophic cascade a trophic cascade? Trends in E c o l o g y and E v o l u t i o n 15: 473-475. Reynolds , C S . 1994. T h e ecological basis for the successful biomanipulation of aquatic communities. A r c h i v fur H y d r o b i o l o g i e 130: 1-33. Rosenheim, J . A . , L . R . Wilhoi t , and A . C A . 1993. Influence o f intraguild predation among generalist insect predators on the suppression o f an herbivore population. Oecologia 96: 439-449. Scheffer, M . 1998. E c o l o g y o f shallow lakes. C h a p m a n & H a l l , L o n d o n . Schluter, D . 1993. Adaptive radiation in stickleback size, shape, and habitat use efficiency. E c o l o g y 74: 699-709. Schmitz , O . J . , P . A . Hamback, and A . P . Beckerman. 2000. T r o p h i c cascades in terrestrial systems: A review of the effects of carnivore removals on plants. A m e r i c a n Naturalist 155: 141-153. 83 Shapiro, J. 1990. Biomanipulat ion: T h e next phase- making it stable. H y d r o b i o l o g i a 200/201: 13-27. Stewart-Oates, A . , J .R. Bence, and C W . Osenberg. 1992. Assessing effects of unreplicated perturbations: N o simple solutions. E c o l o g y 73: 1396-1404. Strong, D . R . 1992. A r e trophic cascades all wet? Differentiation and donor-control in speciose ecosystems. E c o l o g y 73: 747-754. V a m o s i , S . M . , T . Hatfield, and D . Schluter. 2001. A test of ecological selection against young-of-the-year hybrids of sympatric sticklebacks. Journal o f F i s h B i o l o g y 57: 109-121. Vanderploeg , H . A . 1981. Seasonal particle-size selection by Diaptomus sicilis in offshore L a k e M i c h i g a n . Canadian Journal o f Fisheries and Aquat ic Science 38: 504-517. Walters, C . J . , E . Krause, W . E . N e i l l , and T . G . Northcote. 1987. E q u i l i b r i u m models for seasonal dynamics of plankton biomass in four oligotrophic lakes British C o l u m b i a , Canada. Canadian Journal o f Fisheries and Aquat ic Sciences 44: 1002-1017. Watson, S., E . M c C a u l e y , and J . A . D o w n i n g . 1988. Contrasting patterns of net and nannoplankton production and biomass among lakes. Canadian Journal o f Fisheries and Aquatic Sciences 45: 915-920. Watson, S., E . M c C a u l e y , and J . A . D o w n i n g . 1992. S i g m o i d relationship between phosphorus, algal biomass, and algal community structure. Canadian Journal o f Fisheries and Aquatic Sciences 49: 2605-2610. Wetzel , R . G . , and G . E . L ikens . 2000. L i m n o l o g i c a l analyses. Springer, N e w Y o r k . 84 White , T . C . R . 1993. T h e inadequate environment: Nitrogen and the abundance of animals. Springer-Verlag , Ber l in . Y o d z i s , P. 1988. T h e indeterminacy o f ecological interactions as perceived through perturbation experiments. E c o l o g y 69: 508-515. Zar, J . H . 1996. Biostatistical analysis. Prentice H a l l , N e w Jersey. 85 APPENDIX I List of references f rom which data were obtained for m y analyses of published results o f trophic cascade experiments (Figures 3.5 and 3.7). Ponds: B u r k e & B a y n e (1986) Prog. F ish-Cultur . 48: 177-183; H a l l et al. (1970) L i m n o l . Oceanogr . 15:839-928; Hambright et al. (1986) C a n . J. F i s h . Aquat . S c i . 43: 1171-1176; Hambright (1994) Limnol .Oceanogr .39 :897-912 ; Hurlbert & M u l l a (1981) H y d r o b i o l . 83:125-151; L a w s & W e i s b u r g (1990) Prog. F ish-Cul tur . 52: 1-8; M e i j e r et al. (1990); H y d r o b i o l . 191:275-284; M i l s t e i n et al. (1988) Aquae . Fisher. Res. 19:127-137; Q i n & C u l v e r (1996) H y d r o b i o l . 321:109-118; Spencer & K i n g (1984) C a n . J . F i s h . Aquat . Sc i . 41: 1851-1855; T h i s study; Witeska (1995) A c t a H y d r o b i o l . 37: 121-129 Enclosures and Mesocosms: B e k l i o g l u & M o s s (1995) Freshw. B i o l . 33:497-509; B e k l i o g l u & M o s s (1998) Aquat . E c o l . 32:229-240; Bertolo et al. (1999) Oecologia 121:55-65; Bertolo et al. (1999) Freshw. B i o l . 41:795; Byers & V i n y a r d (1990) O e c o l o g i a 83:352-357; Christoffersen et al. (1993) L i m n o l . Oceanogr. 38:561-573; Cr isman & Beaver (1990) H y d r o b i o l . 200/201:177-185; Drenner et al. (1990) H y d r o b i o l . 208:161-167; Hansson et al. (1998) Proc. R. Soc. L o n d . B 265:901-906; Hurlbert et al. (1972) Science 175:639-641; Lancaster & Drenner (1990) C a n . J. F ish . Aquat . Sci . 47: 471-479; Lazzaro et al. (1992) C a n . J. F i s h . Aquat . Sc i . 49:1466-1473; L e i b o l d (1989) A m . Nat . 134:922-949; L y n c h 86 (1979) L i m n o l . Oceanogr . 24:253-272; M c Q u e e n et al. (1992) A r c h . H y d r o b i o l . 125:1-24; M a t v e e v et al. (2000) Freshw. B i o l . 44:375-385; M a z u m d e r et al. (1990) J. Plankt. Res. 12:109-124; N o w l i n & Drenner (2000) Oecologia 122:421-426; Proulx et al. (1996) E c o l o g y 77:1556-1572; Persson (1997) O i k o s 79:137-146; Ramcharan et al. (1996) C a n . J. F i s h . Aquat . Sc i . 53:2819-2828; Richardson & T h r e l k e l d (1993) C a n . J. F i s h . Aquat . Sc i . 50: 29-42; Schindler (1992) C a n . J. F ish . Aquat . Sc i . 49:2498-2506; Slusarczyk (1997) H y d r o b i o l . 342/343:215; Stephen et al. (1998) Freshw. B i o l . 39:699; Threlkeld (1988) L i m n o l . Oceanogr . 33:13620-1375; Turner & Mit telbach (1992) C a n . J. F ish . Aquat . Sci . 49: 1908-1915; V a n n i (1987) E c o l o g y 68:624-635; V a n n i & F i n d l a y (1990) E c o l o g y 71:921; V a n n i & L a y n e (1997) E c o l o g y 78:21-40. Whole-Lakes: B e n n d o r f et al. (1988) L i m n o l o g i c a 19:97-110; Carpenter & K i t c h e l l (1993) Trophic cascade in Lakes . Cambridge U n i v . Press; D o n k et al. (1990) H y d r o b i o l . 200/201:275-301; Giussani et al. (1990) H y d r o b i o l . 200/201:357-366; H a n s s o n et al. (1998) Ecosystems 1:558-574 and references therein; Jeppessen et al. (1990) H y d r o b i o l . 200/201:205-227; Langeland (1990) H y d r o b i o l . 200/201:535-540; L y n c h e et al. (1990) H y d r o b i o l . 200/201: 251; M c Q u e e n et al. (1989) E c o l . M o n o g r . 59:289-309; Persson et al. (1993) O i k o s . 66:193-208; R i e m a n n et al. (1990) H y d r o b i o l . 200/201:241-250; Sondergaard et al. (1990) H y d r o b i o l . 200/201:229-240; van der M o l e n & Boers (1991) Freshw. B i o l . 35:189; V a n n i et al. (1990) Nature 344: 333-335. 87 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0090066/manifest

Comment

Related Items