Revisiting Werner et al. 1983

In a paper published in Ecology in 1983, Earl Werner, James Gilliam, Donald Hall and Gary Mittelbach, showed experimentally that small bluegill sunfish, which are vulnerable to predation by bass, showed an observable shift in foraging behaviour when bass are present choosing to forage in less-profitable habitats. Thirty-six years after the paper was published I asked Earl Werner and Gary Mittlebach, about their motivation to do this study, memories of field work and what we have learnt since about the effect of predation risk on habitat use in fish.

Citation: Werner, E. E., Gilliam, J. F., Hall, D. J., & Mittelbach, G. G. (1983). An experimental test of the effects of predation risk on habitat use in fish. Ecology, 64(6), 1540-1548.

Date of interview: Questions sent by email on 15th January 2019; responses received by email on 4th February 2019.

 

Hari Sridhar: I would like to start by asking you about your motivation for doing this study, in relation to your previous research. At this point, you had already done many years of research on sunfishes, including research directly related to this study. Can you tell us how you got interested in sunfish, especially the bluegill sunfish, and the origins of this particular study published in Ecology in 1983?

Earl Werner & Gary Mittelbach: When you look back thirty years, it’s hard to remember why you did anything, let alone a particular experiment. However, our recollections about the motivation of this experiment are reasonably clear; the question posed in the experiment arose from observations regarding the size class distribution of the bluegill that we had quantified in natural lakes; smaller classes typically were found in the littoral vegetation while larger classes frequented many habitats including open water. The bluegill was the dominant fish in these lakes and a species we had done a lot of work on. This conundrum was deepened by results emerging from Mittlebach’s thesis; a carefully parameterized optimal foraging model predicted that all size classes in a local lake should shift seasonally to feed in the open water on zooplankton. The larger fish conformed very well to the predicted optimal behavior, but the small size classes continued to feed in the then suboptimal littoral habitat. Was that due to competition, predation risk, inadequacies of the model, or something else? We had hypothesized that predation risk was responsible and did this experiment to find out, and the fish gave us their answer.

 

HS: Stepping back a bit, how did you get interested in ecology, particularly freshwater community ecology?

EW: My interests in aquatic ecology certainly has its roots in the countless hours I spent along, and in, a stream that ran through the small farm I grew up on. A more formal introduction began with being hired as a graduate student by Don Hall and Bill Cooper to help tear down the large Cornell Pond Experiment(Limnology and Oceanography, 1970) which led to becoming a co-author on the paper and the realization that a childhood interest could become a career.

GM: Like Earl, my childhood growing up in rural Iowa shaped my love of the outdoors and nature. But it was when I started graduate school at Michigan State and encountered the wealth of lakes, ponds and streams found in Michigan that I truly developed a passion for aquatic ecology.

 

HS: This paper has four authors, how did this group come together and what did each person bring to the study?

EW & GM: Hall and Werner were faculty and had worked together on the Cornell Pond study. Mittelbach and Gilliam were graduate students at Kellogg at the time, both doing their theses on fish ecology. The four of us each contributed different ideas and talents that led to a wonderful collaboration. Hall was an expert in zooplankton ecology, Werner had worked on bluegill foraging and community structure of sunfish, Mittelbach was doing his theses on optimal foraging in bluegill, and Gilliam had provided some theory on the timing of niche shifts in size structured populations. Through this collaboration we were able to conduct the study at a depth that would not have been possible for any of us alone. We had a great deal of fun doing the work together and it led to great friendships.

HS: How did you decide to use largemouth bass (Micropterus salmoides) as the predator in these experiments? Did you consider other species?

EW & GM: This was an easy choice; the largemouth bass is by far and away the dominant predator in the local lakes we studied and the major predator on the bluegill.

 

HS:  Could you tell us a little more about the pond you chose to carry out this study, in Kellogg Biological Station? Why this particular pond?

EW & GM: Our study was conducted in one of the 18 experimental ponds located at the Kellogg Biological Station (see photo).  These ponds were constructed in the early 1970s and were designed to be identical in size and shape. After construction the ponds were quickly colonized with native fauna (e.g., plankton, aquatic insects) and flora (phytoplankton, macrophytes). For our experiment (conducted in 1979), we selected a pond that had a uniform border of cattail (Typha) vegetation around the pond perimeter and that had little submersed vegetation in the pond center, as we sought to create open-water and vegetated habitats for the fish. We removed by hand any remaining submersed vegetation in the pond center so as to create three distinct habitat types – open water, bare bottom, and vegetation – to simulate the natural habitats of a lake. We then divided the pond in half with fine-meshed netting that was anchored securely to the pond bottom (see photo). This was our first attempt at partitioning a pond and we wondered how well it would hold up in storms and the like (fortunately, it did!).

HS: Do you continue to do field work at the Kellogg Biological Station?  When was the last time you visited this place? Does the pond still exists and if yes, in what ways do you think it has changed?

EW & GM: We all continued to do field work at KBS after this study, but for varying amounts of time. During the 1980s, Werner did a number of studies in the KBS ponds examining species interactions in anurans (frogs). He left KBS in 1985 to join the faculty at the University of Michigan. Hall remained on the MSU faculty in East Lansing until his retirement and along with his students continued some work in the KBS ponds for the next decade. Mittelbach and Gilliam, who were graduate students in 1979, went on to jobs at Ohio State University and SUNY Albany (and then North Carolina State) respectively. Mittelbach came back to KBS as a faculty member in 1986, where he continued to work in the KBS ponds until his retirement in 2018.

Yes – the KBS experimental ponds still exist and are available to researchers outside and within MSU.  They have been an important facility for freshwater research for over 40 years.

HS: You say you conducted this study between 15th July and 28th September 1979. What are your most striking memories of this period – what was your daily routine, who helped you with the work etc.? It seems like it must have been a lot of hard work!

EW & GM: It was a tremendous amount of work, especially the sampling and enumeration of the prey from each habitat weekly on both sides of the pond partition  and the stomach analyses of the corresponding samples of each size class of bluegill from both sides of the partition. We recall, however, really looking forward to sampling days as the four of us authors, and several technicians and graduate assistants would gather in the morning at the pond lab at the experimental pond site. First, before disturbing the pond we would sample the open water plankton, the benthos in the center of the pond, and the prey associated with the cattail stems in the littoral zone, all done submersed in the pond in wet suits. Then we would seine both sides of the partition to obtain the fish samples for gut analyses. After that we retired to the pond lab to preserve the fish and spent several hours sorting the prey from the vegetation and benthic samples for preservation. What we remember is the constant stream of repartee, good natured razing and lots of laughter, and sometimes serious science discussions that ensued. It was great fun.

 

HS: I would like to ask you about the people you acknowledge – could you tell us a little more about who they were and how they helped?

EW & GM: Most of the individuals acknowledged were graduate students and post-docs at Kellogg at the time that read and commented on the manuscript. The exceptions were Laura Riley who did the vast majority of the sample processing during the winter and John Gorentz who coded the optimal foraging model for the bluegill foraging predictions.

 

HS: This study was conducted in late 1979 and published in 1983. Do you remember when and where you wrote most of this paper, and how long it took? Were the other authors involved in the writing?

EW & GM: The paper was written at Kellogg and we have no memory of how long it took. It wasn’t quick as the companion and much longer paper on the bluegill foraging behavior was being written simultaneously. Werner did the majority of the writing with much assistance from the other authors. Gilliam wrote the appendix.

 

HS: How did you decide to submit this paper to Ecology? Was it always planned as two companion papers? What do you remember about the peer-review of this paper?

EW & GM: Ecology was one of, if not the, premier journal for empirical papers in the field at the time so we naturally aimed to send it there. We don’t recall any major issues with reviewers.

 

HS: Today, 35 years after the paper was published, I would like you to reflect on validity of the main takeaways from this study in relation to where we stand currently:

“We have demonstrated that the bluegill is able to assess changes in both foraging profitability and predation risk. In the companion paper (Werner et al. 1983) we showed that temporal habitat shifts by all size-classes occurred when foraging rates in another habitat became greater than those of the habitat currently used. In this paper, we further demonstrated that small, vulnerable size-classes of the bluegill showed a marked shift in foraging behavior in the presence of the bass. These data provide experimental support for the hypothesis of Hall and Werner (1977)and Mittelbach (1981) that small bluegills in natural lakes are confined to weed beds because of a behavioural response to the greater predation risk in more open habitats. The quantitative predictions of habitat profitability also indicate that the small bluegills were evaluating this risk in the face of threefold greater foraging rates in the more open habitats. Whether this response also maximizes fitness remains to be tested.”

EW & GM: In retrospect, we were surprised at the generality of this tradeoff, this work, along with Andy Sih’s experiments and Peter Abrams theory, galvanized a large number of ecologists to quantitatively demonstrate the consequences of predation risk to the behavior of many, many taxa. This work clearly led to much refinement of our understanding of the tradeoffs that influence animal behavior, and how animals adaptively balance conflicting demands. But perhaps more exciting to us at the time were the implications to population dynamics and the structure of ecological communities. With respect to population dynamics, since foraging ability and predation risk are strong functions of size over ontogeny, there were many lessons regarding the dynamics of size-structured populations in this experiment. For example, we provided some theory and quantification in the paper regarding the magnitude of the direct mortality effects of predators compared to the indirect effects on mortality through (risk induced) reduced growth rates of prey. With respect to community process, the current active interest in trait-mediated indirect effects was clearly presaged in the results of this experiment. For example, the effect of the predator on behavior of the small size classes through resources to growth of the larger size classes is a trait-mediated indirect effect and motivated our subsequent work in that area. In both of the above examples, the indirect effects have often proven to be as large or larger than the direct effects of the predator, suggesting that these effects are critical to include in ecological theory.

We think that there are a number of larger lessons arising from this experiment that may be useful to graduate students, but they are largely part of the lore of ecological science and not uniquely revealed by this study. First, the combination of, and intimate interaction of,  field observations, theory, and experiment was very powerful and certainly contributed to the influence of this paper. In this context it is helpful for graduate students to realize that a lack of fit to a theoretical model that you have some confidence in is not a reason for depression or despair. Systematic deviations from theory always seem to lead somewhere interesting and important as you think deeper about the processes involved. This was the case in the application of the optimal foraging model to the natural lake pattern of bluegill habitat use where the model seemed to work for some size classes but not others.

 

HS: In the paper you say “The question also arises as to why the small fish exhibit such individual variation in their use of the vegetation in the presence of the predator (Fig. 1). We do not know if certain fish consistently spend more time in the vegetation than do others, i.e., if individuals tend to be risk averse or risk prone, or if this variation simply represents short-term (days, weeks)changes in habitat use by all individuals. This is an important problem, as these two hypotheses lead to very different ideas concerning individual behavior and fitness. We would expect that if risk-prone individuals existed, they would incur higher mortality rates but would also grow faster because of their use of the richer habitats. Thus, there should be a relation between size and habitat use. We noted earlier that there was no relation between body size among the small fish (ranging from 35 to 55 mm in length) and the fraction of their diet that came from the vegetation. The question of this individual variation in behavior deserves more detailed study.”

Has such work, on individual behavioural variation in sunfish, been done subsequently?

EW & GM: It’s very interesting to read these thoughts today, as they turn out to be quite forward looking. Today, the study of consistent individual differences in behavior within a species is a major research focus, often described as “animal personalities” or “behavioural syndromes”. At the time of our experiment (1979), however, this wasn’t yet a field of study and we were in fact puzzled as to why there was so much individual variation in the diets of the small bluegill (“why didn’t all individuals exhibit the ‘optimal’ behavior?”). Now we know much more about such individual personality differences and why a mixture of “shy” and “bold” phenotypes might coexist within a population. Interestingly, one of the pioneers of the study of animal personalities, David Sloan Wilson, who was for a time on the faculty at Kellogg, published an early study of “shy-bold” behavior in pumpkinseed sunfish some 10-years after our publication (Wilson et al. 1993, J. Comp. Psychol.). A more recent experimental study conducted in the KBS ponds documented the fitness consequences of boldness in juvenile and adult largemouth bass (Ballew et al. 2017, American Naturalist).

 

HS: In the paper you say “The observation that juvenile diets of several sunfishes are more similar than adult diets (Laughlin and Werner 1980, Mittelbach1984) is likely the result of this habitat restriction of juveniles. Thus if vegetation is relatively rare and/or resources in it low, risk of predation can create significant competitive bottlenecks for these species at this point in their life histories. Identifying these bottlenecks is obviously central to considerations of niche packing in such systems, but the effects of these sorts of bottlenecks have been virtually unexplored. Our understanding of community structure in fish and other organisms with size-structured populations will be very limited until we systematically explore the consequences of these sorts of interactions.

Has such work been done subsequent to this study?

EW & GM: Yes.  The study of juvenile competitive bottlenecks in size-structured populations was a major research focus for Werner, Mittelbach and colleagues for a number of years after our pond experiment (e.g., Werner and Gilliam 1984 ARES; Mittelbach and Osenberg 1993 Ecology). It would also be fair to say that this early pond study helped inspire the tremendously productive research program on size-structured interactions in fish and other animals by Drs. Lennart Persson and Andre deRoos (deRoos and Persson 2013 Princeton Univ. Press).

 

 

HS: In the 35 years following this paper, have you ever read this paper again? If yes, in what context?

EW & GM: Probably, and this excise prompted a reread.

 

HS: Would you count this as a favourite among all the papers you have published and why?

EW & GM: Neither of us would consider it our personal favourite but it would rank highly for both of us among the papers we have published.

 

HS: What would you say to a student who is about to read this paper today? Would you guide his or her writing in anyway? Would you point them to other papers they should read alongside? Would you add any caveats?

EW & GM: What often comes to mind in this context is the impact this paper had, and it continues to be highly cited, despite some methodological flaws. It was published at about the same time as Hurlbert’s influential paper on pseudoreplication. Our work was a classic case of pseudoreplication – a pond divided in half with one treatment assigned to one half and another treatment to the other half. We think it is interesting for a student to consider what criteria they would use in taking lessons from studies that are flawed in some respect. How does one evaluate what is lasting and useful despite methodological flaws and why? We subsequently replicated such experiments across ponds, but it is daunting to imagine replicating this experiment given the amount of work that went into one divided pond. Scientific knowledge increases by incremental steps. This experiment was a first step in revealing the profound impact of non-lethal effects of predators on their prey. Thus, it was valuable despite its methodological flaws.