The buzz on trophic cascades

Herbivores, by their nature, damage plants in natural ecosystems and in agricultural systems.  And predators, by their nature, do a lot of damage to herbivores, either by eating them, or by harassing them in ways that cause them to change their behavior, or in some cases change their morphology or physiology (these are called nonconsumptive effects).  The indirect effect of a trophic cascade in which predators damage herbivores which damage plants, is that predators can benefit plants by their detrimental effect on herbivores.

Much of the research on nonconsumptive effects has focused on aquatic systems because the predator cues are easy to manipulate in the laboratory.  Simply let a predator hang out in a water tank for a while, and then add the predator tank water to a tank with a possible prey item, and study the prey’s response. But there has been little work on nonconsumptive effects in terrestrial systems.  While there has been some research on how auditory cues emitted by terrestrial predators affect vertebrate herbivores, there has been almost no work on how auditory cues affect invertebrate herbivores.  This is surprising, because invertebrates cause enormous damage to agricultural systems. Evan Preisser and his students wondered whether the beet armyworm caterpillar, Spodoptera exigua, a voracious herbivore on many commercially important crops, responded to buzzing emitted by an important predator, the caterpillar-hunting paper wasp (Mischocyttarus sp.). More important, they tested whether the response was substantial enough to have an impact on caterpillar mortality, and subsequent plant development.

Four Spodoptera caterpillars chomp on a soybean leaf. Credit: Michasia Dowdy, University of Georgia,

Perhaps the biggest challenge was technical.  The researchers needed to come up with a mechanism for delivering an auditory cue to one group of caterpillars that would not be detected by any other nearby group.  They tried various conformations, including separating the cages with soundproofing foam, which, unfortunately, was not soundproof to wasp buzzes.

One of the failed attempts at auditory isolation. Unfortunately, the auditory stimulus was detectable up to two boxes away from the source. Credit: Zachary Lee.

Nothing worked until one of the students suggested using boxes that dry ice was shipped in, reasoning correctly that it should have good insulating properties.  The decibel meter failed to detect any sound from adjacent boxes.

This worked! Credit: Zachary Lee.

Having solved the soundproofing problem, the researchers raised 36 groups of five caterpillars in small cups filled with 25 grams of caterpillar diet. Each cup was placed in a box and subjected to one of three treatments: no-sound control, recorded buzzing of a non-predatory mosquito, or recorded buzzing of a predatory wasp.  The volume was the same for both sound treatments. Each tape went for 12 hours per day, with 2 seconds on, followed by 6 seconds off.  The researchers found that survival was substantially lower for caterpillars that received the wasp treatment (top graph below).  Also, caterpillars that survived the wasp treatment took, on average, longer to develop (bottom graph below), though that difference was not statistically significant.

Survival (top graph ), weight (middle) and time to pupation (bottom) of Spodoptera caterpillars subjected to no sound (green bar), mosquito buzz (yellow) and wasp buzz (red). Different letters above bars indicated statistically significant differences between treatments.

Preisser’s graduate student, Zachary Lee, took the lead in organizing the field experiment.  The researchers wanted to know whether the negative effect of wasp buzzes that they detected in the laboratory had real consequences for agricultural systems.  They surrounded each tomato plant (72  in all) with a mesh bag (to keep the caterpillars in and other insects out), and placed an average of 96 newborn caterpillars on each plant.  Each group of four plants surrounded a speaker that emitted either no sound (control), mosquito buzzing, or wasp buzzing, which were broadcast at levels that caterpillars would experience when an insect was 5 cm away from them.  Each sound was played in a loop of 1 minute on, followed by 10 minutes off, for 12 hours per day.  Lee and his colleagues let the experiment run for 3 weeks, by which time all caterpillars had either pupated or died. They harvested each plant, and calculated the percentage of leaves that were damaged by caterpillars.  Then they dried each plant, including the roots, and weighed them.

Field experiment with four tomato plants positioned equidistant from one central speaker. Each group of four received one of the experimental treatments. Credit: Zachary Lee.

Plant leaves associated with wasp buzzing received the least damage, leaves on control plants received the most damage, and leaves on plants with mosquito buzzing received intermediate damage. Aboveground mass was greater in wasp treated plants than in controls, so the sound of wasp buzzing helps to protect the tomato plants against voracious caterpillar herbivores.

Indirect effects of no sound (green bar), mosquito buzz (yellow) and wasp buzz (red) on tomato plants, via the effects of these treatments on Spodoptera herbivory. Different letters above bars indicated statistically significant differences between treatments.

The researchers did not study caterpillar behavioral changes because these caterpillars are easily disturbed, either freezing or dropping off of plants when approached.  Lee and his colleagues point out that we know very little about how invertebrates, in general, respond to sound cues, as their survey of the literature on prey response to sound cues showed that 181/183 experiments used vertebrate prey.  Given how widespread invertebrates are in agricultural systems, and in ecosystems in general, we need more studies to get a better handle on how invertebrates respond to sound, and most important, how their response influences agricultural systems and ecosystem structure and functioning.

note: the paper that describes this research is from the journal Ecology. The reference is Lee, Z.A., Cohen, C.B., Baranowski, A.K., Berry, K.N., McGuire, M.R., Pelletier, T.S., Peck, B.P., Blundell, J.J. and Preisser, E.L., 2023. Auditory predator cues decrease herbivore survival and plant damage. Ecology, p.e4007. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2023 by the Ecological Society of America. All rights reserved.

Bat benefits

Chiroptophobia, the fear of bats, is widespread throughout the world, but also subject to the cultural biases of different regions.  In Europe, bats were historically associated with the Devil, evil spirits and witchcraft. Dante’s Inferno describes the Devil’s wings as being very much like bat wings in form and texture. Vampirism was well established in Eastern European folklore before Bram Stoker’s depiction of Count Dracula routinely transforming himself into a huge vampire bat. Other regions of the world are historically more nuanced in their perspectives. For example, in Madurai, India, worshippers of the Muni god revere the Indian Flying Fox, Pteropus medius, and protect bat colonies from harm.  In Pudukkottai, Pteropus bats are guardians of sacred groves, while in Bihar these same bats bring wealth. But in the Punjab region of India magicians use bat blood to do malevolent magic, and across the border in some regions of Pakistan, bats are associated with evil witchcraft. This is only the tip of the humans/bats cultural iceberg.  For a thorough consideration, you should go to

A vampire bat flies through the night. Credit: Uwe Schmidt, CC BY-SA 4.0 <

Can ecologists help us resolve this conundrum? The answer is also nuanced.  On the negative side, bats live in dense colonies, are very social and relatively long-lived.  Taken together, these traits allow them to harbor many pathogens, including rabies and coronaviruses, which may be passed on to humans.  On the positive side, bats consume many insects including those that carry diseases.  Recent research has also shown that bats consume insects that eat crops.  Thus, in agricultural ecosystems, there exists a trophic cascade in which bats reduce insect abundance, which leads to an increase in crop production.  Armed with this knowledge, and a recent finding by Tim Divoll that some bats eat insects that defoliate oak and hickory trees, Elizabeth Beilke and Joy O’Keefe decided to explore how important bats were in forested ecosystems.  Does a similar trophic cascade exist, in which bats reduce herbivorous insect abundance, which leads to an increase in tree production?

Underside of oak leaf showing caterpillars hard at work. Credit: Lis Kernan.

To explore the trophic cascade hypothesis, Beilke and O’Keefe set up a three-year experiment (during 2018 – 2020) in the Yellowwood State Forest in Indiana, USA. They built 6 X 7 X 7 meter exclosures that were covered with nylon-mesh netting large enough to allow most insects but small enough to exclude bats. 

Researchers set up a bat exclosure in the forest. A series of ropes and pulleys allowed them to raise the netting each evening and take it down in the morning. Credit: Elizabeth Beilke.

Each experimental unit was a control exclosure without netting, and an experimental exclosure in which the netting was raised during the night to exclude bats, and lowered during the day so that birds could forage.  This allowed the researchers to attribute any treatment effects exclusively to nocturnal animals – basically bats.  They set up seven pairs of exclosures each year; unfortunately one exclosure was destroyed when three trees fell on it during a violent storm. Within each exclosure Beilke and O’Keefe monitored 9 or 10 oak and hickory seedlings during the treatment period.  They counted the number of insects on oak and hickory leaves in May, when the enclosures were set up, and August, when they were taken down.

Basic experimental design. Control plots allowed both bats and birds, while experimental plots allowed birds but excluded bats.

Did bat exclusion increase insect density?  The answer is a resounding “yes” with bat exclusion associated with a 300% increase in insect density in comparison to control plots.

Mean number of insects (+95% confidence intervals) per seedling at the beginning of the field season (left Figure a) and at the end of the field season (right Figure B). Gray dots represent data generated by a statistical model.

Most important, did this increase in insect density lead to greater defoliation of the trees?  Beilke and O’Keefe found that both oaks and hickories suffered greater defoliation when bats were excluded.  The impact on oaks was substantially greater than the impact on  hickories. 

(Figure a – left) Mean defoliation when bats were permitted (top) and excluded (bottom). (Figure b – middle) Mean (+95% Confidence interval) defoliation when bats were permitted (control) or excluded from oak trees. (Figure c – right) Mean (+95% Confidence interval) defoliation when bats were permitted (control) or excluded from hickory trees.

There is some evidence that bats tend to eat more insects that feed on oak trees than insects that feed on hickories. For example, the most common bat in the forest, the eastern red bat, consumed three times more oak-defoliating than hickory-defoliating insect species. Thus bats could be affecting forest composition by preferentially protecting oaks over hickories.  However, given recent declines in bat abundance from white-nose fungus and habitat destruction by humans, losing this protection may be contributing to oak declines in the Eastern United States.

Beilke and O’Keefe point out that bats can negatively influence herbivorous insects directly or indirectly.  Direct effects involve eating herbivorous insects, which are positioned directly on leaves.  Indirect effects can include eating the non-herbivorous adult insects (e.g. butterflies and moths) that produce the herbivorous caterpillars. In addition, some insects are sensitive to the ultrasonic sounds emitted by echolocating bats and may tend to avoid areas populated by bats. Overall, the bat/herbivorous insect/tree trophic cascade results in forests benefitting bats by providing food and places to roost, while bats benefit forested ecosystems by protecting them from herbivory. We now have one more reason to embrace our local bats.

note: the paper that describes this research is from the journal Ecology. The reference is Beilke, E.A. and O’Keefe, J.M., 2023. Bats reduce insect density and defoliation in temperate forests: An exclusion experiment. Ecology104(2): e3903 Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2023 by the Ecological Society of America. All rights reserved.

Tropical trophic cascade slows decomposers

In the rough and tumble natural world, consumers such as lions, lady bugs, llamas and lizards get most of the press, while producers such as peas, pumpkins and phytoplankton come in a close second.  Consumers earn their name because they get their energy from consuming other organisms, while producers produce their own energy (using photosynthesis or chemosynthesis) from inorganic molecules.  Often ignored in this ecosystem structural scheme are decomposers, which get their energy from breaking down the tissue of dead organisms.  They should not be ignored.  Much of the energy transferred through ecosystems passes through decomposers.

One reason they are overlooked is that most decomposers are tiny. Some of the largest decomposers are detritivores, which actually eat the dead materials (detritus), in contrast to other microbial decomposers such as bacteria and fungi.  Shredders are detritivores commonly found in streams and rivers; these aquatic insects eat portions of dead leaves and, in the process, shred them into much smaller pieces that energize other decomposers. Many researchers had noted that shredders were relatively rare in tropical streams, in part because there are many other larger consumers in the ecosystem that are willing to eat dead leaves and any shredders associated with them. Thus Troy Simon and his colleagues expected that shredders, such as the caddisfly, Phylloicus hansoni, would play, at best, a minor role in the streams they studied in the Northern Range Mountains in Trinidad.


A typical headwater stream located in the Northern Range mountains of Trinidad. Waterfalls in the uppermost reaches of these streams act as a barrier to the upstream movement of guppies, but not killifish and crabs, which can move over land during periods of heavy rain. Credit: Joshua Goldberg.

We will discuss interactions between several species in these aquatic systems.  Trees are important producers as they shed leaves into the streams; these leaves are broken down by shredders such as the aforementioned caddisflies and also microbial decomposers.   The major consumers are omnivorous crabs, Eudaniela garmani, which eat leaves and caddisflies (and many other items), and two fish species. Killifish, Anablepsoides hartii, eat caddisflies, other invertebrates and also the occasional small fish (including fish eggs).


Hart’s killifish (Anablepsoides hartii) are primarily insectivorous and major consumers of leaf‐shredding caddisflies. Credit: Pierson Hill.

Guppies, Poecilia reticulata, are much smaller than killifish, maxing out at 32 mm long in comparison to the killifish maximum length of 100 mm.  But guppies are much more omnivorous, feeding on leaves, leaf-shredding insects and even killifish eggs and larvae.


Male (left) and female (right) Trinidadian guppy (Poecilia reticulata). Guppies are omnivorous, feeding broadly on detritus as well as plant and animal prey, including young killifish. Credit: Pierson Hill.

Amazingly, killifish can disperse over land, as can crabs (less amazingly).  This allows them to bypass barrier waterfalls during wet periods, which results in them being the only large consumer species above waterfalls in many Trinidad streams.  Guppies lack killifish dispersal abilities, so they are often confined to stream reaches below significant waterfalls.  These species, and their consumption patterns are highlighted in the figure below.


Diagram of the two detrital-based food webs.  Above the waterfall is the KC reach, named after its two important consumers, killifish and crabs.  Below the waterfalls is the KCG reach, named after its three important consumers, killifish, crabs and guppies. Arrows show direction of energy flow within the ecosystem.

Simon and his colleagues wanted to know how interactions among all of these species influenced the rate of leaf decomposition.  The researchers constructed identical-size leaf packs of recently fallen leaves of the Guarumo tree, Cercropia peltata, and attached them to copper wire frames within each reach of the stream.  They periodically harvested a subset of the packs and measured the amount of decomposition by drying and weighing the leaves, and comparing this weight to the starting weight of the leaf pack.  In addition, they collected all invertebrates > 1 mm long from each leaf pack and identified them to species or genus.

To control the consumers involved in each interaction, Simon and his colleagues constructed underwater electric exclosures which created an electric field that convinced all fish and crabs to exit (and stay out) within 30 seconds of being turned on, but did not influence invertebrates in any detectable way.  Killifish are active day and night, guppies only during the day, and the researchers believed that crabs were active primarily at night. The researchers set up four treatments: control (C) with 24 hour access to consumers, experimental (E) with 24 hour exclusion of consumers, day-only exclusion (D) and night-only exclusion (N).  The researchers expected that the day-only exclusion treatments would selectively exclude guppies, while night-only exclusion would selectively exclude crabs. They then placed the leaf packs into each exclosure, turned on the current, and ran the experiment for 29 days.  Five replicates of each treatment were done above and below the waterfalls.


Electric exclosures established in the stream. Leaf packs were tied to the copper frame and periodically harvested over the 29 days of the experiment. Rectangular tiles shown in treatment frames were part of a separate study. Credit: Troy N. Simon.

We’re finally ready for some data.  The two graphs on the left represent the downstream reach below the waterfalls, where killifish, crabs and guppies are naturally present (KCG).  The two graphs on the right represent the upstream reach above the falls, where only killifish and crabs are naturally present.  There was no evidence in the downstream reach that excluding consumers influenced decomposition rates (top left graph).  However, when consumers were present (C treatment) in the upstream reach, decomposition rates were reduced by about 40% in comparison to treatments when consumers were partially (D and N) or completely (E) excluded (top right graph).


Mean (+SE) for (a,b) decay rate of Cecropia peltata leaves (percentage of mass lost per day) and (c,d)  biomass of Phylloicus hansoni (milligrams of dry mass per gram of Cecropia). 24-hour treatments allow full macroconsumer access [control (C)] or completely exclude macroconsumers [electric (E)]. Twelve-hour treatments exclude access to either diurnally active [day (D)] or nocturnally active [night (N)] macroconsumers. Different letters above the bars indicate statistically significant differences between the treatments.

The two bottom graphs above look at the biomass of the caddisfly, Phylloicus hansoni, which was easily the most abundant macroinvertebrate within the leaf packs.  There was no significant difference in caddisfly abundance below the waterfall regardless of treatment (bottom left graph above).  Above the waterfalls, caddisfly abundance was severely depressed in the controls (C) where killifish were free to feed on them (bottom right graph).

One piece of evidence that killifish ate caddisflies and depressed their abundance was that surviving caddisflies were much smaller in the control treatment leaf packs than in any of the experimental treatment leaf packs.  This suggests that  killifish with unimpeded access to caddisflies were picking off the largest individuals.


Mean (+SE) caddisfly length in mm (y-axis) for each treatment, 

These findings support the hypothesis that a trophic cascade prevails in the KC reach, in which killifish eat caddisflies, thereby slowing down decomposition. But in the KCG reach, guppies eat killifish eggs and larvae and compete with them for resources, thereby reducing killifish abundance, and interfering with the establishment of a trophic cascade.

Lastly, the researchers explored whether the same trophic cascade operated in upper reaches but not in lower reaches of other streams in the area. Surveys of six streams indicate a definite “yes” answer, with Cecropia decay rate and caddisfly biomass much lower in the upper reaches.


(Top) Mean (+SE) decay rate for Cecropia peltata
leaves (percentage of mass lost per day) and (b) caddisfly biomass (milligrams of dry mass per gram of Cecropia) in the landscape study (n = 6 streams). Different letters above bars indicate statistically significant differences  between treatments.

Surveys of each stream indicated that killifish were much more abundant in the upper reaches where guppies were not found, but guppies were much more prevalent in the lower reaches than were killifish.  These findings indicate that this detrital-based trophic cascade, with killifish eating caddisflies and thereby slowing down decomposition, is a general pattern in the upper reaches of these tropical streams.  However, Simon and his colleagues caution us that different streams will have different groups of organisms playing different ecological roles.  Thus the presence of detrital-based trophic cascades will depend on the particulars of which species are present and how abundant they are in a particular stream.

note: the paper that describes this research is from the journal Ecology. The reference is Simon, T. N., A. J. Binderup, A. S. Flecker, J. F. Gilliam, M. C. Marshall, S. A. Thomas, J. Travis, D. N. Reznick, and C. M. Pringle. 2019. Landscape patterns in top-down control of decomposition: omnivory disrupts a tropical detrital-based trophic cascade. Ecology 100(7):e02723. 10.1002/ecy.2723. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2019 by the Ecological Society of America. All rights reserved.


Mystifying trophic cascades

Within ecosystems, trophic cascades may occur when one species, usually a predator, has a negative effect on a second species (its prey), thereby having a positive effect on its prey’s prey. Today’s example considers the interaction between a group of predators (including several fish species, a sea snail and a sea star) their prey (the sea urchin Paracentrotus lividus) and sea urchin prey, which comprise numerous species of macroalgae that attach to the shallow ocean floor. These predators can negatively affect sea urchin populations either by eating them (consumptive effects), or by scaring them so they forage less efficiently (nonconsumptive effects). If sea urchins are less abundant or less aggressive foragers, the net indirect effect of a large population of fish, sea snails and sea stars will be an increase in macroalgal abundance.

Maldonado Halo

A large sea urchin grazing in a macroalgal community.  Notice the white halo surrounding the urchin, indicating that it has grazed all of the algae within that region. Credit: Albert Pessarrodona.

Many humans enjoy eating predatory fish, and we have overfished much of the ocean’s best fisheries including the shallow temperate rocky reefs (4 – 12 m deep) in the northwest Mediterranean Sea. Removing these predators has caused sea urchin populations to explode, overgrazing their favorite macroalgal food source, and ultimately leading to the formation of urchin barrens – large areas with little algal growth, low productivity and a small nondiverse assemblage of invertebrates and vertebrates.


A sea urchin barrens whose macroalgae have been overgrazed by sea urchins. Credit: Albert Pessarrodona

Albert Pessarrodona became interested in this trophic cascade after years of diving in the Mediterranean. He noticed that in Marine Protected Areas, predatory fish abound and there are few visible urchins and lots of macroalgae. In nearby unprotected areas where fishing is permitted, urchins graze out in the open brazenly, and urchin barrens are common. He also wondered whether a second variable – sea urchin size – might play a role in this dynamic. Were large sea urchins relatively immune from predation by virtue of their large size and long spines, allowing them to forage out in the open even if predators were relatively common?


Interactions investigated in this study.  (a) Predators consume either small (left) or large (right) sea urchins (consumptive effects). (b) Sea urchins eat macroalgae. (c) Predators scare small or large sea urchins, reducing their foraging efficiency (nonconsumptive effects). (d) Predatory fish indirectly increase macroalgal abundance.

Pessarrodona and his research team used field and laboratory experiments to explore the relationship between sea urchin size and their survival and behavior in high-predator-risk and low-predator-risk conditions. High-risk was the Medes Islands Marine Reserve, which has had no fishing since 1983 and boasts a large, diverse assemblage of predatory fish, while low-risk was the nearby Montgri coast, which has a similar habitat structure, but allows fishing. The researchers tethered 40 urchins of varying sizes to the sea bottom (about 5m deep) in each of these regions, left them for 24 hours, and then collected the survivors to compare survival in relation to body size in high and low-risk conditions. They discovered that large urchins were much less likely to get eaten than were small urchins, and that the probability of getting eaten was substantially greater in the high-risk site.


Probability of being eaten in relation to sea urchin size (cm) in high-risk (blue line) and low-risk (green line) habitats.

Pessarrodona and his colleagues followed this up by investigating whether the relatively predation-resistant large urchins were less fearful, and thus more likely to forage effectively, even in high-risk sites. Previous studies showed that sea urchins can evaluate risk using chemical cues given off by other urchins injured in a predatory attack, or given off by the actual predators. To explore the relationship between these cues and sea urchin behavior, the researchers put either large or small sea urchins into partitioned tanks with an injured sea urchin. Water flowed from one partition to the other, so the experimental sea urchins received chemical cues from the injured urchins. They also had a group of sea urchins placed in similar tanks without any injured sea urchins as controls. The experimental sea urchins were given seagrass to feed on, and the researchers calculated feeding rates based on how much food remained after seven days.

Small sea urchins were not deterred by the presence of an injured urchin (left graph below), while large sea urchins drastically reduced their feeding rates in response to the presence of an injured urchin (middle graph). This was startling as it flew in the face of the commonsense expectation that small sea urchins (most susceptible to predation) should be most fearful of predator cues. The researchers repeated the experiment (under slightly different conditions) placing an actual predator (a fearsome sea snail) on the other side of the partition. Again, large urchins showed drastically reduced foraging rates (right graph below).


Sea urchin responses to predation risk cues in the laboratory. When exposed to injured urchins – symbolized as having a triangle cut out – (A) small urchins did not reduce their grazing rate, while (B) large urchins drastically curtailed grazing. (C) When exposed to a predatory snail on the other side of a partition, large urchins sharply curtailed grazing. n.s = no significant difference, **P<0.01.

It turns out that large sea urchins are the critical players in this trophic cascade because they do much more damage to algal biomass than do the smaller urchins (we won’t go through the details of that research). The question then becomes how this plays out in natural ecosystems. Do consumptive and non-consumptive effects of predators in high-risk sites reduce sea urchin abundance and reduce the foraging levels of large sea urchins so that macroalgal cover is greatly enhanced? Pessarrodona and his colleagues surveyed high-risk and low-risk sites for sea urchin density and algal abundance. They set up 45 quadrats (40 X 40 cm) at each site, measured each sea urchin’s diameter, and estimated the abundance of each type of algae by harvesting a 20 X 20 cm subsample from each quadrat and drying and weighing the sample.

The findings were striking. Small and large sea urchins were much less abundant at high-risk sites than at low-risk sites (left graph below). At the same time, macroalgae were much more abundant at high-risk sites than at low-risk sites (right graph below).


(Left graph) Density of small and large sea urchins in high-risk and low-risk habitats. (Right graph) Biomass of macroalgae of different growth structures in high-risk and low-risk habitats. Canopy algae are taller than 10 cm, while turf algae are lower stature. Codium algae are generally not grazed by sea urchins. **P<0.01, ***P<0.001.


Summary of interactions.  Arrow width indicates relative importance.

To summarize this system, predators reduce small sea urchin abundance by eating them (consumptive effects), and reduce large sea urchin foraging by intimidating them (nonconsumptive effects). The net indirect effect of predators on macroalgae is a function of these two effects. Large sea urchins are the major macroalgae consumers, but, of course, large sea urchins develop from small sea urchins.

The $64 question is why large sea urchins fear predators so much, while small (more vulnerable) urchins do not. The quick answer is that we don’t know. One possibility is that small sea urchins may be bolder in risky environments since they are more vulnerable to starvation (have fewer reserves), and also have lower reproductive potential since they are likely to die before they get large enough to reproduce. In contrast, large sea urchins can survive many days without food because of their large reserves. In addition, large urchins are close to sexual maturity, and thus may be unwilling to accept even a small risk to their well-being, which could interfere with them achieving reproductive success.

note: the paper that describes this research is from the journal Ecology. The reference is Pessarrodona, A.,  Boada, J.,  Pagès, J. F.,  Arthur, R., and  Alcoverro, T. 2019.  Consumptive and non‐consumptive effects of predators vary with the ontogeny of their prey. Ecology  100( 5):e02649. 10.1002/ecy.2649. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2019 by the Ecological Society of America. All rights reserved.

Kelp consumption curtailed by señorita

Miranda Haggerty was diving through a kelp forest, and noticed that many kelp bore a large number of tiny limpets that were housed in small scars that they (or a fellow-limpet) had excavated on the kelp’s surface. This got her thinking about how these scars might affect the kelp, and equally relevant, whether there were any limpet predators that might lend the kelp a hand (or a mouth) by removing limpets.

Jerry Kirkhart

A limpet grazes on a kelp frond. Credit: Jerry Kirkhart

Feather boa kelp (Egregia menziesii) is a foundation species within the subtidal marine system off the California coast, providing food and habitat for many species that live on or among its fronds. The tiny seaweed limpet, Lottia insessa, specializes on feather boa kelp, grazing on its fronds and living within the scars. Many invertebrates and fish live within the kelp forest, but the most abundant fish is the señorita, Oxyjulis californica. Haggerty wondered whether the señorita might benefit the kelp (directly) by removing limpets, or (indirectly) by scaring limpets away – what ecologists call a trait-mediated indirect interaction.


The señorita – a fearsome predator of limpets.  Credit: Miranda Haggerty

The first order of business was to determine whether the limpets were actually harming the kelp.  Haggerty and her colleagues approached this in two ways.  First they chose 94 kelp plants from kelp forests off the California coast.  From each individual they chose one grazed and one ungrazed frond (each 3 m long). Grazed fronds averaged 5-10 scars and at least 2 limpets per meter of length.  Every three weeks they visited their kelp to score for broken fronds. In 29 of 30 cases, the grazed frond broke before the ungrazed frond (in the remaining cases the entire plant was missing, or both fronds broke and the researchers could not tell which had broken first).


Photo of feather boa kelp showing grazing scars, including one housing a limpet (left).  Diagram of feather boa kelp showing multiple fronds (right).

But the researchers were concerned that perhaps limpets chose to graze on weaker fronds, so the breakage was not caused by grazing scars, but by limpet choice.  To account for this concern, Haggerty and her colleagues chose 43 ungrazed kelp plants, placed three  limpets on one frond, and chose a second, equal-sized frond as an unmanipulated control. Once again, they visited their plants every three weeks, and discovered that grazed fronds broke first in all 20 pairs that the sequence of frond breakage could be determined.  Clearly, limpet grazing is bad news for feather boa kelp.

How does the señorita fit into this system? The researchers designed a laboratory experiment to address this question.  They used 10 large tanks (1700 L), and set up five different experimental treatments to compare direct effects of predation, and indirect effects of predator presence, on limpet grazing, and ultimately on kelp survival. To isolate the direct effects of predation from the indirect effects of predator cues on limpets, Haggerty and her colleagues placed four kelp fronds into fish exclosure cages, which were housed in the large tanks, and placed three limpets onto some of these fronds.  To mimic actual predation (CE treatment in Table below), they removed limpets by hand at a constant rate typical of señorita predation. For the NCE treatment (testing indirect effects of predator presence) they introduced señorita into the large tank so the limpets experienced the predator cues, but were not eaten. The different treatments are summarized in the table below. These experiments ran for one week and each treatment was replicated 10 times.

HaggertyTableFinalEach day the researchers monitored the number of limpets and grazing scars.  After one week, Haggerty and her colleagues counted the number of grazing scars, and measured the breaking strength of each frond by clamping the frond’s end to a table and pulling on the opposite end with a spring scale until it broke. They then recorded the amount of force needed to break the frond.


Clamped kelp frond whose breaking strength has been tested.  Notice that the frond broke at a grazing scar (right). Credit Miranda Haggerty.

Not surprisingly, the predator control (PC) kelp (limpets present without señorita) had the most scars and lost the greatest amount of tissue.  Kelp receiving the consumptive predator effect treatment (CE) had fewer scars and lost less tissue than PC.  But interestingly, kelp receiving NCE and TPE treatments had significantly fewer scars than the CE kelp, and were statistically indistinguishable from each other.  Thus, in the laboratory, the presence of señorita cues (NCE treatment) was more important than actual predation (CE treatment) in reducing kelp scarring and tissue consumption (top and middle graph below).  As a result, the NCE treated kelp were stronger (had greater breaking strength) than were the CE treated kelp (bottom graph below).


Mean (+ standard error) number of grazing scars (top), mass of tissue consumed (middle) and breaking strength (bottom) of kelp in response to five experimental treatments. CE = consumptive effect, NCE = non-consumptive effect, TPE = total predator effect, PC = predator control, LC = limpet control. Different letters above bars indicate significant differences between the means when comparing treatments.

Haggerty and her colleagues replicated this experiment, with a few experimental design modifications, in a field setting.  As with the laboratory experiment we’ve just discussed, the researchers found a very strong non-consumptive effect. The researchers suspect that these limpets respond to chemical cues emitted by their señorita predators. They could not respond to many types of sensory cues because they lack auditory organs, and the experimental design prevented fish from transmitting any shadows (visual cues) or vibrational cues. In addition previous studies have shown that some limpet species use chemoreception for predator avoidance, foraging and homing. However, the nature of this chemical cue is yet to be discovered for this predator-prey system.


Schooling señorita. Credit: Miranda Haggerty

Trophic cascades occur when the effects of one species on another species cascade down through the ecosystem. In this case, señorita predators directly and indirectly reduce limpet density, which increases the survival of kelp – a foundation species for this ecosystem. The researchers point out that this trophic cascade only occurs in the southern feather boa kelp range, because señorita are absent further north.  We don’t know if limpets have other predators in the northern range, but we do know that the kelp are structurally more robust further north, so they (and the ecosystem) may be relatively immune to limpet-induced destruction.

note: the paper that describes this research is from the journal Ecology. The reference is Haggerty, M. B., Anderson, T. W. and Long, J. D. (2018), Fish predators reduce kelp frond loss via a trait‐mediated trophic cascade. Ecology, 99: 1574-1583. doi:10.1002/ecy.2380. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2018 by the Ecological Society of America. All rights reserved.

Life and death in a diminutive ecosystem

Ecosystems are big things, as they encompass an entire community of organisms and the nonliving factors (such as nutrients and water) that interact with the community. So we’re accustomed to thinking about the Serengeti as an ecosystem, as it includes (among many things) the large animals, such as lions, wildebeest and buffalo that live there, the animals and plants they eat, and the soils and nutrients that feed these plants.

But ecosystems can also be tiny. Let’s think about an individual tank bromeliad, Quesnelia arvensis, which can hold up to 3 liters of water in tanks formed where individual leaves come together. Gustavo Romero has identified over 140 species of invertebrates that live within these natural tanks, including large predators such as damselfly and tabanid larvae, and many species of smaller predators (mesopredators) including a diverse group of chironomid midges. The larger predators eat the smaller predators, while predators of both sizes eat a very diverse group of detritivores – animals that feed on the remains of dead organisms. The terrestrial fauna in the immediate vicinity are spiders. Visitors from the surrounding forest ecosystem include 12 bird species and 6 frog species, which forage on larvae within the bromeliads.

Cantorchilus longirostris on bromeliad (Quesnelia arvensis) leaf

Long-billed marsh wren perches on the tip of bromeliad leaf.  This bird can use its long beak to probe for invertebrates living within the bromeliad tank. Credit: Crasso Paulo Bosco Breviglieri

Crasso Paulo Bosco Breviglieri and his colleagues had previously done research demonstrating how insectivorous birds hanging out near bromeliads inhibited dragonflies from ovipositing (laying eggs) within the bromeliad tank. As these birds were much larger than the animals living within the tanks, Breviglieri and Romero hypothesized that the birds would focus on eating the largest items offered to them by this ecosystem. By removing the largest items (the top predators), birds increase the biomass of the prey of these top predators, including detritivores. Thus bird predation should indirectly increase decomposition rate and nutrient availability.

Breviglieri food web

Effects of birds and frogs on bromeliad trophic cascades. Solid arrows are direct effects and dashed arrows are indirect effects (for example frogs eat top predators, thereby indirectly increasing mesopredators).  Wider arrows are stronger effects.

Trophic cascades, a process in which the effects of consumption within an ecosystem cascade down from higher to lower feeding levels, can be difficult to study. The problem is that one favorite approach is to remove predators (the top trophic level) and see if prey abundance increases while the food of these prey decreases, and so on. This is extremely challenging when top predators are lions or wolves and ecosystem area encompasses thousands of kilometers, but much easier when predators are birds or frogs, and each ecosystem is a tank bromeliad. Simply put a cage over a tank bromeliad and presto!, no birds or frogs can get in.

Dr. Crasso Paulo B. Breviglieri building the cages that isolated the bromeliads

Breviglieri with a caged bromeliad. Credit: Jennifer Tezuka

Breviglieri and Romero collected 30 tank bromeliads from the forest, and meticulously cleaned each plant to remove all organisms and organic matter. They filtered and homogenized the water from the bromeliads, and returned 1 liter of water to each plant so that each plant began the experiment with the same quantity of water and microorganisms. The researchers then added equal numbers of organisms to each bromeliad from all of the trophic levels, ranging from apex predators such as damselflies down to detritivores, such as shredders, which eat dead plant leaves and begin the break down process. They also added 10 leaves to each tank for detritivore consumption and further decomposition.

For their experiment, Breviglieri and Romero had three different treatments, with 10 bromeliads per treatment: (1) caged, with each bromeliad enclosed within a steel mesh that allowed insects through but restricted birds and frogs, (2) open-cage control, with each bromeliad only partially enclosed so predators had free access, (3) uncaged control. They returned these to the field at 40 meter intervals, and allowed 155 days to pass.

Larva of zygoptera on bromeliad (Quesnelia arvensis)leaf

Bromeliad with a damselfly larva (top predator) that for unknown reasons has climbed out of the tank onto a leaf.  A bird flew to a nearby perch, but the alert damselfly dove back down into the tank, earning a 9.6 from the judges. Credit: Crasso Paulo Bosco Breviglieri

After 155 days, Breviglieri and Romero collected all of the bromeliads, and identified, counted and weighed (dry weight) all of the organisms. They discovered that the dry mass of invertebrates was much greater in the caged treatments than either control (Figure A). The abundance of apex predators (damselflies and tabanids) did not increase; but the size of individuals increased dramatically (Figure B). Mesopredators increased in abundance (Figure C), while shredder abundance declined sharply (Figure D). Shredder larvae forage on sediment and are a favorite damselfly food item, so it is not surprising that shredders declined, given the sharp increase in damselfly size, and presumably appetite.


Lower shredder abundance in the caged bromeliads led to a sharp decline in decomposition rates (left graph below). In theory, this should make fewer nutrients available to the bromeliads and reduce bromeliad growth. In contrast to expectations, caged bromeliads actually grew more leaves (right graph below), despite the reduction in decomposition rates. Breviglieri and Romero remind us that the greater mass of larvae were producing a much greater mass of fecal matter and prey carcasses, both of which are very nutrient rich. Also, higher predation rates can cause some insects to mature and leave their tank at a smaller size, consuming fewer nutrients while in the larval form, and leaving more nutrients for each plant to use for its own growth.


Decomposition rate measured as detrital mass lost (left graph), and growth rate measured as new leaves grown by the bromeliads (right graph), for caged, open-caged and uncaged controls.

Clearly, there are many unanswered questions about this trophic cascade. For example, why don’t the number of top predators increase in abundance when birds and frogs are excluded? When I asked him this question, Breviglieri suggested that two processes could explain this finding. First, top predators eat smaller larvae of their own species. Second, female insects can chemically sense the presence of predators in these bromeliads, and refrain from ovipositing in plants hosting large predators.

Perhaps most important, can we extend the conclusions from these small ecosystems to larger ecosystems? In nature there are many analogous ecosystems in which predators have strategies for crossing boundaries and influencing ecosystem processes. For example, many birds dive into lakes searching for fish and invertebrates. Moving in the opposite direction, banded-archerfish spit out water jets to dislodge invertebrates from adjacent vegetation into the water, and crocodiles leave rivers to grab and consume convenient gnus. In these systems, as in bromeliads, predators cross ecosystem borders to feed, and it is important for us to understand if there are any general patterns in how these visitors from the outside affect ecosystem functioning.

note: the paper that describes this research is from the journal Ecology. The reference is Breviglieri, Crasso Paulo Bosco, and Gustavo Q. Romero. 2017. Terrestrial vertebrate predators drive the structure and functioning of aquatic food webs. Ecology. doi:10.1002/ecy.1881.  It was published online on June 12, and should appear shortly in print. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2017 by the Ecological Society of America. All rights reserved.

Timely trophic cascades

While many of us appreciate oysters as delectable delights, we may underestimate the environmental benefits they also bring to the table. As filter feeders, they remove vast quantities of organic debris from the water, and as reef builders they protect our shorelines from violent wave action.


Oyster reef. Credit: WFSU, Public Media

Of course, humans are not the only animals to enjoy eating oysters. For example, along portions of the Florida coast dominated by the reef-building oyster Crassostrea virganica, the mud crab, Panopeus herbstii, is a major consumer of juvenile oysters. In some locations, the average abundance of these voracious crabs can exceed 10 adults/m2 of reef. But all is not food and gravy for these crabs, as lurking in nearby burrows are equally voracious crab-eating toadfish, Opsanus tau. When toadfish are detected, the mud crabs will hide within the protective matrix of oyster shells and sediment that form the reef.


A mud crab hiding among a cluster of oysters. Credit: WFSU, Public Media

By consuming mud crabs, toadfish are indirectly protecting oysters from being eaten. Ecologists call this a consumptive effect (CE). But David Kimbro and his colleagues have also shown than toadfish, by their mere presence, can also protect oysters by scaring the crabs into hiding. Since, in this case, they are not consuming the crabs, ecologists call this a non-consumptive effect (NCE). Together, CEs and NCEs should both increase oyster survival. More surviving oysters lead to higher overall feeding by oysters, which lead to more oyster poop, and more organic matter deposited into the sediment below. Ecologists call this type of relationship a trophic cascade, because the effects on one species cascades down through the ecosystem. In this case, increasing toadfish will decrease crabs, thereby increasing oysters and sediment organic matter. Conversely, decreasing toadfish should increase crabs, thereby decreasing oysters and sediment organic matter.


Toadfish/mud crab/oyster/sediment organic matter (SOM) cascade. Dotted arrows are indirect effects

Kimbro and his colleagues wanted to explore this trophic cascade in more detail. They set up an experiment with 24 artificial reefs (made out of natural materials, except for the surrounding fence), which included 35 L of live oysters. They supplied each reef with 0, 2, 4, 6, 8 or 10 live crabs, and provided half of the reefs with a caged toadfish. They then measured oyster survivorship in relation to crab density in the presence or absence of predators.


Setting up an artificial reef. Credit: WFSU, Public Media

The graphs below summarize their findings. The first thing to notice is that mud crabs were bad news for oysters, as survivorship plummeted when mud crabs were abundant. However, early in the experiment (graphs A and B) having a toadfish around helped out considerably. Oysters survived much better in the presence of toadfish (triangles and dotted curve) than they did without toadfish (circles and solid curve). But by the middle of the experiment (Graphs C and D), the toadfish no longer helped. Interestingly, by the end of the experiment (Graph E) the toadfish was once again helping the oyster’s cause, as survivorship was again greater in the presence of toadfish than in its absence. Realize that the difference between the dotted and solid curve is a measure of the NCE, as the toadfish are not eating the crabs (because they are caged). So we can conclude that there was a strong NCE early on, which waned in the middle of the experiment and then returned by the end of the experiment.


A second finding is that the reef grew (expanded) when there were no crabs present, but that even two crabs were enough to reduce reef growth to zero. In addition sediment organic matter was greatest when there were either none or only two crabs present in the reef. Four or more crabs in the reef reduced the deposition of sediment organic matter. These findings were not influenced by the presence or absence of toadfish.

This is a complicated system, but we (and toadfish, crabs and oysters) live in a complicated world. And there are several other complications that I have not even mentioned! We might argue that the crabs may habituate (get accustomed) to these toadfish, so that by the middle of the experiment, the toadfish NCE had worn off. That begs the question of why the NCE returned towards the end of the experiment. Kimbro suggests that at the beginning of the experiment, the novelty of the predator cue probably caused strong NCEs. But by the middle of the experiment, the crabs became hungry and chose to forage regardless of predator cue. Finally, towards the end of the experiment, the crabs, having filled up on juvenile oysters, opted to hide rather than forage when toadfish were present. Whatever the reason, these findings caution us that if we want to understand trophic cascades, we need to consider the dimensions of both space and time.

note: the paper that describes this research is from the journal Ecology. The reference is Kimbro, D. L., Grabowski, J. H., Hughes, A. R., Piehler, M. F., & White, J. W. (2017). Nonconsumptive effects of a predator weaken then rebound over time. Ecology 98(3): 656-667. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2017 by the Ecological Society of America. All rights reserved.