Holling 1959 & Weed et al. 2017

Citations:

Holling, C.S. 1959. The components of predation as revealed by a study of small

Mammal predation of the European pine sawfly. The Canadian Entomologist 91: 293-320.

Weed, A.S., Ayres, M.P., Liebhold, A.M., and Billings, R.F. 2017. Spatio-temporal

Dynamics of a tree-killing beetle and its predator. Ecography 40: 221-234.

Blog author: Bailey McNichol

Author Background:

Crawford Stanley “Buzz” Holling was a Canadian ecologist and an Emeritus Scholar and Professor at the University of Florida. He was one of the founders of the field of ecological economics, and throughout his career merged systems theory and ecology with simulation modeling and policy review. He was also one of the ecologists (along with Brian Walker and others) that introduced important concepts including resilience and adaptive management to the field of ecology. The paper on predation of pine sawflies by small mammals came out of his dissertation research in Canada.

The lead author on the companion paper, Aaron Weed, is an ecologist in the Inventory and Monitoring Division of the National Park Service. He is the lead on long-term monitoring of aquatic and terrestrial resources for the Northeast Temperate Network. His research interests lie at the intersection of the ecology and management of both native and invasive insects and the effects of disturbance on forest ecology. He has worked extensively on the role of density-dependent vs. density-independent factors on population fluctuations of insects, from local to landscape levels.

Holling paper:

Summary of paper, main questions, and results:

Holling begins the paper with a discussion on predator-prey interactions, including the effects of prey selection on predation, shifts in prey populations that occur in the absence of predators, and the role of an environment’s carrying capacity for prey in determining the importance of predation over the large-scale. He indicates that while previous work has offered logical explanations and insights into these dynamics, none of the studies have described the mechanisms responsible for controlling populations. He then introduces the predator-prey relationships between 3 small mammal predators – the masked shrew, short-tail shrew, and the deer mouse – and the prey species the European pine sawfly as a relatively easy model system, given the uniformity of the ecosystem and ease of identifying successful predations events on the pine sawfly cocoons. 

This paper focuses exclusively on direct effects of predation, leaving interactions between indirect population controls (parasites, disease, and predation) for a separate paper. Holling introduced and discussed five components of predation (based on Leopold 1933) that may affect prey mortality: 1) density of prey; 2) density of predators; 3) characteristics of prey; 4) density and quality of alternative foods for predator; and 5) characteristics of the predator. He discussed these in relation to the study, concluding that the simplicity of the experimental design allows us to only majorly consider the prey and predator densities as having an effect on predation. The experiment showed that the number of sawfly cocoons opened by each of the three mammal species increased as the density of cocoons increased until the maximum daily consumption per animal per day was reached. He then discusses four classes of prey characteristics: 1) caloric value of prey; 2) length of prey exposure; 3) attractiveness of the prey; and 4) strength of stimulus used by predators to find prey. Holling’s lab experiments only directly assessed the 4^thcharacteristic by experimentally varying the depths of sand covering the sawfly cocoons (and thus the olfactory detection of cocoons by the deer mice) and found that the maximum number of cocoons opened per day decreased as the depth that cocoons were buried at increased.

Overall, he concludes that his study is novel in that it shows the importance of both: a) a change in the numbers of prey consumed per predator (“functional response”), and b) a change in the density of predators (“numerical response”). He also argues that density-independent factors (e.g., climate) can affect numbers of animals within a population but cannot regulate populations (this, then, only occurring due to a density-dependent interaction). Holling ends the paper with a discussion of (and his own support for/against) hypothesized reasons that populations fluctuate, including Thompson’s and Nicholson’s mathematical theories of population regulation and the influence of types of predation.

Summary of experimental methods: 

To account for the environmental densities of predators, prey, and “destroyed” prey (i.e., pine sawflies that have been eaten), Holling measures each of these groups in terms of their numbers per acre. He assessed the numbers of each of the 3 small mammal predators through capture-recapture techniques, where animals are trapped, marked, released, and then population sizes are estimated based on the number of animals that are recaptured. The number of pine sawfly cocoons (available prey) was estimated shortly after the larval drop from the trees and involved sub-sampling leaf litter and duff within a given radius of the crowns of host Scots and jack pine trees. Sawfly cocoons were then collected in September prior to adult emergence to assess the amount of predation that occurred (via destroyed cocoons), along with examination of the stomach contents of trapped mammals. A viral pesticide treatment was also applied within the pine stands, with varying concentrations applied to control populations of the sawflies. Holling also performed laboratory experiments with three male deer mice, varying the densities and depths of sawfly cocoons and introducing two alternative food sources (less palatable dog biscuits and more palatable sunflower seeds) to assess preference and predation rates.

Weed et al. paper:

While Holling drew most of his conclusions from field measurements of predation and experimental lab studies (on a count of predation density per area basis), Weed et al. used a modeling approach to assess factors that influence the stability of predator-prey populations over spatiotemporal scales. They were interested in determining the role of local-level processes on landscape-level patterns of abundance of both predator and prey populations. The prey species in their study was the southern pine beetle (SPB), a primary bark beetle that can kill living pine tree hosts when it occurs at high population densities, and the predator was the clerid beetle (Thanasimus dubius). Although numerous generalist predators can prey upon SPB, the clerid beetle is by far its most important predator and has been shown to have a significant control on population levels throughout SPB’s native range. Previous studies have shown that clerid beetles have a cyclical delayed density-dependent effect on SPB populations – the prey beetle’s population size increases, predator beetles arrive several weeks later to an area with high densities of SPB, the clerid beetles feed aggressively on SPB, and population levels eventually decrease. However, because these oscillations in predator and prey population levels do not hold up over longer time scales (i.e., multi-decadal), Weed et al. aimed to: 1) evaluate factors influencing local population stability using mechanistic models, and 2) estimate the spatial synchrony of the clerid beetle-SPB system.

The data that Weed et al. used were the counts of trapped SPB and clerid beetles from the annual springtime trapping survey, during which traps are deployed throughout the southeastern United States to monitor predator and prey population levels so that management can be focused in areas that have high counts of SPB and low counts of clerids. They analyzed data collected over ~16 years in 95 forests using models that described temporal behavior in the per capita growth rates of SPB and clerids, evaluating the current vs. previous abundance of both the prey and predator species, and including an error term for density-independent effects (i.e., weather). They also characterized the spatial cross-correlation between SPB and clerids, as well as effects of forest composition and structure – specifically, the influence of the density of pines (the required host tree for SPB) – on abundances of both species.

The authors found that population fluctuations of SPB and clerid beetles were tightly temporally linked throughout the region, and that although populations were subjected to density-dependent feedbacks, the dynamics were not periodic as previously found. Synchrony in abundances of both species was highly correlated on an annual basis in 65% of forests, but there was some temporal variation in the degree and pattern of this synchrony. The model including immediate effects of predator and prey abundance best described the population dynamics of both species; prior SPB abundance had a strong negative affect on current population growth (although not at large spatial scales), and prior clerid abundance also had an overall negative feedback. This indicates that effect of clerid beetle abundance on SPB populations over longer time scales was quite variable across the 95 forests. Overall, their results demonstrate a clear correlation between the population dynamics of SPB and clerids over a large spatial scale, but refute the idea that populations of clerid beetles have a delayed density-dependent response to SPB densities.

My Thoughts:

While I agree that the Holling study provided a relatively uncomplicated system for assessing predator-prey dynamics between the mammals and pine sawflies, I wish he would have elaborated more on why the Lincoln’s index vs. the Hayne’s method (which is no longer used) for estimating population sizes were used for the deer mice vs. the shrews. This may have been elaborated upon in his dissertation, but these techniques should have been clarified so that the reader could better understand the advantages and limitations of each. His technique for sub-sampling pine sawfly cocoons seemed more rigorous (and straightforward), although it is impossible to get a completely accurate estimate over a large scale. However, the introduction of an external control on the pine sawfly larva populations (spraying of a virus) was a non-trivial confounding variable. Despite the blocking of different areas of the plantation based on pesticide concentration, I think it is very difficult to isolate how much changes in overall predation over the course of the study might have been affected by changes in predator behavior (which he asserts is negligible) as a result in reduced prey health/quality. I was also somewhat skeptical of his comparison of only 3 deer mice included in the lab experiments to predation that occurred in the field – particularly because two of the mice only had access to older cocoons (which he says DID affect consumption), and only a single mouse was used to study the effects of cocoon depth on predation.

I thought that the Weed et al. manuscript was a very robust, novel, and quantitatively driven approach to characterizing predator-prey dynamics of a well-studied system in the forest entomology literature. While I think that their model selection technique and the parameters that they included were more than adequate for addressing their questions, I felt that their broad conclusions on the importance of local-scale weather patterns and host pine density (AKA basal area) were somewhat unsubstantiated. Fine-scale temperature and precipitation patterns may play a role in SPB densities changing, but there is a strong body of empirical literature that refutes this idea, so for Weed at al. to mention this without any data in their models and based on 1-2 citations is tenuous at best. Additionally, although it is a reasonable assumption that pine basal area should have a strong effect on SPB densities (because they need a high density of host trees to cause an outbreak), their results showed a weakly significant effect of average pine basal area on prey population levels.