Dynamical Role of Predators in Population Cycles of a Forest Insect: An Experimental Test

See allHide authors and affiliations

Science  13 Aug 1999:
Vol. 285, Issue 5430, pp. 1068-1071
DOI: 10.1126/science.285.5430.1068


Population cycles occur frequently in forest insects. Time-series analysis of fluctuations in one such insect, the southern pine beetle (Dendroctonus frontalis), suggests that beetle dynamics are dominated by an ecological process acting in a delayed density-dependent manner. The hypothesis that delayed density dependence in this insect results from its interaction with predators was tested with a long-term predator-exclusion experiment. Predator-imposed mortality was negligible during the increase phase, grew during the year of peak population, and reached a maximum during the period of population decline. The delayed nature of the impact of predation suggests that predation is an important process that contributes significantly to southern pine beetle oscillations.

Ecologists have been trying to solve the puzzle of population cycles for at least three-quarters of a century (1). One class of ecological system that seems particularly prone to population oscillations is insects attacking forest trees (2, 3). Because these insects cause widespread economic damage, the causes of their outbreaks have been a focus of intensive research. Despite this effort, however, the biological mechanisms that drive oscillations are not yet well understood even in the best-studied systems (2,4). Here we present results of a long-term field experiment designed to test the hypothesis that cycles in one forest insect, the southern pine beetle (SPB) Dendroctonus frontalis, are driven by the beetle's population interaction with its predators (we use the term “predators” in the broad sense that includes parasitoids, but not pathogens).

During the 1980s, SPB outbreaks in pine forests of the southern United States were thought to be driven by exogenous (density-independent) factors, namely, fluctuations in climate (5, 6). However, our analysis of SPB activity in eastern Texas, USA, during 1957 to 1987 did not reveal any statistically significant effects of climatic variables on the rate of population change (7). Time-series analysis indicated that SPB fluctuations were driven primarily by endogenous (density-dependent) factors: ∼80% of the variance in the rate of population change was explained by a joint action of current and lagged population densities. The evidence for second-order dynamics [that is, delayed density dependence; see (8) for the definition of process order] was strong, because regression of the rate of population change on lagged density was highly significant (P < 0.0001) and it alone explained 55% of the variance (7). First-order endogenous factors (those that act in an undelayed manner) or exogenous influences are not unimportant; the former may prevent oscillations from getting out of hand, whereas the latter add stochastic irregularity. However, to understand why SPB populations oscillate, we should look to those mechanisms that act in a delayed density-dependent manner, because theory states that lags in regulation promote the possibility of cycles (9).

Several ecological mechanisms can generate second-order dynamics: maternal effects (10), food quantity (11) or quality (12), pathogens (13), and specialist predators or parasitoids (14, 15). Although time-series analysis cannot distinguish between these alternatives, it suggests how to formulate rival hypotheses in quantitative and testable terms [the predictions of the experiment described below were published in (7)].

The question we addressed experimentally was, what is the dynamical role of predation in the SPB cycle? A demonstration that predators inflict substantial (or even overwhelming) mortality at any particular point in time does not tell us whether predators are responsible for the oscillation or not. We need to determine how the predator impact changes with time, or more precisely, with cycle phase. Three broad outcomes can be distinguished, corresponding to the hypotheses that predators are (i) an exogenous, (ii) a first-order endogenous, or (iii) a second-order endogenous factor (Fig. 1).

Figure 1

Possible dynamical effects of predation. In all graphs, the dotted line indicates SBP population density during the course of a single oscillation, peaking in year 4. The solid line indicates the survival rate that determines the course of the oscillation (for simplicity, we assumed fecundity to be constant). The broken line indicates the survival rate when predators are excluded, and the separation between the solid and broken lines measures the predation impact. (A) The expected or mean predation impact does not vary with density. If predator impact has a large stochastic component, then predators will act as an exogenous factor; if predation impact does not vary with time, then predators are a null factor. (B) Predation acting as a first-order process, with the greatest impact occurring during the peak year. (C) Predation acting as a second-order process, with the greatest impact occurring during the period of population collapse. If predation were the dynamical factor completely responsible for population change, then the broken line in (C) would be completely flat.

In the first case, there is no dynamical feedback between prey density and the predation impact. The average predator-induced mortality may be very high and still predators would have no dynamical impact, simply reducing the intrinsic rate of population increase to a lower value. Fluctuations in predator-imposed mortality will affect prey density in a stochastic manner, but they cannot drive a regular oscillation. In the second case, predators respond to changes in prey population without a significant lag time. The dynamical role of predators, therefore, is stabilizing rather than causing oscillations (16). Generalist predators may act in this manner, reducing the amplitude of oscillations or preventing diverging oscillations. Only in the third case, when acting in a delayed density-dependent manner, are predators actually causing the oscillation. Note that the three scenarios represent extremes of a continuum, because it is possible for the predator community to act in a mixed manner (for example, a mixture of generalist and specialist predators would act in a manner intermediate between cases 2 and 3).

To determine which of the three scenarios (or some combination of them) characterizes the predation impact in the SPB system, we performed a long-term study that measured predation impact by experimentally excluding all natural enemies of the SPB (17). The 5-year-long study covered a complete increase-peak-decrease cycle (Fig. 2). In 1990 and 1991 the survival of SPB brood inside cages did not differ from that outside cages (Fig. 3A), indicating negligible predation impact during the increase phase (18). Predators imposed detectable mortality during the peak year (1992), but numerically the strongest effect of predation was observed during the first year of decline, 1993 (19). We observed a qualitatively similar pattern in the effect of predators on the SPB ratio of increase (Fig. 3B); but this measure of predation was statistically significant during both decline years, and not during the peak year. Thus, both measures indicate that the predator complex acts primarily as a second-order (that is, delayed) process, with perhaps an admixture of a weaker first-order impact. The second-order effect is probably due to arthropod natural enemies, including several species of parasitoid wasps and predacious beetles (20). One predator, the clerid beetle Thanasimus dubius, appears to be a particularly promising subject for further investigation. This predator is a specialist on bark beetles, capable of inflicting significant mortality on SPB (21), and its densities exhibit oscillations that are phase-shifted with respect to those of SPB (Fig. 2). A particularly interesting feature of this predator is its tendency to go into an extended diapause (22). It is known that long developmental delays can have a destabilizing effect on dynamics (23,24).

Figure 2

Population numbers of the SPB (circles, solid line) and one of its important natural enemies, the clerid beetleThanasimus dubius (triangles, dashed line), during 1989 to 1994, as measured by a network of pheromone-baited traps within Kisatchie National Forest.

Figure 3

(A) Survival of bark beetles, measured by the proportion of eggs surviving to become emerging adults: protected from predation (inside cages, broken line) and exposed to predation (outside cages and on control trees, solid line). The dotted line indicates the course of the outbreak (from Fig. 2). (B) Natural logarithm of the SPB ratio of increase, defined as the number of emerging adults divided by the number of attacking adults (the previous generation). Same notation as in (A). For statistical tests, see (18).

Our finding that predators in the SPB system act as a second-order process should be tempered by two caveats. First, our results do not preclude the possibility that other mechanisms (for example, maternal effects, food quantity or quality, and pathogens) also contribute to the delayed density-dependent pattern of SPB dynamics. Nevertheless, given the consistent and forcible impact of predators (50% decrease in survival and 50 to 70% decrease in the ratio of increase), it is clear that they play an important role in driving SPB oscillations. A twofold survival differential per generation translates into a 32- to 64-fold differential per year (because there are five to six SPB generations per year).

Second, our experiment was designed to determine the dynamical role of the whole predator complex. Thus, we do not yet know which particular enemies play an especially important role in causing SPB oscillations. Currently, our results implicate T. dubius as a particularly numerous and effective predator of the SPB. However, the SPB is a native “pest” of pines, and there is a diverse predator community associated with it (20). Only continuing empirical work coupled with modeling can yield quantitative estimates of the relative importance of different SPB predators.

Ecologists have used three general approaches to investigate potential mechanisms that can explain population cycles: general ecological theory based on mechanistic models (25, 26), analyses of time-series data (8), and field experiments (27, 28). No single approach in isolation can resolve the issue of why a particular population exhibits density oscillations. As our study and another recent study (28) illustrate, greatest progress may be achieved when all three approaches are used synergistically in investigations of population cycles.

  • * To whom correspondence should be addressed. E-mail: peter.turchin{at}


View Abstract

Stay Connected to Science

Navigate This Article