Chain Reactions Linking Acorns to Gypsy Moth Outbreaks and Lyme Disease Risk

See allHide authors and affiliations

Science  13 Feb 1998:
Vol. 279, Issue 5353, pp. 1023-1026
DOI: 10.1126/science.279.5353.1023


In eastern U.S. oak forests, defoliation by gypsy moths and the risk of Lyme disease are determined by interactions among acorns, white-footed mice, moths, deer, and ticks. Experimental removal of mice, which eat moth pupae, demonstrated that moth outbreaks are caused by reductions in mouse density that occur when there are no acorns. Experimental acorn addition increased mouse density. Acorn addition also increased densities of black-legged ticks, evidently by attracting deer, which are key tick hosts. Mice are primarily responsible for infecting ticks with the Lyme disease agent. The results have important implications for predicting and managing forest health and human health.

Oak trees (Quercusspp.) produce large autumnal acorn crops (masting) every 2 to 5 years, producing few or no acorns during intervening years (1-4). Acorns are a critical food for white-footed mice,Peromyscus leucopus (1, 4-6). Mice are important predators of pupae of the gypsy moth, Lymantria dispar (1, 6-10). This introduced insect periodically undergoes outbreaks (11, 12) that defoliate millions of hectares of oak forests, decreasing tree growth, survival, and mast production (13). An abundance of acorns draws white-tailed deer, Odocoileus virginianus, into oak forests (14,15). Mice and deer are the primary hosts of the black-legged tick,Ixodes scapularis, which is the vector of spirochete bacteria (Borrelia burgdorferi) that cause Lyme disease in humans (16-18). Here we report the results of experimental removal of mice and addition of acorns, which demonstrate how acorn production is connected to gypsy moth outbreaks and Lyme disease risk.

Masting is associated with increased survival and breeding of mice in winter and spring (19), with peak densities occurring the following midsummer (1, 4, 6). High mouse density correlates with high predation rates on moth pupae (1,6), which may prevent low-density moth populations from increasing (1, 6-8). Conversely, mast crop failure correlates with low mouse densities and low rates of pupal predation the following summer (1, 4, 6), which may initiate moth outbreaks (7,9).

Moth populations at our research site reached peak densities in 1990, declined by four orders of magnitude to 0.2 egg masses ha−1 by 1992, and remained between 6 and 38 egg masses ha−1 in 1993–1994 (1). A large red oak (Q. rubra) acorn crop in autumn 1994 led to high mouse densities in summer 1995 (1). We took advantage of low moth and high mouse densities to remove mice during moth pupation, testing the chain of interactions linking acorns to mice to moths. Mice were removed from three grids of approximately 2.7 ha but were left unmanipulated on three control grids (20). Mouse densities did not differ between control and experimental grids in June 1995, just before mouse removal (Fig. 1; P = 0.18, paired t test) (21). Continuous live trapping reduced mouse densities on experimental grids to less than half those on control grids by the midpoint of a 32-day removal period in June–July coincident with female moth pupation (Fig.1; P = 0.018, one-tailed unpaired t test).

Figure 1

White-footed mouse densities on three control (○) and three experimental (•) grids as mean (±SE) MNKA ha−1, April 1995–October 1996, showing (i) 1995 densities before, during, and after the period of mouse removal from experimental grids (black bar) at the time of native female gypsy moth pupation (white bar); (ii) 1995 densities before and during acorn additions (arrows) to experimental grids, October–November 1995, when there were very low numbers of acorns produced on control grids; and (iii) 1996 densities after acorn additions to experimental grids in 1995. High 1995 mouse densities were associated with autumn 1994 masting, and mouse densities typically decline during winter (1).

Densities of late-stage moth larvae (22) did not differ between treatments at the start of the experiment (Fig.2A). Predation on female pupae was estimated by monitoring survival of the native population and by recording attacks on freeze-dried pupae (23). On control grids with high mouse densities, no living female pupae were found, and 100% of freeze-dried pupae were attacked by predators in 2 to 4 days, which is much less than the 13 days required for eclosion to the adult stage. Over 99% of attacks on freeze-dried pupae that could be attributed to vertebrates or invertebrates were caused by vertebrates, and 97% of vertebrate attacks where the predator species was identifiable were made by mice. In contrast, on experimental grids, 42% of native female pupae survived for 13 or more days, and 22% of freeze-dried pupae were unattacked at 14 days; 77% of these attacks were caused by vertebrates, with 89% being mouse attacks. The number of successfully eclosed female pupae and resulting egg masses on trees (24) was respectively 45-fold (Fig. 2B) and 43-fold higher (Fig. 2C) on experimental than on control grids. Comparison of control grids in 1995 and 1994 showed that oak masting in 1994 led to a 15-fold increase in July mouse densities, a 34-fold increase in mouse predation on freeze-dried pupae, and a decrease by a factor of 26 in moth egg mass densities (25).

Figure 2

Densities of gypsy moth life stages on or under burlap bands on trees on control grids (open bars) versus experimental grids (solid bars) where mice were removed. The bars show grid means and within-grid SEs. Across-grid control and experimental means (±SE) and statistical comparisons are also shown for each graph. (A) Number of living late-stage larvae per tree just before mouse removal. Control grids, 4.17 (±1.51); experimental grids, 3.36 (±0.68); P = 0.83, Mann-Whitney U test. (B) Number of female pupae per tree successfully eclosing to adults after mouse removal. Control grids, 0.008 (±0.005); experimental grids, 0.370 (±0.115); P = 0.02, one-tailed Mann-Whitney U test. (C) Number of egg masses per tree after mouse removal. Control grids, 0.006 (±0.003); experimental grids, 0.245 (±0.095); P = 0.02, one-tailed Mann-Whitney U test.

The increase in moth density that resulted from simulating mast failure by removing mice was similar in magnitude to that observed at the start of natural moth outbreaks, and the decrease in moth density on control grids was similar in magnitude to that previously observed after masting-induced increases in mouse density (1, 6).

Lyme disease in the northeastern and north central United States is transmitted to humans by black-legged ticks infected with B. burgdorferi (16, 26). Adult ticks feed and mate on white-tailed deer before dropping to the ground in autumn, laying eggs the following spring or early summer (17, 27). Larvae hatch in midsummer and are free from infection with B. burgdorferibecause of extremely low rates of transovarial transmission (28). White-footed mice are primarily responsible for infecting ticks with B. burgdorferi during the larval blood meal (29, 30). Larvae then molt to nymphs that overwinter on the forest floor. In spring or early summer 1 year after egg hatch, infected nymphs seek vertebrate hosts, including humans, and may transmit B. burgdorferi to the host at this blood meal (16, 17). The abundance of infected nymphs is the primary determinant of Lyme disease risk (16). Nymphs molt into adults that seek a deer host in the autumn. The location of deer in autumn determines the location of egg-laying adults and thus where host-seeking larvae should occur the following summer (1,31, 32).

In the autumn of mast years, deer spend more than 40% of their time in oak stands feeding on acorns but spend less than 5% of their time there in non-mast years (15). Larval tick density in oak forests reaches peak levels the summer after mast production but is low during the summer after mast failure (1), corresponding to predictions based on habitat use by deer. Increased densities of mice in oak forests during the summer after masting coincide with peak densities of larval ticks (1). Because mice are the principal reservoirs for Lyme disease spirochetes, high densities of infected nymphal ticks and a high risk of exposure to Lyme disease should occur 2 years after heavy acorn production (32).

We took advantage of mast crop failure in the autumn of 1995, when acorn production was lower by a factor of 18 than in 1994, to add acorns to the three experimental grids but not to the three control grids (33), testing the chain of interactions linking acorns to mice, deer, and ticks. We added more than 811,000 acorns (>3500 kg) to experimental grids at densities of 60 m−2 of oak canopy, approximating the 1994 acorn crop. We also simulated food caching by periodically supplementing mouse nest boxes on experimental grids with acorns, leaving boxes on control grids unsupplemented. Mouse density and reproductive status were monitored, and each month we measured the numbers of host-seeking ticks and ticks infesting mice (34). Although mice had been removed from the experimental grids in June–July, densities had returned to the levels measured on control grids by early October 1995, before acorn additions (Fig. 1; P = 0.98, unpaired t test).

Acorn addition significantly increased mouse densities from March–August 1996 (Fig. 1; P = 0.032, one-tailed F test), with approximately three- to sevenfold greater densities on experimental grids than on control grids in March–May. Densities converged on control values after August. From February to May, 75% of the adult mice on experimental grids ( n = 72) were in breeding condition versus 59% ( n = 17) on control grids ( P = 0.09, one-tailed χ2test). Although the low numbers of mice on control grids limited our ability to detect reproductive differences, the increase in mouse density caused by acorn addition was evidently mediated by both enhanced survival and reproduction. Our small-scale acorn additions may have had less effect on mouse populations than do natural masting events that typically occur over thousands of hectares (2,3). For example, predators could have been attracted to the locally elevated mouse densities (35) or mice could have emigrated to surrounding areas with lower mouse densities (32).

Densities of host-seeking ticks in August 1996, the time of peak larval host-seeking activity, were over eight times higher on acorn addition grids than on control grids (Fig.3A). Although deer habitat use was not directly monitored, larval tick distribution is largely determined by deer distribution the previous autumn (1). Consequently, adding acorns would have increased the time deer spent in autumn feeding on acorns in experimental grids as compared with control grids. The number of attached larval ticks per mouse was 40% higher on acorn-addition grids as compared with control grids (Fig. 3B). Adding acorns not only increased the densities of mice up to 9 months later but also increased larval tick burdens on mice because of the effects of acorn additions on deer habitat use and the effects of deer habitat use on the abundance of host-seeking larval ticks (36).

Figure 3

Tick densities on control grids (open bars) and experimental grids (solid bars) in August 1996 after acorn additions to experimental grids in October–November 1995. The bars show grid means and within-grid SEs. Across-grid control and experimental means (±SE) and statistical comparisons are also shown for each graph. (A) Number of host-seeking larval ticks per square meter. Control grids, 1.07 (±0.21); experimental grids, 8.59 (±2.93);P = 0.017, one-tailed paired t test. (B) Number of larval ticks per mouse. Control grids, 24.67 (±9.67); experimental grids, 34.58 (±13.64); P = 0.046, one-tailed paired t test.

Our results provide strong support for the idea that a chain of events links acorns to gypsy moth outbreaks and Lyme disease risk. The experiments demonstrate first that acorns determine overwinter survival, reproduction, and the resulting density of mice. Second, that high or low mouse density, at low gypsy moth population density, can respectively suppress or release moth populations through altered pupal predation. Third, that acorns determine larval tick densities by effecting the use of oak forests by deer, resulting in high densities of both host-seeking uninfected ticks and ticks parasitizing mice at the time when spirochete-infected mice are most abundant.

It may be feasible to predict the risk of contracting Lyme disease from infected nymphal ticks in oak forests on the basis of masting events, with the risk being greatest 2 years after an abundant acorn crop. Similarly, suppression or initiation of moth outbreaks may be predictable from mast production or failure when moth populations are at low densities. However, because other mortality agents, not mice, appear to control moth populations at higher densities (12, 37), outbreak initiation by mast failure and the collapse of mouse populations is probably necessary, but may not always be sufficient, to cause moth populations to rise to levels that cause defoliation. An additional, important long-term feedback to Lyme disease may exist, because moth defoliation reduces acorn production and can reduce oak abundance in forests (13). Our studies indicate that attempting to simultaneously prevent moth outbreaks and minimize Lyme disease risk, by using silvicultural practices that alter acorn production, would be unlikely to succeed because decreasing the likelihood of moth outbreaks could increase the risk of Lyme disease and vice versa.

Ecologists have hotly debated the relative importance of direct versus indirect species interactions as a cause of contingent ecological outcomes (38). Our studies clearly demonstrate that both gypsy moth dynamics and Lyme disease risk have contingent outcomes arising from a complex chain of strong pairwise interactions among taxonomically diverse species that are all interconnected within an ecosystem.

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


View Abstract

Stay Connected to Science

Navigate This Article