Global transport of organisms by humans provides novel resources to wild species, which often respond maladaptively. Native herbivorous insects have been killed feeding on toxic exotic plants, which acted as ‘ecological traps’1,2,3,4. We document a novel ‘eco-evolutionary trap’ stemming from the opposite effect; that is, high fitness on an exotic resource despite lack of adaptation to it. Plantago lanceolata was introduced to western North America by cattle-ranching. Feeding on this exotic plant released a large, isolated population of the native butterfly Euphydryas editha from a longstanding trade-off between maternal fecundity and offspring mortality. Because of this release—and despite a reduced insect developmental rate when feeding on this exotic—Plantago immediately supported higher larval survival than did the insects’ traditional host, Collinsia parviflora5. Previous work from the 1980s documented an evolving preference for Plantago by ovipositing adults6. We predicted that if this trend continued the insects could endanger themselves, because the availability of Plantago to butterflies is controlled by humans, who change land management practices faster than butterflies evolve6. Here we report the fulfilment of this prediction. The butterflies abandoned Collinsia and evolved total dependence on Plantago. The trap was set. In 2005, humans withdrew their cattle, springing the trap. Grasses grew around the Plantago, cooling the thermophilic insects, which then went extinct. This local extinction could have been prevented if the population had retained partial use of Collinsia, which occupied drier microhabitats unaffected by cattle removal. The flush of grasses abated quickly, rendering the meadow once again suitable for Euphydryas feeding on either host, but no butterflies were observed from 2008 to 2012. In 2013–2014, the site was naturally recolonized by Euphydryas feeding exclusively on Collinsia, returning the system to its starting point and setting the stage for a repeat of the anthropogenic evolutionary cycle.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
P. R. Ehrlich introduced M.C.S. to E. editha in 1967; C. D. Thomas, H. L. Billington, D. Ng, L. E. Gilbert and J. L. B. Mallet helped to initiate the project. D. D. Murphy performed the 2010 census. C.L. Boggs, R. A. Steward, J. L. B. Mallet and C. S. McBride critiqued the manuscript. The 2013 Integrative Biology Faculty Merit Review Committee, University of Texas at Austin, provided incentive to complete the study. J. Schneider and B. Schneider, the Drudge family, Clear Creek Tahoe and The Nature Conservancy allowed access to the site.Reviewer information
Nature thanks T. Oliver, M. Friberg and the other anonymous reviewer(s) for their contribution to the peer review of this work.
Extended data figures and tables
a, The butterfly habitat is a single, isolated, spring-fed wet meadow in the centre of the photograph, surrounded by non-habitat for the butterflies: dry sagebrush scrub and coniferous forest. b, Distribution in the meadow-edge ecotone of the principal hosts, Plantago and Collinsia, plus the minor host Penstemon. c, Typical difference in phenology between Plantago and Collinsia in May 2014. In the foreground are red, senescent Collinsia plants that are edible to the insects but which will die within a few days; behind them is a single green, budding Plantago that will remain edible until after all E. editha larvae have entered diapause. d, Hatching egg clutch on Collinsia cotyledon in hot, dry microhabitat.
Extended Data Fig. 2 Changes in distribution of early stages of E. editha (eggs or larvae) from 1982 to 2007.
Data were added by hand to the GoogleMaps image. Most stars represent several groups. For example, in 1989, 23 groups were found on Plantago and one on Penstemon. The restricted distribution in that year followed a bottleneck in 1988 after record-breaking cold in January, without the usual insulating snow cover. Schneider’s Meadow is at 1,700 m elevation: nearby towns at lower elevations recorded −25 °C on 1 January 1988 (Minden, 1,444 m elevation) and −20 °C on 18 January (Carson City, 1,424 m). Note the recolonization of Collinsia as the insects expanded back into the distribution of Collinsia in 1990 and 1993. Larval groups recorded in 1988 and 1989 were clustered around an attractive nectar source (Wyethia sp.); it is possible that adults attracted to this nectar in 1988 had survived as larvae on Collinsia in 1987–1988 and then, as adults, laid eggs in 1988 on Plantago adjacent to nectar. This possibility prevents us from making a definite conclusion that the population would have become extinct if eggs laid in 1987—before the bottleneck—had been placed only on Collinsia. Data for 2005 exist and closely resemble those for 2002.
Extended Data Fig. 3 Effects of cessation of grazing: Plantago plants embedded, whereas Collinsia plants are unaffected.
Data are provided in Table 2. a, Plantago at Schneider in 1984, exposed to full sunlight and physically acceptable to ovipositing E. editha. b, Meadow edge in May 2007, after cattle removal. In the foreground is Plantago habitat with thick grasses; in the background is Collinsia habitat not grassed-in, with barren spaces between the sagebrush. c, Collinsia in May 2007, unaffected by the embedding that simultaneously affected the Plantago plants shown in d and e. Embedding in grasses not only cooled the Plantago plants (Extended Data Table 2) but also rendered them hard to find, both by butterflies seeking oviposition sites and by larvae seeking food.
a, Natural egg clutch laid in May 2007 on Plantago. The plant is pushing through winter thatch, and would have been unlikely to be acceptable to ovipositing butterflies before cattle removal, when plants similar to the one in Extended Data Fig. 3a were available. b, Communal web spun after recolonization. Second-instar larvae on Collinsia at Schneider in May 2014. This is a single group of larvae, probably stemming from a single oviposition event; there were nine such groups, all on Collinsia. Unexpectedly, this group is not on the most exposed Collinsia available.
Extended Data Fig. 5 The return of mostly exposed Plantago plants after anthropogenic lushness abated.
Photographs were taken in 2014, but Table 2 shows that they could have been taken in 2008 or subsequently.
A single fire-enhanced Collinsia at McGee Creek (east of Bishop, California) is still blooming. There is a small web of E. editha larvae at its base from a naturally laid egg clutch. The fifteen senescent individual Collinsia lying on the ground represent a haphazard sample gathered from unburned microsites within 2 m of the enhanced individual.
Extended Data Fig. 7 Strength and direction of the oviposition preferences of butterflies sampled at Schneider in 1983 and 2005 + 2007.
The number over each bar is the sample size of biologically independent samples: individual butterflies captured in the field. The discrimination phase is the length of time for which the insect would search, during which it would consistently accept the preferred host and consistently reject the second-ranked host. At the end of this phase, if it does not succeed in ovipositing, the insect enters an acceptance phase after reaching the level of oviposition motivation at which either host would be accepted (whichever was next encountered). Insects in the blue 1–4 column would search for 1–4 h, during which only Collinsia would be accepted. If they failed to find Collinsia within 4 h, they would subsequently accept either host, until actual oviposition occurred. Green central bar shows butterflies without preference. Sample size for 2005–2007 is smaller than in Fig. 1a because we include on Fig. 1a (and here omit) five butterflies for which we determined the direction of preference, but not the strength.