Forest ecosystems in eastern North America face multiple threats including habitat loss and fragmentation, invasions, overabundance of native species, nutrient deposition, and climate change. While each stressor may have independent detrimental effects on native biota, stressors often co-occur and are likely to have synergistic effects. The objective of this project was to assess interactive effects of deer, nutrient addition, and presence of non-native plants, earthworms, and a root-weevil Barypeithes pellucidus Boheman on the demography of rare understory plant species. The four plant species at risk (SAR) selected—Aristolochia serpentaria L, Agrimonia rostellata Wallr, Carex retroflexa Muhl. ex Willd, and Trillium erectum L.—differ in their predicted response to deer and earthworm activity.

Technical Approach

A combination of experimental manipulations and mark recapture observations were used in the field to assess the effects of deer, earthworms, slugs, root weevil abundance, and invasive plants to vital rates of SAR. Stage-specific matrix projection models were developed for each SAR and then life table response experiments (LTRE) were conducted to estimate how different stressors affect contributions of different vital rates to changes in population growth rates.

Investigations were conducted at U.S. Army Garrison West Point, New York, and in common gardens at Cornell University. Twelve field sites were selected at West Point, which varied according to abundance of three target non-native plant species [Alliaria petiolata (M. Bieb.) Cavara & Grande, Berberis thunbergii DC, and Microstegium vimineum (Trin.) A. Camus] and according to earthworm density (eight and four sites with high and low earthworm density, respectively). At each site, two 30 x 30 m plots were established, one open and one protected by a deer-proof fence. From 2009-2012, vegetation, earthworm, slug, and B. pellucidus populations were monitored at each plot. To estimate SAR recruitment and juvenile survival, a seed bank study, as well as seeding and transplant experiments were conducted. Over five years, mark-recapture data of adult individuals in open and deer-fenced plots established in one extant population of each SAR were collected. At Cornell University, slug feeding trials and common garden experiments were conducted to assess interactive effects of earthworm and B. pellucidus on SAR germination and survival.


Density and species composition of monitored stressor organisms (earthworms, slugs, and B. pellucidus) varied across field sites and years. Unexpectedly, it was found that earthworm density and biomass decreased in the fenced plots, indicating a possible, but unforeseen, interaction between earthworms and deer.

In just five years, it was found that all three target non-native plants had significantly lower abundance (frequency, cover, and/or density) in fenced plots, in response to deer exclusion. This is particularly true for the short-lived M. vimineum and A. petiolata, which are annual and biennial, respectively. Simultaneously, native vegetation responded positively to deer exclusion. Results indicate that it may be possible to reduce abundance of non-native plants simply by substantially reducing deer density.

Contrary to expectations, stressor impacts were not always negative and were even beneficial at certain life stages. For example, a common garden study indicated that earthworm presence, most likely through a reduction in leaf litter accumulation, increased germination of the majority of tested plant species. Similarly, A. serpentaria and T. erectum germination in field trials was significantly higher at sites dominated by the invasive M. vimineum than at native vegetation sites. Not surprisingly, it was found that SAR response to stressors was species- and stage-specific. For example, while M. vimineum had a positive effect on germination of two SAR, it had no effect on survival and growth of transplanted seedlings of the same two SAR. Additionally, while fencing had a small effect on survival or growth of transplanted seedlings of all SAR, earthworm density had a distinct effect on each SAR: A. rostellata and C. retroflexa survival and growth were higher at high earthworm density sites, while T. erectum transplants did better at sites with low earthworm density and A. serpentaria seedlings were not affected by earthworm density.

Marked individuals of all species had a positive response to fencing: overall, individuals of all species grew taller, with larger leaves and had higher probabilities of flowering or seed production. Carex retroflexa showed a tendency for higher growth (new culm production) in the open plot and for higher reproduction (seed production) in the fenced plot. This was the only SAR for which above-ground mammal herbivory was not detected in the open plot. For the three remaining SAR, browsing was significantly lower in the fenced plot. Attack of adult A. serpentaria, A. rostellata, and T. erectum in the fenced plot was most likely due to rodent feeding as evidenced by the angular cut of stems. These three species also showed frequent attack by a variety of invertebrate herbivores.

Population growth rate (λ) of three of the species (A. rostellata, C. retroflexa, and T. erectum) was >1, indicating that populations are projected to grow. Nevertheless, λ for all species was higher in fenced than open plots. Aristolochia serpentaria growth rate was <1 in the open plot and increased to just one after five years of fencing, indicating that deer effects might be of special relevance for maintenance of this threatened species. LTRE analysis indicated a greater effect of fencing compared to that of earthworm density, but effects manifested themselves through different life cycle pathways for each SAR.


Biodiversity conservation projects should critically consider effects of multiple stressors; however, these results emphasize the importance of deer as a structural force and indicate that reductions in deer populations are imperative (and may be sufficient) to restore and preserve populations of rare plant species.

The ecological carrying capacity for local deer herds in different habitats is a function of vegetation composition, soil fertility, climate, previous land use history, and many other factors, including biological invasions. It is difficult for local land managers to separate effects of these different factors without detailed experimental manipulations. To address this problem, a method using indicator species was developed to assess deer browse pressure. In this study, “sentinel” oak seedlings (Quercus rubra L.) protected from deer herbivory continued to grow while those exposed to deer suffered major mortality. This approach allows land managers who have identified conservation goals as the main objective of management activities to set deer population targets based on actual deer browse pressure, rather than relying on questionable historic deer abundance estimates or hunter satisfaction surveys. This approach allows land managers to assess local browse pressure (10 - 50 ha) but is not sufficient over larger landscapes. Furthermore, while it is anticipated that in most areas of North America deer populations are too high to allow any oak recruitment, the protection of browse sensitive species will likely require large deer reduction efforts and future development of different and more sensitive indicator species to assess deer browse pressure at lower deer densities. Until such reductions are in place, vulnerable plant populations may need to be fenced to allow their recovery.