Effects of herbivory by a translocated butterfly on plant size and seed set of Lonicera involucrata

Assisted migration consists of the introduction of a species to previously inhabited areas or to new suitable regions. Such introductions have been touted as a viable tool for conserving earth's biodiversity. However, both the likely success of assisted migrations and the impacts on local communities are hotly debated. Empirical data on the local impacts of assisted migration are particularly lacking. We examined the short and long time-scale effects of herbivory on Lonicera involucrata (Richards) Banks ex. Spreng (Caprifoliaceae) after an introduction of Euphydryas gillettii Barnes (Lepidoptera: Nymphalidae: Melitaeini) to Gunnison County, Colorado via an assisted migration in 1977. The plant is the primary larval hostplant for the butterfly. We quantified plant seed production, plant survival, and population stage structure in two sets of observational experiments. We found that herbivory by E. gillettii increased L. involucrata reproduction on an annual time scale, independent of plant size and local microhabitat characteristics. Over the time since the butterfly’s introduction, herbivory by E. gillettii resulted in a plant population structure biased towards smaller plants in the butterfly introduction and satellite sites compared to sites without the butterfly. Our results highlight the importance of studying effects of assisted migrations on native populations at different temporal scales. As assisted migration becomes an indispensable tool for species conservation, our work adds to the understanding of the multi-trophic impacts of assisted introductions on local populations and communities.

Methods
Euphydryas gillettii was introduced to a 1.75 ha site at the Rocky Mountain Biological Laboratory (RMBL) in Gunnison County, Colorado USA (38°57′5” N, 106°59′6” W, 2912 m asl) in 1977 from Granite Creek, Wyoming USA (43°41’42” N, 110° 26’55” W, 2130 m asl) (Holdren & Ehrlich, 1981). The introduction site is south of the native range. 

The short-term effects of E. gillettii herbivory on L. involucrata were studies at the original butterfly introduction site (“Main”) at RMBL (Holdren and Ehrlich 1981) and a 0.75ha site (“Avalanche”), located 0.6 km southeast of Main. Avalanche was naturally colonized by the butterfly in 2002 (Boggs et al., 2006). We marked 78 L. involucrata individuals in Main and 9 plants in Avalanche with forestry flagging and aluminum metal tags. For each plant, we recorded the number of E. gillettii larval webs present in early September 2004 as the larvae were entering diapause, along with evidence of herbivory by animals other than E. gillettii. In 2004, we recorded plant parameters that are correlated with the likelihood of E. gillettii herbivory on L. involucrata (Bonebrake et al., 2010). These included exposure of the plant to sun, (1 = morning, 2 = afternoon, 3 = both), soil wetness in the first half of July and early September (1 = dry surface, 2 = moist surface, 3 = standing water), and presence or absence of P. engelmannii within 1 m of the plant. The index of soil wetness used in the analysis was the mean of the July and September values. In addition, we recorded measures of plant size: total number of shoots in 2004 and plant volume in 2004 and 2005. Plant volume was calculated as a rectangular cube, based on plant height, plant width and plant length measured with a measuring tape.In 2005, we collected all ripe fruits from each marked plant and counted the number of seeds. Occasionally, seed dispersers collected the fruits before we did, as evidenced by scars present on the base of the bracts. We had earlier recorded aborted fruits on each plant, so we could determine whether the scar represented aborted or dispersed fruits. For cases of missing fruits, we recorded the number missing and estimated total seed set by the plant based on the average for other fruits on that plant.

To examine long-term effects of larval herbivory on individual plants and plant population size structure, we used three additional sites without E. gillettii. All five study sites contained similar vegetation. Both the Main and Avalanche experimental sites consisted of east-facing meadows with beaver ponds and streams at the base of Gothic Mountain. The Avalanche site lacked the Picea engelmannii trees present in the Main site, due to high avalanche activity. The three control sites without butterflies were Trail 401, Copper Creek and North Main. The North Main site was on the same east-facing slope of Gothic Mountain as Main and Avalanche. Trail 401 and Copper Creek consisted of montane meadows, which were drier and at slightly higher elevation than the other three sites. Within each of these five sites we randomly selected 2 to 4 10m x 10m subsites, depending on the size of the site, with a total of 7 subsites in the treatment and seven in the control sites. On July 6-8, 2011, we collected data on all L. involucrata individuals in each of the long-term study subsites. Using a tally-counter, we separately recorded the number of live and dead shoots on an individual plant. Since L. involucrata has rhizomes, individual plants were denoted as shoot clusters that were located within 15cm of each other. We assigned each L. involucrata individual to a size class based on its live shoot count. Most plants with less than 10 live shoots had no dead shoots, and plants with less than 25 shoots typically were not flowering. Therefore, the first two size classes encompassed plants containing 1–10 and 11–25 live shoots. We arbitrarily delineated the next three size classes to include plants having 26–150, 151–500, and greater than 500 live shoots. We recorded instances of flowering, number of dead shoots, large mammal or insect herbivory, and whether the perimeter of the leaves was slightly reddened, indicating a pathogen infection. Based on the surrounding topography and the position and path of the sun, we assessed whether the plant received morning or afternoon sunlight. We classified the soil around each plant according to the soil wetness scale described above. Using a measuring tape or stick, we recorded the standing height of each L. involucrata plant to the nearest centimeter. Additionally, we approximated the height of the tallest surrounding vegetation, excluding P. engelmannii, to the nearest 0.5m. We estimated the percent cover of any vegetation within a meter that was taller than the L. involucrata individual and noted the presence and estimated height of any P. engelmannii within 1m of the plant.