From ecological menace to roadside attraction: 28 years of evidence support successful biocontrol of purple loosestrife

Introduction and spread of non-native plants provide ecologists and evolutionary biologists with abundant scientific opportunities. However, land managers charged with preventing ecological impacts face financial and logistical challenges to reduce threats by introduced species.  The available toolbox (chemical, mechanical, or biological) is also rather limited.  Failure to permanently suppress introduced species by mechanical and chemical treatments may result in biocontrol programs using host-specific insect herbivores.  Regardless of the chosen method, long-term assessment of management outcomes not just on the target species, but also on associated biota, should be an essential component of management programs. However, data to assess whether management results in desirable outcomes beyond short-term reductions of the target plant are limited.  Here we use implementation of a biocontrol program targeting a widespread wetland invader, Lythrum salicaria (purple loosestrife), in North America to track outcomes on the target plant over more than two decades in New York State.  After extensive testing, two leaf-feeding beetles (Galerucella calmariensis and G. pusilla; hereafter ‘Galerucella’), a root-feeding weevil (Hylobius transversovittatus) and a flower-feeding weevil (Nanophyes marmoratus) were approved for field releases.  We used a standardized monitoring protocol to record insect abundance and L. salicaria stem densities and heights in 1m2 permanent quadrats at 33 different wetlands and followed sites for up to 28 years.  As part of this long-term monitoring in 20 of these wetlands, we established a factorial experiment releasing either no insects (control), only root feeders, only leaf beetles, or root- and leaf-feeders. We document reduced L. salicaria occupancy and stem densities following insect releases over time, irrespective of site-specific differences in starting plant communities or L. salicaria abundance.  We could not complete our factorial experiment because dispersal of leaf beetles to root-feeder-only and control sites within five years invalidated our experimental controls. Our data show that it took time for significant changes to occur, and short-term studies may provide misleading results, as L. salicara stem densities initially increased before significantly decreasing. Several decades after insect releases, pre-release predictions of significant purple loosestrife declines have been confirmed.

Methods
We worked at 33 sites (mix of federal, state, and private ownership) varying in size from a few hundred square meters to >10 ha across New York State.  Initially, most sites (old fields, wetlands, marshes, and impoundments) were dominated by L. salicaria, although some releases occurred where managers documented only small patches.  Water levels varied through time, reflecting natural variation in precipitation patterns, beaver activity, and intentional manipulations by wetland managers and private landowners.  We monitored each site for a period of 15 to 28 years, starting the year insects were first released.  Releases included one or several of the following species: two leaf-feeding beetles (Galerucella calmariensis and G. pusilla), a root-feeding weevil (Hylobius transversovittatus), and a flower-feeding weevil (Nanophyes marmoratus).  

At each site we established 1 to 15 1 m2 permanent quadrats spaced 5–50 m apart.  Presence of L. salicaria was a pre-requisite for initial quadrat placement, as the region had well established long-term presence of L. salicaria and populations were stable and not undergoing short-term fluctuations.  The number of quadrats and distance between quadrats was dictated by L. salicaria population size.  We used 1.5 m tall PVC (2.53 cm diameter) poles to mark each corner of each quadrat, and recorded GPS position and hand-drew maps indicating distance and azimuth to help relocate quadrats.  Over the 28 years, individual poles and occasionally all poles of a quadrat disappeared; we used GPS and field notes to re-establish quadrats as close to the original position as possible. 

We initially visited all sites twice annually. At each site, we recorded detailed data about L. salicaria growth and reproduction, presence and abundance of released insects and co-occurring plant species. In June we counted L. salicaria stem density, estimated L. salicaria cover (%), and recorded Galerucella abundance using timed counts (1 min each for eggs, adults, and larvae), and made visual assessments of leaf area removed (%).  We did not record presence for H. transversovittatus as larvae feed in roots, and adults are night active, preventing detection without destructive sampling.  We also did not record presence of N. marmoratus.  However, during our last sampling in 2019, we recorded presence of flower feeders and assessed root-feeder presence at all sites, except those that were permanently flooded, by excavating rootstocks and checking for larval feeding damage. In September, in each quadrat we recorded L. salicaria stem density and cover (%) and, for the five tallest stems, height, number of inflorescences, length of the tallest inflorescence, and number of flower buds in the central 5cm of that inflorescence.  Depending on insect releases, we recorded data for variable time periods at each site between 1996 and 2007, and again at all 33 sites in August and September 2019. 

Data was entered into excel, formatted, and then analyzed using R.  Please see the associated manuscript for more details.