A proposal to use plant demographic data to assess potential weed biological control agents impacts on non-target plant populations

A proposal to use plant demographic data to assess potential weed biological control agents... BioControl (2018) 63:461–473 https://doi.org/10.1007/s10526-018-9886-4 A proposal to use plant demographic data to assess potential weed biological control agents impacts on non-target plant populations . . . Bernd Blossey Andrea Davalos Wade Simmons Jianqing Ding Received: 14 November 2017 / Accepted: 25 April 2018 / Published online: 28 April 2018 The Author(s) 2018 Abstract Weed biocontrol programs aim to reduce assessments, are essential to guide weed biocontrol the spread and population growth rate of the target programs. We propose to add use of plant demography plant while stabilizing or increasing populations of (an assessment of how environmental factors and those native species considered under threat by ecological interactions, for example competition, invasive plants. This goal is not unique to weed disease or herbivory, may affect plant populations by biocontrol but applies to all other invasive plant altering survival, growth, development and reproduc- management techniques, though such information is tive rates of plant individuals) during host specificity rarely collected. Without this information, success of risk assessments of potential biological control agents. management interventions can be ambiguous, and Demographic models can refine assessments of poten- regulatory agencies, the public, policy makers, funders tial impacts for those plant species that experience and land managers cannot be held accountable for some feeding or larval development during host chosen treatments. A fundamental reform, including specificity testing. Our proposed approach to focus use of demographic studies and long-term on impact on plant demography instead of attack on plant individuals is useful in appropriately gauging threats potential weed biocontrol agents may pose to Handling Editors: Mark Schwarzla¨nder, Cliff Moran and S. non-target species after field release. Raghu. Keywords Demography  Host specificity  Non Electronic supplementary material The online version of target effects  Risk assessment  Trapa natans L. this article (https://doi.org/10.1007/s10526-018-9886-4) con- tains supplementary material, which is available to authorized Weed biocontrol users. B. Blossey (&)  W. Simmons Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA Introduction e-mail: bb22@cornell.edu Biological weed control programs aim to find organ- A. Da´valos Biological Sciences, SUNY Cortland, 1215 Bowers Hall, isms able to reduce spread and population growth rate Cortland, NY 13045, USA of target plants, while avoiding non-target impacts. The track record of weed biocontrol over the past J. Ding century is decidedly mixed, since only a third of all College of Life Sciences, Henan University, weed biocontrol programs achieve at least partial Kaifeng 475004, Henan, China 123 462 B. Blossey et al. suppression of targets (Crawley 1989; Fowler et al. 2000; Moran et al. 2005). Many biocontrol agents fail to establish, or fail to control host plants (Crawley 1989; McFadyen 1998). On the other hand, while occasionally contested, the safety record of weed biocontrol is superior to other management methods, while economic and ecological benefits can be enor- mous and continue to accrue (Moran et al. 2005; Suckling and Sforza 2014). Following publications of high profile cases of non- target attack by Rhinocyllus conicus Fro¨lich (Cur- culionidae) and Cactoblastis cactorum Berg (Pyrali- Fig. 1 Schematic design of typical proposed host specificity dae), respectively, changes in decision making testing protocol for potential weed biocontrol agents. Depending processes in regulatory agencies, particularly in the on life history and feeding mode of the herbivore under USA, shifted to a greater reliance on fundamental consideration, test conditions may vary. Pool 1 includes all plant species proposed for host specificity testing that are tested using host-range data, a change that further threatens release highly constrained no-choice conditions (Screen 1). Those even of highly specific agents (Hinz et al. 2014). The species that could not be eliminated in the first screening step irony of this change in risk perception is that specific constitute pool 2 which are tested using more sophisticated and successful agents of the past would have difficul- designs, such as multiple-choice tests using potted plants or ties passing through current approval processes larger cages (screen 2). Species in pool 3 include plant species that were still attacked under the more sophisticated test (Groenteman et al. 2011; Hinz et al. 2014). At a time conditions, or where larvae completed development. Tests when it is becoming increasingly evident that many utilized during screen 3 depend on herbivore feeding niches and invasive species control methods, particularly chem- logistical and regulatory considerations but include use of ical management, are unable to achieve lasting control common gardens, multiple choice tests without containment, etc. Only those species that were still attacked under the most and may in fact threaten non-target species (Ketten- realistic conditions possible in a particular program would then ring and Adams 2011; Pearson et al. 2016), we argue be considered candidate species for demographic analyses that it is time for fundamental reform of risk assess- (screen 4). Note that the pool of species shrinks with each test, ment and decision making processes in invasive plant while the realism of testing conditions and their ecological relevance increases management and weed biocontrol that is guided by appropriate scientific information and open dialogue, highly specific herbivores. We propose to utilize plant not fear (Blossey 2016b). demography (an assessment of how environmental We propose that adoption of modern scientific tools factors and ecological interactions, for example com- focusing on demographic impacts of herbivores could petition, disease or herbivory, may affect plant pop- constitute a breakthrough development in maintaining ulations by altering survival, growth, development and safety while increasing the ability to select effective reproductive rates of plant individuals) (Salguero- herbivores. We consider it paramount to shift non- Gomez et al. 2015) to assess potential threats of target risk assessments away from damage to individ- candidate biocontrol agents to non-target species. This uals to population level effects expected after field approach aims to provide a means by which to releases. We envision that traditional reductionist evaluate potential impacts to non-target plant popula- approaches (no-choice, small herbivore confinements, tions. Our proposal constitutes a significant shift in the followed by multiple-choice or potted plant experi- way weed biocontrol researchers, review panels and ments) will continue to be the mainstay of host others may look at the approval and risk assessment specificity testing. These tests are valuable because the process—but it is a scientifically valid and biologi- vast majority of test plant species will not be attacked cally meaningful one. We are not concerned by host even under constrained conditions (Fig. 1). However, use but by negative impacts to populations of non- in many programs often a few test plant species remain target species. This is not a reduction in protections that may be fed upon, are accepted for oviposition, or afforded to native species as it continues to safeguard even allow larval development (albeit at a greatly all native species or valuable introduced species that reduced rate compared to original host plants) by 123 A proposal to use plant demographic data to assess potential weed biological control… 463 have attained cultural, ornamental or agricultural ranges are always narrower than experimentally significance. We argue that to ‘‘safeguard’’ means determined fundamental host ranges. that populations of non-target organisms are main- tained and do not suffer demographic declines due to biocontrol agent introductions. Cosmetic damage or Evidence for threats of weed biocontrol agents even substantial damage to, or death of, individuals to (rare) native plant species does not necessarily indicate demographic or ecolog- ical consequences. Reports of weed biocontrol agents attacking non- The shift we propose will find resistance based on target species do exist, including spillover events with risk perceptions regarding safety of non-target plants substantial temporary defoliation of non-target species due to concerns that herbivores introduced to control (Blossey et al. 2001; Louda et al. 1997, 2003; Paynter introduced plants will (1) attack (rare) native species et al. 2008; Pemberton 2000; Suckling and Sforza leading to declines in populations of these species; and 2014). Comprehensive reviews assessing weed bio- (2) that diet restriction (i.e. specificity) of weed control outcomes (Blossey et al. 2001; Suckling and biocontrol agents are ‘‘fluid’’ and change over time, Sforza 2014), conclude that [ 90% of all biocontrol leading to attack and unintended negative conse- agents never attack non-target species. The majority of quences for native species. We will briefly review non-target feeding is attributed to spillover events and Suckling and Sforza (2014) report such attacks on 128 evidence for these concerns before further developing our proposal to use demography in host specificity risk non-target plants. Host specificity testing appears assessments. However, we first provide a primer on unable to predict identity of these species, but physical terminology used to describe plant–herbivore interac- proximity may explain some of it (Blossey et al. 2001). tions because we believe that some differences in However, occasional or prolonged host use appears perceived risk perception are semantic. highly predictable using fundamental host range data (Paynter et al. 2015). Fewer than ten biocontrol agents have established populations on non-target species, a Terminology used in describing plant–herbivore risk that was known, and accepted by regulatory interactions and weed biocontrol programs agencies, at time of their introduction (Blossey et al. 2001). Of these, only three, R. conicus, C. cactorum Ecologists typically refer to diets of herbivores using and Trichosirocalus horridus (Panzer) (Curculion- terms like specialists or generalists (Smilanich et al. idae) may have effects that reduce populations and 2016) or more specifically monophagous (feeding on a growth rates of non-target species (Louda et al. 1997; single or few species within a genus), oligophagous Suckling and Sforza 2014; Takahashi et al. 2009). (utilizing several plant species, typically in different None of these herbivores would be approved under genera), and polyphagous (using different plant current decision making frameworks (McFadyen species in different genera and families) (Bernays 1998; Suckling and Sforza 2014). and Chapman 1994). In contrast, weed biocontrol Detailed documentation of non-target plant species researchers typically focus on herbivores using a occasionally attacked by biocontrol agents offer single plant species. Furthermore, ‘‘use’’ in the assurances that significant non-target effects have ecological and evolutionary literature typically refers not gone unrecognized or unreported—in this case to plants chosen for oviposition and allowing larval absence of evidence indicates evidence of absence of development in the field, recognized as realized host such effects and not just lack of effort. A recent range in weed biocontrol. Experimental host-speci- literature survey of threats by insect herbivores to rare ficity testing aims to (1) elucidate the fundamental host plants concluded that with exception of R. conicus and range (plant species acceptable for feeding, oviposi- C. cactorum, ‘‘currently this threat is either seldom tion and larval development using no-choice tests in realized (perhaps because of extensive pre-release the absence of the original host), and (2) provide screening in modern biocontrol programs) or else additional data using less constrained and increasingly seldom documented’’ (Ancheta and Heard 2011). ecologically realistic testing procedures to allow forecasting of realized host ranges. Realized field host 123 464 B. Blossey et al. Lack of evidence for evolution of dietary Species accumulation curves on novel host plants preferences in weed biocontrol agents plateau in approximately 100 years for generalists and 500–10,000 years for specialists (Bernays and Gra- Permitting processes for biocontrol agent releases may ham 1988). However, the vast majority of phy- differ widely among countries (Sheppard and Warner tophagous insects show ‘‘phylogenetic 2016), but host specificity tests are widely standard- conservatism’’ retaining their association with plant ized (Wapshere 1974). Despite further refinements taxa over millions of years with \ 10% of speciation proposed and implemented in subsequent years events including a shift to a different plant family (Briese 2005; Clement and Cristofaro 1995; Sheppard (Winkler and Mitter 2008). Biocontrol agents passing et al. 2005; USDA 2016), this sequence of testing has through host range testing, as far as we can tell from largely remained state-of-the-art, providing over- decades of observation and study, appear particularly whelmingly safe weed biocontrol agents. There is no ‘‘conservative’’. evidence that fundamental host ranges of biocontrol We now return to our argument that use of agents have evolved (Arnett and Louda 2002; Maro- demographic models should be a desired and required hasy 1996; Paynter et al. 2004; Sheppard et al. 2005; tool during risk assessment of biocontrol agents. We van Klinken and Edwards 2002), despite dire warnings are not the first to propose such new tools (Louda et al. (Simberloff and Stiling 1996). There is, however, 2005a; Raghu et al. 2006; Sauby et al. 2017), although evidence for evolution of improved performance on we believe we are the first to ask that this becomes part non-target plants (McEvoy et al. 2012) and we of pre-release risk assessments. We will briefly acknowledge that few formal assessments have been introduce concepts of demographic modeling and made. then provide examples how demography has, and can However, occasional use, even if predicted, of be utilized in weed biocontrol. To the best of our species that are not targets of weed biocontrol, and knowledge, no biocontrol program has used demo- frequent citation of the few species with anticipated graphic information to assess risks to non-targets large negative impacts, appears to be registered by before field releases, so we will rely on post-release non-biocontrol scientists as evidence for a poor track analyses and a hypothetical scenario involving a record and evolutionary malleable diet breadth of current target of biocontrol research, water chestnut, insect herbivores used in weed biocontrol. We argue Trapa natans L. (Lythraceae) to illustrate our that the problem arises due to the disconnect between proposal. perception of labile or rapidly evolving host ranges in weed biocontrol agents and available evidence. Weed biocontrol researchers have increased efforts to Using demography to evaluate biocontrol agent improve testing procedures to mitigate constraints of risks to non-target plants laboratory conditions that affect insect behavior to improve predictions of realized host ranges (Clement Demography and matrix population models (Caswell and Cristofaro 1995; Fowler et al. 2012; van Klinken 2001) are now common tools in biology (Caswell and and Edwards 2002). While such improvements are Salguero-Gomez 2013) and their use in invasion essential to increase reliability of predictions regard- biology and weed biocontrol is increasing (Carval- ing realized host ranges, the focus on improving heiro et al. 2008; DeWalt 2006; Eckberg et al. 2014; testing conditions appears to have prevented a discus- Kerr et al. 2016; McEvoy and Coombs 1999; Shea and sion among scientists and regulatory agencies regard- Kelly 1998; Swope et al. 2017). Technical background ing the appropriate meaning of safeguarding other is provided elsewhere (Caswell 2001; Caswell et al. species (Hinz et al. 2014). 2011; Caswell and Salguero-Gomez 2013; Williams Clearly, diets of insect herbivores change over time et al. 2001), but briefly development of demographic and both generalists and specialists may acquire new models requires estimating vital rates, the transition hosts (Futuyma and Agrawal 2009). Range expansions probabilities from one life stage to another. For plants through human aided introduction of novel plants or with clear developmental stages, a stage-based model insects provide enormous ecological and evolutionary requires estimation of transitions from seed, to opportunities for herbivores to adopt new hosts. seedling, to rosette to flowering plant, to seed output, 123 A proposal to use plant demographic data to assess potential weed biological control… 465 back to seed and seed bank (Davis et al. 2006; Shea Shea et al. 2005; Swope et al. 2017). These examples and Kelly 1998). Local abiotic conditions, competi- are important in recognizing that demographic tion, herbivores, stochasticity, density dependence and approaches are already an important part of the other processes may affect survival and the probability evaluation process in weed biocontrol. We will not that an individual will transition from one stage to the review these here but focus instead on efforts to help next. Vital rates can be inferred in the field by assess demographic impacts of weed biocontrol agents monitoring cohorts of marked individuals. on non-target plants after release. Twenty years have Demographic models can aide in assessments of passed since the initial widespread criticism regarding potential impacts of proposed biocontrol agents on safety of (weed) biocontrol (Louda et al. 1997; non-target plants that could not be excluded using Pemberton 2000; Simberloff and Stiling 1996). At traditional testing sequences (Fig. 1). We propose use least ten biocontrol agents have established popula- of experiments, for example by manipulating herbi- tions on non-target species and [ 120 non-target vore access or attack rates and then measuring stage plants are reported to be attacked (Blossey et al. specific reductions in survival, recruitment, growth, 2001; Suckling and Sforza 2014), thus we expected to biomass, or seed output of non-target plants, which find numerous publications outlining demographic can be done in common gardens, or other confine- consequences, or at least attempts to evaluate conse- ments when insects are not approved for release. quences of such attacks. Our Web of Science searches uncovered few studies, which may indicate that they Constructing and populating models with data, and analyzing model performance under different scenar- either were not conducted, did not get published, or ios (often referred to as perturbation analysis) allows were deemed unimportant to conduct or fund. We comparisons of contributions made by different vital therefore focus on R. conicus and C. cactorum, species rates for overall population growth rates (Caswell that according to categorization by Suckling and 2000). The outcome of these exercises is the ability to Sforza (2014) have ‘‘massive’’ non-target impacts, forecast population growth rates (k), population plus post-release evaluations of Mogulones crucifer fluctuation and potential extinction risk, and the Pallas (Curculionidae), a species approved for release sensitivity of growth rates to small changes in vital against houndstongue (Cynoglossum officinale L.) rate values, regardless of which management action is (Boraginaceae) in Canada (Catton et al. 2016). applied (Kerr et al. 2016). We recognize that demo- Apparently, studies evaluating demographic effects graphic approaches during evaluation of potential of T. horridus beyond documentation of attack on a biocontrol agents will have to contend with many non-target plants do not exist, therefore we exclude different obstacles, the smallest among them may be this species. lack of familiarity of biocontrol scientists with demo- A high-profile paper regarding non-target effects of graphic modeling (Blossey 2016b). But this is a small R. conicus (Louda et al. 1997) tabulated attack rates on price to pay for the ability to improve predictability of native thistles, but fell short of documenting demo- impacts to targets or risks to non-target organisms. graphic effects, which were strongly implied due to Furthermore, a fast growing and increasingly utilized seed limitation and large demographic impacts by open access database, COMPADRE, provides a native seed feeders on Cirsium altissimum (L.) Spreng potentially important resource to inform construction (Asteraceae) (Guretzky and Louda 1997). Additional of appropriate models for species of interest (Sal- investigations clearly documented demographic guero-Gomez et al. 2015). threats by R. conicus (Louda et al. 2005b) based both on field and laboratory data. But effects appear context-dependent and do not occur every year and Retrospective demographic analyses for target in every location (Rand and Louda 2004; Rose et al. and non-target effects in weed biocontrol 2005). In addition, some native thistles show positive population growth rates even in the presence of and Demographic modeling has been used to understand attack by R. conicus (DePrenger-Levin et al. 2010). success or failure of weed biocontrol programs in Furthermore, results of demographic models to assess reducing target plant population growth rates (Buck- population growth rates for Platte thistle, Cirsium ley et al. 2004; DeWalt 2006; Shea and Kelly 1998; canescens Nutt. concluded that impacts may be 123 466 B. Blossey et al. substantial, but variable in space and time and not as models incorporating more than presence of herbivore catastrophic as previously feared (Rose et al. 2005). attack and other ‘‘stressors’’ are we able to gauge While R. conicus should have never been approved for impacts appropriately. As in the case of R. conicus, release, current evidence is of widespread attack on anticipated ‘‘massive’’ impacts of C. cactorum are, native Cirsium species, but evidence for predicted according to published studies, not currently materi- massive negative demographic non-target effects alizing in the field. (sensu Suckling and Sforza 2014) has not been Risk assessment after release of M. crucifer, a root presented at this time. feeding weevil that attacked some Boraginaceae, The accidental introduction of C. cactorum to including some US native and rare plants, during host North America (Pemberton 1995) raised concerns specificity testing (De Clerck-Floate and Sch- over safety of native North and Central American warzla¨nder 2002), provides a good example of an Opuntia spp. (Cactaceae) (Vigueras and Portillo application of matrix population models. Canadian 2001), particularly for rare endemics, such as Opuntia authorities granted release permits and M. crucifer corallicola Small where only 12 known individuals established and began to spread in British Columbia, existed in the Florida Keys (Johnson and Stiling 1996). prompting fears about non-target attacks upon arrival Follow-up work, including using plant demography, in the USA (Andreas et al. 2008). Additional host over the past two decades has delivered a more refined specificity testing, including field tests in British view of realized threats. While initial introductions to Columbia, also documented non-target attack by M. Nevis and St. Kitts in the Lesser Antilles to control crucifer but found minor adult feeding and infrequent weedy native Opuntia spp. was ill advised, a survey larval development, despite ability of the weevil to 50 years after C. cactorum releases showed that the complete development under no-choice conditions targeted native species Opuntia triacantha (Willd.) (De Clerck-Floate and Schwarzla¨nder 2002). Sweet and Opuntia stricta (Haw.) Haw. remain under Subsequent monitoring showed that M. crucifer did biological control while the native tree pear Consolea not establish at sites where C. officinale was absent rubescens (Salm-Dyck ex DC) Lem. (Cactaceae) was (Catton et al. 2015) and attack of non-target species not attacked and the cultivated and naturalized Opun- tapered off within a few meters (Catton et al. 2014), tia cochenillifera (L.) Mill showed limited feeding by including during spillover events. Furthermore, C. cactorum (Pemberton and Liu 2007). In the detailed demographic work on Hackelia micrantha Southeastern USA, C. cactorum has spread rapidly, (Eastw.) J. L. Gentry (Boraginaceae), a native plant resulting in variable impacts depending on cactus species regularly attacked in the field, demonstrated species, often resulting in size decreases and reduction that while population growth rates for C. officinale in relative growth rates (Sauby et al. 2017). Jezorek were reduced below replacement rates (k \ 1), H. et al. (2012) summarized these findings as follows: micrantha benefitted from C. officinale reductions ‘‘although C. cactorum should still be considered a (Catton et al. 2016). These results indicate that while threat, particularly for rare opuntioids, overall survival individual H. micrantha are being attacked and allow along the west central Florida coast is currently high larval development of M. crucifer, the species is safe and plants that are able to survive C. cactorum attack and suffers no harm at the population level (Catton are not being reduced in size, possibly because they et al. 2016). Similar demographic experiments with possess traits that render them more tolerant of C. rare plants that are part of the fundamental host range cactorum damage. Our findings suggest that an of M. crucifer could help evaluate real (vs. feared) assumption of severe negative effects of an invasive threats to other US native Boraginaceae. species, based on its effects in other regions or over These examples showcase the value of detailed short periods of time, may not always be justified’’. In demographic studies to assess how attack by biocon- the case of the rare endemic O. corallicola, detailed trol agents may, or may not, contribute to harm, or studies and restoration efforts revealed that salinity, endangerment of non-target species. Only through moisture conditions, hurricanes, trampling by deer, such detailed work are we able to separate anecdotal and stem rot over the past two decades were more observation of attack from contributions of many important demographic threats than C. cactorum factors (habitat loss and fragmentation, inbreeding (Stiling et al. 2000). Only by developing detailed depression, succession, disturbance, climate, abiotic 123 A proposal to use plant demographic data to assess potential weed biological control… 467 conditions, competition, other natural enemies, etc.) growth and reproduction of T. natans at different that affect plant demography simultaneously. larval densities (0–50 L per rosette) (Ding and To the best of our knowledge, no biocontrol Blossey 2005; Ding et al. 2006b). We combine program has used a demographic analysis to assess herbivore impact data with demographic plant data herbivore impact on non-target plants pre-release or as obtained in outdoor aquatic mesocosms to build part of a release petition. Raghu et al. (2006) proposed demographic models that project future plant popula- to use modeling a priori, but this recommendation tion dynamics under different beetle herbivory sce- followed rejection of a herbivore by Australian narios. These data were initially collected and the authorities due to minor feeding on a non-target plant. demographic model built to assess utility of each We believe weed biocontrol researcher should strive herbivore as a biocontrol agent. By considering, for to make this standard practice when promising species sake of argument, T. natans as a non-target species, we fail traditional testing sequences (Fig. 1). Embracing can show how biocontrol programs can use demogra- this approach can lead to important collaborations phy in risk evaluation for non-target species. Full with those concerned about native species and descriptions of the herbivore impact studies are academics with specialized knowledge (Clewley available elsewhere (Ding and Blossey 2005; Ding et al. 2012). et al. 2006b) and results and details of our experimen- To further develop our proposal to use plant tal design to collect demographic data in aquatic demography models in pre-introduction risk assess- mesocosms using four different plant populations ments, we now turn to T. natans. We present data collected in Massachusetts (MA), Rhode Island (RI), collection procedures used to develop a demographic New York (NY), and Virginia (VA), USA are detailed model for T. natans. For the sake of argument, we in Supplementary Materials, Section 2. assume T. natans to be a non-target species, and we further assume that two different chrysomelid beetles Model development are potential biocontrol agents that failed to be cleared in traditional host specificity screening. We incorpo- We evaluated differences in population growth rate of rate the feeding impacts of each herbivore into the T. natans with periodic matrix population models. demographic model to evaluate the risk each agent Periodic matrix models are suited to explore within may pose to plant demography. year transitions of annual organisms, such as T. natans (Caswell 2001). These models partition life history transitions into m phases defined by a matrix (B ) that Evaluation of two herbivores attacking T. natans projects the population into the next phase (h; where h =1…m). Population over entire cycles is given by The Eurasian T. natans is a floating aquatic annual the product of period matrices: A ¼ B B ...B ; m m1 1 plant invasive in North America where it is attacked by where A is the annual transition matrix. We partitioned the native water lily leaf beetle, Galerucella nym- T. natans life cycle into three phases (Fig. 2): fall- phaeae L. while the extremely similar Galerucella spring (seedbank or new seeds germinate), spring– birmanica Jacoby (Chrysomelidae) attacks T. natans summer (rosettes grow small or large), and summer- in Asia (Ding and Blossey 2005; Ding et al. 2006a, b). fall (reproduction). We classified individuals in each Both herbivores are multivoltine and while G. life stage as seedbank, seeds, small or large plants and nymphaeae is oligophagous with multiple host races determined plant size as a function of surface area. We (Cronin et al. 1999), G. birmanica is host-specific to T. used census data from the common garden to estimate natans, although it occasionally lays a few eggs on transition and germination rates and published data to water shield, Brasenia schreberi J. Gmelin (Cabom- calculate seedbank longevity (Kunii 1988) (Supple- baceae), the only other plant reported to be attacked in mentary Materials, Table S3). We calculated annual the field (Ding et al. 2006a) (natural histories of T. population growth rate (k) given by the dominant natans, B. schreberi, G. nymphaeae and G. birmanica eigenvalue of A and used bootstrap methods to are provided in Supplementary Materials, Section 1). estimate 95% confidence intervals for each population Both herbivores were evaluated as potential biocontrol (defined as the 2.5 and 97.5% quantiles from a agents in experiments that assessed their impact on distribution based on resampling values with 123 468 B. Blossey et al. Fig. 2 Seasonal life cycle diagram and periodic matrices for T. seedbank or previous fall until spring, Bs plant survival and natans. Each row represents a season and each circle a life stage growth from spring to fall in the same year and Bu reproduction (SB: seedbank, S: seed, Sm: small plant, Lg: large plant). of small and large plants. Parameters are defined in Supple- Periodic matrix Bf represents germination of seeds from mental Materials, Table S3 replacement holding sample size constant for 1000 parameters (FSM and FLG, fertility for small and large iterations) (Caswell 2001). We also conducted elas- plants, respectively), but in this case fertility values ticity analyses on each periodic matrix to evaluate were drawn from a Poisson distribution. which transition most influenced growth rate (Caswell To account for density dependent effects, which 2001; Smith et al. 2005). result in decreased T. natans fertility and plant size (Groth et al. 1996), we modeled fertility values as a Non-target threat simulation density-dependent parameter, such that number of seeds produced by small (FSM) or large (FLG) plants We examined how presence and typical attack of G. was dependent on the number of T. natans plants (P) in nymphaeae or of G. birmanica affects demography of the previous season (s) T. natans (Ding and Blossey 2005; Ding et al. 2006b). ðÞ 1ðzP Þ s1 We modeled effects of each scenario on T. natans F ¼ ðÞ 1ðzP Þ s1 1 þ e populations over ten time steps (t) and 10,000 iterations (i) and parametrized the model with values we set z, the regression parameter, to 0.0005, indicat- estimated from the common garden and the literature ing weak density-dependent effects. To model effects (Supplementary Materials, Table S3). We modeled of management scenarios, we estimated average stochastic variation following a two-step procedure in number of seeds produced by small or large plants order to incorporate temporal variation and parametric (FSM or FLG, respectively) and weighted the value by uncertainty in model predictions (McGowan et al. rate of fertility decrease according to scenario. At each 2011). We modeled temporal variation in survival time step we estimated annual growth rate (k) as ratio rates and transition terms for each realization i and between population size at current year (N ) and t?1 time step t as a beta distributed random variable with previous year (N ). Values of k [ 1 indicate increas- parameters a and b derived from the mean survival ing populations while values of k \ 1 indicate i i (or transition) rate, l , such that l = a /(a ? b ) and declining populations. We conducted all analysis i i i i i r = l (1 - l )/(a ? b ? 1). We incorporated using package popbio (Stubben and Milligan 2007) i i i i i parametric uncertainty in the replication loop by in R Core Team (2016). sampling l from a beta distribution, and r from an i i inverse Gaussian distribution with mean m (m = 0.001) and shape parameter k (k = 0.0001). We followed the same two-step approach for fertility 123 A proposal to use plant demographic data to assess potential weed biological control… 469 Results of demographic analyses hypothetical scenario of T. natans as a potential non- target, G. birmanica would be a grave threat to Plant area, seed output and germination varied signif- continued existence of the species. If the goal were to icantly among populations and number of seeds per safeguard T. natans, our results show that of these two plant was positively correlated with plant area (Sup- extremely similar herbivores, G. nymphaeae would be plementary Materials, Fig. S1; Tables S1, S2). a safe biocontrol agent while releases of G. birmanica Asymptotic population growth rates varied signifi- should not be considered. cantly across populations (Supplementary Materials, Section 3). Elasticity analyses indicated that matrix Implications for demographic assessments of non- elements representing germination of new seed and target effects growth into large plants had the greatest influence on k (Supplementary Materials, Fig. S3). We considered a hypothetical example of T. natans as a non-target plant and evaluated potential threats by Results from modelled simulations two herbivores that can successfully feed, oviposit and develop on the species. Under current decision making Our model predicted that T. natans populations will processes, biocontrol scientists or regulators would continue to grow until carrying capacity or habitat not consider G. nymphaeae a safe biocontrol agent requirements do not allow further expansion, despite because fundamental and field host range include T. continued attack by G. nymphaeae (Fig. 3). This natans. Field evidence from [ 100 years of associa- indicates that if T. natans were a non-target species, G. tion with T. natans in North America, and results from nymphaeae would not constitute a demographic threat our demographic modeling efforts, however, clearly to T. natans populations despite regular feeding, indicate that G. nymphaeae is no demographic threat oviposition and larval development. In contrast, our to T. natans. Furthermore, there is no evolutionary model predicts that T. natans populations will be tendency, despite enormous opportunity, to improve greatly reduced when attacked by G. birmanica. After larval performance on the novel host, and female ten years simulated populations increased from 1000 choice retains its preference for oviposition on the plants to [ 40,000 plants when attacked by G. original host, even when larvae developed on T. nymphaeae, but declined to near zero individuals natans (Ding and Blossey 2009). The traditional risk under G. birmanica attack (Fig. 3). Thus, under our assessment process, as currently being implemented, would eliminate a potentially important biocontrol agent due to safety concerns regarding attack on T. natans, but this would be fundamentally unjustified given the realized demographic impact of G. nym- phaeae. Our demographic assessments correctly pre- dict what is evident in the field: G. nymphaeae does not affect T. natans populations, while G. birmanica attack can lead to rapid and severe population growth rate declines. Discussion Reports of non-target attack by weed biocontrol agents after their release has a chilling effect on agent approvals in ongoing programs, funding and recruit- Fig. 3 Simulated population size (continuous line) and popu- ment (Moran and Hoffmann 2015). Land managers, lation growth rate (dashed line) of a T. natans population under ecologists, conservationists and weed biocontrol sci- attack by either the leaf beetle G. nymphaeae or G. birmanica. entists spent enormous amounts of time discussing Model data are integrated means derived by collecting demographic data for four different T. natans populations implications, focusing on the two high profile cases of 123 470 B. Blossey et al. R. conicus and C. cactorum, while critics implied that native plants. Pivoting to new risk assessment proce- non-target impacts are potentially widespread, but not dures and lines of evidence will take some time but we recognized (Louda et al. 2003). Almost 20 years have argue it is essential and well justified. passed since the initial publications and while we have We acknowledge that biological control, like every been provided with some evidence for negative other management technique, is not risk free and that demographic consequences by R. conicus, the same ecological surprises may occur. For example, it cannot be said for C. cactorum. Importantly, there is appears that identity of plant species attacked during not a single publication documenting negative popu- outbreak or spillover events is unpredictable (Blossey lation level impacts on any other non-target plant et al. 2001). But these are temporary events with no species by weed biocontrol agents. Observation of demographic consequences for attacked plant species herbivorous biocontrol agents feeding on other plants and do not constitute host shifts. Assessing potential are widespread, indicating that evidence of their non- non-target effects is the ethical thing to do, but without target feeding is being recognized, collected and use of demographic information on targets and non- reported, but their attack appears inconsequential at targets biocontrol scientists and practitioners are the population level. vulnerable to accusations of inappropriate conduct Unless there are unrecognized, or unpublished data and may get blamed for population declines or on more widespread negative demographic conse- extinctions, whether these accusations are true or not. We argue that we should acknowledge our quences by approved weed biocontrol agents, the feared threats to native plants currently do not appear responsibilities for safeguarding the continued exis- to exist. This is entirely overlooked by regulators, tence of populations of native or non-target species, conservationists and ecologists who may equate and that demographic approaches provide a powerful feeding on non-target plants with demographic threats. tool for evaluating ecological risk. To the best of our The reliance on data about fundamental host range knowledge, with the exception of R. conicus or C. tests has been called risk-averse, but it ignores the cactorum, there is no weed biocontrol agent that has increasing realized impacts of invasive plants on negatively affected populations of non-target plants. native biota (Downey and Paterson 2016), and Given that hundreds of control agents have been increasing herbicide use by land managers that result released over 100 years across the world, this track in widespread and documented detrimental long-term record is exemplary (Winston et al. 2014). But we also effects to species of conservation concern (Kettenring may miss out on very promising herbivores, because and Adams 2011). We emphasize that society and the current regulatory climate makes scientifically citizens are entitled to have questions regarding questionable and poorly justified decisions (Cristofaro performance of invasive species management methods et al. 2013; Fowler et al. 2012; Groenteman et al. 2011; answered to make informed decisions about priorities, Hinz et al. 2014). risk acceptance and funding (Blossey 2016a, b), but Our proposal to utilize demographic approaches in this requires appropriate assessments and interpreta- forecasting agent efficacy, and potential impacts on tion, not fear mongering and it should apply to all non-target species, would go a long way in improving management methods (Kerr et al. 2016; Pearson et al. the standing of the discipline—one grounded in theory 2016). and applying modern tools—while retaining our Use of herbivore impact studies and demographic exemplary track record in safeguarding native species. models in the way we have described here would be an This does not make weed biocontrol inherently more important tool to evaluate efficacy and safety of risky. On the contrary, we would be able to focus on potential biocontrol agents, yet it is completely effective agents with a proven track record of impact unutilized despite its promise (Blossey 2016b). Use on demography of invasive plants, and lack of of plant (and herbivore) demography would be an demographic impact on non-targets. Delivering this important advancement in improving scientific rigor information to society and decision makers and and predictability of weed biocontrol programs, albeit regulators will constitute part of the accountability we acknowledge it will not be an easy transition. For we should require from all of those engaged in far too long, the focus has been narrowly on funda- invasive species management and stewardship (Blos- mental host ranges and not allowing any attack on sey 2016a, b; Hare and Blossey 2014). 123 A proposal to use plant demographic data to assess potential weed biological control… 471 Acknowledgements We appreciate support and assistance we Carvalheiro LG, Buckley YM, Ventim R, Fowler SV, Memmott have received from those supplying rosettes from local J (2008) Apparent competition can compromise the safety populations. We appreciate help with field work and common of highly specific biocontrol agents. Ecol Lett 11:690–700 garden by Audrey Bowe and Jennifer Price for help with model Caswell H (2000) Prospective and retrospective perturbation development. Comments by several anonymous reviewers analyses: their roles in conservation biology. Ecology provided important feedback that improved our manuscript. 81:619–627 Caswell H (2001) Matrix population models: construction, Open Access This article is distributed under the terms of the analysis, and interpretation. Sinauer Associates, Creative Commons Attribution 4.0 International License (http:// Sunderland creativecommons.org/licenses/by/4.0/), which permits unre- Caswell H, Salguero-Gomez R (2013) Age, stage and senes- stricted use, distribution, and reproduction in any medium, cence in plants. J Ecol 101:585–595 provided you give appropriate credit to the original Caswell H, Neubert MG, Hunter CM (2011) Demography and author(s) and the source, provide a link to the Creative Com- dispersal: invasion speeds and sensitivity analysis in peri- mons license, and indicate if changes were made. odic and stochastic environments. Theor Ecol 4:407–421 Catton HA, Lalonde RG, De Clerck-Floate RA (2014) Differ- ential host-finding abilities by a weed biocontrol insect create within-patch spatial refuges for nontarget plants. Funding Funding for this work was provided by the New Environ Entomol 43:1333–1344 York State Department of Conservation. Catton HA, Lalonde RG, De Clerck-Floate RA (2015) Nontar- get herbivory by a weed biocontrol insect is limited to spillover, reducing the chance of population-level impacts. References Ecol Appl 25:517–530 Catton HA, Lalonde RG, Buckley YM, De Clerck-Floate RA (2016) Biocontrol insect impacts population growth of its Ancheta J, Heard SB (2011) Impacts of insect herbivores on rare target plant species but not an incidentally used nontarget. plant populations. Biol Conserv 144:2395–2402 Ecosphere 7:e01280. https://doi.org/10.1002/ecs2.1280 Andreas JE, Schwarzla¨nder M, de Clerck-Floate R (2008) The Clement SL, Cristofaro M (1995) Open field tests in host- occurrence and potential relevance of post-release, non- specificity determination of insects for biological control of target attack by Mogulones cruciger, a biocontrol agent for weeds. Biocontrol Sci Technol 5:395–406 Cynoglossum officinale in Canada. Biol Control Clewley GD, Eschen R, Shaw RH, Wright DJ (2012) The 46:304–311 effectiveness of classical biological control of invasive Arnett AE, Louda SM (2002) Re-test of Rhinocyllus conicus plants. J Appl Ecol 49:1287–1295 host specificity, and the prediction of ecological risk in Crawley MJ (1989) The successes and failures of weed bio- biological control. Biol Conserv 106:251–257 control using insects. Biocontrol News Inf 10:213–223 Bernays EA, Chapman RF (1994) Host-plant selection by Cristofaro M, De Biase A, Smith L (2013) Field release of a phytophagous insects. Contemporary topics in entomol- prospective biological control agent of weeds, Ceratapion ogy. Springer, New York basicorne, to evaluate potential risk to a nontarget crop. Bernays E, Graham M (1988) On the evolution of host speci- Biol Control 64:305–314 ficity in phytophagous arthropods. Ecology 69:886–892 Cronin G, Schlacher T, Lodge DM, Siska EL (1999) Blossey B (2016a) The future of biological control: a proposal Intraspecific variation in feeding preference and perfor- for fundamental reform. In: van Driesche R, Simberloff D, mance of Galerucella nymphaeae (Chrysomelidae: Blossey B, Causton C, Hoddle M, Marks C, Heinz K, Coleoptera) on aquatic macrophytes. J N Am Bent Soc Wagner D, Wagner K (eds) Integrating biological control 18:391–405 into conservation practice. Wiley, Chichester, UK, Davis AS, Landis DA, Nuzzo V, Blossey B, Gerber E, Hinz HL pp 314–328 (2006) Demographic models inform selection of biocontrol Blossey B (2016b) Measuring and evaluating ecological out- agents for garlic mustard (Alliaria petiolata). Ecol Appl comes of biological control introductions. In: van Driesche 16:2399–2410 R, Simberloff D, Blossey B, Causton C, Hoddle M, Marks De Clerck-Floate R, Schwarzla¨nder M (2002) Host specificity of C, Heinz K, Wagner D, Wagner K (eds) Integrating bio- Mogulones cruciger (Coleoptera: Curculionidae), a bio- logical control into conservation practice. Wiley, Chich- control agent for houndstongue (Cynoglossum officinale), ester, UK, pp 161–188 with emphasis on testing of native North American Bor- Blossey B, Casagrande R, Tewksbury L, Landis DA, Wieden- aginaceae. Biocontrol Sci Technol 12:293–306 mann RN, Ellis DR (2001) Nontarget feeding of leaf-bee- DePrenger-Levin ME, Grant TA, Dawson C (2010) Impacts of tles introduced to control purple loosestrife (Lythrum Rhinocyllus conicus the introduced biocontrol agent, salicaria L.). Nat Areas J 21:368–377 (Coleoptera: Curculionidae), on the seed production and Briese DT (2005) Translating host-specificity test results into population dynamics of Cirsium ownbeyi (Asteraceae), a the real world: the need to harmonize the yin and yang of rare, native thistle. Biol Control 55:79–84 current testing procedures. Biol Control 35:208–214 DeWalt SJ (2006) Population dynamics and potential for bio- Buckley YM, Rees M, Paynter Q, Lonsdale M (2004) Modelling logical control of an exotic invasive shrub in Hawaiian integrated weed management of an invasive shrub in rainforests. Biol Invasion 8:1145–1158 tropical Australia. J Appl Ecol 41:547–560 123 472 B. Blossey et al. Ding J, Blossey B (2005) Impact of the native water lily leaf Kettenring KM, Adams CR (2011) Lessons learned from inva- beetle, Galerucella nymphaeae (Coleoptera: Chrysomeli- sive plant control experiments: a systematic review and dae), attacking introduced water chestnut, Trapa natans,in meta-analysis. J Appl Ecol 48:970–979 the northeastern United States. Environ Entomol Kunii H (1988) Longevity and germinability of buried seed in 34:683–689 Trapa sp. Mem Fac Sci Shimane Univ 22:83–91 Ding J, Blossey B (2009) Difference in preference and perfor- Louda SM, Kendall D, Connor J, Simberloff D (1997) Eco- mance of the water lily leaf beetle, Galerucella nymhaeae logical effects of an insect introduced for the biological populations on native and invasive aquatic plants. Environ control of weeds. Science 277:1088–1090 Entomol 38:1653–1660 Louda SM, Pemberton RW, Johnson MT, Follett PA (2003) Ding J, Blossey B, Du Y, Zheng F (2006a) Galerucella bir- Nontarget effects—the Achilles’ heel of biological control: manica (Coleoptera: Chrysomelidae), a promising poten- retrospective analyses to reduce risk associated with bio- tial biological control agent of water chestnut, Trapa control introductions. Annu Rev Entomol 48:365–396 natans. Biol Control 36:80–90 Louda SM, Rand TA, Arnett AE, McClay AS, Shea K, Ding J, Blossey B, Du Y, Zheng F (2006b) Impact of Galeru- McEachern AK (2005a) Evaluation of ecological risk to cella birmanica (Coleoptera: Chrysomelidae) on growth populations of a threatened plant from an invasive bio- and seed production of Trapa natans. Biol Control control insect. Ecol Appl 15:234–249 37:338–345 Louda SM, Rand TA, Russell FL, Arnett AE (2005b) Assess- Downey PO, Paterson ID (2016) Encompassing the relative ment of ecological risks in weed biocontrol: input from non-target risks from agents and their alien plant targets in retrospective ecological analyses. Biol Control biological control assessments. BioControl 61:615–630 35:253–264 Eckberg JO, Tenhumberg B, Louda SM (2014) Native insect Marohasy J (1996) Host shifts in biological weed control: real herbivory limits population growth rate of a non-native problems, semantic difficulties or poor science? Int J Pest thistle. Oecologia 175:129–138 Manag 42:71–75 Fowler SV, Syrett P, Hill RL (2000) Success and safety in the McEvoy PB, Coombs EM (1999) Biological control of plant biological control of environmental weeds in New Zealand. invaders: regional patterns, field experiments, and struc- Aust Ecol 25:553–562 tured population models. Ecol Appl 9:387–401 Fowler SV, Paynter Q, Dodd S, Groenteman R (2012) How can McEvoy PB, Higgs KM, Coombs EM, Karacetin E, Starcevich ecologists help practitioners minimize non-target effects in LA (2012) Evolving while invading: rapid adaptive evo- weed biocontrol? J Appl Ecol 49:307–310 lution in juvenile development time for a biological control Futuyma DJ, Agrawal AA (2009) Macroevolution and the bio- organism colonizing a high-elevation environment. Evol logical diversity of plants and herbivores. Proc Nat Acad Appl 5:524–536 Sci USA 106:18054–18061 McFadyen REC (1998) Biological control of weeds. Annu Rev Groenteman R, Fowler SV, Sullivan JJ (2011) St. John’s wort Entomol 43:369–393 beetles would not have been introduced to New Zealand McGowan CP, Runge MC, Larson MA (2011) Incorporating now: a retrospective host range test of New Zealand’s most parametric uncertainty into population viability analysis successful weed biocontrol agents. Biol Control 57:50–58 models. Biol Conserv 144:1400–1408 Groth AT, LovettDoust L, LovettDoust J (1996) Population Moran VC, Hoffmann JH (2015) The fourteen international density and module demography in Trapa natans (Tra- symposia on biological control of weeds, 1969-2014: del- paceae), an annual, clonal aquatic macrophyte. Am J Bot egates, demographics and inferences from the debate on 83:1406–1415 non-target effects. Biol Control 87:23–31 Guretzky JA, Louda SM (1997) Evidence for natural biological Moran VC, Hoffmann JH, Zimmermann HG (2005) Biological control: insects decrease survival and growth of a native control of invasive alien plants in South Africa: necessity, thistle. Ecol Appl 7:1330–1340 circumspection, and success. Front Ecol Environ 3:77–83 Hare D, Blossey B (2014) Principles of public trust thinking. Paynter QE, Fowler SV, Gourlay AH, Haines ML, Harman HM, Hum Dimens Wildl 19:397–406 Hona SR, Peterson PG, Smith LA, Wilson-Davey JA, Hinz HL, Schwarzla¨nder M, Gassmann A, Bourchier RS (2014) Winks CJ, Withers TM (2004) Safety in New Zealand Successes we may not have had: a retrospective analysis of weed biocontrol: a nationwide survey for impacts on non- selected weed biological control agents in the United target plants. N Z Plant Prot 57:102–107 States. Invasive Plant Sci Manag 7:565–579 Paynter Q, Martin N, Berry J, Hona S, Peterson P, Gourlay AH, Jezorek H, Baker AJ, Stiling P (2012) Effects of Cactoblastis Wilson-Davey J, Smith L, Winks C, Fowler SV (2008) cactorum on the survival and growth of North American Non-target impacts of Phytomyza vitalbae a biological Opuntia. Biol Invasion 14:2355–2367 control agent of the European weed Clematis vitalba in Johnson DM, Stiling PD (1996) Host specificity of Cactoblastis New Zealand. Biol Control 44:248–258 cactorum (Lepidoptera: Pyralidae), an exotic Opuntia- Paynter Q, Fowler SV, Gourlay AH, Peterson PG, Smith LA, feeding moth, in Florida. Environ Entomol 25:743–748 Winks CJ (2015) Relative performance on test and target Kerr NZ, Baxter PWJ, Salguero-Gomez R, Wardle GM, plants in laboratory tests predicts the risk of non-target Buckley YM (2016) Prioritizing management actions for attack in the field for arthropod weed biocontrol agents. invasive populations using cost, efficacy, demography and Biol Control 80:133–142 expert opinion for 14 plant species world-wide. J Appl Ecol Pearson DE, Ortega YK, Runyon JB, Butler JL (2016) Sec- 53:305–316 ondary invasion: the bane of weed management. Biol Conserv 197:8–17 123 A proposal to use plant demographic data to assess potential weed biological control… 473 Pemberton RW (1995) Cactoblastis cactorum (Lepidoptera: Simberloff D, Stiling P (1996) How risky is biological control? Pyralidae) in the United States: an imigrant biological Ecology 77:1965–1974 control agent or an introduction of the nursery industry? Smilanich AM, Fincher RM, Dyer LA (2016) Does plant Am Entomol 41:230–232 apparency matter? Thirty years of data provide limited Pemberton RW (2000) Predictable risk to native plants in weed support but reveal clear patterns of the effects of plant biocontrol. Oecologia 125:489–494 chemistry on herbivores. New Phytol 210:1044–1057 Pemberton RW, Liu H (2007) Control and persistence of native Smith M, Caswell H, Mettler-Cherry P (2005) Stochastic flood Opuntia on Nevis and St. Kitts 50 years after the intro- and precipitation regimes and the population dynamics of a duction of Cactoblastis cactorum. Biol Control threatened floodplain plant. Ecol Appl 15:1036–1052 41:272–282 Stiling P, Rossi A, Gordon D (2000) The difficulties of single R Core Team (2016) R: a language and environment for sta- factor thinking in restoration: replanting a rare cactus in the tistical computing. R Foundation for Statistical Comput- Florida Keys. Biol Conserv 94:327–333 ing, Vienna, Austria. http://www.r-project.org/ Stubben CJ, Milligan BG (2007) Estimating and analyzing Raghu S, Wilson JR, Dhileepan K (2006) Refining the process of demographic models using the popbio package in R. J Stat agent selection through understanding plant demography Softw 22:1–23 and plant response to herbivory. Aust J Entomol Suckling DM, Sforza RFH (2014) What magnitude are observed 45:308–316 non-target impacts from weed biocontrol? PLoS ONE Rand TA, Louda SM (2004) Exotic weed invasion increases the 9(1):e84847. https://doi.org/10.1371/journal.pone. susceptibility of native plants attack by a biocontrol her- 0084847 bivore. Ecology 85:1548–1554 Swope SM, Satterthwaite WH, Parker IM (2017) Spatiotem- Rose KE, Louda SM, Rees M (2005) Demographic and evolu- poral variation in the strength of density dependence: tionary impacts of native and invasive insect herbivores on implications for biocontrol of Centaurea solstitialis. Biol Cirsium canescens. Ecology 86:453–465 Inv 19:2675–2691 Salguero-Gomez R, Jones OR, Archer CR, Buckley YM, Che- Takahashi M, Louda SM, Miller TEX, O’Brien CW (2009) Castaldo J, Caswell H, Hodgson D, Scheuerlein A, Conde Occurrence of Trichosirocalus horridus (Coleoptera: DA, Brinks E, de Buhr H, Farack C, Gottschalk F, Hart- Curculionidae) on native Cirsium altissimum versus exotic mann A, Henning A, Hoppe G, Roemer G, Runge J, Ruoff C. vulgare in North American tallgrass prairie. Environ T, Wille J, Zeh S, Davison R, Vieregg D, Baudisch A, Entomol 38:731–740 Altwegg R, Colchero F, Dong M, de Kroon H, Lebreton USDA (2016) Technical advisory group for biological control JD, Metcalf CJE, Neel MM, Parker IM, Takada T, Val- agents of weeds manual, Interim edn. United States verde T, Velez-Espino LA, Wardle GM, Franco M, Vaupel Department of Agriculture, Washington, DC, USA JW (2015) The COMPADRE plant matrix database: an van Klinken RD, Edwards OR (2002) Is host specificity of weed open online repository for plant demography. J Ecol biocontrol agents likely to evolve rapidly following 103:202–218 establishment? Ecol Lett 5:590–595 Sauby KE, Kilmer J, Christman MC, Holt RD, Marsico TD Vigueras AL, Portillo L (2001) Uses of Opuntia species and the (2017) The influence of herbivory and weather on the vital potential impact of Cactoblastis cactorum (Lepidoptera: rates of two closely related cactus species. Ecol Evol Pyralidae) in Mexico. Fla Entomol 84:493–498 7:6996–7009 Wapshere AJ (1974) A strategy for evaluating the safety of Shea K, Kelly D (1998) Estimating biocontrol agent impact with organisms for biological weed control. Ann Appl Biol matrix models: Carduus nutans in New Zealand. Ecol Appl 77:200–211 8:824–832 Williams BK, Nichols JD, Conroy MJ (2001) Analyses and Shea K, Kelly D, Sheppard AW, Woodburn TL (2005) Context- management of animal populations. Academic Press, New dependent biological control of an invasive thistle. Ecology York 86:3174–3181 Winkler I, Mitter C (2008) The phylogenetic dimension of Sheppard AW, Warner KD (2016) Societal values expressed insect/plant interactions: a summary of recent evidence. In: through policy and regulation concerning biological con- Tilmon K (ed) Specialization, speciation, and radiation: the trol releases. In: van Driesche R, Simberloff D, Blossey B, evolutionary biology of herbivorous insects. University of Causton C, Hoddle M, Marks C, Heinz K, Wagner D, California Press, Berkeley, pp 240–263 Wagner K (eds) Integrating biological control into con- Winston RL, Schwarzla¨nder M, Hinz HL, Day MD, Cock MJW, servation practice. Wiley, Chichester, pp 247–263 Julien MH (eds) (2014) Biological control of weeds: a Sheppard AW, van Klinken RD, Heard TA (2005) Scientific world catalogue of agents and their target weeds, 5th ed. advances in the analysis of direct risks of weed biological USDA Forest Service, Forest health technology enterprise control agents to nontarget plants. Biol Control team, Morgantown, West Virginia, USA. FHTET-2014-04 35:215–226 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BioControl Springer Journals

A proposal to use plant demographic data to assess potential weed biological control agents impacts on non-target plant populations

Free
13 pages

Loading next page...
 
/lp/springer_journal/a-proposal-to-use-plant-demographic-data-to-assess-potential-weed-SOamHZ3jVi
Publisher
Springer Netherlands
Copyright
Copyright © 2018 by The Author(s)
Subject
Life Sciences; Entomology; Plant Pathology; Agriculture; Animal Ecology; Animal Biochemistry; Behavioral Sciences
ISSN
1386-6141
eISSN
1573-8248
D.O.I.
10.1007/s10526-018-9886-4
Publisher site
See Article on Publisher Site

Abstract

BioControl (2018) 63:461–473 https://doi.org/10.1007/s10526-018-9886-4 A proposal to use plant demographic data to assess potential weed biological control agents impacts on non-target plant populations . . . Bernd Blossey Andrea Davalos Wade Simmons Jianqing Ding Received: 14 November 2017 / Accepted: 25 April 2018 / Published online: 28 April 2018 The Author(s) 2018 Abstract Weed biocontrol programs aim to reduce assessments, are essential to guide weed biocontrol the spread and population growth rate of the target programs. We propose to add use of plant demography plant while stabilizing or increasing populations of (an assessment of how environmental factors and those native species considered under threat by ecological interactions, for example competition, invasive plants. This goal is not unique to weed disease or herbivory, may affect plant populations by biocontrol but applies to all other invasive plant altering survival, growth, development and reproduc- management techniques, though such information is tive rates of plant individuals) during host specificity rarely collected. Without this information, success of risk assessments of potential biological control agents. management interventions can be ambiguous, and Demographic models can refine assessments of poten- regulatory agencies, the public, policy makers, funders tial impacts for those plant species that experience and land managers cannot be held accountable for some feeding or larval development during host chosen treatments. A fundamental reform, including specificity testing. Our proposed approach to focus use of demographic studies and long-term on impact on plant demography instead of attack on plant individuals is useful in appropriately gauging threats potential weed biocontrol agents may pose to Handling Editors: Mark Schwarzla¨nder, Cliff Moran and S. non-target species after field release. Raghu. Keywords Demography  Host specificity  Non Electronic supplementary material The online version of target effects  Risk assessment  Trapa natans L. this article (https://doi.org/10.1007/s10526-018-9886-4) con- tains supplementary material, which is available to authorized Weed biocontrol users. B. Blossey (&)  W. Simmons Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA Introduction e-mail: bb22@cornell.edu Biological weed control programs aim to find organ- A. Da´valos Biological Sciences, SUNY Cortland, 1215 Bowers Hall, isms able to reduce spread and population growth rate Cortland, NY 13045, USA of target plants, while avoiding non-target impacts. The track record of weed biocontrol over the past J. Ding century is decidedly mixed, since only a third of all College of Life Sciences, Henan University, weed biocontrol programs achieve at least partial Kaifeng 475004, Henan, China 123 462 B. Blossey et al. suppression of targets (Crawley 1989; Fowler et al. 2000; Moran et al. 2005). Many biocontrol agents fail to establish, or fail to control host plants (Crawley 1989; McFadyen 1998). On the other hand, while occasionally contested, the safety record of weed biocontrol is superior to other management methods, while economic and ecological benefits can be enor- mous and continue to accrue (Moran et al. 2005; Suckling and Sforza 2014). Following publications of high profile cases of non- target attack by Rhinocyllus conicus Fro¨lich (Cur- culionidae) and Cactoblastis cactorum Berg (Pyrali- Fig. 1 Schematic design of typical proposed host specificity dae), respectively, changes in decision making testing protocol for potential weed biocontrol agents. Depending processes in regulatory agencies, particularly in the on life history and feeding mode of the herbivore under USA, shifted to a greater reliance on fundamental consideration, test conditions may vary. Pool 1 includes all plant species proposed for host specificity testing that are tested using host-range data, a change that further threatens release highly constrained no-choice conditions (Screen 1). Those even of highly specific agents (Hinz et al. 2014). The species that could not be eliminated in the first screening step irony of this change in risk perception is that specific constitute pool 2 which are tested using more sophisticated and successful agents of the past would have difficul- designs, such as multiple-choice tests using potted plants or ties passing through current approval processes larger cages (screen 2). Species in pool 3 include plant species that were still attacked under the more sophisticated test (Groenteman et al. 2011; Hinz et al. 2014). At a time conditions, or where larvae completed development. Tests when it is becoming increasingly evident that many utilized during screen 3 depend on herbivore feeding niches and invasive species control methods, particularly chem- logistical and regulatory considerations but include use of ical management, are unable to achieve lasting control common gardens, multiple choice tests without containment, etc. Only those species that were still attacked under the most and may in fact threaten non-target species (Ketten- realistic conditions possible in a particular program would then ring and Adams 2011; Pearson et al. 2016), we argue be considered candidate species for demographic analyses that it is time for fundamental reform of risk assess- (screen 4). Note that the pool of species shrinks with each test, ment and decision making processes in invasive plant while the realism of testing conditions and their ecological relevance increases management and weed biocontrol that is guided by appropriate scientific information and open dialogue, highly specific herbivores. We propose to utilize plant not fear (Blossey 2016b). demography (an assessment of how environmental We propose that adoption of modern scientific tools factors and ecological interactions, for example com- focusing on demographic impacts of herbivores could petition, disease or herbivory, may affect plant pop- constitute a breakthrough development in maintaining ulations by altering survival, growth, development and safety while increasing the ability to select effective reproductive rates of plant individuals) (Salguero- herbivores. We consider it paramount to shift non- Gomez et al. 2015) to assess potential threats of target risk assessments away from damage to individ- candidate biocontrol agents to non-target species. This uals to population level effects expected after field approach aims to provide a means by which to releases. We envision that traditional reductionist evaluate potential impacts to non-target plant popula- approaches (no-choice, small herbivore confinements, tions. Our proposal constitutes a significant shift in the followed by multiple-choice or potted plant experi- way weed biocontrol researchers, review panels and ments) will continue to be the mainstay of host others may look at the approval and risk assessment specificity testing. These tests are valuable because the process—but it is a scientifically valid and biologi- vast majority of test plant species will not be attacked cally meaningful one. We are not concerned by host even under constrained conditions (Fig. 1). However, use but by negative impacts to populations of non- in many programs often a few test plant species remain target species. This is not a reduction in protections that may be fed upon, are accepted for oviposition, or afforded to native species as it continues to safeguard even allow larval development (albeit at a greatly all native species or valuable introduced species that reduced rate compared to original host plants) by 123 A proposal to use plant demographic data to assess potential weed biological control… 463 have attained cultural, ornamental or agricultural ranges are always narrower than experimentally significance. We argue that to ‘‘safeguard’’ means determined fundamental host ranges. that populations of non-target organisms are main- tained and do not suffer demographic declines due to biocontrol agent introductions. Cosmetic damage or Evidence for threats of weed biocontrol agents even substantial damage to, or death of, individuals to (rare) native plant species does not necessarily indicate demographic or ecolog- ical consequences. Reports of weed biocontrol agents attacking non- The shift we propose will find resistance based on target species do exist, including spillover events with risk perceptions regarding safety of non-target plants substantial temporary defoliation of non-target species due to concerns that herbivores introduced to control (Blossey et al. 2001; Louda et al. 1997, 2003; Paynter introduced plants will (1) attack (rare) native species et al. 2008; Pemberton 2000; Suckling and Sforza leading to declines in populations of these species; and 2014). Comprehensive reviews assessing weed bio- (2) that diet restriction (i.e. specificity) of weed control outcomes (Blossey et al. 2001; Suckling and biocontrol agents are ‘‘fluid’’ and change over time, Sforza 2014), conclude that [ 90% of all biocontrol leading to attack and unintended negative conse- agents never attack non-target species. The majority of quences for native species. We will briefly review non-target feeding is attributed to spillover events and Suckling and Sforza (2014) report such attacks on 128 evidence for these concerns before further developing our proposal to use demography in host specificity risk non-target plants. Host specificity testing appears assessments. However, we first provide a primer on unable to predict identity of these species, but physical terminology used to describe plant–herbivore interac- proximity may explain some of it (Blossey et al. 2001). tions because we believe that some differences in However, occasional or prolonged host use appears perceived risk perception are semantic. highly predictable using fundamental host range data (Paynter et al. 2015). Fewer than ten biocontrol agents have established populations on non-target species, a Terminology used in describing plant–herbivore risk that was known, and accepted by regulatory interactions and weed biocontrol programs agencies, at time of their introduction (Blossey et al. 2001). Of these, only three, R. conicus, C. cactorum Ecologists typically refer to diets of herbivores using and Trichosirocalus horridus (Panzer) (Curculion- terms like specialists or generalists (Smilanich et al. idae) may have effects that reduce populations and 2016) or more specifically monophagous (feeding on a growth rates of non-target species (Louda et al. 1997; single or few species within a genus), oligophagous Suckling and Sforza 2014; Takahashi et al. 2009). (utilizing several plant species, typically in different None of these herbivores would be approved under genera), and polyphagous (using different plant current decision making frameworks (McFadyen species in different genera and families) (Bernays 1998; Suckling and Sforza 2014). and Chapman 1994). In contrast, weed biocontrol Detailed documentation of non-target plant species researchers typically focus on herbivores using a occasionally attacked by biocontrol agents offer single plant species. Furthermore, ‘‘use’’ in the assurances that significant non-target effects have ecological and evolutionary literature typically refers not gone unrecognized or unreported—in this case to plants chosen for oviposition and allowing larval absence of evidence indicates evidence of absence of development in the field, recognized as realized host such effects and not just lack of effort. A recent range in weed biocontrol. Experimental host-speci- literature survey of threats by insect herbivores to rare ficity testing aims to (1) elucidate the fundamental host plants concluded that with exception of R. conicus and range (plant species acceptable for feeding, oviposi- C. cactorum, ‘‘currently this threat is either seldom tion and larval development using no-choice tests in realized (perhaps because of extensive pre-release the absence of the original host), and (2) provide screening in modern biocontrol programs) or else additional data using less constrained and increasingly seldom documented’’ (Ancheta and Heard 2011). ecologically realistic testing procedures to allow forecasting of realized host ranges. Realized field host 123 464 B. Blossey et al. Lack of evidence for evolution of dietary Species accumulation curves on novel host plants preferences in weed biocontrol agents plateau in approximately 100 years for generalists and 500–10,000 years for specialists (Bernays and Gra- Permitting processes for biocontrol agent releases may ham 1988). However, the vast majority of phy- differ widely among countries (Sheppard and Warner tophagous insects show ‘‘phylogenetic 2016), but host specificity tests are widely standard- conservatism’’ retaining their association with plant ized (Wapshere 1974). Despite further refinements taxa over millions of years with \ 10% of speciation proposed and implemented in subsequent years events including a shift to a different plant family (Briese 2005; Clement and Cristofaro 1995; Sheppard (Winkler and Mitter 2008). Biocontrol agents passing et al. 2005; USDA 2016), this sequence of testing has through host range testing, as far as we can tell from largely remained state-of-the-art, providing over- decades of observation and study, appear particularly whelmingly safe weed biocontrol agents. There is no ‘‘conservative’’. evidence that fundamental host ranges of biocontrol We now return to our argument that use of agents have evolved (Arnett and Louda 2002; Maro- demographic models should be a desired and required hasy 1996; Paynter et al. 2004; Sheppard et al. 2005; tool during risk assessment of biocontrol agents. We van Klinken and Edwards 2002), despite dire warnings are not the first to propose such new tools (Louda et al. (Simberloff and Stiling 1996). There is, however, 2005a; Raghu et al. 2006; Sauby et al. 2017), although evidence for evolution of improved performance on we believe we are the first to ask that this becomes part non-target plants (McEvoy et al. 2012) and we of pre-release risk assessments. We will briefly acknowledge that few formal assessments have been introduce concepts of demographic modeling and made. then provide examples how demography has, and can However, occasional use, even if predicted, of be utilized in weed biocontrol. To the best of our species that are not targets of weed biocontrol, and knowledge, no biocontrol program has used demo- frequent citation of the few species with anticipated graphic information to assess risks to non-targets large negative impacts, appears to be registered by before field releases, so we will rely on post-release non-biocontrol scientists as evidence for a poor track analyses and a hypothetical scenario involving a record and evolutionary malleable diet breadth of current target of biocontrol research, water chestnut, insect herbivores used in weed biocontrol. We argue Trapa natans L. (Lythraceae) to illustrate our that the problem arises due to the disconnect between proposal. perception of labile or rapidly evolving host ranges in weed biocontrol agents and available evidence. Weed biocontrol researchers have increased efforts to Using demography to evaluate biocontrol agent improve testing procedures to mitigate constraints of risks to non-target plants laboratory conditions that affect insect behavior to improve predictions of realized host ranges (Clement Demography and matrix population models (Caswell and Cristofaro 1995; Fowler et al. 2012; van Klinken 2001) are now common tools in biology (Caswell and and Edwards 2002). While such improvements are Salguero-Gomez 2013) and their use in invasion essential to increase reliability of predictions regard- biology and weed biocontrol is increasing (Carval- ing realized host ranges, the focus on improving heiro et al. 2008; DeWalt 2006; Eckberg et al. 2014; testing conditions appears to have prevented a discus- Kerr et al. 2016; McEvoy and Coombs 1999; Shea and sion among scientists and regulatory agencies regard- Kelly 1998; Swope et al. 2017). Technical background ing the appropriate meaning of safeguarding other is provided elsewhere (Caswell 2001; Caswell et al. species (Hinz et al. 2014). 2011; Caswell and Salguero-Gomez 2013; Williams Clearly, diets of insect herbivores change over time et al. 2001), but briefly development of demographic and both generalists and specialists may acquire new models requires estimating vital rates, the transition hosts (Futuyma and Agrawal 2009). Range expansions probabilities from one life stage to another. For plants through human aided introduction of novel plants or with clear developmental stages, a stage-based model insects provide enormous ecological and evolutionary requires estimation of transitions from seed, to opportunities for herbivores to adopt new hosts. seedling, to rosette to flowering plant, to seed output, 123 A proposal to use plant demographic data to assess potential weed biological control… 465 back to seed and seed bank (Davis et al. 2006; Shea Shea et al. 2005; Swope et al. 2017). These examples and Kelly 1998). Local abiotic conditions, competi- are important in recognizing that demographic tion, herbivores, stochasticity, density dependence and approaches are already an important part of the other processes may affect survival and the probability evaluation process in weed biocontrol. We will not that an individual will transition from one stage to the review these here but focus instead on efforts to help next. Vital rates can be inferred in the field by assess demographic impacts of weed biocontrol agents monitoring cohorts of marked individuals. on non-target plants after release. Twenty years have Demographic models can aide in assessments of passed since the initial widespread criticism regarding potential impacts of proposed biocontrol agents on safety of (weed) biocontrol (Louda et al. 1997; non-target plants that could not be excluded using Pemberton 2000; Simberloff and Stiling 1996). At traditional testing sequences (Fig. 1). We propose use least ten biocontrol agents have established popula- of experiments, for example by manipulating herbi- tions on non-target species and [ 120 non-target vore access or attack rates and then measuring stage plants are reported to be attacked (Blossey et al. specific reductions in survival, recruitment, growth, 2001; Suckling and Sforza 2014), thus we expected to biomass, or seed output of non-target plants, which find numerous publications outlining demographic can be done in common gardens, or other confine- consequences, or at least attempts to evaluate conse- ments when insects are not approved for release. quences of such attacks. Our Web of Science searches uncovered few studies, which may indicate that they Constructing and populating models with data, and analyzing model performance under different scenar- either were not conducted, did not get published, or ios (often referred to as perturbation analysis) allows were deemed unimportant to conduct or fund. We comparisons of contributions made by different vital therefore focus on R. conicus and C. cactorum, species rates for overall population growth rates (Caswell that according to categorization by Suckling and 2000). The outcome of these exercises is the ability to Sforza (2014) have ‘‘massive’’ non-target impacts, forecast population growth rates (k), population plus post-release evaluations of Mogulones crucifer fluctuation and potential extinction risk, and the Pallas (Curculionidae), a species approved for release sensitivity of growth rates to small changes in vital against houndstongue (Cynoglossum officinale L.) rate values, regardless of which management action is (Boraginaceae) in Canada (Catton et al. 2016). applied (Kerr et al. 2016). We recognize that demo- Apparently, studies evaluating demographic effects graphic approaches during evaluation of potential of T. horridus beyond documentation of attack on a biocontrol agents will have to contend with many non-target plants do not exist, therefore we exclude different obstacles, the smallest among them may be this species. lack of familiarity of biocontrol scientists with demo- A high-profile paper regarding non-target effects of graphic modeling (Blossey 2016b). But this is a small R. conicus (Louda et al. 1997) tabulated attack rates on price to pay for the ability to improve predictability of native thistles, but fell short of documenting demo- impacts to targets or risks to non-target organisms. graphic effects, which were strongly implied due to Furthermore, a fast growing and increasingly utilized seed limitation and large demographic impacts by open access database, COMPADRE, provides a native seed feeders on Cirsium altissimum (L.) Spreng potentially important resource to inform construction (Asteraceae) (Guretzky and Louda 1997). Additional of appropriate models for species of interest (Sal- investigations clearly documented demographic guero-Gomez et al. 2015). threats by R. conicus (Louda et al. 2005b) based both on field and laboratory data. But effects appear context-dependent and do not occur every year and Retrospective demographic analyses for target in every location (Rand and Louda 2004; Rose et al. and non-target effects in weed biocontrol 2005). In addition, some native thistles show positive population growth rates even in the presence of and Demographic modeling has been used to understand attack by R. conicus (DePrenger-Levin et al. 2010). success or failure of weed biocontrol programs in Furthermore, results of demographic models to assess reducing target plant population growth rates (Buck- population growth rates for Platte thistle, Cirsium ley et al. 2004; DeWalt 2006; Shea and Kelly 1998; canescens Nutt. concluded that impacts may be 123 466 B. Blossey et al. substantial, but variable in space and time and not as models incorporating more than presence of herbivore catastrophic as previously feared (Rose et al. 2005). attack and other ‘‘stressors’’ are we able to gauge While R. conicus should have never been approved for impacts appropriately. As in the case of R. conicus, release, current evidence is of widespread attack on anticipated ‘‘massive’’ impacts of C. cactorum are, native Cirsium species, but evidence for predicted according to published studies, not currently materi- massive negative demographic non-target effects alizing in the field. (sensu Suckling and Sforza 2014) has not been Risk assessment after release of M. crucifer, a root presented at this time. feeding weevil that attacked some Boraginaceae, The accidental introduction of C. cactorum to including some US native and rare plants, during host North America (Pemberton 1995) raised concerns specificity testing (De Clerck-Floate and Sch- over safety of native North and Central American warzla¨nder 2002), provides a good example of an Opuntia spp. (Cactaceae) (Vigueras and Portillo application of matrix population models. Canadian 2001), particularly for rare endemics, such as Opuntia authorities granted release permits and M. crucifer corallicola Small where only 12 known individuals established and began to spread in British Columbia, existed in the Florida Keys (Johnson and Stiling 1996). prompting fears about non-target attacks upon arrival Follow-up work, including using plant demography, in the USA (Andreas et al. 2008). Additional host over the past two decades has delivered a more refined specificity testing, including field tests in British view of realized threats. While initial introductions to Columbia, also documented non-target attack by M. Nevis and St. Kitts in the Lesser Antilles to control crucifer but found minor adult feeding and infrequent weedy native Opuntia spp. was ill advised, a survey larval development, despite ability of the weevil to 50 years after C. cactorum releases showed that the complete development under no-choice conditions targeted native species Opuntia triacantha (Willd.) (De Clerck-Floate and Schwarzla¨nder 2002). Sweet and Opuntia stricta (Haw.) Haw. remain under Subsequent monitoring showed that M. crucifer did biological control while the native tree pear Consolea not establish at sites where C. officinale was absent rubescens (Salm-Dyck ex DC) Lem. (Cactaceae) was (Catton et al. 2015) and attack of non-target species not attacked and the cultivated and naturalized Opun- tapered off within a few meters (Catton et al. 2014), tia cochenillifera (L.) Mill showed limited feeding by including during spillover events. Furthermore, C. cactorum (Pemberton and Liu 2007). In the detailed demographic work on Hackelia micrantha Southeastern USA, C. cactorum has spread rapidly, (Eastw.) J. L. Gentry (Boraginaceae), a native plant resulting in variable impacts depending on cactus species regularly attacked in the field, demonstrated species, often resulting in size decreases and reduction that while population growth rates for C. officinale in relative growth rates (Sauby et al. 2017). Jezorek were reduced below replacement rates (k \ 1), H. et al. (2012) summarized these findings as follows: micrantha benefitted from C. officinale reductions ‘‘although C. cactorum should still be considered a (Catton et al. 2016). These results indicate that while threat, particularly for rare opuntioids, overall survival individual H. micrantha are being attacked and allow along the west central Florida coast is currently high larval development of M. crucifer, the species is safe and plants that are able to survive C. cactorum attack and suffers no harm at the population level (Catton are not being reduced in size, possibly because they et al. 2016). Similar demographic experiments with possess traits that render them more tolerant of C. rare plants that are part of the fundamental host range cactorum damage. Our findings suggest that an of M. crucifer could help evaluate real (vs. feared) assumption of severe negative effects of an invasive threats to other US native Boraginaceae. species, based on its effects in other regions or over These examples showcase the value of detailed short periods of time, may not always be justified’’. In demographic studies to assess how attack by biocon- the case of the rare endemic O. corallicola, detailed trol agents may, or may not, contribute to harm, or studies and restoration efforts revealed that salinity, endangerment of non-target species. Only through moisture conditions, hurricanes, trampling by deer, such detailed work are we able to separate anecdotal and stem rot over the past two decades were more observation of attack from contributions of many important demographic threats than C. cactorum factors (habitat loss and fragmentation, inbreeding (Stiling et al. 2000). Only by developing detailed depression, succession, disturbance, climate, abiotic 123 A proposal to use plant demographic data to assess potential weed biological control… 467 conditions, competition, other natural enemies, etc.) growth and reproduction of T. natans at different that affect plant demography simultaneously. larval densities (0–50 L per rosette) (Ding and To the best of our knowledge, no biocontrol Blossey 2005; Ding et al. 2006b). We combine program has used a demographic analysis to assess herbivore impact data with demographic plant data herbivore impact on non-target plants pre-release or as obtained in outdoor aquatic mesocosms to build part of a release petition. Raghu et al. (2006) proposed demographic models that project future plant popula- to use modeling a priori, but this recommendation tion dynamics under different beetle herbivory sce- followed rejection of a herbivore by Australian narios. These data were initially collected and the authorities due to minor feeding on a non-target plant. demographic model built to assess utility of each We believe weed biocontrol researcher should strive herbivore as a biocontrol agent. By considering, for to make this standard practice when promising species sake of argument, T. natans as a non-target species, we fail traditional testing sequences (Fig. 1). Embracing can show how biocontrol programs can use demogra- this approach can lead to important collaborations phy in risk evaluation for non-target species. Full with those concerned about native species and descriptions of the herbivore impact studies are academics with specialized knowledge (Clewley available elsewhere (Ding and Blossey 2005; Ding et al. 2012). et al. 2006b) and results and details of our experimen- To further develop our proposal to use plant tal design to collect demographic data in aquatic demography models in pre-introduction risk assess- mesocosms using four different plant populations ments, we now turn to T. natans. We present data collected in Massachusetts (MA), Rhode Island (RI), collection procedures used to develop a demographic New York (NY), and Virginia (VA), USA are detailed model for T. natans. For the sake of argument, we in Supplementary Materials, Section 2. assume T. natans to be a non-target species, and we further assume that two different chrysomelid beetles Model development are potential biocontrol agents that failed to be cleared in traditional host specificity screening. We incorpo- We evaluated differences in population growth rate of rate the feeding impacts of each herbivore into the T. natans with periodic matrix population models. demographic model to evaluate the risk each agent Periodic matrix models are suited to explore within may pose to plant demography. year transitions of annual organisms, such as T. natans (Caswell 2001). These models partition life history transitions into m phases defined by a matrix (B ) that Evaluation of two herbivores attacking T. natans projects the population into the next phase (h; where h =1…m). Population over entire cycles is given by The Eurasian T. natans is a floating aquatic annual the product of period matrices: A ¼ B B ...B ; m m1 1 plant invasive in North America where it is attacked by where A is the annual transition matrix. We partitioned the native water lily leaf beetle, Galerucella nym- T. natans life cycle into three phases (Fig. 2): fall- phaeae L. while the extremely similar Galerucella spring (seedbank or new seeds germinate), spring– birmanica Jacoby (Chrysomelidae) attacks T. natans summer (rosettes grow small or large), and summer- in Asia (Ding and Blossey 2005; Ding et al. 2006a, b). fall (reproduction). We classified individuals in each Both herbivores are multivoltine and while G. life stage as seedbank, seeds, small or large plants and nymphaeae is oligophagous with multiple host races determined plant size as a function of surface area. We (Cronin et al. 1999), G. birmanica is host-specific to T. used census data from the common garden to estimate natans, although it occasionally lays a few eggs on transition and germination rates and published data to water shield, Brasenia schreberi J. Gmelin (Cabom- calculate seedbank longevity (Kunii 1988) (Supple- baceae), the only other plant reported to be attacked in mentary Materials, Table S3). We calculated annual the field (Ding et al. 2006a) (natural histories of T. population growth rate (k) given by the dominant natans, B. schreberi, G. nymphaeae and G. birmanica eigenvalue of A and used bootstrap methods to are provided in Supplementary Materials, Section 1). estimate 95% confidence intervals for each population Both herbivores were evaluated as potential biocontrol (defined as the 2.5 and 97.5% quantiles from a agents in experiments that assessed their impact on distribution based on resampling values with 123 468 B. Blossey et al. Fig. 2 Seasonal life cycle diagram and periodic matrices for T. seedbank or previous fall until spring, Bs plant survival and natans. Each row represents a season and each circle a life stage growth from spring to fall in the same year and Bu reproduction (SB: seedbank, S: seed, Sm: small plant, Lg: large plant). of small and large plants. Parameters are defined in Supple- Periodic matrix Bf represents germination of seeds from mental Materials, Table S3 replacement holding sample size constant for 1000 parameters (FSM and FLG, fertility for small and large iterations) (Caswell 2001). We also conducted elas- plants, respectively), but in this case fertility values ticity analyses on each periodic matrix to evaluate were drawn from a Poisson distribution. which transition most influenced growth rate (Caswell To account for density dependent effects, which 2001; Smith et al. 2005). result in decreased T. natans fertility and plant size (Groth et al. 1996), we modeled fertility values as a Non-target threat simulation density-dependent parameter, such that number of seeds produced by small (FSM) or large (FLG) plants We examined how presence and typical attack of G. was dependent on the number of T. natans plants (P) in nymphaeae or of G. birmanica affects demography of the previous season (s) T. natans (Ding and Blossey 2005; Ding et al. 2006b). ðÞ 1ðzP Þ s1 We modeled effects of each scenario on T. natans F ¼ ðÞ 1ðzP Þ s1 1 þ e populations over ten time steps (t) and 10,000 iterations (i) and parametrized the model with values we set z, the regression parameter, to 0.0005, indicat- estimated from the common garden and the literature ing weak density-dependent effects. To model effects (Supplementary Materials, Table S3). We modeled of management scenarios, we estimated average stochastic variation following a two-step procedure in number of seeds produced by small or large plants order to incorporate temporal variation and parametric (FSM or FLG, respectively) and weighted the value by uncertainty in model predictions (McGowan et al. rate of fertility decrease according to scenario. At each 2011). We modeled temporal variation in survival time step we estimated annual growth rate (k) as ratio rates and transition terms for each realization i and between population size at current year (N ) and t?1 time step t as a beta distributed random variable with previous year (N ). Values of k [ 1 indicate increas- parameters a and b derived from the mean survival ing populations while values of k \ 1 indicate i i (or transition) rate, l , such that l = a /(a ? b ) and declining populations. We conducted all analysis i i i i i r = l (1 - l )/(a ? b ? 1). We incorporated using package popbio (Stubben and Milligan 2007) i i i i i parametric uncertainty in the replication loop by in R Core Team (2016). sampling l from a beta distribution, and r from an i i inverse Gaussian distribution with mean m (m = 0.001) and shape parameter k (k = 0.0001). We followed the same two-step approach for fertility 123 A proposal to use plant demographic data to assess potential weed biological control… 469 Results of demographic analyses hypothetical scenario of T. natans as a potential non- target, G. birmanica would be a grave threat to Plant area, seed output and germination varied signif- continued existence of the species. If the goal were to icantly among populations and number of seeds per safeguard T. natans, our results show that of these two plant was positively correlated with plant area (Sup- extremely similar herbivores, G. nymphaeae would be plementary Materials, Fig. S1; Tables S1, S2). a safe biocontrol agent while releases of G. birmanica Asymptotic population growth rates varied signifi- should not be considered. cantly across populations (Supplementary Materials, Section 3). Elasticity analyses indicated that matrix Implications for demographic assessments of non- elements representing germination of new seed and target effects growth into large plants had the greatest influence on k (Supplementary Materials, Fig. S3). We considered a hypothetical example of T. natans as a non-target plant and evaluated potential threats by Results from modelled simulations two herbivores that can successfully feed, oviposit and develop on the species. Under current decision making Our model predicted that T. natans populations will processes, biocontrol scientists or regulators would continue to grow until carrying capacity or habitat not consider G. nymphaeae a safe biocontrol agent requirements do not allow further expansion, despite because fundamental and field host range include T. continued attack by G. nymphaeae (Fig. 3). This natans. Field evidence from [ 100 years of associa- indicates that if T. natans were a non-target species, G. tion with T. natans in North America, and results from nymphaeae would not constitute a demographic threat our demographic modeling efforts, however, clearly to T. natans populations despite regular feeding, indicate that G. nymphaeae is no demographic threat oviposition and larval development. In contrast, our to T. natans. Furthermore, there is no evolutionary model predicts that T. natans populations will be tendency, despite enormous opportunity, to improve greatly reduced when attacked by G. birmanica. After larval performance on the novel host, and female ten years simulated populations increased from 1000 choice retains its preference for oviposition on the plants to [ 40,000 plants when attacked by G. original host, even when larvae developed on T. nymphaeae, but declined to near zero individuals natans (Ding and Blossey 2009). The traditional risk under G. birmanica attack (Fig. 3). Thus, under our assessment process, as currently being implemented, would eliminate a potentially important biocontrol agent due to safety concerns regarding attack on T. natans, but this would be fundamentally unjustified given the realized demographic impact of G. nym- phaeae. Our demographic assessments correctly pre- dict what is evident in the field: G. nymphaeae does not affect T. natans populations, while G. birmanica attack can lead to rapid and severe population growth rate declines. Discussion Reports of non-target attack by weed biocontrol agents after their release has a chilling effect on agent approvals in ongoing programs, funding and recruit- Fig. 3 Simulated population size (continuous line) and popu- ment (Moran and Hoffmann 2015). Land managers, lation growth rate (dashed line) of a T. natans population under ecologists, conservationists and weed biocontrol sci- attack by either the leaf beetle G. nymphaeae or G. birmanica. entists spent enormous amounts of time discussing Model data are integrated means derived by collecting demographic data for four different T. natans populations implications, focusing on the two high profile cases of 123 470 B. Blossey et al. R. conicus and C. cactorum, while critics implied that native plants. Pivoting to new risk assessment proce- non-target impacts are potentially widespread, but not dures and lines of evidence will take some time but we recognized (Louda et al. 2003). Almost 20 years have argue it is essential and well justified. passed since the initial publications and while we have We acknowledge that biological control, like every been provided with some evidence for negative other management technique, is not risk free and that demographic consequences by R. conicus, the same ecological surprises may occur. For example, it cannot be said for C. cactorum. Importantly, there is appears that identity of plant species attacked during not a single publication documenting negative popu- outbreak or spillover events is unpredictable (Blossey lation level impacts on any other non-target plant et al. 2001). But these are temporary events with no species by weed biocontrol agents. Observation of demographic consequences for attacked plant species herbivorous biocontrol agents feeding on other plants and do not constitute host shifts. Assessing potential are widespread, indicating that evidence of their non- non-target effects is the ethical thing to do, but without target feeding is being recognized, collected and use of demographic information on targets and non- reported, but their attack appears inconsequential at targets biocontrol scientists and practitioners are the population level. vulnerable to accusations of inappropriate conduct Unless there are unrecognized, or unpublished data and may get blamed for population declines or on more widespread negative demographic conse- extinctions, whether these accusations are true or not. We argue that we should acknowledge our quences by approved weed biocontrol agents, the feared threats to native plants currently do not appear responsibilities for safeguarding the continued exis- to exist. This is entirely overlooked by regulators, tence of populations of native or non-target species, conservationists and ecologists who may equate and that demographic approaches provide a powerful feeding on non-target plants with demographic threats. tool for evaluating ecological risk. To the best of our The reliance on data about fundamental host range knowledge, with the exception of R. conicus or C. tests has been called risk-averse, but it ignores the cactorum, there is no weed biocontrol agent that has increasing realized impacts of invasive plants on negatively affected populations of non-target plants. native biota (Downey and Paterson 2016), and Given that hundreds of control agents have been increasing herbicide use by land managers that result released over 100 years across the world, this track in widespread and documented detrimental long-term record is exemplary (Winston et al. 2014). But we also effects to species of conservation concern (Kettenring may miss out on very promising herbivores, because and Adams 2011). We emphasize that society and the current regulatory climate makes scientifically citizens are entitled to have questions regarding questionable and poorly justified decisions (Cristofaro performance of invasive species management methods et al. 2013; Fowler et al. 2012; Groenteman et al. 2011; answered to make informed decisions about priorities, Hinz et al. 2014). risk acceptance and funding (Blossey 2016a, b), but Our proposal to utilize demographic approaches in this requires appropriate assessments and interpreta- forecasting agent efficacy, and potential impacts on tion, not fear mongering and it should apply to all non-target species, would go a long way in improving management methods (Kerr et al. 2016; Pearson et al. the standing of the discipline—one grounded in theory 2016). and applying modern tools—while retaining our Use of herbivore impact studies and demographic exemplary track record in safeguarding native species. models in the way we have described here would be an This does not make weed biocontrol inherently more important tool to evaluate efficacy and safety of risky. On the contrary, we would be able to focus on potential biocontrol agents, yet it is completely effective agents with a proven track record of impact unutilized despite its promise (Blossey 2016b). Use on demography of invasive plants, and lack of of plant (and herbivore) demography would be an demographic impact on non-targets. Delivering this important advancement in improving scientific rigor information to society and decision makers and and predictability of weed biocontrol programs, albeit regulators will constitute part of the accountability we acknowledge it will not be an easy transition. For we should require from all of those engaged in far too long, the focus has been narrowly on funda- invasive species management and stewardship (Blos- mental host ranges and not allowing any attack on sey 2016a, b; Hare and Blossey 2014). 123 A proposal to use plant demographic data to assess potential weed biological control… 471 Acknowledgements We appreciate support and assistance we Carvalheiro LG, Buckley YM, Ventim R, Fowler SV, Memmott have received from those supplying rosettes from local J (2008) Apparent competition can compromise the safety populations. We appreciate help with field work and common of highly specific biocontrol agents. Ecol Lett 11:690–700 garden by Audrey Bowe and Jennifer Price for help with model Caswell H (2000) Prospective and retrospective perturbation development. Comments by several anonymous reviewers analyses: their roles in conservation biology. Ecology provided important feedback that improved our manuscript. 81:619–627 Caswell H (2001) Matrix population models: construction, Open Access This article is distributed under the terms of the analysis, and interpretation. Sinauer Associates, Creative Commons Attribution 4.0 International License (http:// Sunderland creativecommons.org/licenses/by/4.0/), which permits unre- Caswell H, Salguero-Gomez R (2013) Age, stage and senes- stricted use, distribution, and reproduction in any medium, cence in plants. J Ecol 101:585–595 provided you give appropriate credit to the original Caswell H, Neubert MG, Hunter CM (2011) Demography and author(s) and the source, provide a link to the Creative Com- dispersal: invasion speeds and sensitivity analysis in peri- mons license, and indicate if changes were made. odic and stochastic environments. Theor Ecol 4:407–421 Catton HA, Lalonde RG, De Clerck-Floate RA (2014) Differ- ential host-finding abilities by a weed biocontrol insect create within-patch spatial refuges for nontarget plants. Funding Funding for this work was provided by the New Environ Entomol 43:1333–1344 York State Department of Conservation. Catton HA, Lalonde RG, De Clerck-Floate RA (2015) Nontar- get herbivory by a weed biocontrol insect is limited to spillover, reducing the chance of population-level impacts. References Ecol Appl 25:517–530 Catton HA, Lalonde RG, Buckley YM, De Clerck-Floate RA (2016) Biocontrol insect impacts population growth of its Ancheta J, Heard SB (2011) Impacts of insect herbivores on rare target plant species but not an incidentally used nontarget. plant populations. Biol Conserv 144:2395–2402 Ecosphere 7:e01280. https://doi.org/10.1002/ecs2.1280 Andreas JE, Schwarzla¨nder M, de Clerck-Floate R (2008) The Clement SL, Cristofaro M (1995) Open field tests in host- occurrence and potential relevance of post-release, non- specificity determination of insects for biological control of target attack by Mogulones cruciger, a biocontrol agent for weeds. Biocontrol Sci Technol 5:395–406 Cynoglossum officinale in Canada. Biol Control Clewley GD, Eschen R, Shaw RH, Wright DJ (2012) The 46:304–311 effectiveness of classical biological control of invasive Arnett AE, Louda SM (2002) Re-test of Rhinocyllus conicus plants. J Appl Ecol 49:1287–1295 host specificity, and the prediction of ecological risk in Crawley MJ (1989) The successes and failures of weed bio- biological control. Biol Conserv 106:251–257 control using insects. Biocontrol News Inf 10:213–223 Bernays EA, Chapman RF (1994) Host-plant selection by Cristofaro M, De Biase A, Smith L (2013) Field release of a phytophagous insects. Contemporary topics in entomol- prospective biological control agent of weeds, Ceratapion ogy. Springer, New York basicorne, to evaluate potential risk to a nontarget crop. Bernays E, Graham M (1988) On the evolution of host speci- Biol Control 64:305–314 ficity in phytophagous arthropods. Ecology 69:886–892 Cronin G, Schlacher T, Lodge DM, Siska EL (1999) Blossey B (2016a) The future of biological control: a proposal Intraspecific variation in feeding preference and perfor- for fundamental reform. In: van Driesche R, Simberloff D, mance of Galerucella nymphaeae (Chrysomelidae: Blossey B, Causton C, Hoddle M, Marks C, Heinz K, Coleoptera) on aquatic macrophytes. J N Am Bent Soc Wagner D, Wagner K (eds) Integrating biological control 18:391–405 into conservation practice. Wiley, Chichester, UK, Davis AS, Landis DA, Nuzzo V, Blossey B, Gerber E, Hinz HL pp 314–328 (2006) Demographic models inform selection of biocontrol Blossey B (2016b) Measuring and evaluating ecological out- agents for garlic mustard (Alliaria petiolata). Ecol Appl comes of biological control introductions. In: van Driesche 16:2399–2410 R, Simberloff D, Blossey B, Causton C, Hoddle M, Marks De Clerck-Floate R, Schwarzla¨nder M (2002) Host specificity of C, Heinz K, Wagner D, Wagner K (eds) Integrating bio- Mogulones cruciger (Coleoptera: Curculionidae), a bio- logical control into conservation practice. Wiley, Chich- control agent for houndstongue (Cynoglossum officinale), ester, UK, pp 161–188 with emphasis on testing of native North American Bor- Blossey B, Casagrande R, Tewksbury L, Landis DA, Wieden- aginaceae. Biocontrol Sci Technol 12:293–306 mann RN, Ellis DR (2001) Nontarget feeding of leaf-bee- DePrenger-Levin ME, Grant TA, Dawson C (2010) Impacts of tles introduced to control purple loosestrife (Lythrum Rhinocyllus conicus the introduced biocontrol agent, salicaria L.). Nat Areas J 21:368–377 (Coleoptera: Curculionidae), on the seed production and Briese DT (2005) Translating host-specificity test results into population dynamics of Cirsium ownbeyi (Asteraceae), a the real world: the need to harmonize the yin and yang of rare, native thistle. Biol Control 55:79–84 current testing procedures. Biol Control 35:208–214 DeWalt SJ (2006) Population dynamics and potential for bio- Buckley YM, Rees M, Paynter Q, Lonsdale M (2004) Modelling logical control of an exotic invasive shrub in Hawaiian integrated weed management of an invasive shrub in rainforests. Biol Invasion 8:1145–1158 tropical Australia. J Appl Ecol 41:547–560 123 472 B. Blossey et al. Ding J, Blossey B (2005) Impact of the native water lily leaf Kettenring KM, Adams CR (2011) Lessons learned from inva- beetle, Galerucella nymphaeae (Coleoptera: Chrysomeli- sive plant control experiments: a systematic review and dae), attacking introduced water chestnut, Trapa natans,in meta-analysis. J Appl Ecol 48:970–979 the northeastern United States. Environ Entomol Kunii H (1988) Longevity and germinability of buried seed in 34:683–689 Trapa sp. Mem Fac Sci Shimane Univ 22:83–91 Ding J, Blossey B (2009) Difference in preference and perfor- Louda SM, Kendall D, Connor J, Simberloff D (1997) Eco- mance of the water lily leaf beetle, Galerucella nymhaeae logical effects of an insect introduced for the biological populations on native and invasive aquatic plants. Environ control of weeds. Science 277:1088–1090 Entomol 38:1653–1660 Louda SM, Pemberton RW, Johnson MT, Follett PA (2003) Ding J, Blossey B, Du Y, Zheng F (2006a) Galerucella bir- Nontarget effects—the Achilles’ heel of biological control: manica (Coleoptera: Chrysomelidae), a promising poten- retrospective analyses to reduce risk associated with bio- tial biological control agent of water chestnut, Trapa control introductions. Annu Rev Entomol 48:365–396 natans. Biol Control 36:80–90 Louda SM, Rand TA, Arnett AE, McClay AS, Shea K, Ding J, Blossey B, Du Y, Zheng F (2006b) Impact of Galeru- McEachern AK (2005a) Evaluation of ecological risk to cella birmanica (Coleoptera: Chrysomelidae) on growth populations of a threatened plant from an invasive bio- and seed production of Trapa natans. Biol Control control insect. Ecol Appl 15:234–249 37:338–345 Louda SM, Rand TA, Russell FL, Arnett AE (2005b) Assess- Downey PO, Paterson ID (2016) Encompassing the relative ment of ecological risks in weed biocontrol: input from non-target risks from agents and their alien plant targets in retrospective ecological analyses. Biol Control biological control assessments. BioControl 61:615–630 35:253–264 Eckberg JO, Tenhumberg B, Louda SM (2014) Native insect Marohasy J (1996) Host shifts in biological weed control: real herbivory limits population growth rate of a non-native problems, semantic difficulties or poor science? Int J Pest thistle. Oecologia 175:129–138 Manag 42:71–75 Fowler SV, Syrett P, Hill RL (2000) Success and safety in the McEvoy PB, Coombs EM (1999) Biological control of plant biological control of environmental weeds in New Zealand. invaders: regional patterns, field experiments, and struc- Aust Ecol 25:553–562 tured population models. Ecol Appl 9:387–401 Fowler SV, Paynter Q, Dodd S, Groenteman R (2012) How can McEvoy PB, Higgs KM, Coombs EM, Karacetin E, Starcevich ecologists help practitioners minimize non-target effects in LA (2012) Evolving while invading: rapid adaptive evo- weed biocontrol? J Appl Ecol 49:307–310 lution in juvenile development time for a biological control Futuyma DJ, Agrawal AA (2009) Macroevolution and the bio- organism colonizing a high-elevation environment. Evol logical diversity of plants and herbivores. Proc Nat Acad Appl 5:524–536 Sci USA 106:18054–18061 McFadyen REC (1998) Biological control of weeds. Annu Rev Groenteman R, Fowler SV, Sullivan JJ (2011) St. John’s wort Entomol 43:369–393 beetles would not have been introduced to New Zealand McGowan CP, Runge MC, Larson MA (2011) Incorporating now: a retrospective host range test of New Zealand’s most parametric uncertainty into population viability analysis successful weed biocontrol agents. Biol Control 57:50–58 models. Biol Conserv 144:1400–1408 Groth AT, LovettDoust L, LovettDoust J (1996) Population Moran VC, Hoffmann JH (2015) The fourteen international density and module demography in Trapa natans (Tra- symposia on biological control of weeds, 1969-2014: del- paceae), an annual, clonal aquatic macrophyte. Am J Bot egates, demographics and inferences from the debate on 83:1406–1415 non-target effects. Biol Control 87:23–31 Guretzky JA, Louda SM (1997) Evidence for natural biological Moran VC, Hoffmann JH, Zimmermann HG (2005) Biological control: insects decrease survival and growth of a native control of invasive alien plants in South Africa: necessity, thistle. Ecol Appl 7:1330–1340 circumspection, and success. Front Ecol Environ 3:77–83 Hare D, Blossey B (2014) Principles of public trust thinking. Paynter QE, Fowler SV, Gourlay AH, Haines ML, Harman HM, Hum Dimens Wildl 19:397–406 Hona SR, Peterson PG, Smith LA, Wilson-Davey JA, Hinz HL, Schwarzla¨nder M, Gassmann A, Bourchier RS (2014) Winks CJ, Withers TM (2004) Safety in New Zealand Successes we may not have had: a retrospective analysis of weed biocontrol: a nationwide survey for impacts on non- selected weed biological control agents in the United target plants. N Z Plant Prot 57:102–107 States. Invasive Plant Sci Manag 7:565–579 Paynter Q, Martin N, Berry J, Hona S, Peterson P, Gourlay AH, Jezorek H, Baker AJ, Stiling P (2012) Effects of Cactoblastis Wilson-Davey J, Smith L, Winks C, Fowler SV (2008) cactorum on the survival and growth of North American Non-target impacts of Phytomyza vitalbae a biological Opuntia. Biol Invasion 14:2355–2367 control agent of the European weed Clematis vitalba in Johnson DM, Stiling PD (1996) Host specificity of Cactoblastis New Zealand. Biol Control 44:248–258 cactorum (Lepidoptera: Pyralidae), an exotic Opuntia- Paynter Q, Fowler SV, Gourlay AH, Peterson PG, Smith LA, feeding moth, in Florida. Environ Entomol 25:743–748 Winks CJ (2015) Relative performance on test and target Kerr NZ, Baxter PWJ, Salguero-Gomez R, Wardle GM, plants in laboratory tests predicts the risk of non-target Buckley YM (2016) Prioritizing management actions for attack in the field for arthropod weed biocontrol agents. invasive populations using cost, efficacy, demography and Biol Control 80:133–142 expert opinion for 14 plant species world-wide. J Appl Ecol Pearson DE, Ortega YK, Runyon JB, Butler JL (2016) Sec- 53:305–316 ondary invasion: the bane of weed management. Biol Conserv 197:8–17 123 A proposal to use plant demographic data to assess potential weed biological control… 473 Pemberton RW (1995) Cactoblastis cactorum (Lepidoptera: Simberloff D, Stiling P (1996) How risky is biological control? Pyralidae) in the United States: an imigrant biological Ecology 77:1965–1974 control agent or an introduction of the nursery industry? Smilanich AM, Fincher RM, Dyer LA (2016) Does plant Am Entomol 41:230–232 apparency matter? Thirty years of data provide limited Pemberton RW (2000) Predictable risk to native plants in weed support but reveal clear patterns of the effects of plant biocontrol. Oecologia 125:489–494 chemistry on herbivores. New Phytol 210:1044–1057 Pemberton RW, Liu H (2007) Control and persistence of native Smith M, Caswell H, Mettler-Cherry P (2005) Stochastic flood Opuntia on Nevis and St. Kitts 50 years after the intro- and precipitation regimes and the population dynamics of a duction of Cactoblastis cactorum. Biol Control threatened floodplain plant. Ecol Appl 15:1036–1052 41:272–282 Stiling P, Rossi A, Gordon D (2000) The difficulties of single R Core Team (2016) R: a language and environment for sta- factor thinking in restoration: replanting a rare cactus in the tistical computing. R Foundation for Statistical Comput- Florida Keys. Biol Conserv 94:327–333 ing, Vienna, Austria. http://www.r-project.org/ Stubben CJ, Milligan BG (2007) Estimating and analyzing Raghu S, Wilson JR, Dhileepan K (2006) Refining the process of demographic models using the popbio package in R. J Stat agent selection through understanding plant demography Softw 22:1–23 and plant response to herbivory. Aust J Entomol Suckling DM, Sforza RFH (2014) What magnitude are observed 45:308–316 non-target impacts from weed biocontrol? PLoS ONE Rand TA, Louda SM (2004) Exotic weed invasion increases the 9(1):e84847. https://doi.org/10.1371/journal.pone. susceptibility of native plants attack by a biocontrol her- 0084847 bivore. Ecology 85:1548–1554 Swope SM, Satterthwaite WH, Parker IM (2017) Spatiotem- Rose KE, Louda SM, Rees M (2005) Demographic and evolu- poral variation in the strength of density dependence: tionary impacts of native and invasive insect herbivores on implications for biocontrol of Centaurea solstitialis. Biol Cirsium canescens. Ecology 86:453–465 Inv 19:2675–2691 Salguero-Gomez R, Jones OR, Archer CR, Buckley YM, Che- Takahashi M, Louda SM, Miller TEX, O’Brien CW (2009) Castaldo J, Caswell H, Hodgson D, Scheuerlein A, Conde Occurrence of Trichosirocalus horridus (Coleoptera: DA, Brinks E, de Buhr H, Farack C, Gottschalk F, Hart- Curculionidae) on native Cirsium altissimum versus exotic mann A, Henning A, Hoppe G, Roemer G, Runge J, Ruoff C. vulgare in North American tallgrass prairie. Environ T, Wille J, Zeh S, Davison R, Vieregg D, Baudisch A, Entomol 38:731–740 Altwegg R, Colchero F, Dong M, de Kroon H, Lebreton USDA (2016) Technical advisory group for biological control JD, Metcalf CJE, Neel MM, Parker IM, Takada T, Val- agents of weeds manual, Interim edn. United States verde T, Velez-Espino LA, Wardle GM, Franco M, Vaupel Department of Agriculture, Washington, DC, USA JW (2015) The COMPADRE plant matrix database: an van Klinken RD, Edwards OR (2002) Is host specificity of weed open online repository for plant demography. J Ecol biocontrol agents likely to evolve rapidly following 103:202–218 establishment? Ecol Lett 5:590–595 Sauby KE, Kilmer J, Christman MC, Holt RD, Marsico TD Vigueras AL, Portillo L (2001) Uses of Opuntia species and the (2017) The influence of herbivory and weather on the vital potential impact of Cactoblastis cactorum (Lepidoptera: rates of two closely related cactus species. Ecol Evol Pyralidae) in Mexico. Fla Entomol 84:493–498 7:6996–7009 Wapshere AJ (1974) A strategy for evaluating the safety of Shea K, Kelly D (1998) Estimating biocontrol agent impact with organisms for biological weed control. Ann Appl Biol matrix models: Carduus nutans in New Zealand. Ecol Appl 77:200–211 8:824–832 Williams BK, Nichols JD, Conroy MJ (2001) Analyses and Shea K, Kelly D, Sheppard AW, Woodburn TL (2005) Context- management of animal populations. Academic Press, New dependent biological control of an invasive thistle. Ecology York 86:3174–3181 Winkler I, Mitter C (2008) The phylogenetic dimension of Sheppard AW, Warner KD (2016) Societal values expressed insect/plant interactions: a summary of recent evidence. In: through policy and regulation concerning biological con- Tilmon K (ed) Specialization, speciation, and radiation: the trol releases. In: van Driesche R, Simberloff D, Blossey B, evolutionary biology of herbivorous insects. University of Causton C, Hoddle M, Marks C, Heinz K, Wagner D, California Press, Berkeley, pp 240–263 Wagner K (eds) Integrating biological control into con- Winston RL, Schwarzla¨nder M, Hinz HL, Day MD, Cock MJW, servation practice. Wiley, Chichester, pp 247–263 Julien MH (eds) (2014) Biological control of weeds: a Sheppard AW, van Klinken RD, Heard TA (2005) Scientific world catalogue of agents and their target weeds, 5th ed. advances in the analysis of direct risks of weed biological USDA Forest Service, Forest health technology enterprise control agents to nontarget plants. Biol Control team, Morgantown, West Virginia, USA. FHTET-2014-04 35:215–226

Journal

BioControlSpringer Journals

Published: Apr 28, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off