Adaptive management studies of invasive plants on non-agricultural lands typically employ an empirical approach based on designed field experiments that permit rigorous statistical analysis of results to quantify outcomes and assess the efficacy of management practices. When habitat restoration is the primary goal of a project, traditional plot-based study designs (e.g., the randomized complete-block design) are sometimes infeasible (this is often true in aquatic habitats) or inappropriate (e.g., when the goal is to assess effects of management practices on survival or resprouting of individual plants, such as trees or shrubs). Moreover, the assumptions of distribution-specific parametric statistical methods such as ANOVA often cannot be convincingly verified or are clearly untenable when properly assessed. For these reasons, it is worthwhile to be aware of alternative study designs that do not employ plots as experimental units and nonparametric statistical methods that require only weak distributional assumptions. The purpose of this paper is to review several of these alternative study designs and nonparametric statistical methods that we have found useful in our own studies of invasive aquatic and terrestrial plants. We motivate each statistical method by a research question it is well suited to answer, provide corresponding references to the statistical literature, and identify at least one R function that implements the method. In the Supplementary Material, we present additional technical information about the statistical methods, numerical examples with data, and a set of complete R programs to illustrate application of the statistical methods.

Among the most widely distributed species globally, common reed [Phragmites australis (Cav.) Trin. ex Steud.] has generated extensive interest in invasive plant science and management because its introduced strains are highly invasive and often form monocultures that alter ecosystem properties. In desert wetlands in Las Vegas, NV, USA, where management goals included reducing hazardous P. australis fuels and increasing native plant diversity, we assessed variation in P. australis cover, the degree of native plant colonization, and soil seedbanks after P. australis management treatments (cutting, glyphosate–imazapyr herbicide) and wildfires across gradients in soil properties. Based on change in P. australis cover during six measurement events over 24 mo, 24 study sites formed three groups: (1) decreasing cover, where initially high P. australis cover (60% to 85%) decreased to <5% following multiple cutting or herbicide treatments; (2) sustaining low cover, where wildfire or clearing was associated with initially low P. australis cover which remained low (<30%) after multiple herbicide applications; and (3) sustaining high cover (45% to 100% initially and remaining at 30% to 100%), including sites unmanaged or treated/burned only once. High soil salinity correlated with low postmanagement P. australis cover. No native plants were detected in the sustaining high P. australis cover group, despite natives occurring in the seedbank. Where management reduced P. australis cover, minimal native plant colonization did occur. Secondary invasion by other non-native plants was nearly absent. Our results suggest that if P. australis can be initially cleared, multiple herbicide applications can persistently keep cover low, especially on drier, saline soils. Slow native plant colonization suggests that a phased approach may be useful to initially reduce P. australis cover, keep it low via repeated treatments, and actively revegetate sites with native species tailored to the moisture–salinity gradient across P. australis–invaded habitats.

Non-native plants negatively impact ecosystems via a variety of mechanisms, including in forested riparian areas. Japanese knotweed [Polygonum cuspidatum Siebold & Zucc.] and its hybrids (referred to as Polygonum spp. hereafter) are widely spread throughout North America and can impact flora and fauna of riparian habitats. Thus, information improving our ability to understand and predict the potential spread and colonization of Polygonum spp. is valuable. One dispersal mechanism is hydrochory (i.e., dispersal by water), including the downstream dispersal of viable stems that can facilitate rapid invasion within a watershed. We used passive integrated transponder (PIT) telemetry in experimental releases of Polygonum spp. stems to track the downstream transport of Polygonum spp. in a small (second-order) stream in northern New Hampshire, USA, in the summers of 2021 and 2022. A total of 180 (90 each year) Polygonum spp. stems were released at three sites within the stream reach, with 185 (∼98%) being recaptured at least once, with a total of 686 recaptures. Individual relocated stems moved a maximum distance of 30 to 875 m downstream in 2021 and 13 to 1,233 m in 2022 during regular flows; however, a high-streamflow event in July 2021 flushed out all remaining stems downstream of the study area. Generalized additive mixed models (GAMMs) identified site-specific differences in stem movement rates and a general reduction in movement rates with increased duration of time elapsed since post-release. In general, Polygonum spp. stems moved farther downstream in sites with lower channel sinuosity, although other fine-scale habitat factors (e.g., water depth, habitat type, and presence of wood and debris jams) likely contribute to the ability for Polygonum spp. to further disperse or otherwise be retained within the channel. Thus, stream morphology and stream flow are likely to affect where Polygonum spp. stems will be retained and potentially reestablish. Predictive tools identifying areas of higher probability of hydrochory-based dispersal could help to focus removal efforts when employed or to identify riparian habitats at highest risk for spread.

The southern African shrub boneseed [Chrysanthemoides monilifera subsp. monilifera (L.) Norl.] is a perennial shrub that is a significant threat to natural ecosystems and is listed as a Weed of National Significance in Australia. In Western Australia (WA) it has spread across peri-urban and natural environments. We assembled a single standardized database containing more than 2,050 presence records for individual plants and 135 absence records at a local population level. We further refined the populations into 89 sites that require different management trajectories due to topography and capacity of land managers to implement control. Forty-nine of these sites were in urban regions and 40 sites were in regional WA. We split these 89 sites into three near-term management goals: watch (12), extirpate (68), and contain (9). The 12 watch sites are those where all available evidence suggests that there have been no new inputs into the seedbank for 15 yr. The 68 sites marked for extirpation are those where delimitation is already achieved or easily achievable, where there have been minimal seed inputs into the soil seedbank in recent years due to consistent surveillance and control, and where surveys for new plants are likely to be efficient to conduct. Finally, for nine sites in urban regions around Perth, we recommend containment in the near term with a longer-term goal to achieve delimitation and extirpation. To achieve the objective of state-level eradication, a coordinated and sustained campaign involving three components—delimitation of all sites, prevention of further inputs into the soil seedbank, and systematic field surveys to remove plants—must commence without delay. While resourcing requirements for delimitation and overall program management are not possible to estimate, our prior experience suggests that it will take at least 1,900 h of on-ground surveying by experienced personnel to achieve extirpation of C. monilifera subsp. monilifera in WA.

We evaluated herbicides for controlling the annual grass ventenata [Ventenata dubia (Leers) Coss.], with particular interest in indaziflam, a preemergence cellulose biosynthesis inhibitor. In 2016, indaziflam was applied postemergence alone and in mixture with glyphosate, imazapic, propoxycarbazone-sodium, or rimsulfuron to an improved pasture in southwestern Montana. A non-sprayed control was included for comparison purposes. Canopy cover of each species was assessed annually for 7 yr; cover was grouped by life-form and longevity, and species richness was calculated. Five years (2021) after treatment, the seedbank was assessed. Our results indicated that treatments including indaziflam reduced V. dubia cover 1 to 3 yr and even up to 6 yr after application, with V. dubia cover being zero or close to zero. However, at 7 yr (2023) after treatment, V. dubia was low across all treatments, including the non-sprayed control. Perennial grasses and forbs and annual forbs were generally unaffected by any treatment and did not increase in cover over the 7 yr, even though V. dubia decreased. Two years after treatment, species richness was lowest in treatments that included indaziflam, but at 7 yr, species richness was similar across all treatments. Indaziflam depleted the monocot and dicot seedbank, with fewer than 5 seedlings of any species emerging from treatments that included indaziflam, while other treatments resulted in 60 to 165 seedlings per sample (40 cm³ of soil). In summary, at our study site, a single application of indaziflam controlled V. dubia for 6 yr, appeared to deplete the seedbank at 5 yr, and cover of perennial and annual vegetation and species richness was unaffected. By the end of the study, though, V. dubia cover appeared to be influenced by factors other than herbicide treatments, possibly variable precipitation over time, an exclusion of grazing, and competitive perennial grasses dominating the site.

Fast and efficient identification is critical for reducing the likelihood of weed establishment and for appropriately managing established weeds. Traditional identification tools require either knowledge of technical morphological terminology or time-consuming image matching by the user. In recent years, deep learning computer vision models have become mature enough to enable automatic identification. The major remaining bottlenecks are the availability of a sufficient number of high-quality, reliably identified training images and the user-friendly, mobile operationalization of the technology. Here, we present the first weed identification and reporting app and website for all of Australia. It includes an image classification model covering more than 400 species of weeds and some Australian native relatives, with a focus on emerging biosecurity threats and spreading weeds that can still be eradicated or contained. It links the user to additional information provided by state and territory governments, flags species that are locally reportable or notifiable, and allows the creation of observation records in a central database. State and local weed officers can create notification profiles to be alerted of relevant weed observations in their area. We discuss the background of the WeedScan project, the approach taken in design and software development, the photo library used for training the WeedScan image classifier, the model itself and its accuracy, and technical challenges and how these were overcome.

Invasive plant taxa are generally regulated at the species level, without considering infra- or interspecific variation. However, cultivars or hybrids can pose a lower risk of invasion, for example, due to sterility. We evaluate six general approaches to regulating cultivars and hybrids: (1) Globally Guilty by Association; (2) Nationally Guilty by Association; (3) Guilty until Proven Innocent; (4) Negotiated Guilt; (5) Claimed to be Innocent; and (6) Innocent until Proven Guilty. We discuss these approaches in the context of South Africa (which has a typified Negotiated Guilt approach). Following negotiations since 2001 between the South African horticultural industry/green industry and legislators, an unofficial consensus list of “presumed sterile” cultivars and hybrids was produced in 2014 containing 187 entities from 34 taxa. In 2020, this was reduced to 157 entities from 16 taxa. But the evidence supporting the original lists and the subsequent revisions was not published. To address this issue, we developed a generic pro forma (template) for reporting sterility based on observations and/or experiments on: flowering, fruiting, pollen, and seeds; the potential for vegetative propagation; and the potential for genetic changes (including hybridization and reversion to fertility). We recommend that such information be incorporated into risk analyses conducted specifically for infra- and inter specific entities, and only if the risk of a harmful invasion is demonstrated to be acceptably low or can be easily mitigated should such entities be exempted from regulation. This will be time-consuming, but, by setting out the evidence clearly, the approach is transparent and provides a clear route for stakeholders to seek exemptions for entities of importance. In conclusion, although we suspect the simplicity of the Negotiated Guilt approach is desirable to many stakeholders, and is the approach currently adopted in South Africa, we recommend a shift toward the Guilty until Proven Innocent approach.