Management Implications
Insects, pathogens, and other tree-afflicting pests are accumulating and spreading in many forests, often with poorly known effects on invasive plants. In temperate forests in Ohio, USA, we found that cover of the non-native liana Celastrus orbiculatus (Oriental bittersweet; hereafter Celastrus) proliferated after Quercus velutina (black oak) tree mortality associated with the invasive fungal pathogen Bretziella fagacearum (oak wilt) and its attempted containment (severing root connectivity and sanitation removal of infected trees). Over 5 yr between 2020 and 2024, Celastrus cover increased by an order of magnitude (reaching up to 22% average cover by 2024) in sites with oak wilt, while it remained comparatively low (1%) in reference forest sites without evidence of oak wilt. By 2024, Celastrus cover was as high as 60% among oak wilt sites, and the liana formed a mat-like covering on native understory plants. In these forests with spatially and temporally dynamic oak wilt patches, we suggest that Celastrus be managed in two contexts: (1) proactively as incipient populations with low cover in understories of non–oak wilt affected patches to potentially suppress the response of Celastrus should oak wilt arrive and (2) by reducing established Celastrus populations already with high cover in existing oak wilt patches to enable forest recovery from oak wilt to be dominated by native plants. Invasive plant management may need to be increasingly paired with forest management of and adaptation to tree-afflicting pests that may catalyze non-native plant invasions.
Introduction
Introduced, tree-afflicting pests, such as insects and pathogens, could make forests more invasible by non-native plants (Baron and Rubin Reference Baron and Rubin2021; Burnham and Lee Reference Burnham and Lee2010; Eschtruth and Battles Reference Eschtruth and Battles2009). Additionally, management attempting to contain or slow the spread of tree-afflicting pests could inadvertently accelerate non-plant invasions (Hausman et al. Reference Hausman, Jaeger and Rocha2010). However, how these tree-afflicting pests or their attempted containment may be associated with non-native plants is unclear for many pests. The limited research available for three of the most major introduced pests of trees in eastern North American forests exemplifies potential variation in associations of tree pests with non-native plants. After invasion by the beetle emerald ash borer (Agrilus planipennis), non-native plants increased after cutting of ash (Fraxinus spp.) trees for attempted quarantine of the invading insects (Hausman et al. Reference Hausman, Jaeger and Rocha2010). However, without attempted cutting for quarantine of this pest, non-native plants have not increased (Abella et al. Reference Abella, Hausman, Jaeger, Menard, Schetter and Rocha2019) or have increased where already present (e.g., Baron and Rubin Reference Baron and Rubin2021; Dolan and Kilgore Reference Dolan and Kilgore2018; Hoven et al. Reference Hoven, Gorchov, Knight and Peters2017). After invasion by the sap-sucking insect hemlock woolly adelgid (Adelges tsugae), non-native plants already present increased in one study (Eschtruth and Battles Reference Eschtruth and Battles2009) and not in another, but new invasions occurred (Small et al. Reference Small, Small and Dreyer2005). After arrival of beech bark disease (a complex of an invasive insect and fungal pathogen), no non-native plants were recorded in diseased forest sites (Cale et al. Reference Cale, McNulty, Teale and Castello2013). With existing invasions of pests that afflict trees expanding and new introductions accumulating, further understanding their potential associations with invasive plants is a research priority (Gougherty et al. Reference Gougherty, Elliott, LaRue, Gallion and Fei2023).
We examined change in cover of the non-native liana Oriental bittersweet (Celastrus orbiculatus Thunb.; hereafter Celastrus) after ongoing death and sanitation removal of black oak (Quercus velutina Lam.) trees associated with the fungal pathogen oak wilt (Bretziella fagacearum). Oak wilt was first documented in the United States in the 1940s and was reported in 24 eastern states by 2009 (Juzwik et al. Reference Juzwik, Appel, MacDonald and Burks2011). Trees of the red oak group (subgenus Erythrobalanus) are most susceptible to oak wilt and can die the first growing season after infection (Juzwik et al. Reference Juzwik, Appel, MacDonald and Burks2011). The pathogen can spread belowground through root grafts and overland via flying beetles transporting it to wounds on trees. A protocol intended to contain the spread of oak wilt includes severing root connectivity with neighboring trees followed by sanitation cutting and removal of infected trees (Juzwik et al. Reference Juzwik, O’Brien, Evenson, Castillo and Mahal2010).
In the oak wilt–affected forests of our study area, Celastrus was the dominant non-native plant species and the focus of our study. Originally from temperate climates in Asia, Celastrus is thought to have been introduced to North America (New York State) by the 1880s as a horticultural plant and has since spread to at least 34 states and 5 Canadian provinces (McKenzie-Gopsill and MacDonald Reference McKenzie-Gopsill and MacDonald2021). We addressed the following questions: (1) Does cover of Celastrus vary among forest sites receiving onetime oak wilt containment treatments 1 to 10 yr prior or change through time spanning our 5-yr study period (2020 to 2024)? (2) Is variation in Celastrus cover correlated with tree canopy cover?
Materials and Methods
Study Area and Oak Wilt Treatments
Within the 45,000-ha Oak Openings region in northwestern Ohio, USA, we performed the study in the 200-ha Wildwood Preserve (41.6814°N, 83.6739°W), managed by Metroparks Toledo. Oak forests in the preserve contain overstory trees 80 to more than 120 yr old of mostly Q. velutina with some white oak (Quercus alba L.). Total live basal area of these Quercus species in undisturbed forests is typically 30 to 50 m2 ha−1 with 90% to 97% tree canopy cover. Understories typically contain vascular plants of all growth forms, including native shrubs, tree seedlings, and herbaceous plants with mixtures of forbs, graminoids, and ferns (Abella et al. Reference Abella, Sprow, Walters and Schetter2021). The sandy soils are classified as the Ottokee (mixed, mesic Aquic Udipsamments) and Oakville (mixed, mesic Typic Udipsamments) series. Climate is temperate, averaging 86 cm yr−1 of precipitation (Toledo Airport weather station, National Centers for Environmental Information, Asheville, NC, USA). Within our 2020 to 2024 study period, early to midsummer (May through July) precipitation was 108% of the 26-cm (1955 to 2024) average, being 92% (2020), 133% (2021), 110% (2022), 87% (2023), and 116% (2024).
Oak wilt was first noted in the preserve after 2010 based on symptoms exhibited by trees, signs of fungal presence on likely infected trees (Juzwik et al. Reference Juzwik, Appel, MacDonald and Burks2011), and on tissue samples collected from symptomatic trees testing positive for oak wilt (C. Wayne Ellett Plant and Pest Diagnostic Clinic, Ohio State University, Reynoldsburg, OH, USA). Thereafter, in scattered sites throughout the preserve, groups of two to three or individual mature trees of the susceptible Q. velutina showing oak wilt sign and symptoms began dying, often within the same year that symptoms were detected. In 2015, Metroparks Toledo began implementing an oak wilt containment protocol consisting of: (1) using a vibratory plow blade, mounted on a Ditch Witch RT125Q Quad Ride-On Tractor (Charles Machine Works, Perry, OK, USA), to establish a containment trench line (1.5-m deep and 4-cm wide) in the soil designed to sever root connectivity for 30 m around symptomatic trees and encircling the groups of two to three or individual symptomatic trees, (2) sanitation cutting (using chainsaws with cuts made just above ground level) of symptomatic trees within the containment line, and (3) chipping of wood and slash of the infected trees and removal of the chipped material (Juzwik et al. Reference Juzwik, O’Brien, Evenson, Castillo and Mahal2010). Although assessing effectiveness of this attempted containment would require a below- and aboveground fungal distribution and transport investigation beyond the scope of our study, prior research in similar oak forests in Minnesota concluded that the containment protocol slowed the spread of oak wilt from infection centers for at least 4 to 6 yr (Juzwik et al. Reference Juzwik, O’Brien, Evenson, Castillo and Mahal2010).
Data Collection and Analysis
We defined an oak wilt treatment site as the canopy gap centered on a single tree or group of two to three trees sanitation cut and encircled by a containment line following the protocol described earlier. We designated oak wilt treatments by their age according to the growing season immediately following the dormant season completion of the onetime treatments. We named these treatments as old (established in 2015; age 6 yr in 2020 when our study began and age 10 yr in 2024 when our study ended), middle-aged (2018; age 3 yr in 2020 and 7 yr in 2024), and young (2020; age 1 yr in 2020 and 5 yr in 2024). Sites containing dead trees consistent with oak wilt symptoms that had not received oak wilt treatments were not available to sample, because managers wished to avoid leaving areas with potentially unabated spread of oak wilt. This situation of unavailability of invaded, untreated sites is common in invasive species science and management but can enable comparison of invaded, treated sites with uninvaded, reference sites (McNair et al. Reference McNair, Frobish, Rice and Thum2024). We used this type of design in our study. Thus, while oak wilt and its attempted containment are an inseparably combined influence in our study, we were able to sample mature oak forest sites without evidence of oak wilt as untreated reference forest. We randomly selected 7 sites for sampling for each of the three oak wilt treatment ages and for reference forests, totaling 28 sites. Sample sites among treatments and reference forests were interspersed across the landscape, averaged 0.3 km apart, had an extent of 1.0 km, and were intermixed on the same soil series (Ottokee and Oakville).
During peak plant cover in June to July, we measured Celastrus cover within a circular, 100-m2 plot centered on the single or central stump of removed Q. velutina trees at each of the 21 oak wilt treatment sites. In each of the 7 sites in reference forests, we centered the plot on the bole of the nearest live Q. velutina tree to the randomly selected point. In 2020, stump diameters of focal, sanitized trees ranged from 29 to 130 cm in plots within treatment sites and from 45 to 107 cm for live, mature trees in reference forest plots. In all 28 plots in 2020, 2022, and 2024, we visually categorized areal cover of Celastrus as 0.1%, 0.25%, 0.5%, and 1%; 1% intervals to 10% cover; and 5% intervals above 10% cover up to the maximum 100% areal cover. Nearly all (typically 99% to 100%) Celastrus cover occurred as plants growing unsupported or mat-like on other understory plants or subcanopy tree saplings, rather than as climbing plants on overstory trees (Figure 1). Taxonomic identification of Celastrus in the study area included collecting and depositing two specimens (T.L. Walters #4077 and #4535) in a herbarium (Cleveland Museum of Natural History, Department of Botany, Cleveland, OH, USA). We also recorded tree canopy cover (defined as live foliage on stems above a height of 3 m) averaged per plot from sighting tube measurements (densitometer manufactured by Geographic Resource Solutions, Arcata, CA, USA) at the center and four cardinal directions along the perimeter of each plot.
Figure 1. Example plot showing the increase in cover of the invasive Celastrus orbiculatus in a young containment treatment for the fungal pathogen oak wilt, Wildwood Preserve, Ohio, USA. On this plot in June 2020 (the first growing season and 3 mo after oak wilt sanitation treatment), C. orbiculatus had 0.25% cover, increasing to 2% in 2022 and 10% in 2024. In the June 2024 photo, C. orbiculatus formed a mat-like covering (visible in the foreground and topped with twining C. orbiculatus) on native woody and herbaceous understory plants and had also climbed the black cherry (Prunus serotina Ehrh.) tree in the foreground on the right. Photos by SRA.
We analyzed log10-transformed Celastrus cover using a generalized linear mixed model including oak wilt treatment (three treatment ages and reference forest), sample year (2020, 2022, and 2024), their interaction, and plot as the repeated-measures subject. We performed the analysis in SAS v. 9.4 (SAS Institute, Cary, NC, USA) using PROC GLIMMIX with autoregressive structure and Tukey tests for multiple comparisons. We then examined association between tree canopy cover and Celastrus cover through time using repeated-measures correlation (Marusich and Bakdash Reference Marusich and Bakdash2021).
Results and Discussion
When our study began in 2020, Celastrus inhabited nearly all plots (25 of 28, 89%), being absent from only one young oak wilt treatment plot and two reference forest plots without evidence of oak wilt. By 2024, all 28 plots contained Celastrus. Mean Celastrus cover varied with the main effects of oak wilt treatment (F(3, 24) = 4.0, P = 0.019), study year (F(2, 48) = 57.7, P < 0.001), and their interaction (F(6, 48) = 3.3, P = 0.009; Figure 2). In 2020, among different-aged oak wilt treatments, mean Celastrus cover ranged from 0.3% (young treatments: 1 yr old in 2020) to 1.0% (old treatments: 5 yr old in 2020), compared with 0.4% in reference forest. Subsequently, Celastrus cover increased sharply through time between 2020 and 2024. The increase was disproportionately high in oak wilt sites, increasing on average by 22 times (old treatments) to 32 times (middle-aged treatments), compared with 3 times in reference forest. By 2024, Celastrus mean cover ranged from 6% (young) to 22% (old) among oak wilt treatments, compared with 1% in reference forest. Maximum covers among plots (all of which were in old oak wilt treatments) increased sharply from 3% in 2020 to 20% in 2022 and 60% in 2024.
Figure 2. Variation in mean cover of invasive Celastrus orbiculatus across oak wilt containment treatments and study years, Wildwood Preserve, Ohio, USA. Error bars are +1 standard error of the mean. Means without shared letters differ at P < 0.05. Ages of oak wilt treatments during the 2020–2024 study period were 1–5 (young), 3–7 (middle), and 6–10 yr (old). Plots in reference forest did not display evidence of oak wilt and were untreated for oak wilt.
Celastrus cover was not correlated with tree canopy cover across all years and plots (repeated-measures r = 0.14, P = 0.316, df = 55) nor across only oak wilt plots (repeated-measures r = 0.15, P = 0.334, df = 41). Nearly all or all (99% to 100%) of the Celastrus cover was in understories as a shrub- or mat-like growth form rather than as climbers on tree boles.
Results suggest that Celastrus was present at low cover in reference forest, and the inseparable influence in our study of oak wilt and its attempted containment acted as a catalyst for a major increase in Celastrus cover. As an influence of oak wilt was killing mature oaks to create canopy gaps, it may seem surprising that Celastrus cover was not correlated with tree canopy cover. Average tree canopy cover narrowly ranged from 51% to 60% among oak wilt treatments (compared with 91% in reference forest) and changed little (by |2–9|%) between 2020 and 2024. The lack of correlation between tree canopy cover and Celastrus cover could result from the qualitative presence of a canopy gap serving as a release event stimulating Celastrus growth, then Celastrus cover continuing to increase under minimally temporally varying canopy cover, resulting in little correlation. As mentioned previously, our study is not intended to partition the potential relative influences of oak wilt–related tree mortality from disturbance associated with its attempted containment, but prior research with Celastrus and its traits suggest that the appearance of canopy gaps in the presence of Celastrus seedlings was likely a major contributor to Celastrus’s increase (Pavlovic and Leicht-Young Reference Pavlovic and Leicht-Young2011). Although Celastrus may not form persistent soil seedbanks, the species’ shade tolerance enables persistence of seedlings in shaded, forest understories (Ellsworth et al. Reference Ellsworth, Harrington and Fownes2004). These seedlings grow slowly in shade but can initiate rapid growth if light increases (McNab and Meeker Reference McNab and Meeker1987). As Celastrus fruits are dispersed by animals (e.g., birds), which may be attracted to oak wilt openings, the increase in Celastrus cover we observed could stem from accelerated growth of existing seedlings as well as new recruitment (Greenberg et al. Reference Greenberg, Smith and Levey2001; McNab and Loftis Reference McNab and Loftis2002).
To what extent oak wilt as a disturbance is unique in facilitating Celastrus invasion is not clear. Prior research has reported increases in Celastrus after logging (Silveri et al. Reference Silveri, Dunwiddie and Michaels2001) and wind disturbance (Berg et al. Reference Berg, McNab and Zarnoch2023). A potential difference between these typically more discrete disturbance types and oak wilt is that oak wilt apparently kills trees on a more continual basis across the landscape (Juzwik et al. Reference Juzwik, Appel, MacDonald and Burks2011). As these oak wilt–created canopy gaps form, Celastrus cover can keep increasing within them for at least 10 yr based on our results and similar to continued increases across two decades observed after tree windthrow in North Carolina (Berg et al. Reference Berg, McNab and Zarnoch2023). In our study, this temporal pattern resulted in persistently increasing Celastrus cover in existing oak wilt sites, as well as Celastrus increasing as new oak wilt sites continually formed, cumulatively increasing the amount of Celastrus cover across the landscape.
Celastrus can negatively affect native plant communities by reducing growth or killing trees that the liana climbs and by shading and outcompeting native understory plants (McNab and Meeker Reference McNab and Meeker1987). The sharp increase in Celastrus we observed at oak wilt sites suggests that treating Celastrus could aid forest adaptation to oak wilt. Treating Celastrus (e.g., using herbicide; McKenzie-Gopsill and MacDonald Reference McKenzie-Gopsill and MacDonald2021) proactively in reference forests where it occurred at low cover could potentially temper the increase that would otherwise occur if or when oak wilt arrived. Although treating Celastrus mats that have already formed on top of other understory plants following oak wilt presence may be more challenging, large-diameter stems of Celastrus can be controlled by cutting (McKenzie-Gopsill and MacDonald Reference McKenzie-Gopsill and MacDonald2021). As Celastrus cover increased sharply through time in oak wilt treatments, co-managing oak wilt and Celastrus may help favor native plants in oak wilt patches. Our study highlights how forest changes from a tree-afflicting pest and its attempted containment were followed by acceleration of non-native plant invasion.
Acknowledgments
We thank Zuri Carter, Tim Schetter, and Jay Wright (Metroparks Toledo) for facilitating the study; Metroparks Toledo interns Ashley Fink and Elizabeth Stahl for help sampling in 2022; LaRae Sprow, Jason Diver, Tim Gallaher, and Metroparks staff for planning and implementing oak wilt treatments; Josh Brenwell for GIS support; and two anonymous reviewers for comments on the article.
Funding statement
This study was funded by Metroparks Toledo through a contract to Natural Resource Conservation LLC.
Competing interests
The authors declare no known conflicts of interest.
Open access
