No consistencies in abundance–impact relationships across herbaceous invasive species and ecological impact metrics

During the last two centuries, invasive species have become a ubiquitous problem in most ecosystems, causing wide-scale loss of biodiversity (David et al., 2017; Pyšek 2020) and economic damage (Diagne recipient communities. The impact is defined as measurable change an ecosystem property that can be attributed to individual (Ricciardi 2013). A plethora case studies identified impacts plant on native (Hejda 2009; Tekiela & Barney, Vilà 2011), animal (Fletcher 2019; Schirmel 2016) soil microbial communities (Huangfu Stefanowicz 2019), properties (Stefanowicz 2017, 2018), fire regime (Fusco 2019) functioning services (Castro-Díez Monserat Hulme, 2017). As consequence globalisation, more alien are come, increase number non-native does not show any sign saturation (Seebens Despite exponentially increasing studies, success poorly explored rarely quantified, while understanding mechanisms behind their limited (Barney 2013; Hulme O'Loughlin for at least three simple reasons. First, direct assessment highly time- budget-consuming; therefore, only tiny proportion known been subject (Hulme Second, published assessments refer single (Tekiela 2017), making synthesis difficult. Third, research largely restricted examine few measures, thus potentially missing often complex effects abiotic biotic features community (Crystal-Ornelas Lockwood, 2020; choice measures bias our perception invasion; broader overall effect invasion requires set ecological metrics These arguments also explain ability predict particular before they widespread considerable resident widely applied framework quantifying states product abundance per capita define magnitude direction (Bradley Parker 1999). Therefore, threat local communities, it crucial understand how populations Sofaer 2018). scale linearly or nonlinearly with may imply low- high-density threshold (Sapsford Yokomizo 2009). Especially high (i.e. occur beyond critical abundance) help managers choose optimal timing control eradication (Panetta Gooden, Individual strongly supported significant link between cover, density, biomass specific such changes richness, diversity traits (Gooden Livingstone 2015; Thomsen 2011). But measure correlates positively negatively) species, shape linear nonlinear) strength steepness curve) relationships vary considerably among depending response metric being measured target 2015). Likewise, syntheses considered might species; thus, general abundance–impact relationship (but see Bradley 2019). multispecies needed reveal different measures. size individual) determined by morphological physiological significantly affect environment (Drenovsky 2012). If constant, abundance. However, nonlinear expected if varies density Regarding case-specific evidence supports functional species. Life span growth form cited essential distinguishing strongest impact. For example, rhizomatous perennials mat-forming creeping had effect, annuals minimal composition Mediterranean habitats (Fried 2014). Height, leaf area (SLA), flowering phenology and/or seed mass were shown important other Martin te Beest meta-analysis found traits, stature, life pollination syndrome, confer means regardless invaded biome (Pyšek there major need study whether similar cause which this sense. In summary, depends intraspecific variability along gradient, characteristics disturbances (Thiele 2010). perspective incorporates function invader's could step forward considering context dependencies phenomena. Such shed light key question: simply, abundant those having bigger (Sofaer Another pending ‘outstanding question’ type consistent across shape, these study, objective was identify 11 herbaceous East-Central Europe properties. We used ‘space-for-time substitution’ approach observational comparison non-invaded sites) assess chosen This provides valuable wealth information several taxa, allowing us test generalise hypotheses suggested experiments small-scale studies. space-for-time has its drawbacks: allow rule out possibility pre-invasion differences environmental (e.g. resistance) influence (Thomaz Nevertheless, recent using time comparisons sites pre- post-invasion situations) combining substitutions time-series data suggest spatial patterns reliably reflect temporal dynamics (Pearse Stotz selected 16 commonly literature regarding wide variety structure After completing quantitative assessment, we each metric. Then, tested better understanding. Our objectives (1) investigate assessing same wise taking together draw conclusions; (2) explore cover–impact metrics, (3) latter, competitive related acquisition strategies, productivity competitiveness, (Navas Violle, 2009), already emphasised determining 2014, 2012; occurred natural semi-natural terrestrial regenerating old fields Romania Hungary; occupied large enough areas delimitate seven 30 m × 50 sampling protocol (see Kovács-Hostyánszki 2022); locally sample plots low invader cover. included comply conditions countries under investigation. studied following: Ambrosia artemisiifolia L., Asclepias syriaca Erigeron annuus L. (Desf.), canadensis Gaillardia aristata Pursh, Helianthus tuberosus Impatiens glandulifera Royle, Reynoutria japonica Houtt., Rudbeckia laciniata Solidago gigantea Aiton, Symphyotrichum lanceolatum Nesom. agg (Appendix S1: Table S1). Central European (except G. aristata) therefore representative invasions temperate zone Europe. surveyed various regions Hungary S1 Figure S1), where one achieved dominance To infer substitution chose pairs species: ‘invaded sites’, ‘uninvaded no cover (<5%) absent. Within sites, sampled variable S1) relationship. site pairs, uninvaded possible vegetation, succession status, former land-use, elevation, slope exposure S2). assumed absent from solely due dispersal limitations. purposely without presence exclude potential confounding effect. method correlations variables, but cannot establish cause–effect attributes follow Global Invader Impact Network 2015) neighbouring because many vegetation very different. average distance invaded–uninvaded ranged hundred meters A. artemisiifolia, S. gigantea, syriaca) thousand E. annuus, I. glandulifera, R. laciniata). total, 154 types ranging dry, riparian lowland rivers all 3 (154 = 462 altogether; Appendix Fieldwork carried 2017 2019. All once, May June, during peak make identification easier. Percent bare ground vascular visually estimated. Since percent hereafter, use term ‘cover’ instead ‘abundance’ when referring data. Species shrub tree layers present excluded analyses. characterise trait height SLA clonality extracted databases below). (‘native height’) 10 random points then averaged. (leaf unit mass, mm2/mg) values came measurements international differ regional (Sonkoly J., unpublished data). survey (364 altogether). collected five mature nonsenescent leaves calculated based photographs ImageJ software (Abràmoff 2004). dried 48 h 65°C weighed. (thousand weights grams; Török 2013, 2016), (‘annual clonal spread’) CLO-PLA database (lateral spread [cm/y], Klimešová samples randomly cores site: 100 mL core upper 5 cm spud. mixed sent analyses University Debrecen, Faculty Agricultural Food Sciences Environmental Management, Hungary. They provided pH, organic carbon (C) humus, phosphorus (P) nitrogen (N) content samples. six focal relationships: height, SLA, annual spread, duration individuals averaged them level. above), field guides (Király, Sârbu both grouped into four categories: basic characteristics, (Table 1). 2015), measurement implemented within brief period Plant plot level, pH nutrient analysed level according above). calculation wanted quantify contribution functionality (Thomsen 2016). compositions characterised (measured situ disregarding above) community-weighted mean (CWM; i.e. species-level weighted cover) spread. consider space sites. calculate main representing axes strategies de Bello, Westoby, 1998). describe following indices: evenness, divergence (Villéger 2008) Rao's quadratic entropy (RaoQ, Botta-Dukát, 2005). invasion. ‘impact metrics’ throughout manuscript. quantified ratios (RR) vs. plots, equation: RR ln (Cinvaded/Cuninvaded), C (Murphy Romanuk, equation sizes meta-analyses (Hedges 1999) competition (Brown Cahill, 2022). absolute value quantifies amount impact; shows negative indicate decrease invasion). base logarithm allows interpreting easily: −2 quarter, −1 half, 1 two-fold increase, 2 fourfold compared explicit pairings paired, site, averages reference (Cuninvaded) RRs. content, species-wise 95% confidence intervals characteristic studied, RRs zero. interval contain zero, 5% reported generalised additive models (GAMMs; Zuur 2009) varied gradient adequate choices relax assumption linearity permitting explanatory variables take straight line curves complexity. Because definition, missing, forced zero Each separately. Two GAMMs RRs) variable. model 1, taken together. Thus, nested terms. While 2, wise, so identity By comparing models, decide contributes Akaike's criterion (AIC) models. Model predictor) presented graphically interpreted parsimonious (lower AIC, delta AIC > 2) than together). always certain metric, projected curve horizontal line. effective degree freedom (edf) curvature An edf ≤ translated variable, >> denotes assumptions (normally distributed residuals constant variance) verified diagnostic plots. basis dimension smooth (Wood, Section 5.9), ‘k-index’ check. gam function. AIC-based selections compare MOB partitioning better, trees shown, criteria indicated. lower), grouping presented, indicated explained involved analysis. Note assumes impact). Although fully satisfied data, apply simplification did find handle unknown (Hothorn Zeileis, performed R 4.1.1. statistical (R Core Team, 2021). CWMs FD package, version 1.0–12.1 (Laliberte Legendre, GAMM fitted mgcv 1.8.36 2006), model-based done Linear Tree (lmtree) interface partykit 1.2–15 graphical visualisation, ggplot2 3.3.5 (Wickham, worked available Dryad (Fenesi 2023). Considering together, 9 showed less rich lower (Figure slightly higher CWM lateral functionally even RaoQ, N, P differed strikingly metrics. richness 1a), showing diversity, H. tuberosus, japonica, laciniata, On contrary, ones 1b). species-specific differences. tallest (H. 1e). 1f). 1g). perennial invaders (R. 2h). Among evenness RaoQ half 1i–l). Only revealed (N, P, C) syriaca, canadensis, aristata, lanceolatum, neutral results 1m–o). invasion-driven 1p). When depended cases (12 16, 2), 2). however, impact: N matters relationships, since predictors 13 exceptions 2h,l,p). detected exceeded 75% 2a), uniform 2b). except 2a). exception, increased decreased 2c). positive (vegetation spread; 2e–h) separately, impacted (in respectively, 2e–g). clear thresholds case: above 40% 2e). until reached 90% drastic 2i). individually, results: lanceolatum. proved analyses; clearly exception 2j). unanimously analysing separately 2k). communities' index complexity, together: initial 30% followed slight 80%, pronounced threshold. curve: decre