First record of Leucanthemum vulgare (Lam. 1778) (Asterales: Asteraceae), ox-eye daisy in Limpopo province of South Africa

Leucanthemum vulgare , Lam. (Asteraceae) (ox-eye daisy) is a perennial herb and an aggressive invader native to Eurasia. Today, the species is found in all continents except for Antarctica, and is an invasive in many countries. In this study, we present records of invading populations of Leucanthemum vulgare in the Limpopo province, South Africa, covering over 21.717 m 2 of critically endangered grassland. Management guidelines for Leucanthemum vulgare are provided.


Introduction
Invasive alien species represent one of the major factors of global loss of biodiversity and disruption of natural ecosystems (Sala et al. 2009).The oxeye daisy, Leucanthemum vulgare, Lam.(Asteraceae) is considered one of the invasive plant species with the greatest negative impact on global biodiversity (Ahmad et al. 2019a).Leucanthemum vulgare stricto sensu diploid (2n = 18) is closely related to Leucanthemum ircutianum (Turcz.)DC tetraploid (2n = 36) (Stutz et al. 2018), both are part of the L. vulgare species complex (L.vulgare aggregate), L. ircutianum is part of L. vulgare sensu lato, hereafter Leucanthemum vulgare s.l.(Greiner et al. 2012;Oberprieler et al. 2018).Leucanthemum vulgare is a perennial herb native to Eurasia, and an incredibly robust and widespread invasive species (Stutz et al. 2018;Ahmad et al. 2019a).It was introduced and naturalized into many countries around the globe (Clements et al. 2004;Stutz et al. 2018;Ahmad et al. 2019a), primarily for ornamental reasons, as a seed contaminant (Clements et al. 2004;McDougall et al. 2018), and for phytoremediation (Noori et al. 2018).Post introduction, L. vulgare tends to escape cultivation into the wild, where it establishes viable populations (Khuroo et al. 2010).The species has naturalised in most countries (Khuroo et al. 2010;McDougall et al. 2018;Stutz et al. 2018;Ahmad et al. 2019a).Even in its natural range such as in Russia and Finland, for L. vulgare is regarded as an environmental weed (Jantunen and Saarinen 2002).
Leucanthemum vulgare is able to establish in a wide range of habitats, such as high-altitudinal mountainous landscapes (McDougall et al. 2018;Ahmad et al. 2019a), valleys (Khuroo et al. 2010), and disturbed areas like grazing pasture, open meadows, roadsides, and forest openings (Cofer et al. 2008;Stutz et al. 2016Stutz et al. , 2018;;McDougall et al. 2018;Ahmad et al. 2019a, b, c).The distribution of L. vulgare globally attests to this ability to easily spread and occupy this wide range of habitats.Leucanthemum vulgare poses a major threat to natural landscapes because its invasion reduces plant species diversity (Clements et al. 2004).Leucanthemum vulgare can also form dense, extensive populations in pastures where it is generally avoided by grazing cattle, and its shallow roots often promote soil erosion in areas it has invaded (Clements et al. 2004).A recent Species Distribution Modelling (SDM) study by Ahmad et al. (2019a), reported L. vulgare to have suitable habitats in only in the winter rainfall parts of South Africa.The finding of a thriving population in the north most part of South Africa is intriguing.In this study, the records of wild populations of L. vulgare invading the Woodbush Granite Grassland (WGG) in the Limpopo province, South Africa, are presented, as well as recommendations on how to manage the invasion.

Materials and methods
Multiple field surveys were conducted in the Haenertburg and surrounding areas of the Greater Tzaneen district municipality, in the Limpopo province, South Africa, in 2018-2019 (Moshobane et al in press).The purpose of the field surveys was to find and document the presence of invasive alien species in South Africa, with a special focus on species whose populations are limited or emerging invaders, in order to detect them early before the populations expand.This work was done as a programme of the Biological Invasions Directorate (formerly known as Invasive Species Programme) of the South African National Biodiversity Institute (SANBI) (Wilson et al. 2013).The recent major population of L. vulgare was reported by the third author, as the custodian of the WGG.The species was identified based on morphological features and verified by a Dr. Marinda Koekmoer taxonomist based at Pretoria National herbarium of the South African National Biodiversity Institute.Herbarium specimens were deposited at the Larry Leach Herbarium of the University of Limpopo (KRA 0529539, 0529545, 0529546).The occurrence map of L. vulgare in South Africa was prepared using the Herbarium records (Figures 1, 2) and records from Southern African Plant Invaders Atlas (SAPIA) records (Henderson 1998).The occurrence records for L. vulgare linked to preserved specimens (n = 16067) were obtained from the Global Biodiversity Information Facility (GBIF 2021, https://www.gbif.org)and they were used to produce habitat suitability map  for South Africa (Figure 1).The R-software version 3.6.4(R Core Team 2020) R package sdm version 1.0-89 (Naimi and Araújo 2016), was used.We initially considered 19 bioclimatic variables that were downloaded from the WorldClim platform (www.worldclim.org) at approximately 1 km × 1 km spatial resolution (Fick and Hijmans 2017;Booth 2018).We validated the models by means of the Area Under the Curve (AUC) of a Receiver Operating Characteristic plot (ROC) (Fielding and Bell 1997), the True Skill Statistic (TSS) (Allouche et al. 2006) and the randomised correlation (COR) (Thuiller et al. 2009).Whereby, the AUC, An AUC value of 1 indicates perfect performance, whereas an AUC value of 0.5 indicates that the model performs no better than a random model.AUC > 0.7 generally indicates good model accuracy (Pearson et al. 2006), and A TSS value below 0 indicates a no better than random model performance, and a value of 1 indicates perfect performance (Allouche et al. 2006), and lastly COR values closer to 1 indicated which predictor variables contributed highly to the respective species suitability using the selected six methods (Thuiller et al. 2009).The

Results and discussion
We found large populations of L. vulgare in two sites in Tzaneen district municipality, Limpopo province, South Africa.The first population was found in 2019 along the riverbed, on the R71 outside of Haenertsburg village.For the newly discovered populations of L. vulgare, which were found in October 2020 in the Haenertsburg graveyard, within the critically endangered Woodbush Granite Grassland (WGG), we estimated the occupied area to be 21.717m 2 , and the number of individuals to be 200 plants within the WGG.Despite the findings from Ahmad et al. (2019a), which indicated the suitable habitats to be only in the southern parts of South Africa, herbarium records indicate that the species occurs in six provinces (Supplementary material Appendix 1), the earliest record in South Africa was in 1884 in Kwa-Zulu Natal which is 950 km away from the reported point in Limpopo Province.The present study reports the most northern records of L. vulgare.
The discovery of L. vulgare in the WGG (Figure 3) is a great cause for concern because this species is a common grassland invader (Coulson et al. 2001).In particular, where L. vulgare invades, it spreads rapidly and significantly alters existing plant species' structure and composition, and decreases species richness at all spatial scales (Ahmad et al. 2019b).Furthermore, Ahmad et al. (2019c) showed the significant impact of L. vulgare invasion on key soil properties such as soil organic carbon (OC) and total nitrogen (N) between invaded and uninvaded areas.Upon inspection of inflorescences, beetles (coleoptera), butterflies (Lepidoptera), flies and mosquitoes (Diptera), ants (Hymenoptera), and bees (Hymenoptera) were found to visit the flowers of L. vulgare.Most notably were the spiders' traps, with a honeybee caught by a spider on one of the flowers found (Figure 3).These findings are consistent with other studies that have observed the same phenomenon (Suttle 2003).
The invasiveness of L. vulgare is widely reported in literature, and its invasive is noted in several countries around the world.The potential global invasiveness of this species is already established in a study by Ahmad et al. (2019a), as well as its impacts (Khuroo et al. 2010;Stutz et al. 2018;McDougall et al. 2018;Ahmad et al. 2019a, b, c).Leucanthemum vulgare is seen as one of the worst invasive alien species in protected areas across the globe (Foxcroft et al. 2017;McDougall et al. 2018;Ahmad et al. 2019a).For example, L. vulgare has invaded the Kosciuszko National Park in New South Wales (McDougall et al. 2018) and many other protected areas around the world (Foxcroft et al. 2017).The results obtained from different algorithms used for SDMs high excellent performance in projecting potential distribution of the target grass species (Table 1).The high values for AUC, TSS indicate the reliability of our models, and thus they were all incorporated into the ensemble models (Figure 1).The habitat suitability modelled in this study shows that the species has greater chances of expansion and it may colonise larger areas.The results show that L. vulgare has suitable areas in Limpopo province, this result is contrary to the findings reported by Ahmad et al. (2019a).This could be attributable to the algorithm used or the number of records included in the modelling.

Management
Initiating a rapid response and the development of an area based eradication and control strategy in order to curb the spread and associated impacts of this global invader (Clements et al. 2004;Ahmad et al. 2019b, c).Although mowing was shown to be ineffective in controlling L. vulgare as it is resistant to mowing (Wahlman and Milberg 2002), mowing may reduce seed production (Coulson et al. 2001).Root-feeding weevil Cyphocleonus trisulcatus Herbst is recommended as a biological control agent for L. vulgare (Stutz et al. 2020), and tribenuron is recommended as an effective herbicide (Olszyk et al. 2013).Therefore we recommend, both mowing and herbicide treatment.
In South Africa, invasive alien species are regulated under National Environmental Management: Biodiversity Act (NEMBA), which provides the list of invasive alien species (Moshobane et al. 2019).We recommend that this species should be listed in South Africa, possibly as category 1a, and that eradication plans be implemented.
The management implications of the present study are therefore especially relevant for: (i) Preventing new introductions, through regulating the species in NEMBA regulations for invasive species.(ii) Prioritizing controls on regulated species that could be placed on an eradication plan while the populations are still manageable at a low cost.(iii) Educating the public about the regulations that manages invasive species.(iv) Implementing an early detection and rapid response programme.(v) Conduct a risk assessment.(vi) Genetic analysis research should be conducted to ascertain the species identification.

Figure 1 .
Figure 1.Habitat suitability map and distribution map for L. vulgare in South Africa (colour scale, 0.1 less suitable 0.8 more suitable).The black dots indicate the occurrence records of L. vulgare in South Africa.
present study considered three algorithms to test the performance of the model namely boosted regression tree (BRT: Friedman 2001), generalised linear model (GLM: McCullagh 1989), random forest (RF: Breiman 2001), and support vector machine (SVM: Vapnik 1995), maximum entropy (MaxEnt: Elith et al. 2006).In essence, the algorithms combines to the best fit the occurrence records and predictor variables to generate the model.The resultant map, is an indicator of habitat suitability (Phillips et al. 2006; Phillips and Dudík 2008).

Figure 3 .
Figure 3. Invasion of Leucanthemum vulgare at Haenertburg graveyard and WGG: (A) Extent of invasion; (B) Spread from graveyard into WGG; and (C) Honeybee trapped by spider on Leucanthemum vulgare inflorescences.Photograph by M.C.Moshobane.

Table 1 .
Results of evaluation of the performance of models.(AUC and TSS values closer to one indicate excellent model output) Methods performance run at 100 boostrap replication.