THE EFFECTS OF AN INVASIVE ALIEN PLANT ( CHROMOLAENA ODORATA ) ON LARGE AFRICAN MAMMALS

Alien plants have invaded most ecosystem types (terrestrial, fresh water and marine) and are responsible for the loss of irreplaceable natural services on which humankind relies. They alter food quantity, quality and accessibility, and may result in declines in native species richness, which may ultimately result in e�tinc� tion. For an effective management of invasive alien plants, it is important to understand the effects that such plants have on all levels of biodiversity. However, the effects that invasive alien plants, such as the Triffid weed (Chromolaena odorata), have on mammalian biodiversity, especially large mammalian species, are not well�known, although they play major ecological roles in areas such as nutrient cycling. Also, little is known about the recovery of the ecosystem following alien plant removal. This study investigated the effects of C. odorata invasion on large mammalian herbivores in Hluhluwe�iMfolozi Park and whether clearing of this plant helped in rehabilitating the habitat. We used track counts to estimate and compare species richness, di� versity and abundance indices for large mammalian species between areas with differing C. odorata invasion durations (ca 2 years, ca 10 years, ca 20 years), areas with differing clearing times (cl < 2 years, cl 3–5 years) and an area without any history of C. odorata invasion as a control. The results from this study show that large mammalian species utilised the uninvaded and the cleared areas more than the invaded areas. Species richness, abundance and diversity decreased with increasing invasion duration and cleared areas showed an increasing species richness and abundance. We conclude that this invasive alien plant modifies habitats and their removal does aid in the restoration of the ecosystem.


Introduction
Invasion by e�otic species is the second larg� est threat after habitat destruction to global biodi� versity (Schei, 1996).The impact of invasives can cause changes in ecosystem functioning (Vitousek, 1990) and a reduction in productivity for native spe� cies due to competitive interactions with the invader (Gould & Gorchov, 2000;Miller & Gorchov, 2004).Declines in native species richness occur in a vari� ety of communities following invasion (D'Antonio & Vitousek, 1992).However, quantitative data showing direct effects of invasives are rare, with predictions of e�tinctions linked to invasives of� of� ten based on anecdotes, speculation and limited observation (Gurevitch & Padilla, 2004).
Triffid weed, Chromolaena odorata (L.) R.M. King & H. Rob. (Asteraceae), is native to the Neotropics and occurs in suitable areas below 1000 m (Zachariades & Goodall, 2002).The plant is allelopathic (Fitter, 2003) and is one of the worst weeds in the southern African subregion (Goodall & Erasmus, 1996).In KwaZulu�Natal province, South Africa, it has invaded nearly all conservation areas, and in Hluhluwe�iMfolozi Park (HiP) where the current study has been conducted, it poses a threat to the natural vegetation (Macdonald, 1983;Te Beest et al., 2017).There has been some success in its clearing in the HiP (Dew et al., 2017).
Alien species may effect various impacts on an ecosystem, from genetic impacts, impacts affecting individual organisms, impacts affecting populations, impacts affecting community structure and impacts affecting ecosystem processes (Parker et al., 1999;see Davis, 2009 for a review).These studies usually investigate the effects of alien invasions on invertebrates (e.g.Samways & Taylor, 2004;Mgobozi et al., 2008) or on the vegetation (e.g.Te Beest et al., 2015) but seldom on mammals.This is despite large mam� This is despite large mam� mals having major ecological roles in ecosystems via nutrient cycling (Holland & Detling, 1990;McNaughton et al., 1997) and in determination of vegetation structure and composition (Augustine & McNaughton, 1998), which ultimately affect the abundance and diversity of most other ta�a (Milc� hunas et al., 1998).
There is little documented information to show whether the ecosystem recovers after the removal of invasive alien species.Rozen�Rechels et al. (2017), using transect counts in HiP from 2004 and 2014, showed that the ungulates in the HiP do respond negatively to C. odorata, with grazers been more affected than browsers.Here we take a different approach to see what the long�term changes are in response to the clearing of C. odorata.

Material and Methods
HiP lies between latitudes 28º 00' to 28º 26' S and longitudes 31º 43' to 32º 09' E and is appro�imately 960 km 2 in size (Howison, 2017) (Fig. 1).Using historical maps of the distribution of C. odorata in the park (provided by Ezemvelo KZN Wildlife), si� treatments with differing C. odorata invasion durations and differing clearing times were selected.These treatments included one that had re� cently been invaded (ca 2 years); two that had been invaded for a longer period (ca 10 years and ca 20 years); a treatment that had recently been cleared (cl < 2 years); a treatment that had been cleared for a longer period (cl 3-5 years); and finally, as a control, a treatment without any history of C. odorata inva� sion.The treatments were located in the North East part of the park, where C. odorata invasion was the highest (Rozen�Rechels et al., 2017).All plots were in a section dominated by white Stinkwood trees (Celtis africana Burm.f.) and were very similar in vegetation composition, but differing in C. odorata invasion duration and clearing times.
We used tracking plots to estimate large mam� malian species richness and diversity and abun� dance.In each of the si� treatments, tracking plots were created along tracks that showed evidence of mammalian utilisation.Tracking plots were repli� cated eight times in each treatment, totalling 48 plots for the study.Located appro�imately 100 m apart and in different stands of C. odorata, each plot was created by removing the vegetation in a space 3 × 1 m.The plots were raked and smoothed and fine soil, of the same type as in the vicinity, was added to provide a good tracking surface and to ensure that the animals could not visually dis� tinguish the plots from their usual road track.Us� ing a Global Positioning System unit, the position of each tracking plot was recorded.All tracking plots were e�amined for tracks at least twice a week from February to June 2006, making a total of 27 tracking days.This was at the end of the rainy season and therefore rain did not interfere with the results.At each tracking plot, the num� ber of track sets left by each mammal species was recorded, after which the plots were raked and smoothed in preparation for the following observation (e.g.Wilkie & Finn 1990;Engeman & Allen 2000;Engeman & Evangilista 2006).To avoid pseudo�replication, an attempt was made to determine whether track sets belonged to the same animal that may have wondered around the plot or to two different animals.We calculated a relative inde� of species abundance for each species following Wilkie & Finn (1990).A non�metric multidimensional scal� A non�metric multidimensional scal� ing (nMDS) was constructed to graphically repre� sent similarities or dissimilarities between treat� ments.The nMDS plots samples that are similar in their species composition are closer together and samples that differ in species composition are plotted further apart in a two�dimensional ordina� tion space.An Analysis of Similarities (ANOSIM; R) was used to test for significant differences in the animal species composition of the treat� ments.We calculated the Shannon and Simpson's diversity indices to further e�amine the differences in mammal species composition between the treatments.All these calculations were carried out using the PRIMER v5 software (PRIMER�E, Ltd., Plymouth, U.K.; Clarke & Warwick, 2001).The Kruskal�Wallis test ANOVA (H) was used to test for significant differences between the treatments using the diversity indices and the relative indices of abundances, at 5% level of significance.

Results
We recorded 3469 tracks from 17 large mammal species.Table 1 shows the number of species recorded, the number of tracks recorded per species, the relative indices of abundance for each species, and the percentage of tracks per species, recorded per treatment.We found signifi � We found signifi� cant differences (p > 0.05) between the treatments with regard to diversity indices (Fig. 2), with the pairwise comparison showing that the ca 10 & 20 yrs treatments were significantly different to the control (Table 2).The nMDS shows that there are differences in animal species composition between the selected treatments (Fig. 3).All si� treatments were different in species composition (Global ANOSIM, R = 0.946, p = 0.001; pairwise ANO� SIM.Table 3).The pairwise ANOSIM test statistic R�values were all close to one (Table 3), suggesting that similarities between samples from one treat� ment are much less than any similarities between samples from other treatments.Based on this, the hypothesis of no differences is then rejected.

Discussion
Chromolaena odorata is a dense, sometimes impenetrable bush, and our data suggest that this has a negative effect on large mammal communities.Although in a limited area within HiP we observed that mammalian species richness and diversity were highest in the control area after which it decreased with increasing C. odorata invasion duration, with the area that had been invaded for the longest time (ca 20 years) displaying the lowest number of species.This suggests that the longer this invasive plant is left to establish itself, the less attractive the habitat becomes to many large mammals.These findings are supported by a multitude of studies which have found invasive plants to have a negative influence on native organisms (although mostly plants) and pose a threat to diversity (Richardson et al., 1994;Pyšek & Pyšek, 1995;Cole & Landres, 1996;Rozen�Rechels et al., 2017).
Although the treatments had the same vegetation types, they differed in species composition, which is attributed to the presence of C. odorata.The nMDS shows that the ca 20 yrs treatment displays the most different species composition in comparison to the others.This highlights the long�term impact of C. odorata.The abundances of some species (Table 1) are similar in invaded and un�invaded treatments, suggesting indifference in the use of these areas, or that these ungulates are perhaps using the thick C. odorata for shelter from predators.However, the absence of buffalo Syncerus caffer tracks in the invaded treatments suggests that, although they could pro� that, although they could pro� vide shelter, these areas were not suitable habitat for other purposes.Buffalos are unselective bulk grazers (Sinclair, 1977), but they have been found to adjust their selection of grass species and feed� ing localities in relation to seasonal changes which change the quantity and quality of food (Macan� dza et al., 2004).It is assumed that they were using alternative and more suitable habitats.Our data on buffalos corroborate Rozen�Rechels et al. (2017) who, using transect sampling, showed grazers to be more affected than browsers.Lion (Panthera leo) and spotted hyaena (Crocuta crocuta) tracks were recorded only in densely invaded treatments (ca 10 years and ca 20 years), which could be attributed to the lion's selecting areas where prey would be easier to catch (Hopcraft et al., 2005).
Our results suggest that C. odorata invasions change the habitat for use by large mammals by outcompeting native plant species either through their allelopathic characteristics (Zachariades & Goodall, 2002;Fitter 2003), or by shading, or by competition for moisture or nutrients.Although diversity improves with the clearing of the plant, the species composition remains different to pre� invasion state.This is concerning, as it appears, that the habitat may remain changed for a long time after clearing.Chromolaena odorata has also been found to have adverse effects on some of the high�profile mammalian species in South Africa, such as the Nile crocodile Crocodylus niloticus Laurenti, 1768 in the GSLWP (Leslie & Spotila, 2001) and possibly the black rhinoceros Diceros bicornis (Linnaeus, 1758) in HiP (Reid et al., 2007).In the same study area, Mgobozi et al. (2008) found the progressive invasion of C. odorata brings with it changes in spider abundance, assemblage patterns, diversity and estimated species richness.However, spider assemblages are restored after clearing.The same trend occurred with small mammals in the area (Dumalisile, 2007).We conclude that C. odorata invasions negatively affect large mammal species' habitat use.C. odorata has been well managed and controlled in HiP but remains widespread and thus there is a risk of reinvasion (Te Beest et al., 2017).Follow up clearance of invasive alien plants is essential to restore ecosystem functioning (Te Beest et al., 2017).More studies on the effects of alien invasives on different levels of biodiversity are a necessity.

Fig. 3 .
Fig. 3. nMDS showing large mammal assemblages in ar� nMDS showing large mammal assemblages in ar� showing large mammal assemblages in ar� ar� eas with differing C. odorata invasion durations, differing C. odorata clearing (cl) times, and in the control area without any history of C. odorata invasion in Hluhluwe� iMfolozi Park, South Africa.

Table 1 .
The total number of tracks recorded, the total number of species recorded, the total percentage of tracks recorded per species in treatments with differing C. odorata invasion and clearing (cl) times and in the control area without any history of C. odorata invasion in Hluhluwe�iMfolozi Park.Relative indices of abundance for each species in each treatment are given in brackets

Table 3 .
ANOSIM results showing significant differences in large mammal composition in areas with differing C.