Tropical Grasslands (2008) Volume 42, 88–95 88

Interference potential of the perennial grasses Eragrostis curvula, Panicum maximum and eriantha with Parthenium hysterophorus

M. VA N DER LAAN1, C.F. REINHARDT1, In a laboratory study, seeds of the 3 grass R.G. BELZ2, W.F. TRUTER1, L.C. FOXCROFT3 species were exposed to pure parthenin in a AND K. HURLE2 germination bioassay. Based upon germination 1 University of Pretoria, Department of and early radicle development, E. curvula was Production and Science, Pretoria, the least sensitive and P. maximum the most sen- South sitive to parthenin. Therefore, if P. maximum was 2 University of Hohenheim, Institute of to be used in a control or rehabilitation program, Phytomedicine, Department of Weed Science, it might be diffi cult to establish P. maximum Stuttgart, Germany stands from seed in areas already infested with 3 Scientifi c Services, Kruger National Park, parthenium. While transplanting seedlings of the Skukuza, grass would avoid the seed germination problem and give the grass a head-start, the practicality of this method on a large scale is open to question. Abstract

The successful invasiveness of Parthenium Introduction hysterophorus (parthenium) is attributed to its competitive ability and high allelopathic The invader plant Parthenium hysterophorus potential. The compound, parthenin, has been (common name: parthenium) is globally rec- implicated as a major allelochemical in the plant. ognised for its threat to agricultural activities, A fi eld trial was established in Kruger National biodiversity and human health. The success of Park (South Africa) to investigate the interference parthenium as an invasive species is attributed to between parthenium and 3 indigenous grass its broad ecological adaptation (Hedge and Patil species, namely: Eragrostis curvula, Panicum 1982), strong competitive ability (Parsons and maximum and Digitaria eriantha. Grass seedlings Cuthbertson 1992), high fecundity and absence were transplanted from a glasshouse into fi eld of natural predators and diseases. Parthenium plots after failure to establish the seed in situ. Par- readily invades disturbed areas, such as along thenium seedlings were introduced at densities roadsides, and degraded rangeland (Haseler of 5 and 7.5 /m2 from areas adjacent to 1976; Pandey and Dubey 1989). It predominantly the fi eld trial. invades pastures in , being vigorous in both developing and established pastures (Navie P. maximum displayed best overall growth per- et al. 1996). Nath (1988) reported that the weed formance and was able to completely suppress reduced forage production in grasslands by parthenium growth with time. The other 2 grass up to 90%. Parthenium is currently invading species performed less favourably, in terms of several prominent game reserves in South Africa, both growth rate and ability to suppress parthe- including Hluhluwe-iMfolozi and the Kruger nium. The ability of P. maximum to interfere National Park (Strathie et al. 2006). effectively with parthenium growth indicates that It is hypothesised that parthenium, in addition to this species has good potential for use as an antag- its competitive ability, is able to impede the recruit- onistic or rehabilitative species in containing the ment and/or growth of naturally occurring plant spread of the weed. species through the release of allelo chemicals. Phenolics and sesquiterpene lactones are impli- Correspondence: C.F. Reinhardt, Department of Plant Production and Soil Science, University of Pretoria, Pretoria cated as the 2 major chemical groups in parthenium 0002, South Africa. E-mail: [email protected] allelopathy. In particular, the sesquiterpene Interference between parthenium and grasses 89

lactone, parthenin, has been observed to exhibit then only garden refuse has been dumped, stop- dose-dependent toxicity effects on a range of test ping with the commencement of the trial. The plants, including aquatic species (Patil and Hedge trial site was cleared of vegetation and debris and 1988; Belz et al. 2007). Reinhardt et al. (2004; 36 plots, each measuring 4 m2, were demarcated 2006) observed that parthenin is sequestered at in a completely randomised design. Following high volumes in the capitate-sessile trichomes on failure to establish the grasses in situ from seed the surfaces of the leaves, and that a single parthe- in December 2003, E. curvula, P. maximum and nium plant can potentially introduce around 270 D. eriantha seedlings were raised in seedling trays mg of parthenin into the environment in a single in the University of Pretoria glasshouses. Seeds growing season. obtained commercially were sown into each tray Few studies have investigated the mecha- cell to form a tuft consisting of 5–8 seedlings. nisms of interference of P. hysterophorus with Once the seedlings attained a height of 4–7 cm, other plant species. Interactions between plants each species was transplanted into fi eld plots at are often the result of complex combinations 16 tufts/m2. Tufts were planted 25 cm apart in of specifi c mechanisms (Welden and Slauson rows parallel to 4 dripper lines (2 plant rows, each 1986; Callaway et al. 1991; Chapin et al. 10 cm off the dripper line) that spanned the plots 1994). Although the fundamentals of competi- at 50 cm intervals. P. hysterophorus seedlings, tion and allelopathy are generally understood as growing in the immediate vicinity of the fi eld isolated mechanisms, little is known about the trial, were transplanted into the plots at densities relative contributions of these 2 mechanisms to of 5 and 7.5 plants/m2. For the 5 plants/m2 den- overall interference between plants (Ridenour sity, parthenium plants were established between and Callaway 2001). Ecologists have identifi ed pairs of grass tufts along the dripper lines, and the importance of defi ning the individual effects additional plants were planted between the grass more precisely (Ridenour and Callaway 2001), rows that ran along the dripper lines for the but diffi culties in separating the effects experi- 7.5 plants/m2 density. Plots without parthenium mentally have hampered better understanding served as controls for the respective grass spe- (Fuerst and Putnam 1983). cies. A gravitational drip-irrigation system was The objective of this study was to investigate installed to supplement the region’s unreliable the growth of P. hysterophorus and 3 indigenous rainfall and as a safeguard against failure of the perennial grass species, Eragrostis curvula, trial. In the second growth season, parthenium Panicum maximum and Digitaria eriantha, at control plots at the 5 and 7.5 plants/m2 densities differing parthenium densities. These grasses were included. A wire fence enclosed the trial to were selected because they were inherently dif- prevent interference from animals. ferent in terms of growth form, vigour and Eight representative grass tufts and 6 repre- adaptability. In addition to the fi eld trial con- sentative parthenium plants were harvested from ducted in the Kruger National Park, a laboratory each plot after 18- and 14-weeks growth for germination bioassay was conducted to assess the 2003–2004 and 2004–2005 growth seasons, the sensitivity of these grass species to the allelo- respectively, and dry matter yields determined. chemical parthenin, as this may have implications Relative grass dry matter, as a percentage of when trying to establish an antagonistic or reha- the control, was logarithmically transformed bilitative species in a parthenium stand. to achieve normal distribution of the data and analysed using SAS®. P. hysterophorus dry Materials and methods matter accumulation data were analysed without transformation. A general linear model (GLM) Field trial of ANOVA was used, and means were separated In the 2003–2004 summer growing season, a by the multiple comparison Tukey test at P<0.05 2-year fi eld trial was established on an abandoned (unless otherwise stated). dumpsite (24°98′S, 31°60′E; elevation 263 m asl), which had been invaded by P. hysterophorus at Laboratory bioassay Skukuza in Kruger National Park, South Africa. The soil on the experimental site is classifi ed as Grass seed used for the germination bioassay was a solonetz form. Dumping of general refuse at from the same seed lot as for the fi eld trial. Pure the site ceased around 12 years earlier, and since parthenin was supplied by our research partner 90 M. van der Laan, C.F. Reinhardt, R.G. Belz, W.F. Truter, L.C. Foxcroft and K. Hurle based at the University of Hohenheim, Stuttgart, E. curvula, 8 days for D. eriantha and 10 days Germany. Extraction and purifi cation of parthenin for P. maximum, because of differences in rates were done from parthenium plants growing in the of germination between the species. Non-linear University glasshouses as described by Belz et al. regression analysis was conducted to fi t logistic (2007). A dose-response bioassay was conducted dose-response curves according to Streibig (1988) using a series of 12 parthenin concentrations using SPSS® regression models. ED50 (dose ranging from 0 to 500 µg/ml. Each concentra- causing 50% of the total response) values for tion in the series, including the control, contained radicle length and germination were calculated 1% acetone as solvent for parthenin. Owing to from equations for best-fi t regression lines, and differences in pre-determined seed germination compared using the F test for lack-of-fi t based on between the grass species, 10, 25 and 30 seeds of analysis of variance (P<0.05). E. curvula, P. maximum and D. eriantha, respec- tively, were placed into 9 cm diameter Petri dishes Results containing one layer of fi lter paper. The same volume of treatment solution (5 ml) was added to Field trial the Petri dishes, and 5 replications were used for each concentration. Petri dishes were sealed with Grass performance. Based on dry matter accumu- parafi lm, and placed in a growth chamber and lation, P. maximum performed the most favourably seeds allowed to germinate in the dark at 20/30°C over the 2 growth seasons, with E. curvula and (12/12h). Germination percentage and radicle D. eriantha both growing poorly, irrespective of length measurements were taken after 5 days for parthenium density (Figure 1). D. eriantha and

 A     PARTHENIUM  PARTHENIUM  PARTHENIUM 

$RYMASSGTUFT    %CURVULA 0MAXIMUM $ERIANTHA 'RASSSPECIES

 B    PARTHENIUM  PARTHENIUM  PARTHENIUM  $RYMASSGTUFT   %CURVULA 0MAXIMUM $ERIANTHA 'RASSSPECIES

Figure 1. Mean grass dry matter accumulation for: (a) the fi rst (2003–2004); and (b) the second (2004–2005) growth seasons on plots containing 0, 5 or 7.5 plants/m2 of Parthenium hysterophorus (parthenium). Interference between parthenium and grasses 91

E. curvula produced little growth in 2003–2004 (2004–2005), only the main effect for grass but growth increased during the 2004–2005 species was signifi cant (P<0.05) (Figure 2b). Aver- season (Figure 1b), with above-ground DM yields aged across both parthenium densities, E. curvula on control plots being 406% and 233%, respec- performed signifi cantly worse than P. maximum. tively, higher than for the 2003–2004 season. At the conclusion of the trial, selected weed den- Grass DM accumulation data, expressed as sities employed in the fi eld experiment were a percentage of control, revealed that neither of low relative to naturally established parthenium the main effects (grass species and par thenium stands outside the fenced trial area, where infesta- density) was signifi cant in the fi rst growth tion levels of 0–96 plants/m2 have been observed, season (2003–2004), and that the interaction depending on local distribution (less plants in effect was not signifi cant at P<0.05, but was shade; more in full sun) and season. signifi cant at P<0.075 (Figure 2a). P. maximum Parthenium performance. In the fi rst growth performed signifi cantly better than D. eri- season (2003–2004), the grass species main antha at 7.5 par thenium plants/m2. Signifi cant effect was signifi cant (Figure 3a). Parthenium growth differences between the 2 parthenium plants growing with P. maximum produced signif- densities were observed only for D. eriantha, icantly less dry matter than those growing with which indicated highest sensitivity to parthe- E. curvula and D. eriantha at both parthenium nium interference. For the second growth season densities. However, in the second growth season

 A     %CURVULA  0MAXIMUM  $ERIANTHA  

0ERCENTAGEOFCONTROL     0ARTHENIUMDENSITYPLANTSM

 B    %CURVULA  0MAXIMUM  $ERIANTHA     0ERCENTAGEOFCONTROL    0ARTHENIUMDENSITYPLANTSM

Figure 2. Grass dry matter as a percentage of control for: (a) the fi rst (2003-2004); and (b) the second (2004–2005) growth seasons as infl uenced by Parthenium hysterophorus (parthenium) density. 92 M. van der Laan, C.F. Reinhardt, R.G. Belz, W.F. Truter, L.C. Foxcroft and K. Hurle

(2004–2005), when all 3 grasses were well estab- that P. maximum was more effective in interfering lished, the interaction effect was highly signifi cant with parthenium growth than the other 2 species. (Figure 3b). At the lower parthenium density, all On E. curvula and D. eriantha plots, parthenium grass species reduced (P<0.05) growth of the yield per plant was higher at the lower par- parthenium plants compared with parthenium thenium density (5 plants/m2) than at the higher plants grown alone, with the degree of suppres- weed density (7.5 plants/m2), while the oppo- sion being in the order P. maximum > D. eriantha site occurred on P. maximum plots. However, > E. curvula. At the higher parthenium den- differences were signifi cant only for E. curvula. sity, all grass species again reduced parthenium growth but the difference was signifi cant only for P. maximum. Growth of parthenium in con- Laboratory bioassay trol plots was lower (P<0.05) at the higher than the lower density. Parthenin had a dose-dependent phytotoxic effect For the entire duration of the trial, parthenium on all 3 grass species tested, and the dose-response plants growing on P. maximum plots were shorter curves (not presented) had different slopes. and produced less seed than parthenium plants ED50 values calculated from dose-response curves growing with E. curvula and D. eriantha (data not for germination percentage and radicle length presented). Furthermore, a large number of parthe- showed that P. maximum was the most sensitive nium plants died in P. maximum plots, confi rming species, and E. curvula the least (Table 1).

 A





 %CURVULA 0MAXIMUM  $ERIANTHA  $RYMATTERPLANTG 

   0ARTHENIUMDENSITYPLANTSM

 B





 %CURVULA 0MAXIMUM  $ERIANTHA #ONTROL  $RYMATTERPLANTG 

   0ARTHENIUMDENSITYPLANTSM

Figure 3. Parthenium hysterophorus (parthenium) dry matter accumulation for: (a) the fi rst (2003–2004); and (b) the second (2004–2005) growth seasons as infl uenced by companion grasses. Interference between parthenium and grasses 93

Discussion In hindsight, the two weed densities used were relatively low, and therefore, interference effects The suppression of parthenium growth and even were actually studied at the lower or emerging mortality of parthenium seedlings on P. maximum end of the weed infestation scale. Large differ- plots, together with good seed production by the ences in total rainfall during the fi rst (506 mm) grass when co-existing with parthenium, indi- and second (275 mm) growth seasons were cates that this species has potential for use as an most likely responsible for the large differences antagonistic or rehabilitative species in a biolog- observed in parthenium dry mass accumula- ical weed control program. tion between the 2 growth seasons (Figures 3a, In the fi eld experiment, P. maximum domi- 3b). Differential rainfall and temperatures within nated in terms of overall performance measured and between seasons probably govern the large as dry matter accumulation as well as suppres- swings in parthenium levels that were observed in sion of parthenium growth. Although D. eriantha the general area of the trial site, during the experi- was selected because of the widespread occur- mental period and after termination of the trial. rence of this species in the area, and E. curvula Buckley et al. (2004) assert that successful because it is the most widely used sown pasture control of invasive plants requires changing of species in South Africa (Fair 1986) and has been disturbance regimes or the succession trajectory used successfully in mine rehabilitation projects, of the community by creating favourable con- both species showed poor adaptation to the ditions for native competitors and unfavourable local environmental conditions. The mean pH opportunities for weed regeneration. We contend of the trial site in March 2004 was 7.7 (H2O), it is imperative that antagonistic species should be and a more acidic pH preference for E. curvula selected according to environmental compatibility and D. eriantha (Anon. 2004) may have been a in addition to their interference potential with primary factor in the poor growth rate of these the particular invader species. As P. maximum species. P. maximum is more suited to alkaline is a highly palatable grass, a high grazing pres- (Anon. 2004), and its high growth rate may sure on the grass can be expected under natural be attributed to better adaptation to the environ- conditions, which may impact negatively on its mental conditions at the trial site, especially soil ability to interfere successfully with parthenium. pH and soil texture. High temperatures and other P. maximum is considered to be intolerant of environmental factors at Skukuza may also have intensive, frequent grazing, while to the best of infl uenced grass performance. Although an irri- our knowledge, parthenium is not utilised by any gation system was utilised over the 2 growing herbivores in South Africa. Therefore, results of seasons, P. maximum is known to tolerate a wider the present fi eld study might have been different range of moisture regimes than the other 2 grass if the area had been grazed during the study and species (Anon. undated). Since P. maximum per- not left ungrazed. Similarly, results might have formed relatively better than the other 2 grass been different if the grasses had been established species, it is possible that, on the E. curvula and from seed sown in situ. In such a situation, even D. eriantha plots, intra-species (parthenium- the highly interfering grass P. maximum may parthenium) interference dominated, while on experience diffi culty in establishing, and could the P. maximum plots inter-species (P. maximum- even fail totally. Considering the high competi- parthenium) interference was the dominant tiveness of P. maximum under these conditions, factor. further investigation into feasible methods of

Table 1. Phytotoxicity of parthenin on 3 grass species in germination bioassays.

μ 1 Species ED50 ( g/ml)

Radicle length Germination

E. curvula 212.9a2 345.9a D. eriantha 144.7b 184.2b P. maximum 100.6c 96.1c

1 Within parameters, means followed by different letters differ signifi cantly (P = 0.05). 2 ED 50 = Effective dose causing 50% inhibition. 94 M. van der Laan, C.F. Reinhardt, R.G. Belz, W.F. Truter, L.C. Foxcroft and K. Hurle grass establishment from seed under such condi- of parthenin by parthenium into the environment tions may be warranted. When practical, however, may confound attempts to use naturally occur- hand planting of grass seedlings can be utilised to ring species such as the grass P. maximum for guarantee improved establishment in an intensive biological control of the weed. Further research rehabilitation program. is required to progress our understanding of the Allelochemicals from parthenium have been interference mechanisms between parthenium observed to inhibit germination and to stunt seed- and the 3 grasses studied, other grasses, and other ling growth of a wide variety of species (Mersie plant species threatened by this noxious weed. and Singh 1987, 1988; Belz et al. 2007). This aspect must be considered and further investigated in the selection of an antagonistic (interfering) Acknowledgements species for the biological control of parthe- nium. Pure parthenin had a phytotoxic effect on We acknowledge the National Research Founda- all 3 grass species tested under laboratory con- tion for funding this research. ditions. The greater inhibitory effect on radicle growth than on germination, observed in this experiment, has also been reported for a variety References of species by Batish et al. (1997), Reinhardt et al. (2004) and Belz et al. (2007). Parthenin may, ANON. (undated) Agricol Product Guide. (Agricol, Bracken- fell: South Africa). therefore, be considered a rather weak germina- ANON. (2004) Kynoch Pasture Handbook. 1st Edn. (Kejafa tion inhibitor (Belz et al. 2007), but it may play Knowledge Works: Maanhaarrand). a larger role in delaying germination (Kohli et al. BATISH, D.R., KOHLI, R.K., SAXENA, D.B. and SINGH, H.P. (1997) Growth regulatory response of parthenin and its 1996), and hence, could impede seedling estab- derivatives. Plant Growth Regulation, 21, 189–194. lishment. BELZ, R.G., REINHARDT, C.F., FOXCROFT, L.C. and HURLE, K. For E. curvula, Belz et al. (2007) observed ED (2007) Residue allelopathy in Parthenium hysterophorus L. 50 — does parthenin have a leading role? Crop Protection, 26, values for germination percentage and radicle 237–245. length of 491 and 167 µg/ml, respectively. Dif- BUCKLEY, Y.M., REES, M., PAYNTER, Q. and LONSDALES, M. (2004) Modelling integrated weed management of an inva- ferences between those ED50 values and data for E. curvula from the present experiment (Table 1) sive shrub in tropical Australia. Journal of Applied Ecology, 41, 547–560. may be attributed to different experimental con- CALLAWAY, R.M., NADKARNI, N.M. and MAHALL, B.E. (1991) ditions. Belz et al. (2007) reported a signifi cant Facilitating and interfering effects of Quercus douglasii in hormesis effect for E. curvula, with growth stim- central California. Ecology, 72, 1484–1499. CHAPIN, F.S. III, WALKER, L.R., FASTIE, C.L. and SHARMAN, ulation at low parthenin concentrations, and L.C. (1994) Mechanisms of primary succession following inhibition at higher doses. E. curvula also dis- deglaciation at Glacier Bay, Alaska. Ecological Monograph, played radicle growth stimulation in the current 64, 149–175. FAIR, J. (1986) Guide to profi table pastures. (M and J Publica- experiment, but this was not tested for signifi - tions: Harrismith, South Africa). cance. Belz et al. (2007) further observed that FUERST, E.P. and PUTNAM, A.R. (1983) Separating the compet- E. curvula was more sensitive to parthenin than itive and allelopathic components of interference: theoretical Journal of Chemical Ecology 9 the other monocot species tested, namely: Era- principles. , , 937–944. HASELER, W.H. (1976) Parthenium hysterophorus L. in Aus- grostis tef and Echinochloa crus-galli. tralia. PANS, 22, 515–517. Our preliminary investigations into possible HEDGE, B.A. and PATIL, T.M. (1982) Effect of salt stress on the structure and carbon fl ow mechanism in a noxious weed allelopathic effects of the grasses on parthe- Parthenium hysterophorus L. Weed Research, 22, 51–56. nium revealed that root leachates from all 3 grass KOHLI, R.K., RANI, H.P., SINGH, H.P. and KUMAR, S. (1996) species had no phytotoxic effect on parthenium Response of crop seeds towards the leaf leachates of germination or its early development. However, Parthenium hysterophorus L. Indian Journal of Weed Science, 28, 104–106. further investigation is warranted. MERSIE, W. and SINGH, M. (1987) Allelopathic effects of Based on the fi ndings for parthenin production parthenium (Parthenium hysterophorus L.) extract and dynamics in P. hysterophorus leaves (Reinhardt residue on some agronomic crops and weeds. Journal of Chemical Ecology, 13, 1739–1747. et al. 2006; Belz et al. 2007) and on the phyto- MERSIE, W. and SINGH, M. (1988) Effects of phenolic acids and toxic effect of parthenin, it is plausible that this ragweed parthenium (Parthenium hysterophorus) extracts on allelochemical may inhibit or impede the nat- tomato (Lycopersicon esculentum) growth and nutrient and chlorophyll content. Weed Science, 36, 278–281. ural recruitment of various indigenous plant NATH, R. (1988) Parthenium hysterophorus L., a review. species. Furthermore, the production and release Agricultural Reviews, 9, 171–179. Interference between parthenium and grasses 95

NAVIE, S.C., MCFADYEN, R.E., PANETTA, F.D. and ADKINS, S.W. Journal of Plant Diseases and Protection, Special Issue, 20, (1996) The Biology of Australian Weeds 27. Parthenium 427–433. hysterophorus L. Plant Protection Quarterly, 11, 76–88. RIDENOUR, W.M. and CALLAWAY, R.M. (2001) The relative PANDEY, H.N. and DUBEY, S.K. (1989) Growth and popula- importance of allelopathy in interference: the effects of Parthenium hysterophorus tion dynamics of an exotic weed an invasive weed on a native bunchgrass. Oecologia, 126, Linn. Proceedings of the Indian Academy of Science (Plant Sciences), 99, 51–58. 444–450. TRATHIE OOD AN OOI PARSONS, W.T. and CUTHBERTSON, E.G. (1992) Noxious Weeds S , L.W., W , A.R., V R , C. and of Australia. (Inkata Press: Melbourne). M CCONNACHIE, A.J. (2006) Parthenium hysterophorus PATIL, T.M. and HEDGE, B.A. (1988) Isolation and purification (Asteraceae) in southern Africa, and initiation of biological of a sesquiterpene lactone from the leaves of Parthenium control against it in South Africa. Proceedings of the 2nd hysterophorus L. – its allelopathic and cytotoxic effects. International Conference on Parthenium Management, Current Science, 57, 1178–1181. 5–7 December 2005, Bangalore, India. pp.127–133. EINHARDT RAUS ALKER OXCROFT OBBERTSE R , C., K , S., W , F., F , L., R , STREIBIG, J.C. (1988) Herbicide bioassay. Weed Research, 28, P. and HURLE, K. (2004) The allelochemical parthenin is 479–484. sequestered at high level in capitate-sessile trichomes on leaf surfaces of Parthenium hysterophorus. Journal of Plant WELDEN, C.W. and SLAUSON, W.L. (1986) The intensity of Diseases and Protection, Special Issue, 19, 253–261. competition versus its importance: an overlooked distinc- REINHARDT, C., VAN DER LAAN, M., BELZ, R.G., HURLE, K. tion and some implications. Critical Reviews in Botany, 61, and FOXCROFT, L. (2006) Production of the allelochemical 23–44. parthenin in the leaves of Parthenium hysterophorus L.

(Received for publication October 5, 2006; accepted October 10, 2007)