Control measures for the encroacher shrub Seriphium plumosum

Abstract

Seriphium plumosum encroachment in South Africa has converted extensive areas of grassland into less productive shrubland–grassland, but its control is not being seriously addressed at present. Therefore, the short-term response of S. plumosum to different applications of nitrogen (N), phosphate (P), lime, sodium chloride (NaCl) and a soil-applied suspension herbicide, Molopo (active ingredient tebuthiuron), was examined. The lime and P-fertiliser treatments lead to no deaths of S. plumosum for any of the concentrations. The smaller the shrubs, the more sensitive they were to both N and NaCl applications. The minimum N-fertiliser applications of 30, 60, 120, 1 000 and 2 000 kg ha⁻¹ were responsible for 100% death of shrubs with heights of 200, 300, 400, 500 and 600 mm, respectively. Sodium chloride application of only 100 kg ha⁻¹ lead to 100% death of shrubs smaller than 400 mm high. Although high applications of NaCl led to total death of shrubs 600 mm high, the enormous problem of saline/sodic soil accompanying it must not be disregarded. Molopo successfully killed all shrubs up to a height of 600 mm. As the plant reaches maturity, the root:shoot ratio increases significantly. It was proved that S. plumosum encroachment is not linked to overgrazing. These results confirm the vulnerability of S. plumosum in semiarid areas, following changes in soil characteristics, which can be used in the control of this invasive plant. A combination of methods is recommended for S. plumosum control.

Keywords:

• bankrupt bush
• Molopo
• Seriphium plumosum (Stoebe vulgaris)
• shrub encroachment
• slangbos
• vaalbos

Introduction

Although Seriphium plumosum (also known as bankrupt bush, slangbos, vaalbos or Khoi-kooigoed in the Western Cape) is indigenous to South Africa, it has naturalised in other countries in Africa (Angola, Namibia, Mozambique and Zimbabwe; Koekemoer 2001) and is also found in Madagascar and the USA (Schmidt et al. 2002, Badenhorst 2009). Seriphium comprises a total of nine species of which five occur in South Africa and the rest in East Africa, Madagascar and Reunion (Koekemoer 2001). Seriphium plumosum, previously known as Stoebe vulgaris, is currently viewed as an aggressive encroacher species in large parts of the Fynbos and Grassland Biomes of South Africa (du Toit and Sekwadi 2012), such as districts of the Eastern Cape, Free State, Mpumalanga, North West and Gauteng provinces (Krupko and Davidson 1961, Snyman 2009a).

Previous authors have alleged that overgrazing, mismanagement (Roberts 1966, Roux 1969, Wepener 2007) or lack of controlled burning (Trollope 1987) lead to the appearance of S. plumosum, but actually these factors only contribute to the encroachment (Davidson 1962). There are numerous reports of the species’ rapid spread on a farm, even in grassland in good condition (Hattingh 1953, Smit 1955, Wepener 2007). Seriphium plumosum requires summer rainfall of approximately 620–750 mm (Snyman 2009a), which reflects the precipitation boundary between mesic and semiarid grassland (Snyman 2012a). Especially during the last 10 years, encroachment by this unpalatable shrub, which grows to a height of approximately 0.6 m, has assumed a previously inconceivable severity (Snyman 2009a), to such an extent that the CARA legislation (Regulation 16 of the Conservation of Agriculture Resources Act 43) listed it as a proclaimed encroacher plant (Jordaan and Jordaan 2007). It enormously reduces the grazing capacity of rangelands (Moore and van Niekerk 1987, Wepener et al. 2008, Snyman 2012a), because its rapid spread displaces grasses (Story 1952, Roux 1969). A common view is that S. plumosum evolved from Stoebe cinerea by mutations, which changed its character and enabled it to invade grassveld (Roux 1969). This must have happened a long time ago, and it may have been an uncommon plant until recent times, when changing environmental or ecological conditions enabled it to spread rapidly.

The control of S. plumosum is not being seriously addressed at present, partly owing to widely diverging opinions regarding its control and eradication (Jordaan 2009). Unfortunately, existing inexpensive control measures are untried or have not yet withstood the test of time. Therefore, there is an urgent need for an inexpensive, but effective, remedy or technique to control this encroacher plant. Good follow-up action is also important to control those seedlings that emerge later. It is currently impossible to single out a best method of controlling S. plumosum as a variety of factors influence its practical application in different districts, for example topography, accessibility and negotiability of the terrain (Jordaan 2009). The effectiveness of S. plumosum control is also determined by factors such as time of year, clay content of the soil, rainfall and the distribution thereof, and stand density of the shrubs. The control measures chosen must be economically, financially and ecologically justifiable. Unfortunately, no biological control measures are currently known (Snyman 2012b). There are many opinions on mechanical control (chopping action), which is labour intensive and does not show lasting success. The plant will definitely coppice if the stem is not cut underneath the soil surface (Snyman 2010a). A follow-up or post-treatment is also necessary to control those seedlings emerging after the removal of the mother plant (Snyman 2009b). As many additional seeds are dispersed during the chopping-out process, the chopped shrubs must be removed and burnt or the problem can intensify. No scientifically based fire control measures are known (Roux 1969, Snyman 2011), except those applied in the 1940s and 1950s (Hattingh 1953, Smit 1955, Lecatsas 1962). The wrong type of burning can also increase the problem (Snyman 2009aSnyman 2011). Seriphium plumosum can be very successfully controlled chemically (by a granular formation or suspension) with agents even having a residual effect of a few years to control those seedlings that may emerge subsequently (Richter 1989, du Toit and Sekwadi 2012). This is understandably an expensive process when addressed correctly (Snyman 2009b). Thousands of rands are spent annually in South Africa on its chemical control (Snyman 2012b).

This study aimed at quantifying an inexpensive, but ecologically justifiable, remedy that could be applied to different-sized S. plumosum shrubs. It was hypothesised that by changing soil fertility, S. plumosum shrubs will respond negatively in terms of growth to such an extent that death results. The purpose of this study was therefore to modify the plant's edaphic environment in such a way that it can lead to the plant's death.

Material and methods

Study area

The field research was conducted from October 2008 to March 2009 on the farm Eden, close to the small town of Thaba Nchu (29°12′ S, 26°50′ E; altitude 1 450 m), about 80 km east of Bloemfontein in the semiarid region of South Africa. Rain falls almost exclusively during summer (October to April), with a long-term annual average of 630 mm and a mean of 66 rainy days per year (Schulze 1979). Mean monthly maximum temperatures range from 17 °C in July to 33 °C in January, with a mean of 131 frost-days annually (Schulze 1979). Frost occurs from the end of April to the beginning of October. Summer temperatures are moderate with very cold winters, with absolute minimum and maximum temperatures varying between −11 °C and 38 °C.

The study area is situated in the Eastern Free State Sandy Grassland (vegetation type GM4) described by Mucina and Rutherford (2006). The grassland was in good condition and consisted of dense grassland (Snyman 2010b). Grassland condition was determined according to the degradation gradient technique of van der Westhuizen et al. (1999). Dominant perennial species included Cymbopogon pospischilii, Themeda triandra, Digitaria eriantha and Elionurus muticus. With rangeland degradation because of overgrazing the perennial grass cover has diminished, with Eragrostis chloromelas and forage-poor perennials such as Microchloa caffra and Artistida species more abundant. Although the stocking rate was in accordance with the recommended long-term grazing capacity of this veld type (5 ha LSU⁻¹), severe S. plumosum encroachment has taken place in this area, especially over the last 10 years (Snyman 2009a).

Soils in the study area are of the Estcourt soil form (1 200-Nuweplaas family), depending on the topography (Soil Classification Working Group 1991), with a clay content of 14% in the A-horizon. The pH (KCl), carbon:nitrogen ratio, phosphorus, calcium, magnesium, potassium and nitrogen concentrations were 4.80, 3.69, 3.98, 1 102, 208, 1 180 and 1 140 mg kg⁻¹, respectively. The effective depth of the soil is 550 mm, with the A-horizon 30 mm and E-horizon 250 mm.

Fertiliser and herbicide treatments in the field

The fertilisation and herbicide research in the field was conducted on four plots of 10 m×10 m each, which were randomly set out on the foot slope terrain morphological unit (Soil Classification Working Group 1991). The plots were laid out across the slope on an area of 50 m×100 m. The fertiliser treatments included nitrogen (limestone ammonia nitrate [LAN]; 28% N), phosphate (superphosphate; 83 g kg⁻¹ P), lime (dolomitic agricultural lime; calcium 160 g kg⁻¹, magnesium 120 g kg^-1, calcium carbonate equivalent acid 88%, resin 78%) and sodium chloride (sodium 375 g kg⁻¹ minimum and sodium chloride 850 g kg⁻¹ minimum).

The quantities of each fertiliser applied are summarised in Table 1. The different nitrogen (N) and phosphorus (P) treatments were applied separately, but also in combination, whereas sodium chloride (NaCl) and lime were only applied separately. The fertiliser concentrations were calculated for an application area of 0.5 m×0.5 m around a shrub. For shrubs taller than 500 mm and more than 300 mm in diameter, the application area was 0.5 m×?0.5 m surrounding the shrub, whereas for shrubs shorter than 500 mm and smaller than 300 mm diameter, it was 0.2 m×0.2 m around the shrub. The reason for this is that the larger the shrubs, the more widely their root systems spread from the main stem (Cohen 1940) and the more effective fertilisers should be assimilated. Only at the two largest shrub sizes was fertiliser worked in to a depth of about 10 mm with a garden fork, whereas for the other shrub sizes fertiliser was only applied on the soil surface.

Table 1 : Application rates (kg ha⁻¹) of the different fertiliser (kg ha⁻¹ or g fertiliser shrub⁻¹) and soil-applied suspension herbicide (Molopo – active ingredient tebuthiuron) (ml shrub⁻¹) treatments. Fertiliser was applied as calculated for an application area of 0.5 m×0.5 m around the stem of a Seriphium plumosum shrub. LAN = Limestone ammonia nitrate

Lime		Nitrogen		Sodium chloride		Phosphorus		Molopo
(kg lime ha^−1)	(g lime shrub ^−1 )	(kg N ha^−1)	(g LAN shrub^−1)	(kg NaCl ha^−1)	(g NaCI shrub^−1)	(kg P ha^−1	(g super-phosphate shrub^−1)	(ml shrub^−1)
1 000	25	30	2.7	100	2.5	5	1.20	0.5
2 000	50	60	5.4	200	5.0	10	2.50	1.0
5 000	125	120	10.7	300	7.5	15	3.75	2.0
		200	17.9	600	15			4.0
		500	44.6	1 000	25
		1 000	89.2	2 000	50
		2 000	178.4	4 000	100

A soil-applied suspension herbicide, Molopo (5005C, regulation no. L5854, act 36/1947, active ingredient tebuthiuron [a urea compound], 500 g l⁻¹) was used. This was applied to the soil as close as possible to the main stem of each shrub by hand, with the application quantities presented in Table 1.

The fertiliser and herbicide treatments (Table 1) were applied to the following shrub sizes, namely 600, 500, 400, 300 and 200 mm high with respective diameters of 500, 300, 200, 100 and 50 mm. The treatments allocated randomly to the shrubs within a plot were applied on 20 shrubs for each size class per plot. The different treated shrubs were marked with cattle cartages fastened to the shrubs with cable ties. The treatments were applied on 15 October 2008. A total of 478 mm rain fell from the beginning of October 2008 to the end of March 2009.

The soil on which the NaCl and N treatments were applied was analysed at the start of the research and at the end of the trial (March 2009) to quantify a possible salinity/ sodicity effect and an increase in fertility that could have taken place with application. All analyses regarding physical and chemical characteristics were conducted in accordance with standard laboratory techniques outlined by the Non-Affiliated Soil Analysis Work Committee (1990). The sodium adsorption ratio (SAR) was used to determine the sodium to calcium plus magnesium ratio of the soil for the classification of the salinity/sodicity effect (van der Merwe et al. 1975). The experimental layout for the NaCl ratio and N content analysis in the soil was a fully randomised design, with 10 replications for each treatment.

The death percentage of the different sizes of S. plumosum plants was monitored in the different treatments over a six-month period (October 2008 to March 2009). A one-way analysis of variance (ANOVA) for each of the 15, 35, 35, 15 and 20 treatment combinations for lime, N, NaCl, P and Molopo, respectively, was performed to determine any significant differences within a treatment combination (Winer 1974). There were for replications for each treatment.

The mean±SE density of mature S. plumosum plants where the observations were made was 1 424±11 shrubs ha⁻¹.

Root:shoot ratio study in the greenhouse

A root:shoot study was conducted in the greenhouse from March 2008 to March 2009, with respective day and night temperatures of 32 °C (±2 °C) and 18 °C (±2 °C). The ratio was determined from S. plumosum plants (1–4 months and one-year old) established in asbestos pots (210 mm diameter and 550 mm deep) from seed. The pots were filled with the same soil (the upper 200 mm layer) on which the field trial was conducted. Roots were extracted via successive washings of soil through a 2 mm mesh sieve. The remainder of the soil was spread in a shallow tray and fine roots were collected by flotation. The overflow from the tray passed through a 0.5 mm mesh sieve. All phytomass (above and below-ground) was oven-dried at 80 °C for 72 h before weighing. The experimental layout was a fully randomised design with 10 replications for each treatment.

Data analysis

All data collected for soil analysis, shrub deaths and root:shoot ratios were normally distributed for use in the ANOVA. The fertiliser treatments that included different application concentrations were statistically analysed individually. Data were analysed using SAS (DOS program, version 6.04; SAS Institute 2001) and the Number Cruncher Statistical System (2000) software package (Hintze 1997). Significance between treatments was determined using Tukey's test (Mendenhall and Sincich 1996). Descriptive statistics such as means, standard deviations and percentages were employed for the remainder of the data.

Results and discussion

Fertiliser and chemical herbicide

One day after the treatments were applied, 30 mm of rain fell and subsequently 478 mm fell from October 2008 to end of March 2009 when the trial ended. The pure applications of lime and P treatments caused no deaths of S. plumosum plants for any of the concentrations (data not shown). Snyman (2002) also found on semiarid grassland that grasses had no reaction to P application. The fact that these fertilisers have difficulty in leaching through the soil may perhaps also contribute to poor absorption by the plants. If they are worked into the soil, the reaction of the plants might be different.

The smaller the shrubs, the more sensitive they were towards both N and NaCl applications (Table 2). For example, 100% death occurred in plants smaller than 200 mm for all of the N-fertiliser applications. These complete deaths occurred within two weeks after application. Minimum N-fertiliser applications of 30, 60, 120, 1 000 and 2 000 kg ha⁻¹ were responsible for 100% death (F_34,105=5.38, P<0.01) of shrubs with heights of 200, 300, 400, 500 and 600 mm, respectively. The cost of the different treatments is presented in Table 2. The treatment cost increased considerably with the increase in fertiliser application rate from 2 to 96 cents per shrub. The cost for N was calculated at R269.41 for 50 kg LAN and that for 50 kg NaCl was R41.74. A large full-grown S. plumosum shrub clearly needs a high concentration of N fertiliser for total die-back, which may not be economically justifiable. This is supported by Lecatsas (1962), who found that heavy applications of N fertiliser (up to 2 000 kg ha⁻¹) to mature S. plumosum plants on rangeland caused die-back within three weeks. These effects closely resemble those produced by heavy grazing where N is supplied by dung and urine excreted by cattle. The smaller S. plumosum shrubs can, according to this study, be relatively economically controlled with N fertiliser. The growth of S. plumosum seedlings was also investigated by Lecatsas (1962) in hydroponic culture with different N concentrations; S. plumosum growth was depressed at the higher concentrations, which in the same experiment markedly stimulated the growth of Eragrostis curvula.

Table 2 : Seriphium plumosum death (%) in response to different fertilisers and a soil-applied suspention herbicide (Molopo: active ingredient tebuthiuron), as grouped in different height classes. Cost per shrub is also indicated for each treatment. Means (n = 4) within a plant height class with different superscript letters indicate significant (P ≤ 0.01) differences among treatments based on Tukey's test

	Height size (mm)
Concentration	600	500	400	300	200	Cost shrub^−1 (cents)
	N (kg ha ^−1 )
30	0	0	52^c	77^b	100	2
60	0	0	73^b	100^a	100	3
120	0	22d	100^a	100^a	100	6
200	0	48c	100^a	100^a	100	10
500	0	71b	100^a	100^a	100	24
1 000	44^b	100^a	100^a	100^a	100	48
2 000	100^a	100^a	100^a	100^a	100^a	96
	NaCI (kg ha ^−1 )
100	0	0	100	100	100	0.2
200	0	0	100	100	100	0.4
300	0	0	100	100	100	0.6
600	0	48c	100	100	100	1
1 000	52^c	73^b	100	100	100	2
2 000	71^b	100^a	100	100	100	4
4 000	100^a	100^a	100	100	100	8
	Molopo (ml shrub ^−1 )
0.5	0	0	89	100	100	1.5
1	47^c	71^b	92	100	100	3
2	82^b	100^a	100	100	100	6
4	100^a	100^a	100	100	100	12

The fertilised shrubs that did not die in response to the N application also did not visibly grow more lushly than the unfertilised plants. In contrast, Cohen (1935) found that a heavy application of mixed fertiliser containing ammonium sulphate, phosphate and potash could cause S. plumosum to be ‘vigorously vegetative’ and prevented it from flowering. No significant (F _4,45=2.39, P > 0.05) differences in the N levels of all the fertilised and unfertilised soils occurred after six months of application. The soil N content of the unfertilised control was 0.111% versus 0.139% (F _4,45=4.15, P < 0.01) for the highest N fertiliser treatment after a six-month period. The applied N must have either leached from the soil or was taken up by the surrounding grasses. However, even the highest N fertiliser application rate had no transfer effect on the surrounding plants.

Combined application of N and P induced exactly the same results as where only N was applied (data not shown), which again emphasised the ineffectiveness of the P fertiliser alone. In contrast, most researchers agree that the response to phosphate has usually been observed only when applied in conjunction with N (Tainton et al. 2000, Snyman 2002). Other researchers found that P accumulation in degraded rangeland of certain vegetation types is one of the dominant soil indicators with respect to vegetation change (Jordaan 1997, van der Westhuizen 2003, van der Westhuizen 2012). The fact that P is static and does not leach from the soil indicates that it could have a major influence on the process of rangeland improvement.

Sodium cloride caused 100% death (F _34,105=4.22, P < 0.01) of shrubs 400 mm high and smaller at an application of only 100 kg NaCl ha^-1, at a cost of only 0.2 cent per shrub (Table 2). The 500 mm height class suffered 100% death at a minimum of 2 000 kg NaCl ha⁻¹. Though 4 000 kg NaCl ha⁻¹ application led to total death (F _34,105=6.16, P < 0.01) of shrubs 600 mm high, the enormous problem of saline/sodic soil and the cost accompanying it must not be disregarded. The cost to kill a shrub 600 mm high at such an application rate is 8 cents per shrub, which is little less than the use of Molopo. The latter chemical control agent also does not contribute towards a salinity/sodicity problem.

As expected, the SAR increased (F _7,72=6.66, P < 0.01) with an increase in Na application (Table 3). Only the 100 to 300 kg NaCl ha⁻¹ applications did not lead to salinity/ sodicity conditions, with an allowable SAR level of less than 5 for monoculture crops (van der Merwe et al. 1975). In contrast, a relationship between rangeland condition and SAR was found by van der Westhuizen (2003) along a degradation gradient (R ²=0.92) for rangelands of this vegetation type. According to this relation the highest SAR level for rangelands in a very poor condition was 3.1 and less than 1 for rangeland in moderate and good condition. It is therefore very important to monitor the SAR level of the 200 and 300 kg NaCl ha⁻¹ before these treatments could be recommended. Even the 100 kg treatment, where the SAR levels increased to 1 in relation to 0.17 for the control (Table 3), will have a negative influence on ecosystem functioning if it does not return to normal SAR levels over the short term. According to van der Westhuizen (2003, van der Westhuizen (2012) the loss in ecosystem functioning is about 35% in terms of grazing capacity if SAR levels increased to 1. It is also important to note that no significant differences in S. plumosum deaths occurred between the 100, 200 and 300 kg ha⁻¹ NaCl treatments (Table 2), but that the SAR levels differed significantly between the 100 and 200 kg ha⁻¹ treatments (Table 3).

Table 3 : Mean ± SE sodium absorption ratio (SAR) in response to different NaCl applications, as measured on the foot slope at the end of the trial (March 2009). Means (n = 10) with different superscript letters indicate significant (P < 0.01) differences based on Tukey's test

NaCl ratio
Control (no NaCl)	100	200	300	600	1 000	2 000	4 000
0.17±0.03^g	1.00±0.20^f	3.41±0.11^e	4.82±0.09^e	8.55±0.14^d	15.58±0.22^c	21.00±3.61^b	36.13±4.66^a

High concentrations of sodium in relation to calcium plus magnesium result in a dominantly sodium soil, with unfavourable soil structure, as the calcium and magnesium on the clay particles are replaced by sodium. This soil usually swells shut when wet and can let water through only slowly, resulting in waterlogged conditions. Further salinity/ sodicity conditions occur in the upper soil profile, whereas the soil deeper down remains dry. Hard crusts form, which deter germination and slow down root growth (van der Merwe et al. 1975). In a follow-up study it is important that the impact of increased soil salinity/sodicity is quantified over time, and whether the salinity/sodicity conditions can be lifted after a single high-rainfall event or only over a few years of rainfall, for example. This finding will determine whether or not a high salt application rate will influence the ecosystem over the short or long term. The advantage of NaCl application only around the shrub (0.5 m×0.5 m), rather than broadcast application, is that the Na concentration is considerably diluted with every rain shower and therefore decreases the salinity/sodicity conditions per area, but does not influence the functioning of the ecosystem.

Broadcast application of N fertiliser and NaCl is therefore not recommended because various researchers found a change in botanical composition of the grasses with the fertilisation of grassland (Cilliers et al. 1997, Tainton et al. 2000, Snyman 2002, Zeng et al. 2010). By only applying the fertiliser around the stems of shrubs, any influence on the vegetation composition should be avoided. The NaCl can be applied in granular formulation or in suspension by hand. Sodium chloride easily dissolves in water and can be applied, for example, with a dosage gun to individual shrubs. A long-term practice may be to place animal licks in dense S. plumosum stands and in time move them to other dense stands in the camp. Dung excretions may gradually increase the Na level in the soil over a period rendering the soil unfavourable for S. plumosum survival. This is a very long-term aim, but worth considering for the future control of S. plumosum.

The herbicide tebuthiuron (Molopo) successfully killed all shrubs to a height of 600 mm high (F _19,60=8.15, P < 0.01) at an application rate of 4 ml shrub⁻¹. Shrubs smaller than 400 mm high were successfully (F _19,60=7.12, P < 0.01) controlled at a dosage of 2 ml shrub⁻¹. Though an expensive practice, it has the advantage of a residual effect of about three years, which prohibits seedlings from emerging in those areas. Herbicides are most effective, although expensive, and should form the basis of all control measures.

A follow-up action, following the application of a chemical control measure, can also be very effectively and inexpensively implemented with NaCl to control seedling emergence. The average lifespan of individual plants is about 15 years (Roux 1969). Within this period the mother plants should be eradicated to allow the rangeland to produce sustainably.

Root:shoot ratio

As the plant reached maturity, the root:shoot ratio increased significantly (F _4,45=4.17, P < 0.01) in relation to the aboveground phytomass production (Table 4). Unfortunately, the ratio was not monitored for mature plants, but should still increase to a point that the root biomass of the large shrubs should be more than its aboveground phytomass.

Table 4 : Mean ± SE root:shoot ratio for different age Seriphium plumosum plants as measured in the greenhouse. Means (n = 10) with different superscript letters indicate significant (P < 0.01) differences based on Tukey's test

Plant age (months)
1	2	3	4	12
0.11±0.02^e	0.18±0.03^d	0.23 ±0.04^c	0.29±0.04^b	0.66±0.08^a

The roots of shrubs more than 400 mm high and 200 mm in diameter showed a substantial increase in distribution (data not shown). Therefore it was difficult to dig out a plant of this size rapidly with a spade. The root system of shrubs smaller than 400 mm high was not well developed yet and such plants could easily be removed with a spade or even pulled out by hand. The above may be one of the most important reasons why larger shrubs only die-back at a higher concentration of N fertiliser and NaCl. The widely distributed root system enabled the large shrubs to survive and take up nutrients far away from the stem and was therefore not affected by the high concentrations of N and NaCl close to the stem. There is little information on the root distribution generally found in S. plumosum. A root bisect by Gilman (1934) showed Seriphium roots going down 1.5 m in a clay soil with a maximum water-retaining capacity of about 65%. The plant is usually deep-rooted and also produces a large number of thick lateral roots (Gillman 1934). The rootstock, from which branches radiate upwards and outwards, bears a number of dormant buds that are also situated on the base of the stems. From the rootstock a thick, steadily tapering taproot descends vertically. The root system of full-grown plants is extensive with a moderate amount of superficial roots. In the present study many fine rootlets were attached to the stocks and found close to the soil surface.

Conclusions

A better understanding of certain aspects of encroachment by S. plumosum linked with soil fertility was obtained from this study. The death of S. plumosum plants following N and NaCl applications confirms their avoidance of the more fertile (high organic matter) and more saline/sodic soils that characterise moist vlei areas (Snyman 2009a). The fact that the SAR levels increase in this vegetation type as rangeland condition decreases (van der Westhuizen 2003) indicates that S. plumosum encroachment is not linked to overgrazing and will rather increase in rangeland with a moderate to good condition, than rangeland in poor to very poor condition. The fact that marginal soils, withdrawn from cash-crop cultivation, are among the most actively encroached areas might be ascribed to the lower fertility (organic matter, N and C contents) of the soil forming a more favourable habitat for S. plumosum. Additional in-depth habitat data is required to form a better understanding of the population dynamics of S. plumosum so that the species can be successfully controlled and not allowed to further decrease the production potential of grasslands.

Seriphium plumosum responds negatively to N and NaCl in terms of growth, and positively to the herbicide tebuthiuron (Molopo) to such an extent that an application concentration of only 4 ml shrub^-1 killed the plants in all cases, which supports the posted hypothesis as stated in the Introduction. No reaction was found with P or lime, or with a combination of N and P.

It is recommended, from effectiveness, economic and ecological viewpoints, to control S. plumosum encroachment by large plants (taller than 400 mm high) with Molopo or any other product with the same active ingredient, and smaller shrubs with NaCl. The latter is a very inexpensive chemical that can serve very effectively in the control of the smaller shrubs. Treatments higher than 300 kg NaCl ha⁻¹ must be avoided to minimise the negative effect of SAR incensement on the ecosystem. Nitrogen application also fares well to control the smaller shrubs, but is a very expensive practice. Therefore, a combination of methods is recommended. It is impossible to single out any one method of S. plumosum control as various factors influence its practical application in different districts, e.g. topography and access to the area. The effectiveness of S. plumosum control is determined by factors such as time of the year, the chemical product's selectivity, clay content of the soil, density of the stand of the shrub, and rainfall distribution.

Everything possible must be done to prevent S. plumosum encroachment, while the control measures chosen must be economically, financially and ecologically justifiable. Controlling problem plants is not a one-off process but rather a continual attempt at drastic control actions combined with good grassland management in order to establish a sustainable ecosystem. It is important to concentrate initially on those areas with sparse or moderate encroachment to prevent further spreading and subsequently treat denser stands.

Acknowledgements

The research was partially funded by the National Research Foundation (NRF), South Africa.