IDENTIFICATION OF KEY GRASS SPECIES IN VEGETATION UNIT Gm 11: Rand Highveld Grassland IN MPUMALANGA PROVINCE, SOUTH AFRICA

Winston S.W. Trollope & Lynne A. Trollope 22 River Road, Kenton On Sea, 6191, South Africa Cell; 082 200 33373 Email: [email protected]

INTRODUCTION

Veld condition refers to the condition of the vegetation in relation to some functional characteristic/s (Trollope, et.al., 1990) and in the case of both livestock production and wildlife management based on grassland vegetation, comprises the potential of the grass sward to produce forage for grazers and its resistance to soil erosion as influenced by the basal and aerial cover of the grass sward.

The following procedure was used for developing a key grass species technique for assessing veld condition in the vegetation type Gm 11: Rand Highveld Grassland (Mucina & Rutherford, 2006) in Mpumalanga Province – see Figure 1.

Rand Highveld Grassland: Gm11

Figure 1: The Rand Highveld Grassland Gm11 vegetation type located in the highveld area east of Pretoria in Mpumalanga Province (Mucina & Rutherford, 2006).

The identification of the key grass species is based on the procedure developed by Trollope (1990) viz..

Step 1: Identify and list all the grass species occurring in the study area noting their identification characteristics for use in the field.

Step 2: Based on careful observations in the field and consultations with local land users, subjectively classify all the known grass species in the area into Decreaser and Increaser species according to their reaction to a grazing gradient i.e. from high to low grazing intensities, as follows:

DECREASER SPECIES: Grass & herbaceous species that decrease when veld is under or over grazed; 2 INCREASER I SPECIES: Grass & herbaceous species that increase when veld is under grazed or selectively grazed; INCREASER II SPECIES: Grass & herbaceous species that increase when veld is over grazed.

Step 3: Based on experience, observation and local knowledge subjectively allocate forage and fuel factors on a scale of 0 - 10 according to the potential of the different grass species to produce forage for both domestic and wild ungulate bulk grazers and to produce grass fuel to support a prescribed burn. This procedure is used because it conforms to the concept of veld condition being the condition of the vegetation in relation to some functional characteristic and in this case the potential to produce grass forage for domestic livestock and wild ungulates and grass fuel to support a prescribed burn..

Step 4: Using the information from Step 3, develop a technique for assessing the condition of the grass sward based on the classification of the grass species into Decreaser and Increaser species together with their respective forage and fuel factors.

Step 5: Conduct grass surveys over the widest range of grassland, in as many different conditions possible, and with these data identify the key grass species that have the greatest effect on veld condition in terms of their potential to produce forage and grass fuel. The potential for producing forage and grass fuel is calculated for each grass survey by multiplying the respective forage and fuel factors with the percentage relative frequency for each of the herbaceous species recorded during the different grass surveys. The sum of these products is expressed as the forage and fuel scores for each sample site. The selection of the key grass species is done using a multiple regression analysis where the forage and fuel scores are the dependent variables and the percentage frequency for each recorded grass or herbaceous species are the independent variable. The precision of the resultant regression model is then tested in a goodness of fit graph illustrating the relationship between actual and predicted forage and fuel scores.

The reason for selecting the key grass species in terms of their forage and fuel production potential is because the area in the aforementioned vegetation unit is used for both livestock production and in recent times for wildlife management and it is therefore essential that the key grass species be selected to reflect the grazing potential of the veld and its potential to support prescribed burns in the area. The data used for identifying the key grass species was collected by Mr Francois de Wet from ENVIROPULSE, Middelburg, Mpumalanga Province.

RESULTS

Ecological Categories and Forage Factors

In accordance with Steps 1, 2 and 3 the following set of different grass species were recorded in the project area. These were classified into Decreaser and Increaser I and II species according to their perceived reaction to a grazing gradient from high to low grazing intensities based on personal experience, research and consultations with successful livestock farmers in South Africa over the last three decades. In addition Forage & Fuel Factors on a scale of 0 - 10 were allocated to the different grass species recorded in the project area using the expert system developed by Bosch & Gaugh (1991). A total of 88 different grass species were recorded reflecting the botanical composition of the grass sward over a wide range of veld conditions in the project area – see Table 1. 3

Table 1: The different grass species recorded in the vegetation units in Gm 11: Rand Highveld Grassland and classified into Decreaser and Increaser I and II species together with their respective Forage & Fuel Factors reflecting their forage and fuel production potential for bulk grazers (cattle, buffalo) and for prescribed burning on a scale of 0 – 10.

GRASS SPECIES ECOLOGICAL CATEGORY FORAGE FACTOR FUEL FACTOR Andropogon appendiculatus D 8 8 Andropogon chinensis D 4 4 Andropogon schirensis D 4 5 Bewsia biflora D 1 3 Brachiaria brizantha D 5 6 Brachiaria nigropedata D 7 8 Brachiaria serrata D 3 3 Digitaria brazzae D 4 5 D 6 6 Digitaria tricholaenoides D 4 5 Diheteropogon amplectens D 3 3 Eustachys paspaloides D 5 6 Lophacme digitata D 2 3 Monocymbium ceresiiforme D 2 3 Panicum maximum D 8 8 Setaria lindenbergiana D 5 8 Setaria nigrirostris D 7 8 Setaria sphacelata D 6 7 Themeda triandra D 8 8 Alloteropsis semialata I 2 7 Andropogon eucomus I 1 2 Aristida meridionalis I 1 5 Aristida transvaalensis I 1 4 nepalensis I 4 6 Ctenium concinnum I 1 5 Cymbopogon caesius I 2 6 Cymbopogon validus I 2 10 Digitaria setifolia I 1 2 Elionurus muticus I 2 7 Enneapogon pretoriensis I 2 6 Eragrostis curvula I 4 9 Eragrostis gummiflua I 2 6 Hyparrhenia anamesa I 4 10 Hyparrhenia dregeana I 3 10 Hyparrhenia filipendula I 4 10 Hyparrhenia hirta I 4 10 hexandra I 4 6 flavida I 3 6 Loudetia simplex I 2 6 4 Melinis nerviglumis I 1 3 Panicum natalense I 4 6 Paspalum notatum I 3 5 Pennisetum sphacelatum I 1 8 Rendlia altera I 1 3 Schizachyrium sanguineum I 1 8 Schizachyrium ursulus I 1 8 Sporobolus centrifugus I 1 3 Sporobolus pectinatus I 1 2 Trachypogon spicatus I 1 6 Triraphis andropogonoides I 2 6 biseriata I 1 7 Tristachya leucothrix I 6 7 Tristachya rehmannii I 1 10 Urelytrum agropyroides I 1 9 Aristida adscensionis II 1 2 Aristida congesta subsp. barbicollis II 1 2 Aristida congesta subsp. congesta II 1 2 Aristida diffusa II 1 2 Aristida vestita II 1 4 Chloris gayana II 5 8 dactylon II 3 3 II 2 2 Digitaria monodactyla II 1 3 Enneapogon cenchroides II 1 4 Eragrostis capensis II 2 3 Eragrostis chloromelas II 3 4 Eragrostis inamoena II 1 4 Eragrostis lehmanniana II 3 4 Eragrostis nindensis II 2 3 Eragrostis plana II 2 4 Eragrostis pseudosclerantha II 1 4 Eragrostis racemosa II 1 2 Eragrostis rigidior II 2 5 Eragrostis sclerantha II 2 3 Eragrostis viscosa II 2 3 Heteropogon contortus II 5 6 Melinis repens II 1 2 Microchloa caffra II 1 1 Paspalum scrobiculatum II 1 3 Paspalum urvillei II 2 8 Perotis patens II 1 1 Pogonarthria squarrosa II 1 3 Sporobolus africanus II 3 4 Sporobolus festivus II 1 2 Sporobolus ioclados II 1 2 5 Sporobolus stapfianus II 1 2 monachne II 2 3 Trichoneura grandiglumis II 1 2 Bare Ground II 0 0

Identification of Key Grass Species

In accordance with Step 5 multiple regression analyses were conducted to identify the Key Grass Species where the Forage & Fuel Scores were the dependent variables and the percentage frequency recorded for the different grass species in the different sample sites were the independent variables. Two multiple regression analyses were conducted using data from the Gm 11: Rand Highveld Grassland vegetation unit to identify the key grass species that had the greatest statistically significant effect on the grass forage and fuel potentials in this vegetation unit. Firstly a multiple regression analysis was conducted involving all the abundantly occurring grass species in the different sample sites as a means of identifying those species that had a statistically significant effect on the Forage and Fuel Scores. Arising from these results the multiple regression analysis was repeated using the aforementioned significant grass species resulting in the identification of the Key Grass Species for the aforementioned vegetation unit. The following Key Grass Species were identified for the vegetation unit Gm 11: Rand Highveld Grassland for predicting the forage potential of the grass sward and their respective statistical parameters and regression coefficients are presented in Table 4.

Table 4: The Key Grass Species identified in the vegetation unit Gm 11: Rand Highveld Grassland in Mpumalanga Province for predicting the FORAGE potential of the grass sward together with their respective statistical parameters and regression coefficients.

KEY GRASS SPECIES ECOLOGICAL PROBABILITY REGRESSION COEFFICIENT CATEGORY VALUE Digitaria eriantha Decreaser 0.0000** 4.2094 Setaria sphacelata Decreaser 0.0000** 5.0806 Aristida transvaalensis Increaser I 0.0096** -1.6016 Eragrostis curvula Increaser I 0.0000** 2.3155 Hyparrhenia anamesa Increaser I 0.0080** 3.6253 Perotis patens Increaser II 0.0204* -1.3662 CONSTANT 0.0000** 177.289 KEY: ** = P<0.01; * = P<0.05 – Levels of Significance

This forage regression model is based on the following statistics:

Number of cases = 62 Degrees Freedom - Model = 6 Degrees Freedom – Residual = 55 Multiple Correlation Co-efficient (R) = 0.9367 Coefficient of Determination (R2) = 0.8774

The coefficient of determination indicates that the Key Grass Species accounted for 87.7% of the variation in the FORAGE potential of the grass sward in Gm 11: Rand Highveld Grassland in Mpumalanga Province. Except for Perotis patens the Probability Value for the Key Grass Species was less than 0.01 meaning that the key species were significant at the 99% confidence level. In the case of the aforementioned species it was significant at the 95% confidence level. The ability of the Key Grass Species to predict the FORAGE potential of the 6 grass sward as represented by the FORAGE SCORE is illustrated by the relationship of OBSERVED vs PREDICTED FORAGE SCORES presented in Figure 1.

Figure 1: Relationship between Predicted vs Observed FORAGE SCORES using the Key Grass Species identified for the vegetation unit Gm11: Rand Highveld Grassland in Mpumalanga Province.

The resultant Key Grass Model involving six Decreaser, Increaser I and Increaser II grass species was tested with independent data from 20 additional grass surveys and the relationship between Predicted vs Observed FORAGE SCORES is presented in Figure 2.

Figure 2: Relationship between Predicted vs Observed FORAGE SCORES using independent data from 20 additional grass surveys comprising the same Decreaser, Increaser I and Increaser II Key Grass Species from the vegetation type Gm11: Rand Highveld Grassland in Mpumalanga Province.

The results in Figure 2 illustrates very well the effectiveness of the six Key Grass Species to predict the grass FORAGE forage potential of the Rand Highveld Grassland vegetation type.

7 In order to simplify the model for predicting the grass FUEL potential for vegetation unit Gm11: Rand Highveld Grassland the same Key Grass Species that were identified for predicting the grass FORAGE potential were used for predicting the grass FUEL potential of the grass sward. The results of the multiple regression analysis together with the respective statistical parameters and regression coefficients are presented in Table 5.

Table 5: The Key Grass Species to be used in vegetation unit Gm11: Rand Highveld Grassland in Mpumalanga Province for predicting the grass FUEL potential together with their respective statistical parameters and regression coefficients. KEY GRASS SPECIES ECOLOGICAL PROBABILITY REGRESSION COEFFICIENT CATEGORY VALUE Digitaria eriantha Decreaser 0.2928 NS 0.8583 Setaria sphacelata Decreaser 0.0202* 1.8289 Aristida transvaalensis Increaser I 0.4046 NS -1.5580 Eragrostis curvula Increaser I 0.0000** 3.8662 Hyparrhenia anamesa Increaser I 0.5278 NS 2.6020 Perotis patens Increaser II 0.0008** -6.2779 CONSTANT 0.0000** 504.004 KEY: ** = P<0.01; *= P<0.05; NS = Not Significant – Levels of Significance

This forage regression model is based on the following statistics:

Number of cases = 62 Degrees Freedom - Model = 6 Degrees Freedom – Residual = 55 Multiple Correlation Co-efficient (R) = 0.6745 Coefficient of Determination (R2) = 0.4550

The coefficient of determination indicates that the Key Grass Species accounted for 45.5% of the variation in the fuel potential of the grass sward in Gm11: Rand Highveld Grassland in Mpumalanga Province. The significantly lower Coefficient of Determination for predicting the FUEL potential of the grass sward is as a result of three of the Key Grass Species (Digitaria eriantha, Aristida transvaalensis and Hyparrhenia anamesa) in Table 5 having a non-significant effect on the grass fuel potential. However, the Grass Fuel Model as a whole is still accounting for a significant variation in the grass fuel potential (45.5%) therefore justifying the decision to use the same Key Grass Species for predicting the FORAGE and FUEL potential of the grass sward in Gm11: Rand Highveld Grassland in Mpumalanga Province. However, if greater predictive precision is required a new set of Key Grass Species can be identified that have a greater positive and negative effect on the FUEL SCORE. The ability of the Key Grass Species to predict the FUEL potential of the grass sward as represented by the FUEL SCORE is illustrated by the relationship of OBSERVED vs PREDICTED presented in Figure 2.

8

Plot of FUEL

1100

900

700 observed

500

300 300 500 700 900 1100 predicted

Figure 2: Relationship between Predicted vs Observed FUEL SCORES using the Key Grass Species identified for the vegetation unit Gm11: Rand Highveld Grassland in Mpumalanga Province.

Key Grass Technique For Assessing Veld Condition Arising from the identification of the KEY GRASS SPECIES the following simplified technique has been developed for assessing veld condition in the Gm11: Rand Highveld Grassland vegetation type – see Table 6. The associated veld management practices related to assessing the condition of the veld in this vegetation type are also included in Table 6 viz. mean grazing capacity, stocking rate, rotational grazing, rotational resting, livestock ratio and veld burning. 9 TABLE 6: VELD CONDITION ASSESSMENT – Gm11: Rand Highveld Grassland MPUMALANGA PROVINCE

Sample Site: ………………………. GPS: S - …………………….. Date: ……………………….…. Soil Type: …………………………… E - ……………………. RECORDER:………………………. FREQUENCY FORAGE FORAGE FUEL FUEL CATEGORY SPECIES % FACTOR SCORE FACTOR SCORE DECREASER Digitaria eriantha 4 1 Setaria sphacelata 5 2 DECREASER TOTAL ******* ******* INCREASER I Aristida transvaalensis -2 -2 SPECIES Eragrostis curvula 2 4 Hyparrhenia anamesa 4 3 INCREASER I TOTAL ******* ******* INCREASER II Perotis patens -1 -6 SPECIES INCREASER II TOTAL ******* ******* OTHERS CONSTANT 177 CONSTANT 504 TOTAL 100.0 TOTAL TOTAL VELD CONDITION SCORE = Forage Score/ 888 x100% =

CONCLUSIONS (tick appropriate block) 1) VELD CONDITION 4) ROTATIONAL GRAZING 5) ROTATIONAL RESTING FORAGE/ FUEL FORAGE FUEL POTENTIAL Seeding? …….. SCORE tick tick HUG? …….. >600 Very High HPG? …….. Vigour? …….. 501 - 600 High HUG/HPG? …….. Fodder Reserve? …….. 401 - 500 Medium REASON: REASON: 301 - 400 Low <301 Very Low 2) MEAN GRAZING CAPACITY GC = 365/(-31.159 + (1.433 x VCS%)) + ((Rainfall – 6) LIVESTOCK RATIO 7) TREND 419.7)x 0.231) Recommended = 1AU: 6 SSU CATEGORY % UTILIZATION tick GC = ……….. HA/AU Current = …….AU: …....SSU Decreaser Correctly stocked Where: (Danckwerts 1981) COMMENT: Increaser I Under stocked GRAZING CAPACITY = Hectares/Animal Unit; Increaser II Over Stocked VCS = Veld Condition Score expressed as %; Increaser II Selectively Grazed Rainfall = Previous 12 months rainfall – mm. COMMENT:

3) STOCKING RATE Predicted = …………………....ha/AU Current = ……………………ha/AU

8) SOIL EROSION POTENTIAL 9) VELD BURNING FACTOR POTENTIAL EROSION CATEGORY % BURN Point To Tuft Distance LOW MOD HIGH YES NO <3cm 3-5cm >5cm DECREASER Distance – cm = INCREASER I Grass Standing Crop LOW HIGH INCREASER II >1500kg/ha <1500kg/ha FUEL LOAD - kg/ha = kg/ha = OVERALL DECISION TO BURN OVERALL SOIL EROSION POTENTIAL

10 REFERENCES

BOSCH, O.J.H. & GAUGH, H., 1991. The use of degradation gradients for the assessment and ecological interpretation of range condition. J. Grassl. Soc. South. Afr. 8. (4): 18 – 146.

DANCKWERTS, J.E., 1981. A technique to assess the grazing capacity of sweetveld with particular reference to the False Thornveld areas of the Ciskei. MSc Thesis, Dept. Pasture Science, Univ. Natal, Pietermaritzburg, South Africa: 1-245.

MUCINA, L. & RUTHERFORD, M.C., (eds) 2006. The vegetation of South Africa, Lesotho and Swaziland. Strelitzia 19. South African National Biodiversity Institute, Pretoria.

TROLLOPE, W S W, TROLLOPE, L.A. & BOSCH, O J H., 1990. Veld and pasture management terminology in southern Africa. J. Grassld. Soc. South Afr. 7,1:52-61.

TROLLOPE, W S W., 1990A. Development of a technique for assessing veld condition in the Kruger National Park. J. Grassld. Soc. South. Afr.7, 1:46-51.