EFFECTS OF VARIETY, ENVIRONMENT AND MANAGEMENT ON SUGARCANE RATOON YIELD DECLINE

REFEREED PAPER EFFECTS OF VARIETY, ENVIRONMENT AND MANAGEMENT ON SUGARCANE RATOON YIELD DECLINE RAMBURAN S 1, WETTERGREEN T 1, BERRY SD 1 AND SHONGWE B 2 1 South African Sugarcane Research Institute, P/Bag X02, Mount Edgecombe, 4300, South Africa 2 Swaziland Sugar Association Technical Services, PO Box 367, Simunye, Swaziland Sanesh.Ramburan@sugar.org.za BernardS@ssa.co.sz Abstract Factors influencing ratoon yield decline (RYD) are often unclear and difficult to unravel. The objectives of this study were (i) to explore the relative contributions of variety, environment and management to RYD, and (ii) to evaluate a method of detecting statistical differences in RYD between treatments. Four sets of trials, comprising six irrigated variety trials from South Africa and Swaziland (set 1), four variety trials from the rainfed KwaZulu-Natal region (set 2), three cycles of a long-term burning and trashing trial (set 3), and two variety x nematicide trials (set 4) were analysed. Yield data were collected over six or more crops. Cane yield was fitted as a linear or quadratic function of ratoon number for each replicate of each trial. The differences between the treatment linear or quadratic coefficients were analysed using ANOVA. Significant (p<0.05) and highly significant (p<0.001) differences in quadratic coefficients were observed between varieties in two out of six trials in set 1. No significant differences between variety coefficients were observed in trials in set 2. Highly significant differences were observed between the burning and trashing treatments in set 3. In set 4, a significant difference in the linear coefficient was observed between variety x nematicide treatments in one trial only. The RYD patterns generally varied between trials, while variability of treatments/varieties within trials was negligible. Hence, it was apparent that environment may have an overriding effect on RYD, followed by the effects of management and variety, respectively. The statistical method illustrated here may be applied to other studies of RYD. Keywords: environment, management, ratoon yield decline, variety Introduction Commercial sugarcane production entails the successive harvesting of ratoons (regrowth after harvesting) over as many crops as possible. The decline in crop yields with successive ratoons, a phenomenon termed ratoon yield decline (RYD), limits the economic viability of sugarcane production by increasing the frequency of costly replanting operations. RYD is a well-known phenomenon, which may be attributed to a range of factors including pests and diseases (Spaull and Cadet, 2003), increased competition between tillers, stool damage (Swinford and Boevey, 1984) and compaction (van Antwerpen et al., 2000) caused by machinery, weed competition (Milhollon, 1995; Srivastava and Chauhan, 2006), and other management factors. Modern sugarcane varieties, which are complex hybrids of Saccharum spontaneum and Saccharum officinarum, are known to differ in their ability to sustain yields over many ratoons. The difference arises from the relative contributions to the genetic make- 180

up of the variety from the different sugarcane ancestors. Varieties with dominant S. spontaneum traits normally demonstrate greater longevity than those of S. officinarum (Milligan et al., 1996). Consequently, most studies of ratooning in sugarcane have focused on varietal effects. Most studies of ratooning in sugarcane have evaluated the possibility of indirectly selecting for ratooning ability (RA) in younger crops (Jackson, 1992; Milligan et al., 1996). In these studies, RA was usually defined as the ratio of older crop yields to younger crop yields (conventionally second ratoon vs. plant crop yields). This definition for RA is appropriate for industries where the last profitable crop to be harvested is usually the second ratoon (e.g. Louisiana). In industries like South Africa, profitable sugarcane production is possible for as many as eight to ten ratoons (Hoekstra, 1976). The above definition of RA is therefore not appropriate for local conditions. Additionally, the difference in economics and environmental conditions between farms makes it difficult to use a rigid number of ratoons in the definition for RA. In other words, what may be considered as an acceptable number of ratoons on one farm may not be profitable on another. It therefore follows that a description of RA that incorporates a rate of decline rather than a ratio of crop yields at early vs. later ratoons, may be more widely applicable. The factors causing RYD in South Africa are currently unclear. Although it is generally accepted that factors such as pests and diseases (Cadet and Spaull, 2001; Spaull and Cadet, 2003) and stool damage (Meyer, 2005) are of importance, there is a general grower perception that varietal differences in RYD are of greater importance. The reasons for such a perception are also unclear, but may be linked to the long-term yield decline trends in the industry which are thought to be associated with continuous burning, soil degradation and acidification and continuous monocropping. The perception may also be linked to the release of more varieties with S. officinarum traits (high sucrose, low stalk population) as well as the belief amongst growers that the visual appearance of regrowth after harvest is associated with final crop yields. These perceptions have led to numerous questions about the ratooning of new varieties, with many growers suggesting the replanting of older, better ratooning varieties. In light of these developments, it was realised that the relative effects of different crop production factors on RYD had never been evaluated. Information on the relative influences of variety, environment and management practices on RYD is essential to direct any future research. Therefore, the objectives of this study were (i) to evaluate and illustrate the relative contributions of variety, environment and management to RYD, and (ii) to evaluate a method of detecting statistical differences in RYD between treatments to assist future studies. Materials and Methods Trial datasets The approach used in this study was to evaluate the RYD trends of different sets of trials that were conducted under different conditions. Each trial set was conducted with specific objectives. Trial set 1 comprised six fully irrigated variety trials established in South Africa and Swaziland, with the objective of identifying varieties suited to varying environmental and management factors in order to refine recommendations. Trial set 2 comprised four variety trials established in the southern rainfed regions of South Africa, with similar objectives as set 1. Trial set 3 comprised three cycles of a long-term burning and trashing trial. The objectives of these trials are to investigate the effects of burning vs. trashing on sugarcane yield and to quantify various parameters (soil and plant nutrition, soil temperature, soil 181

moisture, soil health issues and the impact of reduced soil disturbance) affected by the presence of a trash blanket. Trial set 4 comprised two variety x nematicide trials, where the objectives were to evaluate the agronomic performances over time of different varieties on poor, sandy (<10% clay) soils as well as to examine their response to annual nematicide (Temik at 20 kg/ha) treatment. The six variety trials in set 1 were all established in 2003 and consisted of six to ten commercial varieties (only varieties that were common to all trials were included in the analysis). The trials were established as RCBDs with six to ten replicates. Trial plots consisted of five or six rows that were 8-10 m long and spaced 1.4 to 1.5 m apart. All trials in this set were harvested on a 12-month cutting cycle over six or seven crops (plant + five or six ratoons). The trials were established under different irrigation systems, with fertiliser and weed management as per commercial practice. The four variety trials in set 2 were established between 2001 and 2005. These trials consisted of eight to ten varieties planted in RCBDs with four to six replicates. Trial plots consisted of five to six rows that were 8-10 m long and spaced 1.0 to 1.5 m apart. Trials in this set were harvested between 12 and 18 months over six to nine crops (plant + five to eight ratoons). All but one of these trials were established under fully rainfed conditions with weed and fertiliser management as per commercial practice. The three cycles of the burning and trashing trial formed part of a long-term trial that was established in 1939 under rainfed conditions. The cycles used for this paper were established in 1977 (BT1), 1991 (BT2) and 2002 (BT3), and consist of 32 plots each, made up of seven rows 18 m long and spaced 1.4 m apart. The main treatments are trashed (T) vs. burnt (B) cane. The burnt treatments are further divided into four plots with tops spread (t) and four plots with tops removed (t 0 ). An additional treatment subdivision (across all plots) is fertilised (F) and not fertilised (F 0 ). For the first cycle the trial was harvested between 12 and 18 months, but, as of the beginning the second cycle, the trial has been harvested on a 12-month cycle (September/October) each year. A fixed fertiliser rate is applied to treatments requiring fertiliser in October/November each year. The cycles BT 1, BT 2, and BT 3 were planted to varieties NCo376, N16 and N27, respectively. The two trials in set 4 were established between 1994 and 2000. These trials consisted of four (VN1) and six (VN2) varieties planted in RCBD (VN1) or split plot (Latin Square, VN2 trial) designs with six replicates. Trial plots consisted of five rows that were 8-10 m long and spaced 1.2 m apart. Trials in this set were harvested at 12 months over six (VN1) or seven (VN2) crops (plant + five to six ratoons). Both trials were established under fully rainfed conditions with weeding and fertiliser management as per commercial practice. Further details of all trials analysed are summarised in Table 1. 182

Table 1. Details of the four trial sets analysed. The trial codes IR, RF, BT and VN refer to the irrigated variety trials, rainfed variety trials, burn vs. trash trials and variety x nematicide trials, respectively. Trial Trial Crops Location Irrigation Soil form Harvest details code harvested Set 1 IR1 Pongola C * Sprinkler Katspruit 12 month, early season 8 Fully IR2 Pongola RS ** Drip Hutton 12 month, late season 7 irrigated IR3 Swaziland C Sprinkler Katspruit 12 month, early season 7 variety IR4 Swaziland C Sprinkler Swartland 12 month, late season 7 trials IR5 Pongola RS Drip Hutton 12 month, early season 8 IR6 Komatipoort RS Drip Glenrosa 12 month, early season 8 Set 2 RF1 Doringkop C Rainfed Shortlands 18 month 7 Rainfed RF2 Empangeni RS Rainfed Shortlands 12 month, early season 7 variety RF3 Scottburgh C Rainfed Glenrosa 12 month, late season 9 trials RF4 Schroeders C Supplementary Inanda 12 month, mid-season 6 Set 3 BT1 Mt Edgcombe RS Rainfed Arcadia 12-18 month, mid season 11 Burn vs. trash BT2 Mt Edgcombe RS Rainfed Arcadia 12 month, mid season 9 trials BT3 Mt Edgcombe RS Rainfed Arcadia 12 month, mid-season 9 Set 4 VN1 New Guelderland C Rainfed Fernwood 12 month, late season 6 Variety x nematicide VN2 La Mercy C Rainfed Fernwood 12 month, late season 7 C = Trial established on a commercial farm, RS = Trial established on a SASRI research farm Data analysis For all trials, net trial plots (outer two rows discarded) were hand-harvested and weighed using a scale mounted on a tractor-operated hydraulic boom, to determine cane yield in tons cane/ha. Each trial was subsequently analysed separately using conventional analysis of variance (ANOVA) procedures. For each trial, the average cane yield of each treatment (variety, burn/trash treatment, variety x nematicide combination) was plotted against the ratoon number to identify the general shape of the RYD curve. Based on this initial assessment, the cane yield of each plot in the trial was fitted as a linear or quadratic function of ratoon number, i.e. curves were fitted at the replicate level. The quadratic function was of the form: Y = Ax 2 + Bx + C where Y = cane yield at ratoon x A = quadratic coefficient (determines whether the curve faces up or down and if it is narrow or wide), B = linear coefficient (determines the axis of symmetry of the curve), C = intercept (determines the y-intercept, i.e. the yield of the plant crop in this study). The linear function was of the form: Y = Ax + B where Y = cane yield at ratoon x A = linear coefficient (rate of yield decline with successive ratoons) B = intercept (y-intercept). Linear or quadratic coefficients were therefore derived for each plot in each trial. Where R 2 values of less than 0.5 were obtained, such coefficients were not included in the analysis and were considered as missing values. The coefficients derived for each plot were subsequently subjected to routine ANOVA to determine whether there were significant differences 183

between the treatments. Significant differences in coefficient values between treatments suggested that treatments differed significantly in terms of RYD. This approach allowed for an evaluation of the relative influences of varieties, environments and management practices on RYD. Results Trial set 1 (irrigated variety trials) The RYD trends for trial set 1 are shown in Figure 1. All trials in this set showed quadratic RYD trends. Trial IR1 (Figure 1a) showed the most rapid rate of RYD and was the only trial with downward facing curves, i.e. the values for the A coefficients of all varieties were negative. When calculating the difference in cane yields between the plant and last ratoon harvested, it was found that the early season trials IR1 (Figure 1a), IR3 (Figure 1c) and IR5 (Figure 1e) lost approximately 51, 45 and 61% of their plant crop yields averaged across varieties. This was in contrast to the 30 and 33% loss of plant crop yields observed in the late season trials IR2 (Figure 1b) and IR4 (Figure 1d), respectively. This suggests a possible influence of time of harvest on RYD. Trial IR6 (Figure 1f) showed the most unusual trends, as yields improved after the fourth ratoon. It was later revealed that the unusual increase in yields after the fourth ratoon coincided with a change in farm management on the research station. The new management placed more emphasis on addressing nutrient deficiencies based on soil sample analyses and also implemented irrigation scheduling practices. Therefore, the unusual improvement in yields after the fourth ratoon in Figure 1f may very well be due to changed management practices. Across all trials in this set, variety N25 consistently produced the highest cane yields compared with the other varieties. However, within each trial, the shape of the N25 curve did not differ drastically from the other varieties, suggesting that there were no explicit differences in RYD trends. The ANOVA for differences between varieties in terms of the A, B and C coefficients for each trial showed that significant differences in the C coefficient were observed in five out of six trials (Table 2). These significant and highly significant differences in the C coefficient were due mainly to the high yields of N25, i.e. the C coefficient represents the vertical positioning of the curve (intercept). The A coefficient, which represents the upward/downward shape of the curve, differed significantly between varieties in three out of six trials. The B coefficient, which most closely represents a rate of decline, differed significantly between varieties in two out of six trials. Overall, varieties showed significant differences in all three coefficients in two out of the six trials evaluated, suggesting that in the majority of cases varieties did not differ statistically in terms of RYD. The shapes of the quadratic curves varied across trials, suggesting that the overall RYD trend was dependent on trial site conditions, i.e. environment and management. This was highlighted by the fact that the same variety often produced varying coefficients across trials. For example, values for the A, B and C coefficient for N25 ranged from -0.9 to 4.2, -4.8 to -38.9, and 134.7 to 186.8, respectively (not shown). Similarly, for a low yielding variety like N40 the A, B and C coefficients varied from -0.5 to 6.9, -7.2 to -46.0, and 118.0 to 163.0, respectively (not shown). These results suggest that RYD is more dependent on environmental conditions and management practices (which determine the overall shape of the RYD curve), while variety influences only slight variation within a particular shape. 184

Figure 1. Quadratic ratoon yield decline trends of varieties tested in trials IR1 (a), IR2 (b), IR3 (c), IR4 (d), IR5 (e), and IR6 (f). All trials were established in 2003 and harvested annually thereafter. Actual (points) and fitted (lines) cane yields for the different varieties are indicated. 185

Table 2. Levels of statistically significant differences in quadratic coefficients between treatments in different trials. Trial set Trial code Coefficients A B C 1 IR1 NS NS * IR2 ** ** *** IR3 *** *** *** IR4 * NS *** IR5 NS NS * IR6 NS NS NS 2 RF1 NS NS NS RF2 NS NS ** RF3 NS NS ** RF4 NS NS *** 3 BT1 *** *** BT2 *** *** BT3 NS *** 4 VN1 * NS VN2 NS NS ** *P<0.05, **P<0.01, ***P<0.001, NS = not significant Trial set 2 (rainfed variety trials) Trials in this set were more variable in terms of RYD patterns than trials in set 1. In trial RF1 (Figure 2a) variety curves were vertically separated, however, the shape of the curves were similar (except for N39). As a result, there were no significant differences between varieties in any of the coefficients (Table 2). The variety curves in RF2 (Figure 2b), RF3 (Figure 2c) and RF4 (Figure 2d) seemed very different from each other visually. However, the ANOVA showed that no significant differences were observed between the varieties in terms of the A and B coefficient. In the same trials, significant and highly significant differences in the C coefficient were observed between varieties. This suggests that the vertical positioning and intercept of the varieties, which are determined by the average cane yields, were significantly different. The lack of significant differences in the A and B coefficients between the varieties was surprising, given the contrasting variety curves obtained in each trial. For example, variety N32 showed a marked rapid decline in yield compared to N27 in RF2 (Figure 2b). Similarly, varieties N33 and N21 actually showed crossover interactions between the early and late ratoons in RF3 (Figure 2c). The lack of significant differences in RYD (at least in the A and B coefficients) in these trials may point to a possible flaw in the statistical methodology employed. It is apparent from an agronomic perspective that the varieties in these trials differed economically in terms of RYD. Evidence for this can be seen through a simple comparison of the yield differences between the plant and last ratoon harvested for each of the varieties in the trials. 186

Figure 2. Quadratic ratoon yield decline trends of varieties tested in trials RF1 (a), RF2 (b), RF3 (c), and RF4 (d). Actual (points) and fitted (lines) cane yields for the different varieties are indicated. Trials consisted of different variety sets. Trial set 3 (burn vs. trash trials) All three trials in this set were fitted to linear RYD curves (Figure 3). In all three trials, the fertilised treatments (solid lines) clustered together and produced flatter curves compared to the unfertilised (broken lines) treatments, which showed more rapid rates of RYD. When calculating the difference in cane yields between the plant and last ratoon harvested, it was found that the fertilised treatments lost on average 17.6, 0 and 36.3% of their plant crop yields for trials BT1 (Figure 3a), BT2 (Figure 3b) and BT3 (Figure 3c), respectively. This was in contrast to the 70.2, 49.9 and 47.3% loss of plant crop yields observed with the unfertilised treatments in trials BT1, BT2, and BT3, respectively. The ANOVA for this trial set (Table 2) showed that there were highly significant differences between treatments for both the A and B coefficients in BT1 and BT2. However, in BT3 there were no significant differences between treatments in the A coefficient (Table 2), which meant that there were similar rates of yield decline, as observed in Figure 3c. It was interesting to note that the burn vs. trash treatments did not produce any contrasting RYD trends, and that the fertiliser treatment was more influential at determining RYD. 187

Figure 3. Linear ratoon yield decline trends of different treatments tested in trials BT1 (a), BT2 (b), and BT3 (c). Actual (points) and fitted (lines) cane yields for the different varieties are indicated. BT1, BT2 and BT3 were planted to varieties NCo376, N16 and N27, respectively. Solid lines represent fertilised treatments, while broken lines represent unfertilised treatments (Bt = Burnt with tops retained; Bt0 = Burnt and tops removed from plot; F = Plot has been fertilised, F0 = No fertiliser has been applied to the plot; T = Plot has been trashed). Trial set 4 (variety x nematicide trials) In this series, treatments from trial VN1 were fitted to linear curves, while treatments from VN2 were fitted to quadratic curves (Figure 4). In trial VN1 (Figure 4a), significant differences were observed between the treatments in terms of the A coefficient (Table 2), suggesting that treatments showed differences in RYD. In fact, many treatments in trial VN1 actually showed slight improvements in yields with subsequent ratoons. No explicit differences in RYD were observed between the control and nematicide treatments for the different varieties in VN1. In trial VN2 (Figure 4b), most of the treatments produced curves of a similar shape. Exceptions to this were the N23 nematicide treatment and the N24 control treatment, which produced slightly upward and downward facing curves, respectively. Although there were no significant differences in the A and B coefficients in VN2, it was clear that the nematicide treatments (solid lines) consistently produced higher cane yields than the control treatments (broken line) for all varieties. This vertical separation of the curves (differences in intercepts) resulted in significant differences in the C coefficient (Table 2). 188

a b Figure 4. Linear and quadratic ratoon yield decline trends of different variety x nematicide treatments tested in trials VN1 (a), and VN2 (b). Actual (points) and fitted (lines) cane yields for the different treatment combinations are indicated. Solid lines represent nematicide treatments (Temik) while broken lines represent no nematicide treatments. Discussion This study is the first to explore the RYD trends of different varieties and crop management factors under various conditions. It was apparent that the RYD trends varied from one trial site to the next, making it impossible to define a specific RYD pattern for sugarcane in the industry. The RYD trend also varied for specific varieties when planted in different trials. This suggests that the RYD of individual varieties is not a set genetic characteristic that remains constant, but rather a factor that is influenced by the environment. Even in trial set 1, where N25 consistently produced the highest cane yields across trials, it was shown that the actual coefficients for N25 varied drastically from one trial to the next. Similarly, in trial set 2, the RYD trend of N12 varied between RF1 and RF3. Furthermore, in trial set 3, it was shown that the RYD trend of the same varieties (NCo376 in BT1, N16 in BT2, and N27 in BT3) varied dramatically when management practices (fertiliser application) were changed. These observations highlight the fact that RYD is primarily dependent on environmental and management factors. Varieties did differ in RYD (as observed in trial sets 1 and 2 in Table 2); however, these differences were negligible compared to the overall effects of environment 189

and management. Follow-up studies should consider performing a combined analysis of the trial sets so that the relative influences of variety and environment can be quantified. The scope of future studies should also include a statistical comparison of the actual coefficient values for the different treatments using means comparison tests, e.g. least significant differences. It was also found that under fully irrigated conditions in trial set 1, the RYD curves for the different varieties clustered together tightly. However, in the rainfed trials in set 2 the variety curves were more separated. This was also evident in the BT trials as well as the VN trials, which were conducted under rainfed conditions. This may suggest that differences in RYD may become more marked under rainfed conditions. Furthermore, all of the crossover interactions (ranking of treatments changing between early and late ratoons) were observed in the rainfed trials only. Most of the trials analysed in this study showed quadratic RYD trends, which gave the best R 2 values. Although providing a more accurate representation of the yield decline than a linear trend, the quadratic coefficients were not practical in terms of interpretation. In many cases, the lack of significant differences in the A and B coefficients, coupled with significant differences between the C coefficients, made it difficult to reach any affirmative conclusions about differences in RYD. In contrast, a linear model was more practical, as significant differences in the A coefficient meant only that treatments differed in terms of their rate of yield decline. However, the linear model produced lower R 2 values when fitted to the data. There was therefore a trade-off between accurately representing the RYD pattern (using quadratic models) and obtaining results that were practically relevant to crop production (using linear models). Most studies of ratooning in sugarcane have focused on production up to the second ratoon. However, as pointed out by Kang et al. (1987), ratoons of sugarcane are confounded with seasonal effects (each ratoon grows through a different season). It is therefore possible that past studies that have evaluated ratooning up to the second ratoon only, may have confounded the ratoon and season effect. Good performance in the second ratoons compared to the first may have been due to better seasons experienced in the second ratoons or better responses of certain varieties to conditions in the second ratoons. In South Africa, good ratooning varieties are those that produce good yields beyond the fifth ratoon. A ratio of plant crop to ratoon crop performance (any defined ratoon beyond the fifth) should determine whether a variety is a good ratooner. However, the problem of confounding the ratoon and seasonal effect still exists. Therefore, a definition of ratooning that takes into consideration both the rate of yield decline, as well as the absolute yields at older ratoons, would be more accurate. Conclusions The recent concerns around ratooning in the industry prompted this exploration of factors influencing RYD. The results showed that: Environmental conditions and management practices are of primary importance in determining RYD trends. The effects of variety are of secondary importance, and influence only the variation within a RYD pattern which is ultimately determined by environment and management. 190

The same variety will show different RYD patterns under different environmental conditions and management practices. A greater number of statistically significant differences in RYD were observed between management practices (BT trials) than between varieties (IR and RF trials). The implications of these results are that: Growers in the industry should place more emphasis on altering the environment through management rather than focusing their efforts on choosing a variety with perceived better ratooning. Growers must be made aware of the relative importance of management vs. variety in maintaining yields over ratoons; a message that SASRI should deliver more explicitly. Any further study of RYD of varieties must occur across contrasting environmental conditions. Future studies of RYD should focus on: In-depth scrutiny of which treatments are significantly different from each other in terms of the coefficients. Determining whether other yield components such as stalk population are responsible for differences in RYD. Integrating the rate of yield decline with an absolute yield at a specific ratoon to better define ratooning ability. Using the crop model to correct for seasonal variability in order to evaluate the true RYD of different treatments. Allowing all sugarcane field trials to proceed beyond the fifth ratoon in order to gather sufficient data on RYD. REFERENCES Cadet P and Spaull VW (2001). Annual and long term benefits of nematode control on yield of sugarcane. Proc S Afr Sug Technol Ass.75: 115. Hoekstra G (1976). Analysis of when to plough out a sugarcane field. Proc S Afr Sug Technol Ass 50: 103-113. Jackson PA (1992). Genotype x environment interaction in sugarcane. II: Use of performance in plant cane as an indirect selection criterion for performance in ratoon crops. Aust J Agric Res 43: 1461-1470. Kang MS, Miller JD, Tai PYP, Dean JL and Glaz B (1987). Implications of confounding genotype x year and genotype x crop effects in sugarcane. Field Crops Res 15: 349-355. Meyer E (2005). Machinery systems for sugarcane production in South Africa. MSc Eng Thesis. 36 pp. Milhollon RW (1995). Growth and yield of sugarcane as affected by Johnsongrass (Sorghum halepense) interference. Am Soc Sug Cane Technol 15: 32-40. Milligan SB, Gravois KA and Martin FA (1996). Inheritance of sugarcane ratooning ability and the relationship of younger crop traits to older crop traits. Crop Sci 36: 45-50. Spaull VW and Cadet P (2003). Impact of nematodes on sugarcane and the benefit of tolerant varieties. Proc S Afr Sug Technol Ass 77: 230-238. 191

Srivastava TK and Chauhan RS (2006). Weed dynamics and control of weeds in relation to management practices under sugarcane (Saccharum species complex hybrid) multi-ratooning system. Indian J Agron 51: 228-231. Swinford JM and Boevey TMC (1984). The effects of soil compaction due to infield transport on ratoon cane yields and soil physical characteristics. Proc S Afr Sug Technol Ass 58: 198-203. van Antwerpen R, Meyer JH and Meyer E (2000). Soil compaction in the South African sugar industry A review. Proc S Afr Sug Technol Ass 74: 101-108. 192