INTRODUCTION
With steadily rising labour costs, mechanical harvesting (MH) of tea (Camellia sinensis) is becoming an increasingly attractive proposition. However, there are concerns over both yield and quality under MH (Bore and Ngetich, Reference Bore and Ngetich2000; Kamunya et al., Reference Kamunya, Jibwa Kale, Wanyoko, Wachira and Bore2010; Madamombe et al., Reference Madamombe, Tesfamariam and Taylor2015; Mwakha, Reference Mwakha1986; Reference Mwakha1990; Owuor et al., Reference Owuor, Othieno, Robinson and Baker1991).
The yield and quality of hand-plucked tea are both dependent on the harvest interval, the method of harvesting, or plucking style, and the skill of the plucker. With ‘coarse’ plucking (3 leaves + a bud or 4 + bud shoots) yield will be greater than with ‘fine’ plucking (2+bud shoots); quality will usually be poorer, though this is not true for all clones (Obanda and Owuor, Reference Obanda and Owuor1995). Fine shoots can be obtained by non-selective plucking with a short harvest interval, or by selective plucking at longer intervals, taking only shoots of the required size. Non-selective plucking at longer intervals will give coarser shoots. To monitor quality, the composition of the harvested leaf, in terms of stages of shoot development, is routinely recorded.
Yield can be considered as the product of shoot number and mean shoot weight. As mentioned above, shoot weight can be controlled by the harvesting regime adopted. Shoot numbers depend on growth rate and shoot generation time, density of plucking points and shoot replacement ratio (e.g., Stephens and Carr, Reference Stephens and Carr1994). Harvest intensity also affects yield: ‘hard’ plucking (shoots plucked close to the base, with no leaves left behind) generally gives greater yield than ‘light’ plucking (shoots plucked above the oldest 1 or 2 leaves) (Chandra Mouli et al., Reference Chandra Mouli, Onsando and Corley2007; Visser, Reference Visser1960), though the effect on the total photosynthetic surface needs to be considered.
In the 1990s, Brooke Bond Tanzania (now Unilever Tea Tanzania) started MH on a commercial scale. At the same time, trials were established to ascertain how best to manage MH and to determine effects on yield and quality. Under non-selective MH, any shoots projecting above the cutting height are removed. This releases axillary buds from dormancy, thereby stimulating the development of a ‘generation’ of new shoots of the same age. These generations need to be well separated either in age or by adjusting cutting height, so that the next generation of shoots is not damaged during harvest, as noted by Madamombe et al. (Reference Madamombe, Tesfamariam and Taylor2015). In this paper, we report results of two experiments in which different cutting heights were compared. To ensure separation of generations, harvest interval with the machine was longer than with hand plucking; Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006) showed that yield was likely to be greater with a longer harvest interval.
Harvesting costs are not considered here, as it was not intended to use the small experimental machine on a commercial scale, but Barbora et al. (Reference Barbora, Sarma, Barua and Barbora1993) showed a 3-fold increase in labour productivity with a similar machine compared to hand plucking. Provided that yield and quality from MH were not much lower than from hand plucking, the greatly reduced labour requirement would favour MH. Thus, the main objective of our experiments was to compare yield and leaf quality under MH regimes and hand plucking.
MATERIAL AND METHODS
The experiments were all located on Ngwazi Estate (8° 32’S, 35° 10’ E, altitude 1840 m asl) in Tanzania. The climate has been described by Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006); the estate is relatively flat and well suited to harvesting with wheeled machines. In the experiments, we used a small, self-propelled, wheeled tea harvesting machine with a 110 cm wide, variable speed, rotating cylinder cutter and precisely adjustable cutting height, supplied by the New Century Corporation, Japan. Two machines were available to ensure that harvesting schedules could be followed despite the occasional need for maintenance and repairs. Two people guided the machine with a third managing leaf collection. The cutting speed used depended on the crop: a faster speed was needed with heavy crop. The machine did not cause noticeable soil compaction.
Two experiments are described here; each was done on a single clone. Clone TRFK31/8 has medium-sized erect and induplicate leaves; BBT133 has smaller semi-erect leaves. Treatments were replicated four times in randomised complete blocks, with plots of single rows of approx. 190 bushes (data from only two replicates of treatment B were recorded in 2002–2003). The experiments were irrigated with overhead sprinklers whenever soil water deficit reached 60 mm, to bring the deficit back to zero. Standard estate fertiliser dressings were applied annually (N: 300 kg ha−1; P2O5: 50 kg ha−1; K2O: 125 kg ha−1). This was supplemented by 60 kg ha−1 of sulphur annually from 1999 onwards and 2.5 kg ha−1 of zinc oxide from 2000, as symptoms of sulphur and zinc deficiency were seen. The experiment with clone BBT133 commenced in October 1995, and that with TRFK31/8 in March 1996; both had to be terminated in August 2003, because of machine breakdowns. The clones were planted in 1971, at 9260 bushes/ha, and pruned at a height of 42 cm in August 1994.
Over the first cycle treatments were as follows:
A Hand plucking under an intensive, non-selective regime, leaving no young shoots above the table surface, harvest interval 2 phyllochrons.
B Machine harvesting, constant cutting height, 58–61 cm above ground level (no table rise), harvest interval 3 phyllochrons.
C Machine harvesting, table rise of 1 cm every 3rd round, harvest interval 3 phyllochrons.
D Machine harvesting, table rise of 1 cm at every round, harvest interval 3 phyllochrons.
From 1995 to 1998, the average harvest intervals in the peak cropping season (October–April) were 24 days (range 16–34 days) for machine harvesting and 17 days (range 11–20 days) for hand plucking.
The experiments were pruned in August 1998 and again in August 2003; pruning height was 42 cm; treatments A, B and C were tipped at 53 cm in November 1998. Treatment D was terminated after pruning in 1998, as yields were low.
In the second cycle, from 1998 onwards, hand plucking continued at an interval of 2 phyllochrons. The MH treatments were changed in the second cycle as follows:
B Zero table rise, target shoot (TS) 3+bud; harvest interval judged to maximise the proportion of the TS.
C Table rise flexible, depending on shoot population before harvest; average table rise intended to be 1 cm per 3 rounds; TS = 3+bud; harvest interval judged to maximise the proportion of the TS.
D Discontinued
The machine harvest intervals in the peak cropping season (October–April) averaged 30 days (range 18–42 days), compared to 15 days (range 11–20 days) for hand plucking.
Shorter term experiments were also done with three other clones (BBT1, BBK35 and TRFCA PC81), with the same treatments. Only quality data from these experiments are presented here.
Recording
Unless otherwise specified, all records were based on whole plots. All harvested fresh leaf was weighed, and plot totals were converted to made tea (kg ha−1) with a conversion factor of 22% for both clones and all treatments. The main growing season in this region is from October to April, so annual yields were computed from September to the following August.
Mean fresh weight per shoot was determined on samples of 100 g taken from the harvested fresh leaf. For MH samples, shoots cannot be counted reliably, as they are often cut more than once by the machine, but the number of terminal buds in the sample gives a good estimate, as every shoot has a bud; thus mean shoot weight was calculated from weight of sample divided by number of terminal buds. Mean shoot weight was recorded at every harvest for both clones from 1999 onwards and also for 5–8 harvest rounds in clone BBT133 in 1996. The average of samples from all rounds was calculated for each plot before statistical analysis. Number of shoots harvested was estimated from total fresh leaf yield divided by mean shoot fresh weight (Burgess et al., Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006).
Quality of green leaf was evaluated by sorting the same 100 g samples as for shoot fresh weights into acceptable leaf and coarse leaf (mature foliage and hard leaf and stalks of the larger shoots). For hand plucked samples, the shoots were also sorted according to number of leaves (1 leaf + bud, 2+bud and so on up to 5+bud). This was not possible for MH leaf as shoots are often cut more than once, so the number of shoots at each stage was counted in a 700 cm² quadrat immediately before harvest; this was done for three bushes per plot, at every round from May 1999 to Dec 2002. Quality of made tea was evaluated by a professional tea taster in the United Kingdom, on samples produced by Teacraft mini-manufacture equipment; this was done between 1995 and 1997. The taster assigned monetary values to the samples, which were expressed relative to a Liptons Yellow Label standard. The number of samples per clone varied as shown in Table 3.
In some plots die-back of shoot stumps after plucking was seen; frequency was recorded on five bushes in every plot using a 400 cm² quadrat in August 2000 and August 2001.
The weight of pruned stems and leaves was recorded after pruning; based on unpublished work by Unilever Tea Kenya, it was assumed that pruning trash had a dry matter content of 40%. From these figures, and weight of fresh leaf with an assumed dry matter content of 22%, total dry matter production (DMP) and harvest index (HI, yield as a proportion of total dry matter) above the pruning level were calculated. This excluded fallen leaves, which were a relatively small proportion of total dry matter in the study by Magambo and Cannell (Reference Magambo and Cannell1981); growth below the pruning level was also excluded.
Intensity of harvest (IoH) was measured directly in 2003, using the method described by Chandra Mouli et al. (Reference Chandra Mouli, Onsando and Corley2007). This involved recording the number of leaves remaining on plucked stumps immediately after harvesting of three bushes per plot, using a 700 cm² quadrat. An indirect measure of harvest intensity was also obtained, by measuring the rise in height of the plucking table (bush surface) with time. Table height was recorded at the start of each pruning cycle, and immediately before pruning at the end of each cycle.
Leaf area index (LAI) was recorded in 1996 and 1997 on three bushes per plot: a 400 cm2 quadrat was placed on the bush after plucking, and all leaves below the quadrat were removed and their areas measured.
Statistical analysis
Analysis of variance and 95% least significant differences (LSD) were used to determine whether means from the MH treatments differed from hand plucking. Individual plot data for table rise in the first cycle and for LAI were not available, so for these measurements, the analysis of variance was based on means for treatments, with clones as replicates.
RESULTS
Effects on yield
Yields are summarised in Table 1. For both clones, yields under hand plucking were lower during the first cycle (1995–98) than the second (1998–2003), perhaps because of zinc and sulphur deficiencies, which were treated from 1999 onwards (see Methods). Yield of clone 31/8 was over 50% greater in the second cycle, while that of BBT133 was 14% greater. In 2000–2001, clone 31/8 yielded over 8000 kg ha−1 under hand plucking, a good yield for this environment.
Table 1. Effect of hand harvesting (A) and the three mechanical harvesting methods (B, C, D) on yields of made tea (kg ha−1 yr−1) from clones BBT133 and 31/8.
The tea was pruned in August 1998 and August 2003; annual yields were recorded from September to August.
For treatment C, the table rise was 1 cm every third round until 1998. From 1998 to 1999 onwards, the rise was intended to average 1 cm per 3 rounds, but a decision on whether to raise the cutting height was made before each harvest, with the aim of maximising yield of fully developed shoots while minimising damage to younger shoots.
In 2002–2003, data from only two replicates for Treatment B were included.
In each row, means followed by the same letter are not significantly different (95% LSD).
During the first cycle the treatment with no table rise (B) gave a significantly larger yield than hand plucking (A) in both clones, averaging 17% higher. In the second cycle, yield from treatment B was only 7% greater than treatment A, and for clone 31/8 the difference was not significant.
In the first cycle, a table rise of 1 cm every third round (C) gave a significant yield increase compared to hand plucking with clone BBT133 (13%), but not with 31/8 (−1%). Under this treatment, with a predetermined time for cutting height rise and a harvest interval of 3 phyllochrons, the planned harvest without a rise sometimes damaged the next generation of immature shoots. Conversely, sometimes a rise was scheduled, but mature shoots could have been taken without raising the cutting height, without damage to the next generation. To avoid these situations, in the second cycle flexible timing for harvest and cutting height rise was introduced. This allowed better management of the shoot generations, and in the second cycle yields from treatment C were greater relative to other treatments (Table 1). Between 1995 and 1998, treatment C gave a similar yield to zero rise (B) with clone BBT133, and less with 31/8, but in the second cycle it yielded significantly more than zero rise with both clones, and 24% more than hand plucking.
Treatment D, with the largest table rise of 1 cm every round, gave significantly lower yield than hand plucking in both experiments, averaging 11% lower. This treatment was stopped after 1998.
Effects on yield components and quality
Under hand plucking in the second cycle, clone 31/8 gave heavier shoots than BBT133, despite a slightly larger percentage of 1+bud and 2+bud shoots (Table 2). In both clones, the increases in yield under MH compared to hand plucking came from significantly heavier shoots. The total number of shoots harvested per year was smaller with zero rise than with hand plucking, significantly so in clone BBT133. Thus, the most likely reason for increased yield with MH is the longer harvest interval, which allowed the shoots to grow to a larger size.
Table 2. Effect of hand harvesting (A) and the two mechanical harvesting methods (B, C) on yield components and shoot types from clones BBT133 and 31/8 between 1999 and 2003.
For each clone, within a column, means followed by the same letter are not significantly different (95% LSD).
Mean weight per shoot is often used as a simple measure of shoot quality, with small shoots usually indicating better quality. Table 2 shows that the larger shoot weights came from a much higher proportion of 4+bud shoots with the machine treatments, reflecting the longer harvest intervals. The proportion of coarse mature leaf, which also degrades quality, was significantly higher with table rise of 1 cm per 3 rounds than with hand plucking, and higher again with zero rise.
Between 1995 and 1998, a number of samples from these experiments, and from other clones under similar treatments, were mini-manufactured and tasted by a professional taster. Results are summarised in Table 3. Clone BBK35 was the most extensively tasted, with three samples per treatment, and there was a significant difference in monetary value between the MH samples and hand plucked tea. However, averaged over all clones, there were no differences in value between the MH treatments or between MH and hand plucking, all being assessed at about 80% of the value of the Liptons Yellow Label standard.
Table 3. Taster's evaluations of mini-manufactured tea from hand plucking (A) and two MH treatments (B and C), for five clones.
On each sampling date, one sample from each treatment was taken.
Values are expressed as a percentage of a Liptons Yellow Label standard.
Within a column, means followed by the same letter are not significantly different (95% LSD).
Intensity of harvest and bush health
A retrospective indication of the IoH is given by the amount of table rise. Table 4 shows that table rise under hand plucking was 6–8 cm yr−1. Annual rise was greater with a rise of 1 cm/round (treatment D) than for hand plucking. Hand plucking and a rise of 1 cm every third round (treatment C) gave very similar table rise in the first cycle, but during the second cycle, with treatment C having some flexibility in raising the cutting height, table rise was slightly less than in the first cycle and significantly less than under hand plucking.
Table 4. Effect of hand harvesting (A) and the two mechanical harvesting methods (B, C) on table rise for clones BBT133 and 31/8 between 1996 to 1998 and 1998 to 2003, and intensity of harvest in 2003.
Within each cycle, means followed by the same letter are not significantly different (95% LSD).
IoH also affects the weight of pruning trash. The lower the table rise, the smaller the expected weight of prunings. Results in Table 5 confirm that weights were smallest with zero rise (B). In 1998, 1 cm rise every round (D) gave the largest weight. In 2003, but not in 1998, treatment C gave significantly smaller trash weights than hand plucking. For treatment B, despite almost no rise there was still some pruning trash, because some growth would have occurred below the table surface but above the pruning height. Pruning trash weight (t ha−1 yr−1) was significantly correlated with table rise (cm yr−1): combining results for both clones and both cycles: Trash = 1.4 + 0.51 × Table rise (r = 0.956***, 12 d.f.).
Table 5. Effect of hand harvesting (A) and the three mechanical harvesting methods (B, C, D) on dry matter production (t ha−1 yr−1) by clones BBT133 and 31/8.
Trash = pruned stems and leaves; HI = harvest index (%).
Within each cycle and column, means followed by the same letter are not significantly different (95% LSD).
Direct measurements of IoH were started in 2003, by counting leaves on the stumps remaining on the bush immediately after harvest as described by Chandra Mouli et al. (Reference Chandra Mouli, Onsando and Corley2007). Results are summarised in Table 4. As expected, zero rise (B) was harder (fewer leaves remaining on the plucked stump) than the other treatments. Treatment C was similar to hand plucking, despite showing less table rise.
There were no deaths of bushes in any treatments, but with zero rise the appearance of the bushes was poor. The canopy looked sparse, because with no table rise there was little replacement of maintenance foliage except by shoots from unplucked origins (aperiodic shoots arising below the table surface). Measurements of LAI confirmed that there was less foliage with zero rise (B: mean LAI for both clones in 1997 = 3.5 ± 0.44) than with other treatments (Treatment C = 4.7 ± 1.34; hand plucking 4.6 ± 0.78). Treatment D, with 1 cm table rise every round, had the highest LAI (6.2 ± 0.78).
In the zero-rise plots, there was considerable die-back after plucking of shoots. Incidence was recorded in August 2000 and August 2001: 59 ± 3.6% of plucking points in clone 31/8 showed die-back, and 52 ± 2.8% in clone BBT133. No die-back was seen under the other treatments.
Dry matter production
DMP above the pruning height was greater with lighter plucking in both cycles (Table 5), with zero rise (B) giving the lowest DMP, and clone 31/8 tending to give slightly higher figures than clone BBT133.
Table 5 also shows estimates of HI. The HI figures are over-estimates of the true values, because fallen leaves and growth below the pruning level are excluded, but they should be correlated with true values. Zero rise (B) gave much the highest HI in both experiments, because of the reduced weight of pruning trash. In the second cycle treatment C gave a significantly higher HI and lower pruning trash weight than hand plucking, even though the difference in total DMP was not very large. Thus, it appears that hard plucking will tend to increase HI.
DISCUSSION
Yield under mechanical harvesting
In both clones, yield under hand plucking in the second cycle was greater than in the first cycle (Table 1). This may be at least partly explained by improved nutritional status. To evaluate yield trends under MH, annual yield can be expressed as a percentage of that under hand plucking. With a table rise of 1 cm every third round (Treatment C) there was an annual yield increase of 3.1 ± 2.9% for clone BBT133 and 4.3 ± 3.5% for 31/8. Thus, it is clear that with appropriate management of harvest interval and cutting height yields under MH can be maintained over eight years. There was no trend in yield under zero table rise (Treatment B) with clone BBT133, while for clone 31/8 there was a slight but non-significant downward trend (−2.5 ± 2.8% per year). Where table rise exceeded that under hand plucking, yield was reduced (Treatment D, rise of 1 cm every round), and this treatment was not continued in the second cycle.
The largest yields were obtained with an extended harvest interval and with table rise limited to no more than about 7 cm/year (Table 4). Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006) showed that extending the harvest interval from 2 to 4 phyllochrons increased yield by up to 25% when harvesting with shears. They also obtained an increase with a longer interval under hand plucking. The best treatment in our experiments gave yields 15–21% higher than from hand plucking, but we did not test longer intervals with hand plucking, and it is possible that a similar yield increase might have been obtained with hand plucking, if the harvest interval had been extended.
Harvesting without a table rise requires the shoot generations to be well separated, so that the next generation of shoots is much smaller than the generation to be harvested and is not damaged during harvest. The introduction of flexible timing for harvest and table rise in treatment C allowed better management of the shoot generations, avoiding damage to immature shoots and increasing yield, while maintaining a similar average rate of table rise.
Barbora et al. (Reference Barbora, Sarma, Barua and Barbora1993) obtained a 15–30% yield increase with MH, but cutting height and IoH were not described. Mwakha (Reference Mwakha1986, Reference Mwakha1990) obtained a large yield increase with seedling tea on a 70-day harvest interval, but a yield decrease with clonal tea. Madamombe et al. (Reference Madamombe, Tesfamariam and Taylor2015) found that yield decreased under a regime with table height raised 1 cm at every third round (equivalent to treatment C in the first cycle in our trials), despite a longer harvest interval with MH. The yield decrease was much larger in the first year of their experiment (42%) than in the second and third years (5–7%), suggesting that long-term damage was not the cause of the yield loss. We did not see a yield loss with treatment C, but in clone 31/8 yield was not increased until the introduction of some flexibility in 1999.
Madamombe et al. (Reference Madamombe, Tesfamariam and Taylor2015) obtained slightly lower yields with a hand-held machine than with a wheeled machine, but did not give table heights or harvest intensity figures. Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006) found that harvesting with shears gave lower yields than hand plucking at the same harvest interval. With shears or hand-held harvesting machines cutting height cannot be controlled with the precision that a wheeled machine allows, so damage to the next generation of shoots may be more likely to occur.
Components of yield
For both clones in the second cycle, shoot weight was greater for treatments B and C than for hand plucking, while shoot numbers were the same or lower (Table 2). The greater shoot weight can be attributed to the longer harvest intervals under MH. Table 2 shows a larger proportion of 3+bud and 4+bud shoots from MH than from hand plucking. The average shoot generation time was thus probably longer, with fewer replacement cycles per year, contributing to the lower total harvested shoot numbers under MH. Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006) also recorded increased shoot weight and reduced shoot number as harvest interval was extended. In our zero rise treatments, die-back would also have reduced harvested shoot number.
In contrast to these results, figures for shoot density on the bush before and after harvest in Madamombe et al. (Reference Madamombe, Tesfamariam and Taylor2015) indicate an increase in harvested shoot numbers from clone PC108 under MH, even though yield was reduced. We also recorded increased shoot numbers in short-term experiments with other clones. Harder plucking is expected to increase shoot number (Chandra Mouli et al., Reference Chandra Mouli, Onsando and Corley2007; Visser, Reference Visser1960), while extending the harvest interval will reduce number, so the net effect of changes in both intensity and interval on harvested shoot number may differ between clones.
Quality
Coarser plucking usually gives higher yields, but coarse shoots with four leaves and a bud also tend to give poorer made tea quality than less developed ‘fine’ shoots (two or three leaves and a bud). Mean shoot weight was greater and the proportion of 4+bud shoots was higher with the machine treatments (Table 2). Owuor et al. (Reference Owuor, Othieno, Robinson and Baker1991) showed that quality was reduced by extending the harvest interval for MH. They also found that quality was better with a 2 cm rise in cutting height between harvests than with a 1 cm rise. Thus, our use of an extended harvest interval with minimum table rise would be expected to reduce quality, but overall our taster found no significant differences in monetary value between MH and hand-plucked leaf (Table 3). However, the maximum number of samples per clone was three, which Corley and Chomboi (Reference Corley and Chomboi2005) found was insufficient to give a reliable estimate of value. Combining all nine sets of samples gave more replication, and showed no differences. Further work on this aspect is clearly needed, but an important point on quality is that the auction prices fetched by mechanically harvested tea from Ngwazi estate were very little different from the prices for hand-plucked tea from other estates in the same group (data not shown). Thus, in commercial practice, MH can give an acceptable tea.
The harvest intervals with hand plucking in our experiments were equivalent to 2 phyllochrons (Burgess and Carr, Reference Burgess and Carr1998), compared to 3–4 phyllochrons with MH. In other work on MH, longer harvest intervals have sometimes been used; for example, Mwakha (Reference Mwakha1986, Reference Mwakha1990) used intervals of up to 70 days, equivalent to 7–10 phyllochrons under Kenyan conditions. This should increase machine productivity, but is likely to result in more coarse leaf being harvested, which could also reduce quality if it is not separated from acceptable leaf. With harvest intervals of 42–70 days in Kenya, Mwakha (Reference Mwakha1990) found between 24% and 45% 4+bud and ‘hard pieces’ (mature leaf) with MH, compared to 6.5% with hand plucking. In our experiments, with shorter intervals, mature leaf and 4+bud shoots ranged from 17% to 25% with different clones under Treatment C, compared to 4% to 11% with hand plucking, but with all treatments there was only a small proportion of mature leaf (Table 2).
Intensity of harvest and bush health
Visser (Reference Visser1960) and Chandra Mouli et al. (Reference Chandra Mouli, Onsando and Corley2007) showed that harder plucking increases yields, and these results confirm this. However, there is a widespread view among tea planters that hard plucking can damage the bush, and in these experiments, under very hard harvesting with almost no table rise for eight years, bush appearance was poor, LAI was only 3.5, and there was die-back from plucked points. Despite these factors, yields remained above those from hand plucking, so the idea that hard plucking is harmful to yield is not supported. Visser (Reference Visser1960) similarly found that hard hand plucking continuously for 18 years gave higher yields than light plucking, despite lower weight of maintenance foliage. He noted that the life of individual maintenance leaves appeared to be extended on hard-plucked bushes.
The effect of plucking intensity can be seen from the trend of yield against table rise. A regression of yield loss, relative to zero rise for the same clone and cycle (from Table 1), against table rise (Table 4) showed a loss of 176 kg ha−1 yr−1 per cm table rise (r = −0.661*, 8 d.f.), with zero loss at about 6 cm rise. Data from Mwakha (Reference Mwakha1986; Reference Mwakha1990) indicate a loss of about 250 kg cm−1. Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006) found that when a step was added to the shears to control cutting height, a larger step gave greater table rise and lower yield. The yield loss per cm of table rise amounted to 187 kg cm−1 and 227 kg cm−1 for two different clones.
For clone 133 over the first cycle, shoot number harvested per year was greater with the smaller table rise of Treatment C (1506 shoots m−2) than with Treatment D (1172 shoots m−2). Visser (Reference Visser1960) also found that hard plucking gave a greater shoot number. Up to a point, therefore, harder plucking increases shoot number, but the shoot numbers with zero rise (treatment B: 1538 shoots m−2) were not greater than for treatment C. In the second cycle, similarly, shoot numbers for treatment B were not greater than for treatment C (Table 2). A possible explanation of this lies in the observation of die-back from plucked points in the zero-rise treatment B, but not in other treatments. With 50% die-back, potential shoot numbers with zero rise could have been as much as double the actual numbers shown in Table 2.
Dry matter production and limits to yield
Yield of tea is generally considered to be ‘sink limited’ – i.e., limited by the ability of shoots to use assimilates from photosynthesis, rather than by the rate of photosynthesis itself (Tanton, Reference Tanton1979). An objective of harvesting research should be to try to increase shoot numbers or weights to reach a point where yield becomes source limited.
The rate of DMP (DMP = pruning trash weight plus yield) was greater with lighter plucking (Table 5). Combining results from both clones and both cycles, regression of pruning trash weight against table rise showed an increase of 506 kg ha−1 yr−1 per cm rise in table height. Similarly, Magambo and Cannell (Reference Magambo and Cannell1981) observed that unplucked tea had greater DMP than tea under harvesting, and Visser (Reference Visser1960) found greater pruning trash weight under light plucking. These observations are consistent with the idea that yield is sink limited: lighter plucking allows more stem growth, which uses more assimilates than are required for shoot growth. Each centimetre of table rise resulted in 176 kg ha−1 yr−1 of crop loss (see above), and 506 kg ha−1 yr−1 more pruning trash, indicating the greater ‘sink strength’ of stems than flush shoots.
Under the hardest MH, treatment B with zero rise, yield was increased relative to hand plucking, but DMP was much lower than for other treatments. With zero rise, potential shoot number must have been greater than for other treatments: the actual numbers harvested were only slightly lower than from treatment C or hand plucking in the second cycle (Table 3), despite die-back of 50% of plucking points in treatment B but not others. Die-back was not restricted to these long-term experiments: it was noticed in experiments with clones BBK35 and TRFCA PC81 within less than two years after the treatments started. Similar die-back was observed on a field scale in South India when seedling tea was subjected to intensive harvesting. Die-back started to occur when yields exceeded about 600 kg ha−1 of made tea per month, and occurred at such yield levels under both hand plucking and harvesting with shears (K. Appachu, personal communication, 1990).
It seems that the very hard plucking has increased potential shoot number in treatment B, but at the same time may have damaged the canopy (low LAI) and thus reduced light interception and photosynthetic production. Burgess et al. (Reference Burgess, Carr, Mizambwa, Nixon, Lugusi and Kimambo2006) considered that an LAI of at least 4 was required to maintain yields, but mean LAI was only 3.5 for treatment B. The result is that source activity is no longer sufficient to support the growth of all shoots, and die-back starts to occur. Madamombe et al. (Reference Madamombe, Tesfamariam and Taylor2015) considered that the lower yield under MH was attributable to reduced photosynthetic production. They showed much lower light interception, indicating a sparser canopy, under MH, but did not mention die-back.
If yield was source limited with zero rise, die-back might be reduced and shoot numbers increased by harvesting with a TS of 2+bud, giving improved quality with little change in yield. Conversely, a larger TS would be expected to give no yield increase, but would result in more die-back and smaller shoot numbers. We hoped to investigate this by modifying Treatment B, but the experiments had to be terminated, because of machine breakdown, before any observations on die-back had been made, and without allowing much time for responses in shoot number.
Possible future work
The explanation of our results given above is that treatment B had greater potential shoot numbers than the other treatments, but that the canopy could not support the potential yield. An objective of tea research should be to increase shoot numbers to the point where yield becomes source limited, to make maximum use of intercepted solar radiation. However, in this case, source limitation may have resulted partly because the canopy was intercepting and using radiation less efficiently. Visually, the deterioration in the canopy appeared to be gradual, so a possibility would be to harvest with no rise for a year, and then to give a rise of, say, 2.5 cm to allow some canopy renewal (described by some tea growers as ‘adding a leaf’). This might allow shoot numbers to increase with treatment B, without die-back. However, Table 2 shows that the yield increase in the first year from treatment B was no greater than the overall mean increase.
Future trials should include a flexible table rise treatment, aiming at an average of 1 cm every three rounds (treatment C), as the best option so far. If, with zero rise, potential shoot number is increased to the point where there is source limitation and die-back, then reducing harvest interval to harvest finer shoots might give greater shoot numbers with less die-back, possibly improving quality without loss of yield. Different TSs should therefore be tested with zero rise, for long enough to allow differences in the amount of die-back to develop. Table rise and IoH should always be recorded. Several different clones should be included, as not all may respond in the same way. Intensive hand plucking as a control will give a good basis for comparison of clones.
CONCLUSIONS
The conclusion from eight years of experiments is that MH with an extended harvest interval can give sustained yields above those from hand plucking. The largest yields were obtained with a harvest interval about double that for hand plucking, and with table rise limited to below about 6 cm per year. The long harvest interval resulted in larger shoots for MH than for hand plucking, but a limited number of taster evaluations of quality of tea showed significant differences for only one clone.
Acknowledgements
We thank Unilever Tea Tanzania for permission to publish, and to an anonymous reviewer for many suggestions for improvement of the paper.
