Sustainability research at Usutu Forests – Swaziland

July 29, 2014

In the following pages, I record the main events and, of course, results of Usutu’s world famous research into long-term productivity of pine plantations. I have written it as a narrative, to tell a story, which is largely chronological. My aim has been for an account accessible to those who don’t want every detail of the science. The latter has been well covered in over 30 scientific papers, book chapters and in the numerous consultancy reports to Usutu and SAPPI.
by Julian Evans

Prof. Evans with research assistant Milton Nkambule, monitoring tree growth at Usutu in Swaziland since 1968.
View over C4 in 1969. How to compare? At left, 11-year-old first rotation trees; right, six-year-old second rotation trees.


Why it began
In the 1960s, reports emerged of yield decline in pine plantations in Australia and New Zealand. It was labelled ‘second-rotation decline’. It created alarm in forestry circles around the world, raising the question: ‘Is intensive plantation forestry an unsustainable silvicultural practice?’ With some of the shortest pine rotations anywhere, Usutu’s patula plantations on poor highveld soils would be a prime candidate for exhibiting ‘second-rotation decline’ if it was a widespread problem. It is perhaps forgotten that when the first rotation was cut for pulpwood at Usutu after the mill opened in 1962, the rotation was only 12 years.

That, in essence, sets the scene. But the research I began at Usutu in 1968 had the additional backdrop of anecdotal evidence from Usutu forest staff. They felt that the young second rotation stands in A1, A12, C6, B12, D12 and D14, but especially the ones in Block A, looked ‘peaky’, somewhat chlorotic and thus suggestive of a decline problem.

One other significant comment needs to be made. The last major planting of the first rotation at Usutu was in 1957/8, when there was a gap of five to six years before second rotation stands began to be established in 1963. This period of five+ years with virtually no new plantations may have led staff to forget how very young stands appeared to the eye. This almost absurd remark actually resonates with experience elsewhere! All of us quickly forget what something is like, or only remember the best, and later when we’ve been alerted to a problem, we see signs of it even if it isn’t there. It’s just like worrying about medical symptoms, so often more in the imagination than reality.

This is the story of what really was happening. And it’s an important story because the research at Usutu is, arguably, the best maintained long-term productivity study of plantation forestry in the world. It has become the forestry equivalent of agriculture’s oldest still running recorded experiment, the famous Broadbalk Field trial in Rothamsted, England, where for 170 years successive crops of wheat have been grown continuously year after year.

Who funded it?
Initially, all that was envisaged was a piece of doctoral research (1968-1971) supported by UK’s then Ministry of Overseas Development; it was part of the vote for Swaziland immediately post-independence. Owing to the nature of the data, a follow-up visit was made four years later, in 1973, to re-assess second-rotation growth at 10 years i.e. mid-rotation. The British Government funded this too, and the British Forestry Commission granted me official leave of absence for two months.

However, what turned out to be the crucial three-month visit in 1977 that laid the foundation for all subsequent visits in later rotations, was my study leave allowance from the Papua New Guinea University of Technology. This was supplemented by some personal funds as part of research for a book entitled Plantation Forestry in the Tropics, as well as the Usutu work. I again contributed in 1984 when a short personal visit was made, after attending an IUFRO symposium in South Africa, to review the potential of re-establishing plots in the third rotation. This is reported in Usutu Forest Research Report No. 63.

UPCo and later SAPPI funded all subsequent visits wholly or very substantially, with the exception of a specialised mensurational study in 1997 supported by UK’s DFID. The years when visits and assessments were made are worth listing to display Usutu’s huge and sustained commitment: 1986, 1990, 1993, 1995, 1997 July, (UK DFID), 1997 Nov. (UPCo), 1999, 2002, 2003, 2005, 2007, 2009, 2011 & 2013.

For every visit, there is a report on file at Usutu.

Continuity of staff and research support
An unusual feature of this research has been the opportunity (and immense privilege) for me to continue the research throughout the 45 years. And alongside this is the fact that since the 1970s, Milton Nkhambule has assisted me on every visit with the on-site mensuration.

However, I must pay full respects to Usutu’s research team, led for many years by Andrew Morris, and to SAPPI’s researchers at Tweedie.

All this work could not have been achieved without the ongoing support of successive forest managers and their staff, beginning with Hora Wilhelmij in 1968 to Mandla Dlamini’s strong encouragement today. Just one anecdote: A couple of weeks after I arrived as a 22-year-old graduate, I was given some maps of Usutu, but they only showed the roads. Although I wouldn’t get lost, they were useless for providing the all important crop information of species and planting year and hence rotation. It took weeks of negotiation to be allowed to borrow a set of very small-scale stock maps, which required poring over with a magnifying glass.

The challenge – experimental approach
Two challenges confronted me when I began research in 1968. There were no mensuration sample plots that had been in first rotation stands that were now second rotation. The tiny network of 12 such sample plots had been deliberately retained beyond rotation age when the surrounding crops were felled. Indeed, they were still standing in 1968 and stood out like ‘a sore thumb’ in the landscape. Thus there was no opportunity to compare second rotation growth on a site where that of the previous stand was known accurately. Some sort of matching or pairing of sites between the two rotations was required.

The second challenge was that the oldest second rotation stands at the time were only five to six years of age, while the youngest first rotation ones still in existence were a 1957/1958 planting, i.e. 11 years old. This difference necessitated the use of stem analysis of the trees to discover their past growth so that comparison between the rotations could be made to exactly the same age. This was felt to be superior to attempting simply to apply a site index or yield class to the stands as a basis for comparison, but at the time, it was laborious. Little was it realised what a fabulous data set the approach would provide.

The need for site assessment
Because of the first challenge of needing to match sites in each rotation, the first year’s research was spent determining what site and environmental factors most influenced the growth of Pinus patula. This site assessment study was based on 61 temporary plots and, using selective multiple regression, showed that soil set, elevation of site, and distance from ridge top accounted for 60-70% of variation in tree height. This significant result meant that all matched pairs of plots were established with these site factors in common. Additionally, such matched pairs were always within the same forest block (A-E) to reduce the likelihood of management-related differences.
It was a good year’s work, but the crucial question still wasn’t answered: how was the second rotation performing compared to the first?

Creating the plot network
Armed with the site assessment information, 92 pairs of plots were established throughout Usutu across all five forest blocks, one plot of each pair in each rotation. This was done using Usutu’s excellent maps with their stocking and topographic data and a soil map put together for me by Swaziland Government soil surveyor, Dr George Murdoch. Once candidate sites were identified, I surveyed myself to the spot, checked the main site parameters, recorded the site and soil details, and established the plot. Every effort was made to ensure that the appearance of the trees did not influence plot location.

However, one limitation arose in this plot network: the plots were necessarily established in the oldest second rotation stands, namely the ones on the sites which were first felled for the mill – they were of p63 and p64 planting. Thus, the plots were predominantly in compartments near the main roads through the forest such as A12, B12 and 13, C4 and C6, D12 and 14 and E8. One result is that the plots are not representative of sites throughout the forest. This is of no consequence provided it is remembered: what they did very well was answer the question whether the second rotation stand was growing better or worse than its predecessor on a comparable site?
[Subsequently, this shortcoming of both compartments with plots and the restriction of just two planting years led to the establishment of additional plots, the XC series, which were mainly p67 stands and in different compartments.]

What was measured?
Plot area was about 0.07ha, which, at the standard spacing of 2.74 x 2.74m, gave 96 trees. Statistically, this was sufficient to allow a simple sampling system of selecting a tree of median DBH and two others, 30% above and 30% below the median – remember this was in the days before even hand held calculators! These three trees were felled for internodal stem analysis. Cross-sectional discs were taken from each internode and ring widths measured. In this way, stand growth history could be recovered to give height and volume development.

By the end of 1970, in all, a total of 550 trees were felled for stem analysis and my late wife, Margaret, measured in excess of a quarter of a million ring widths from the discs cut. And I learnt how to operate a chainsaw!

Stem analysis
In effect, internodal stem analysis recreated representative plot data for the first rotation. It was laborious and time-consuming but allowed the capture of growth patterns and productivity. Writing software in Algol 68 was no less a challenge to turn all the measurements in to underbark volumes and volume increments.

The stem analysis research helped in two other ways. First, it enabled a careful check of stand age by detailed examination of ring development on the bottom-most cross-sectional disc taken near ground level. Secondly, it identified irregularities such as in one compartment, completely missing annual height increments owing to severe hail damage or, more generally, years of poor or good growth usually attributed to variation in rainfall.

The unexpected initial results
I was in for a surprise: when the data of five or six years’ growth of second-rotation patula were compared with what the stem analysis showed for the matched pairs of first-rotation plots, the second rotation was growing significantly better. The second-rotation had established more rapidly.

Over the years, it has become clear why this was so. The first-rotation was established in grass veldt, which is highly competitive, while the second-rotation grew in weed-free, often clean burn sites. The trees simply got away quickly and the improvement was still measurable at six years of age.

But that wasn’t the only surprise; there were two more. Over part of the forest, predominantly in Block A, this improvement did not happen. Indeed, there was significant decline in productivity and, initially, it remained unexplained until the pioneering and elegant research of Andrew Morris revealed the cause.

The initial results threw up another surprise. Although early second-rotation growth was superior to the first-rotation, and significantly so, there was evidence that it wouldn’t last. Thanks again to the detail stem analysis affords, the intercept growth from ages three to five or four to six years was poorer in the second rotation. The impression was that the second-rotation started well but then slowed.

That is as far as the early research reached, and formed the basis of my doctoral thesis.

The intriguing intercept data led to a further visit in 1973 to assess whether the second-rotation really was slowing and, from my own curiosity, to a long visit in 1977. The latter was the visit that laid the foundation for all subsequent visits and research. Both merit brief comment.

Mid-rotation assessment at age 10 years – the 1973 visit
After finishing my PhD, I never expected to return to the network of 92 second-rotation plots. Their locations were only identified on maps – not in any other way. But as soon as they were looked for, more than three years after the last visit, tell-tale signs of the stem analysis logs each with a ‘hat’ of needles were readily seen or the old soil pit found. Indeed, the combination of memory, maps and mud led to 90 of the 92 plots being precisely relocated!

Milton accompanied me to all the sites and so added knowledge of their whereabouts.

At each plot, now nine or 10 years old, layout followed the same procedure: all trees measured for DBH and the three trees (median, & ± 30%) measured for height. Time did not permit full stem analysis, but at the suggestion of Spencer Brook, a sample of 60 trees was felled for detailed measurement to provide data for creating a local volume equation of the generalised form: V = d2H.

So what did we find? The initial improvement had disappeared and the strong block effect – poor growth in Block A – persisted. Overall, the second-rotation was still better but not significantly so, which I felt had been helped by three wetter-than-average years.

I was left with the lingering possibility that by the end of the rotation, the second rotation would indeed show a small decline. So when an opportunity occurred in 1977 to make an extended visit, I grasped it.
The 1973 visit also led to the establishment of 38 plots in the younger 1967 age class, the XC series.

The long 1977 visit
By mid-rotation the second-rotation had slowed compared to the first. It was essential therefore to see whether this disappointing performance persisted to the end of the rotation and so confirm the feared ‘second rotation decline’. Thus, when the Papua New Guinea University of Technology offered me a period of funded study leave, I contacted Usutu. The company provided a house in Mhlambanyatsi for three months in late 1977 and every second rotation plot was visited for an end-of-rotation assessment, actually made at either 13 or 14 years of age.

As well as mensuration, the significance of the long-term productivity plot network was becoming clear and so the location of every plot was surveyed in to a prominent feature – it was long before the days of GPS! The simple survey data using road junctions, very large rocks, watercourses and similar features (along with my and Milton’s memories), allowed relocation of the plots in the third (and later fourth) rotation to within 5m of the original plot centre.

But how had the second rotation done based on this final assessment? Overall, there were small but significant declines in height and volume growth, but these disappeared if the Block A plots were removed from the analysis.

Also, the mid-1970s were dry years and this too would have depressed growth.

Overall, the results did not look like the alarming yield declines recorded 10 years earlier in Australia and New Zealand, where in some cases, yield had collapsed and was one-quarter to one-third poorer. At Usutu, decline was a few per cent and most of it confined to one part of the forest.

The network today
Before moving on to consider the third rotation, it is worth reminding (as noted in a footnote earlier), that the large network of second-rotation plots became the backbone of all future assessments for the third and fourth rotations. They are the most important and best measured plots and are the XB series.

Two other foundations were laid. First, for consistency of comparison between rotations, identical mensurational procedures were followed, including the arcane use of median tree +/- 30% method for selecting trees for height assessment. Secondly, because the second-rotation had been measured three times at ages 5/6, 9/10 and 13/14 years, this was persisted with. It proved a blessing in that it increased visits to Usutu and so allowed me to cope with the vagaries of felling programmes by the company as they quickly became out of synchrony with research needs – a common problem!

Today, almost every XB plot has a GPS location and superimposed on or adjacent to the plot, SAPPI Research has established a long-term productivity plot conforming to their own standards and international guidelines. Also, all data collected has been stored in plot files, copies of which are held in Tweedie, at Usutu and by myself.

Usutu’s second rotation was disappointing, but not a disaster. Now the third rotation would provide an even better opportunity to investigate plantation sustainability because, for the first time, growth of a new crop of pines could be compared with the previous crop on exactly the same site. And not just on a handful of research locations, but across the whole forest in over 70 locations. A few plots had been lost to road building and other exigencies of management e.g. species changes, but the bulk of the second rotation network was available and was used to re-establish long-term productivity plots in the third rotation.

Owing to company felling plans, re-establishment of plots became spread out and across more age classes. As mentioned above, it was decided to stick with just three measurements in the life of each plot in early, mid and late rotation and, to match the ages, this was timed as closely as possible with the previous plot’s assessments.

The proposal to re-establish the plots in the third rotation was accepted by UPCo in 1984 after I had been invited to participate in an IUFRO symposium in Pretoria. A two-day feasibility visit to Usutu was sufficient to show that the laboriously collected data in 1977, providing survey data for the plot locations, would enable re-establishment to within 5m of the old centres.

Throughout the late 1980s and 1990s, visits were made every two to four years to measure the plots in the third rotation following the same protocols and conventions. Each visit lasted two to three weeks and, with Milton’s untiring help, I would assess between 25 and 35 plots. It is perhaps worth commenting that by my doing this, it allowed the responsibility for visit planning, measurement schedules etc. to remain with me and get done without compromising or competing with Usutu’s many other research priorities.

How did the third rotation perform?
The answer is very simple. The third rotation grew as well or better than the second rotation. A few plots were poorer but most were better and thanks to Andrew Morris’s research, the A Block decline was largely eliminated. Phosphate fertilising on Usushwana soils remedied the deficiency that had led to the yield decline.

We can be even more positive about long-term productivity because where poorer growth was found, it usually arose from poor silviculture such as low stocking per hectare, plentiful bugweed, poor respacing etc.

The bar charts of the third rotation
Over the great bulk of the forest, the growth was as in Figure 1. In Figure 2, from plots in Block A, the decline is stabilised. By the fourth rotation, it is no longer in evidence.

An intriguing question
The good third rotation result could conceivably arise because of slight changes in tree shape and form. The novel thought was: might one rotation’s trees require a different volume function for accurate estimation from those of another? This was specifically investigated in 1997 with funding from UK DFID.

The answer was ‘yes’ and ‘no’! The optimum volume function did differ in a small but significant way between rotations, but the shape of the trees hadn’t changed. The coefficients for lnHeight and lnDBH used in the local volume equations were not significantly different, it was the ‘constant’ term that was. This was a surprise and may reflect the poorer stocking in the third rotation or even a slightly stretched girthing tape(!) or similar mechanical factor.

The reason was never fully explained, but it did show the benefit of computing a new local volume equation for each rotation to achieve the greatest accuracy.

By the end of the 1990s, it was clear that the third rotation had grown well and no yield decline of any consequence was in evidence. Usutu could be proud of their plantations and their commitment to research – mensurational and silvicultural, but what about genetics? The fourth rotation would be the test bed.

It is believed that nowhere else in the world have four successive rotations of conifers been monitored continuously for their productivity, and at Usutu, where the new suite of assessments began just after the turn of the millennium, a new ingredient was present. For the first time, most of the new crop being established – the fourth rotation – used genetically improved material.

As an aside, it was always a puzzle that Usutu was relatively late in mounting active genetic improvement when the circumstances of its operations so readily lent themselves to this manner of productivity uplift. It may hark back to the time of Peter Germishuizen and the priorities of the then research advisory committee. But that said, the bulk of the fourth rotation was from selected or orchard grade stock. Would it yield the dividends hoped for?

As a reminder, long-term productivity plots were re-established as in the past and the same measurement protocols adopted. The same team carried out the work: myself and Milton, ably supported by Sabelo Khosa and others (including members of my family). Visits to undertake the assessments increased in frequency to every other year because of ever divergent re-establishment dates to ensure measurement at or close to the same ages as in the past. During more recent visits, plot locations were recorded by GPS and in most cases, SAPPI permanent sample plots were located on or adjacent to the long-term productivity ones.

Has genetic improvement been forthcoming?
My most recent report to SAPPI in April 2013 shows that the fourth rotation is doing well. The summarised results are reproduced below:

Analysis of all available plots (42) shows that the fourth (4R) rotation is growing significantly better than did the second rotation (2R). The improvement is around +7% for both height and volume per hectare.

Based on fewer plots (28) where valid comparison can be made, the fourth rotation is at least as productive as the third (3R) rotation and shows superior height and volume per hectare, but the gains are not statistically significant.  

There are reasons for believing that 4R assessments are conservative and that fourth rotation productivity is even better than the data suggests.

The old second-rotation decline problem in Block A no longer exists: stands in that block have responded well to the prescriptions research advocated.

Plantation forestry operations at Usutu are demonstrably sustainable: yields are being maintained or improved over time. They could be even better with attention to better stocking, better cleaning, weeding and singling operations, and conserving organic matter, underpinned by the continued use of genetically improved planting stock.

I believe that Usutu’s fourth rotation is an example of a forest-wide benefit of using genetically improved material.

Readers are reminded that the number of plots above on which the final fourth rotation comparison was made was lower than hoped owing (a) to the disastrous fires of 2008 destroying much of Blocks D and E as well as parts of Block C, and (b) switching species on a number of plots to Pinus taeda, so negating comparison of successive crops of the same species.

In reviewing 45 years of research, some aspects of the work and some observations during the many visits to Usutu remain with me because we could have done better!

Usutu could be even more productive as a forest with more attention to basic silviculture:

  • achieving uniform stocking on site. In too many cases, wide windrows, poor respacing and poorly managed regeneration has led to less than optimum stocking and stand productivity;
  • conserving organic matter during the harvesting and re-establishment window; and
  • rigorously controlling weed growth.

I could have done better and now regret the following:

  • failing to monitor soil changes over 45 years – a wonderful opportunity has been missed by focusing only on mensuration; and
  • failing to extend the long-term productivity plot network into the forest areas between Blocks D and E and the South African border. This had been mooted, possible sites have been identified on maps, but never implemented.

Fifth rotation assessments?
The long-term productivity research at Usutu has provided a world lead in investigating the sustainability of conifer plantation forestry. There is a strong research case for exploring every opportunity to evaluate fifth rotation performance where such stands occur. The very problems, fires and hail, which have caused such consternation at Usutu in recent years, provide such an opportunity as several fourth rotation stands have been felled and replanted to pine.

Such fifth rotation (5R) stands are to be found in parts of D12 and C22 and include XB and XC series plot locations. Other compartments e.g. D4, D14, D22 and possibly E8, may also now include some 5R pine where long-term productivity plots are sited. It seems probable that by 2016 or 2017, a sufficient number would be old enough to justify a short programme of plot re-establishment. The purpose would be more research focused: is good sustainability continuing into the fifth rotation? Additionally, it would certainly also keep up Usutu’s international reputation. I don’t know anywhere else in the world where there is a fifth rotation of pine and certainly nowhere where the growth of the previous four rotations are known!

A short research visit in 2017 or 2018 to initiate a 5R research programme, as well as perhaps evaluate the eucalypt plots on XB sites, could be rewarding.

XA series plots
For the record, the next time the 22 XA series of plots is due is about 2020-2022 – see 2005 Report concerning this interesting series. The XA plots are where some of my original first rotation plot sites have been assessed at end-of-rotation only. Second and third rotation measurements have been made, the fourth are due around 2020.

Why did the research keep going for 45 years. I think there are two key reasons. First, I was enthusiastic to know the answer to the productivity question and so took what opportunities presented themselves to keep plugging away. Secondly, Usutu has a long history of investment in research. It has been a privilege for me to team up with fellow scientists and researchers at Usutu to pursue this one question.
Usutu comes out well from this story – staff and forest. And, to be honest, I’ve enjoyed myself too!


The finest research support I could ask for. A young Milton Nkambule pointing out signs of three rotations.
Why rotations were short – the mill's demand for pulpwood.

*Prof. Julian Evans OBE is currently President of the UK Institute of Chartered Foresters and is author of some 16 books on forestry including very recently God's Trees, – Trees, Forests and Wood in the Bible, DayOne, Leominster, UK.

Published in April 2014

Notify of
Inline Feedbacks
View all comments
linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram