Rhododendrons on Limestone: An Experimental Test of the Kinsman Hypothesis
Brian A. Carter, B.A.
Birmingham, England

Summary
In an earlier paper in this Journal (see Ref. 3), Dr. David Kinsman proposed the hypothesis that an organic topsoil overlying a limestone substrate would remain acidic as long as precipitation exceeds evapotranspiration, thus facilitating the growth of rhododendrons. The present study attempts to validate this hypothesis, and considers the processes by which an organic topsoil could become basified by the invasion of calcium carbonate solution. The potential role of peat as a protectant against such basification is discussed.

Introduction
In 1999, the Journal American Rhododendron Society reprinted an article by Dr. David Kinsman entitled "Rhododendrons in Yunnan, China - pH of Associated Soils" from the New Plantsman of March 1998. It is one of a small number of recent reports of observations of soil pH in the native habitats of rhododendrons in China. A copy of this paper was sent to the author in his capacity of editor of the publications of the Scottish Rhododendron Society (Scottish Chapter of the A.R.S.) following publication of a paper entitled "Non-Conformist Rhododendrons" by David Rankin in Newsletter No. 40, briefly outlining how his recent work seemed to refute current theories on how wild rhododendrons tolerate lime (see Ref. 5).

Over the last several decades there have been many suggestions put forward to account for the observations of healthy rhododendrons growing freely in limestone areas of China and the Himalaya, when the same species reportedly failed in limestone areas of the U.K. Few of these suggestions have had any status higher than that of a bright idea without supporting evidence, while those qualified to apply scientific investigative techniques have until recently overlooked the problem, perhaps regarding the Chinese observations as merely anecdotal. The recent return of Western botanists and horticulturists to previously closed areas of China and the ensuing field observations of rhododendrons in relation to their substrates have put the problem back on the agenda.

Dr. Kinsman's paper immediately took the writer's attention because he offered a hypothesis with supporting evidence which might resolve the problem. He hypothesized that the soil layer in which the rhododendrons grow in Yunnan is acidic, and will remain acidic whatever the nature of the underlying soils or bedrock, as long as precipitation exceeds evapotranspiration. He offered figures which suggest that this condition is fully met in normal years in a representative area of Yunnan, and in the wetter areas of the U.K. (along the western coast), but not in the drier areas of the U.K., such as the Thames Valley catchment where evapotranspiration normally exceeds precipitation between April and August. Excess evapotranspiration was assumed to permit draw-up of alkaline water which is traditionally presumed to be inimical to rhododendrons.

In "Growing Rhododendrons on Limestone Soils: Is it Really Possible?" by McAleese and Rankin in this Journal (see Ref. 4), it is suggested that the situation is in fact far more complex, with a number of species flourishing in near neutral to alkaline mineral soils in the limestone mountains of Yunnan. These findings do not, in themselves, invalidate Kinsman's hypothesis, since only a relatively small number of Rhododendron species have been observed in these conditions, and many other species apparently prefer more highly organic soils, such as may be observed in forested areas. However, it is important to bear in mind that there may well be more than one mechanism by which rhododendrons may survive in limestone areas.

The essence of a useful hypothesis is that it should be testable, and the Kinsman hypothesis is indeed eminently testable. The author, therefore, designed a series of experiments, utilizing some temporarily spare capacity in the facilities of the Civil Engineering Group of the School of Engineering and Applied Science at Aston University in Birmingham, England, to simulate the precipitation and evapotranspiration conditions throughout a full year in the Thames Valley and in Yunnan, as applied to an organic soil overlying limestone, utilising the climate data quoted in Dr. Kinsman's paper. In the case of Yunnan, these data represent a locality which may not be typical of the area, and Prof. Rankin suggests that precipitation outside the monsoon period would generally be less, but it is a convenient starting point for the simulation, since the winter deficit and summer surplus of precipitation representative of Yunnan is compared with the summer deficit and winter surplus of precipitation representative of the Thames Valley.

The details of the simulation required careful consideration. In nature both precipitation and evapotranspiration are extremely variable, as is the soil. This variability had to be reduced to the simplest cases because of lack of facilities and time, and the materials selected for their availability rather than their authenticity. The materials chosen for these preliminary experiments were a column of moss peat (derived from the decay of sphagnum moss), as widely supplied in garden centres and equally widely utilised by rhododendron growers in the U.K, superimposed on crushed Cheddar limestone, a hard, pure organic detrital limestone of lower carboniferous (Mississippian) age widely available in the U.K. for civil engineering applications. This was considered to be particularly useful because in the U.K. it is known as "Mountain Limestone," for topographic reasons, and in some ways it quite resembles the mode of occurrence in China. The chief differences lie in the highly developed "karst" features of China, a consequence of the extreme rainfall in monsoon conditions, and possibly in the Chinese limestone being a little softer, although the Chinese limestone is clearly durable enough to support much more extreme and impressive topographic relief than any that exists in the U.K. (where outcrops of more than 100 m. are rare.)

After some experimentation and thought it was decided to use freshly distilled water in all these simulations. This had a pH of 6.9, much less acid than the 5.7 or less of natural precipitation. This was justified by its reproducibility and plentifulness, and the inconvenience of producing otherwise pure water acidified to a standard value. It was considered that the acid produced by the peat would be strong enough to be relatively unaffected by the neutral pH of distilled water, and this can be confirmed by a simple calculation.

Equipment and Procedures
Two test cells were constructed out of thick-walled plastic tube, 145 mm ID by 460 mm long. One end was closed off and became the base, and a series of 8 mm diameter apertures were drilled at 50 mm intervals longitudinally, commencing at 150 mm above the base, the level of the interface between the peat and the limestone. The tube was filled to the base of the first aperture with 4 kg of crushed Cheddar limestone sieved to pass 5 mm and be retained at 3.36 mm, and distilled water added until it overflowed. Moss peat saturated with distilled water to the point where any disturbance caused water loss filled the remainder of the tube, and was gently compacted by gradually adding a litre of distilled water. All apertures except the lowest one were closed off with a strip of plastic insulating tape; free drainage was permitted only from the lowest aperture. (See Fig. 1.) The test cells were situated for the duration of the experiment in a controlled atmosphere room at a temperature of 20°C and a relative humidity of 60-65%. Another cell of the same diameter but without side apertures was filled with saturated peat and weighed daily as a reference for evaporation loss. After each weighing its weight was returned to a standard value with distilled water.

Figure 1. Two test cells with a series of apertures were constructed. The tubes were filled to the base of the first aperture with Cheddar limestone and the remainder filled with moss peat. Photo by Brian A. Carter

Each day, a measured application of distilled water was made to the test cells, simulating the daily excess of precipitation over evapotranspiration for the Thames Valley in cell 1 and Yunnan in cell 2 as derived from Dr. Kinsman's paper, with an additional amount to balance the actual evaporation as measured in the reference cell. At no point did the summer deficit in the Thames Valley simulation exceed the reference cell correction, which made it simple to subtract daily deficits from the reference cell correction. During deficit intervals the alkaline water in the base of the Thames Valley cell was replenished daily with alkaline water decanted from an agitated mixture of distilled water and limestone gravel.

The simulations took place in "real time," and at the end of each month the plastic tape was removed and the pH measured through each aperture with a simple metal probe pH meter which had been checked against standard buffers at pH 4.0 and 7.0. Although the pH meter was a cheap instrument mass-produced by Rapitest Ltd and sold in garden centres for simple horticultural purposes, it proved to be reliable and accurate to within 0.1 pH.

While preparations for the simulations were in progress, a far more simple but perhaps ultimately more impressive experiment was commenced. An elepidote hybrid Rhododendron seedling ('Tidbit' X yakushimanum ) was planted in 150 mm of peat in a 12" (300 mm) pot with a 150 mm layer of crushed limestone in its base. This was watered daily and fed with a weak proprietary fertilizer solution once a week during the growing season, and its progress observed.

Observations and Further Experiments
Even before the first pH determinations were made, the simulations yielded interesting observations. It is widely held in the U.K. that peat is very difficult to drain and therefore has a low permeability. However, water added to the surface of the saturated peat was absorbed in seconds, and appeared at the drainage aperture (300 mm lower in the column) in about two minutes, with drainage complete in less than an hour. This sample of peat, when saturated to its maximum stable moisture content, held 85% by weight of water (as a percentage of total weight) and was typical of U.K. moss peats. Possibly the high water retention (a consequence of the particle structure of the peat) had been confused with low permeability.

When the simulations were at the point where the midsummer precipitation deficit of the Thames Valley was in strong contrast with the monsoon precipitation of Yunnan, it quickly became clear that no invasion of the lower levels of the peat by alkaline water could be detected in the Thames Valley simulation. This situation needed further investigation, and two further experiments were quickly devised to try to throw more light on what was happening.

Two 350 mm lengths of 65 mm diameter rigid plastic tube were filled with moss peat at its stable saturated moisture content of 85%. One tube was closed at the base end and stood upright in the controlled atmosphere room and allowed to gradually dry. It was weighed once a week and after eleven weeks was extruded and a moisture content profile determined by sampling. The results are given in Table 2.

The second tube was left open at the base and stood upright in a container, held in place by a packing of 100 mm of crushed Cheddar limestone. Water was added to the limestone until it was level with the surface, and maintained at that level. Thus, as evaporation took place from the surface of the peat the opportunity to draw up alkaline water was maximised. After eleven weeks this specimen was also extruded, and both moisture content and pH profiles were determined by sampling. The results are given in Table 3.

Finally, when the simulations in the main cells had been run for a full year the daily water addition at the surface was terminated, and the water level in the limestone layers was maintained with additions through the lowest aperture of alkaline water for a full year, thus simulating a total drought for that period.

Results
The results of the main simulation are given in full in Table 1A and 1B. The pH readings at 50 mm above the limestone are presented graphically. They are quite unambiguous. The Kinsman hypothesis predicts that under conditions where precipitation exceeds evapotranspiration no upward movement of alkaline water can occur. Under these conditions, no upward movement of alkaline water could be observed. However, the converse of the prediction (implied as a necessary corollary by Dr. Kinsman) is that where evapotranspiration exceeds precipitation, alkaline water can move upwards. Under these conditions no such movement could be observed. There is no indication of significant basification of the peat by ingress of dissolved calcium carbonate from the gravel layer, and if basification occurs at all it must take place so slowly that any possible basic front would advance less than 50 mm in two years.

It is clear from the results of the further experiments that evaporation losses were not uniformly distributed throughout the specimens but took place predominantly from the surface layers. The lower levels of the peat layers were in effect largely insulated from the effects of the simulated drought which was desiccating the upper levels.

The final drought simulation showed little evidence of basification of the peat, and indeed for the first few months the peat layer gradually became a little more acidic. Since this has to be a worst case scenario in the context of these simulations this result is very significant.

Discussion
In view of the results of the subsidiary experiments, the results of the main experiments are not surprising. Saturated peat contains a very considerable reservoir of water. At a moisture content of 85%, a 300 mm layer of peat would be able to yield about 250 kg of water per square metre on evaporation to equilibrium. A substantial proportion of that reservoir would have to be evaporated in a prolonged drought before significant amounts of alkaline water could be drawn up, assuming that any is available in drought conditions .

Once the draw-up of alkaline fluids commences, the considerable availability of acid from the peat and its continuing replenishment by organic breakdown, coupled with the low concentration of alkaline salts in the drawn-up water, ensures that the peat will remain acidic for a considerable but as yet undetermined time.

We should give some consideration to how relevant the above simulations are to actual conditions in Yunnan and the Thames Valley.

The climate and geology in the Thames Valley are well known, and are known in broad outline for Yunnan. However, we have to recognise the importance of variability. The annual climatic cycles are in themselves far from reliable in both areas, and local control of precipitation and evaporation by topography and aspect is very important, particularly in Yunnan under the influence of the monsoon and the rugged mountains.

In Yunnan, plant collectors speak of "dry" and "wet" hillsides with their distinctive vegetation covers reflecting the orographic modification of the all-important monsoon. The valley bottoms vary with the seasons from dry to flooded, the flood waters sometimes carrying a suspension of comminuted carbonate rock material as well as dissolved carbonates (one assumes that this is because of the inclusion of glacial outwash in some cases), and no doubt where the flood waters don't reach, seasonal springs and seepages contribute to the availability of alkaline water where the local dip of the strata permits. The surface soil can vary from scree to an organic blanket, its thickness from millimetres to metres, and underlying it is the limestone bedrock, its apparent massiveness concealing the profound solution-enlarged joint pattern and bedding-plane cavities carrying the true water-table. In effect the limestone has little porosity but is highly pervious. The drainage of surface water into the limestone is likely to be very rapid, and above the bottom lands it should usually be one-way (see Ref. 2).

The Thames Valley limestones are very soft and porous, being mainly cretaceous chalk. The more moderate rainfall does not lend itself to regular flooding, and peat is infrequent. The chalk is extensively mantled by drift deposits, often very thick and variable on a metre scale, and, in the north of the area, the drift is an impermeable clayey till incorporating chalk. The chalk itself is extremely porous and an important aquifer. Droughts can occur in any month of the year but are rarely prolonged. Clearly, the simulation cells in such a restricted experiment will have only the most generalised relationship to reality. Peat in the form used is perhaps not typical in Yunnan and is decidedly uncommon in the Thames Valley, and a permanently saturated crushed limestone subsoil is not likely to occur frequently in the Thames Valley since it would be fairly rapidly dissolved away, although persistent reports of limestone debris in the soils of Yunnan suggest that it is either frequently replenished there, or the soils are very young. However, a beginning must be made somewhere, and the properties of peat are of some importance to rhododendron growers in the U.K. In fact it is in this latter area that the chief significance of the results will be found.

The most valuable contribution of the simulations is that they clearly illustrate the way that the presumed horticultural needs of rhododendrons can be met when planted in an area with shallow basic bedrock (for there are other rocks beside limestone which could yield alkaline water, e.g., ironstones and some volcanic and metamorphic rocks). This is not a theoretical point, for the more simplistic experiment referred to earlier in this paper has been an outstanding success. At the time of writing it is two and a half years since the seedling rhododendron was planted in pure peat overlying limestone gravel, and it is thriving while the peat is quite acidic right down to its contact with the limestone. There is no indication at present that this is a temporary stability and that in the long term the peat will become contaminated; indeed, observations indicate that the top of the crushed limestone layer has been acidified, perhaps by an organic coating on the gravel fragments which have become blackened. Further experiments on a range of rhododendron types has been commenced.

In view of the above results it becomes more than a little difficult to account for the problems widely believed to result from growing rhododendrons in organic soils in limestone areas, as quoted in many basic horticultural texts and specialist volumes. There is no reason to believe that such rhododendrons did not benefit from the conventional provision of an organic soil, with the use of peat being almost universal in the U.K., but the simulations described above strongly suggest that alkaline fluids do not invade the organic layer at all freely. It also seems possible that any such invasion that might take place would be of a temporary nature, subject to neutralisation by continuing decay of the organic matter and flushing by the passage of acidic rainwater. Furthermore, there is a mounting body of evidence that many more Rhododendron species and hybrids are tolerant of neutral or alkaline conditions than were previously thought and that some have a definite need or pronounced tolerance for such conditions (some specialist catalogues, such as that from Glendoick Gardens Ltd., now routinely indicate the plants that are tolerant or actually require added lime). On the other hand, it would be very premature to suggest that all rhododendrons will tolerate alkaline conditions despite research that indicates that calcium carbonate is not toxic (see Ref.1 ), and there is as yet no reason to doubt that many if not most rhododendrons have at least a preference for acidic conditions.

Assuming that the difficulties of rhododendron culture in limestone areas are real, rather than a myth out of the voluminous horticultural folklore, we are faced with a paradox which must be resolved. Assuming that rhododendrons are planted in soil which is initially acidic, it seems necessary to examine the processes by which basification of the soil could proceed. The writer has considered seven possible processes which could lead to the basification of an organic layer over limestone. They are:

1) Draw-up and diffusion of alkaline fluids from below (the traditional scenario) enhanced by the ability of the vascular system of the plants to draw water if there is any subsurface release of water from the roots.

2) Decay of litter from deep-rooted calcicole plants releasing calcium salts.

3) Bioturbation bringing up basic subsoils (e.g. worm casts, animal disturbance etc.).

4) Deposition of aeolian (windblown) base-rich dust from other areas with alkaline soil.

Then there are two processes which might be described as "semi-catastrophic" in that they are vigorous but dependant on temporary but not too uncommon extreme conditions. They are:
5) Lateral influxes of alkaline water as surface run-off or seepages during very heavy rain.

6) Vertical refluxes of alkaline water due to rapid deep penetration of torrential rain into profoundly sun or shrinkage cracked ground (or ground cracked by earthquakes).

Finally there is a suite of more or less catastrophic processes appertaining mainly to very steep ground in the wild:

7) Topsoil flow due to solifluxion, frost heave, landslips, earthquakes etc., plus instability of scree (talus) slopes below crags, not to mention rockfalls from the crags themselves, which in many photographs can be seen to bear recent scars. The effects of avalanches can be equally catastrophic with scouring and chaotic redeposition remixing more mature deposits.

Process group 7) presumably explains the descriptions of obviously metastable soils consisting of carbonate rock debris mixed with organic matter. However, these processes are of little relevance in a horticultural context and may be indistinguishable in the wild from relatively recently colonised ground, so they will not be considered further at this point, although they may be simulated in the production of garden soils, where builders rubble may simulate limestone scree!

The writer has no access to any experimental evidence appertaining to the validity and quantitative potential of these processes. Indeed it seems that little attention is paid in horticultural circles to any but process 1). It would appear that in a horticultural context most of these processes are not very significant and would be minimised by regular irrigation and intelligent design of the beds. More to the point, the amount of Ca2+ which must be introduced in solution in the basification of an organic soil must be quite considerable. Where this is transported in solution, which is known to have a saturated concentration of about 14 parts per million of Ca2+ in calcium carbonate, a most substantial amount of alkaline water must pass into the peat layer to neutralise it. Even if leaching out of the Ca2+ in normal conditions does not take place, basification of a peat layer must take a considerable time, even in very favourable circumstances. The most thought-provoking possibility is aeolian deposition of alkaline dust, which would appear to be a most effective method of basification of organic soils if sufficient dust is available on a regular basis. In the U.K. this process can be observed downwind of limestone quarries and cement works, but there is a possibility that it could operate on something like a regional scale in parts of China.

There is another possibility which must be considered, which is that the reported distress of rhododendrons grown above limestone subsoils in the U.K. is not due to the invasion of alkaline fluids, but has a secondary origin, such as deterioration of roots due to heat and water shortage in drought conditions exacerbated by the very rapid drainage of limestone areas. This does not seem unreasonable, but in any case does not affect the current investigation. It would certainly be useful to investigate this possibility.

Finally, we should consider an "escape clause," the possibility that a highly organic loam such as that provided by experienced horticulturists for growing rhododendrons will not behave in the same way as pure peat. Such loams would probably drain a little more slowly than pure peat, and would have a less extreme water retention, but whether these differences would hold any significance in the face of proper irrigation could only be determined by further experiment, and this is a matter which the writer is now investigating.

It has also become clear that peat itself must be investigated further. The considerable water retention of moss peat is presumably due to the "closed-cell" structure resulting from the non-vascular nature of the mosses, but even sedge peat has comparable water retention, as has leaf mould (which in the sample which the writer has commenced work on holds 67% by weight). Furthermore, the writer has noticed that the permeability of undisturbed sedge peat and leaf mould tends to be anisotropic, sometimes considerably greater in horizontal flow than in vertical, which may be due to the orientation of the original plant matter.

Further Investigations
Although the peat used in this experiment proved to be unsuitable in its properties to conclusively test the Kinsman Hypothesis in a reasonable time, this is compensated for by providing some insights into the value of peat in horticulture. The author has commenced a follow-up project in which similar simulations are being conducted using the very organic loam in which he grows a wide range of species and hybrid rhododendrons, the leaf mould which he routinely produces in his own garden, and for reference a locally available "sharp" silica sand. This may be the subject of a future paper.

Conclusion
The Kinsman hypothesis promised a simple and testable explanation for the occurrence of certain rhododendrons growing naturally in limestone areas of Yunnan, and for the reputed failure of those same species to grow well in the drier limestone areas of the U.K. The work described above, whilst only preliminary in nature, strongly suggests that Kinsman's proposed mechanism works effectively, without in any way impinging on more fundamental arguments as to whether this mechanism is necessary for the growth of many or most rhododendrons in the wild. Observed resistance of saturated peat to invasion from below by calcium carbonate solution strongly suggests that it could be used as a protective layer between a rhododendron planting medium and an alkaline subsoil, given an appropriate irrigation program.

Acknowledgements
The writer wishes to record his gratitude to Prof. R. Kettle, Head of the Civil Engineering Group of the School of Engineering and Applied Science at Aston University, for permission to use the controlled atmosphere room. Thanks are also extended to Dr. P. Hedges of the above School for a helpful discussion of the hydrology of the areas involved; to Prof. D. Rankin for information on his own observations and researches and his most helpful comments; and of course to Dr. D. Kinsman for the copy of his paper which sparked off the above research, and also for helpful correspondence.

References

1. Drehmel, G.; and W. Preil. 1992. Rhododendron und Immergrune Laubgeholze Jahrbuch 23, Bremen, Germany: Deutsche Rhododendron Gesellschaft.
2. Ford, D. C., and P. W. Williams. 1989. Karst Geomorphology and Hydrology
3. Kinsman, David J. J. 1999. Rhododendrons in Yunnan, China - pH of associated soils. Jour. Amer. Rhod. Soc. 53: 1.
4. McAleese, Anthony J., and David W. H. Rankin. 2000. Growing rhododendrons on limestone soils: Is it really possible? Jour. Amer. Rhod. Soc 54: 3.
5. Rankin, David. Non-conformist rhododendrons. Scottish Rhododendron Society Newsletter No. 40.