ABSTRACT
The effects of potassium (K) (0–160 kg/ha/yr) on soil and pasture were measured in five small plot mowing sites on flat and easy slopes in Hawkes Bay hill country from 2004 to 2007. Initial soil QT K levels ranged from 1.6 to 3.8 at each site and increased to a range of 3.5–11.3 at the highest rate of K in the third year. At most sites, clover cover on some occasions increased up to the highest rate of K. Despite this increase, there was only one significant response in pasture production to K at one site in the third year. Possible reasons for the lack of a total DM production response to K application were more grass growth compensating for less clover growth at the nil and lower rates of K or insufficient transfer of nitrogen from the clover to grasses at higher rates of K.
Introduction
Hill pastures where mainly sheep and beef cattle are grazed are grown on soils derived from sedimentary (Brown, Pallic, Melanic, Ultic, Semi-arid soil orders), volcanic ash (Allophanic, Granular) and pumice (Pumice) (Hewitt Citation2010). Potassium (K) is mainly required by legumes to fix nitrogen (N) and supply it to grasses. Most hill pastures are on soils of sedimentary origin where moderate weathering of clay minerals ensures an adequate supply of potassium for pastures. Pallic soils have appreciable amounts of hydrous micas and illite clay which release K for plant uptake (McLaren and Cameron Citation1994). In contrast, Pumice soils have K-containing minerals that are of large particle size and slow to weather so consequently supply little K for plant needs (Metson and Gibson Citation1974).
Toxopeus and Gordon (Citation1971) reported marked one-year responses in pasture production to 63 kg K/ha/yr on seven small plot mowing sites on Pumice soils where QT K levels were 5 or less. O’Connor and Gray (Citation1984) measured significant pasture production responses (10–37%) to K at three small plot mowing sites on Pumice soils with initial soil QT K levels of 4–5 over two years. Over four years at two small plot mowing trials with initial soil QT K levels of 2 and 4 on Pumice soils, Morton et al. (Citation2014) measured significant 28% responses in pasture DM production up to 150 kg K/ha/yr. In a long-term (13-year) trial on a sedimentary Brown soil on North Island hill country near Te Kuiti (average annual rainfall 1400 mm), the farmlet receiving 250 kg superphosphate/ha/yr and stocked at 15 ewes/ha had an average soil QT K level of 4.5 compared with the farmlet with no P and S applied and a stocking rate of 11 ewes/ha which had a soil QT K level of 5.9 (Ledgard et al. Citation1997). Over the trial period, soil QT K levels declined at a greater rate where 250 kg superphosphate/ha/yr was applied compared with no fertiliser applied. This trial under sheep grazing on all hill aspects and slopes showed a 0% and 21% response in pasture production respectively to 100 kg K/ha/yr on the 0 and 250 kg superphosphate/ha/yr farmlets (5% and 22% response respectively in legume production) (Ledgard et al. Citation1997). Officer et al. (Citation1997) found that soil QT K levels in hill country are greatly influenced by the distribution of animal excreta and differential weathering of soil minerals caused by spatial variability in soil moisture and plant uptake.
Because of the limited number of K response trials in hill country, a series of small plot mowing trials was undertaken to provide further information on fertiliser K requirements of hill pastures on the east coast of the North Island of New Zealand.
Methods
Sites
The trials were conducted at five hill country sites in Hawkes Bay. Three of the sites (Waipawa-flat, Puketapu-flat, Wairoa-flat) were on flat land of less than 15° slope and two of the sites on easy slopes of about 20° (Puketapu-easy, Wairoa-easy). The Waipawa and Puketapu sites were on Pallic soils and the Wairoa sites on Pumice soils. All sites were fenced-off from grazing. For the previous three years, the sites had been used for small plot mowing trials measuring pasture production responses to nitrogen (N) and phosphorus (P) fertilisers (Gillingham et al. Citation2007). Hence, especially on the flat sites, they had an atypically high legume cover, mainly white clover (Trifolium repens), (up to 50%) for hill country because of the close mowing reducing the competition from grasses and fencing-off from stock preventing selective grazing of clover. Subterranean clover (Trifolium subterraneum) was also present, especially on the previous plots that received low rates of P. The grasses in the pasture were mainly browntop (Agrostis capillaris) but also included perennial ryegrass (Lolium perenne), and sweet vernal (Anthoxanthum odaratum).
In autumn 2004, soil QT K levels at 0–75 mm depth were 2.7 and 3.8 from the Waipawa-flat and easy site, 1.8 from the Puketapu-flat site and 2.4 and 1.6 from the Wairoa-flat and easy sites. Corresponding site mixed herbage K contents were 2.0%, 1.9%, 1.6%, 2.1% and 1.7% respectively.
Average annual rainfall over the trial period was 869 mm at the Waipawa site, 826 mm at the Puketapu sites and 1348 mm at the Wairoa sites. Rainfall was above average at Waipawa and Puketapu in the first year (2004/2005) and in the second year (2005/2006) at Wairoa. At Wairoa, rainfall was below average in the third year (2006/2007).
Fertiliser treatments
In the first year, potassium chloride was applied at 0, 15, 30, 50 and 80 kg K/ha in July 2004 to plots of 3 m × 2 m in a randomised block design with four replicates. In the second and third years, these rates of K were doubled and applied at the same time so that a greater increase in soil QT K levels could be achieved. All plots received basal applications of P and sulphur (S) and were initially over sown with Pitau white clover.
Measurements
Pasture mass was measured using a Rising Plate Meter at 3–4-week intervals, or as regrowth required, when pasture mass was no greater than 2500 kg DM/ha in the plot with most growth. A standard calibration equation was used to convert plate meter readings into kg DM/ha. This method of measuring pasture mass was used by Gillingham et al. (Citation2007) and has the advantage over the traditional mowing technique of being less labour intensive. The pasture from the plots was then mown to a typical sheep grazing height of 4–5 mm with the clippings blown from the back of the mower on to the stubble so that about 95% of the total clippings were returned to all the plots. Although the more standard method of returning a proportion of the clippings to each plot would have been more accurate, this would have taken considerable time and as shown by the clover cover and pasture K content measurements, did not appear to greatly influence the effect of K.
Before mowing, each plot was visually assessed for clover cover using the same method as Gillingham et al. (Citation2008). Because all the assessments were carried out by the same operator thereby minimising between-plot and site variation, the relative values for clover cover would be expected to be reasonably consistent. Plots at each site had 15–20 soil cores sampled at 0–75 mm depth in autumn 2004, prior to the application of treatments in winter 2004 and subsequently in the spring of each year after treatments had been applied in July. Pasture was also clipped at 4–5 cm height from 15–20 sites in each plot in November of each year and the mixed herbage analysed for K content.
Statistical analysis
Analysis of variance was conducted on all plot data using Genstat. For the analysis of clover cover, the percentage clover in each plot at the start of the trial was incorporated as a covariate in the analysis.
Results
Soil quick test K
With the exception of the Puketapu-flat site in 2004/2005 and the Wairoa-easy site in 2005/2006, increasing the rate of K resulted in significant increases in soil QT K levels (P < .05) at all subsequent samplings (). In the first year, increases in soil QT K levels, although significant, were small, but in the second and third years the increases were larger.
Table 1. The effect of K application rate on soil QT K levels for each site in each year.
Herbage K content
In 2004, there was a significant increase in herbage K content as rate of fertiliser K increased at all sites except Waipawa-easy (). Herbage K content from control plots ranged from 1.77% to 2.34% at each site. In 2005 and 2006, herbage K content increased significantly with rate of K at all sites. Control plot herbage K content ranged from 1.77% to 2.56% in 2005 and 1.45% to 1.85% in 2006.
Table 2. The effect of K application rate on herbage K content in November of each year. LSD is least significant difference in herbage K content between treatments (P < .05).
Clover cover
There were no significant responses in clover cover to K at the Waipawa-easy site (results not shown) with low clover cover (<10% overall) (). At the Waipawa-flat site, only on three occasions was there a significant increase in clover cover to rate of K. Clover cover was most responsive to K at the Puketapu-flat site with significant, mostly large increases measured on nine occasions, especially in 2005/2006 and 2006/2007. At the Wairoa sites, the significant responses to K were also mainly large and measured on seven occasions for the flat site and six occasions for the easy site. All of the significant responses in clover cover to K were measured during spring and early summer when more soil moisture was available to the plants.
Table 3. Clover cover on measurements dates where significant responses (P < .05) to K application rate were measured.
Annual pasture production
In 2004/2005, the differences in pasture DM production between treatments were smaller than in the following years (). Despite larger differences at the Waipawa-flat, and Wairoa- easy sites in 2005/2006, there were no significant differences between treatments at any site. Differences in treatment DM yields were also large at the Waipawa-flat and Puketapu-flat sites in 2006/2007, but only significant at the latter site.
Table 4. Annual pasture production at each rate of K application for each site. LSD is the least significant difference in annual pasture production between treatments (P < .05).
Discussion
The most notable result from these trials was that despite the low soil QT K and herbage K content from the control plots, only at one site in one year (Puketapu-flat in 2006/2007) was there a significant response in annual pasture DM production to rate of K application. The Puketapu-flat site was also the most responsive to K in terms of clover cover. The small number of occasions where significant increases in clover cover to K occurred at the Waipawa-flat site and the lack of significant responses in clover cover at the Waipawa-easy site was matched by the absence of significant responses in pasture DM production. However both the Wairoa-flat and easy sites measured significant increases in clover cover to K on four to five occasions from the second to third years with no corresponding increase in pasture DM production.
Although there were only small absolute differences in soil QT K levels from the initial whole site measurement and the control plots between sites, these levels were all in the responsive range (2–4) where on average, pasture DM responses to K would be expected (Edmeades et al. Citation2010). The control plots at both of the Waipawa sites had soil QT K at the higher end of this range compared with the other three sites where levels were at the lower end. Therefore, at the Waipawa sites there was some relative degree of correlation between soil QT K and lack of response in pasture DM production to rate of K. At the Puketapu-flat site, a significant response in pasture production to K had occurred by the third year which was consistent with the low initial soil QT K level. However the Wairoa sites had the lowest soil QT K levels (2–3) and were on the lower K-supplying Pumice soils which other trials (Toxopeus and Gordon Citation1971; O’Connor and Gray 1984; Morton et al. Citation2014) had shown to be responsive to K fertiliser.
The conventional wisdom in grass/legume pastures is that fertiliser nutrients such as K are applied to provide for the less competitive legume which in turn fixes N for the growth of grasses (Morton and Roberts Citation2017). So in nutrient response trials on pasture, a nutrient response from clover should theoretically result in a corresponding response in total pasture production. Therefore the clovers must be allowed to grow to their optimal level so that they can supply as much N as possible to the grass and there must be efficient below-ground transfer of N from the clovers to the grasses through return of excreta or clippings (about 50% of the total transfer – death and decay of clover nodules and roots accounting for the remainder (Ledgard Citation1991)). Moreover, there needs to be a production measurement technique that reflects the contribution of both clovers and grasses to total pasture growth.
Grasses with their more erect growth habit can suppress the more prostrate clovers through competition for light (Brock and Hay Citation2001) and this could have occurred if the pasture was allowed to grow to excessive masses before sampling. In practice, the majority of the measurements were carried out at pasture masses less than 2000 kg DM/ha with very few greater than 2500 kg DM/ha, so shading of clovers by grasses should not have been a significant factor.
In these trials, nutrients including K were recycled through about 95% of the clippings removed in mowing from each plot being returned by redistribution to all the plots. During grazing, 70–90% of the ingested K and about 50% of the excreted N in pasture is returned unevenly in urine patches by sheep or cattle so the return of clippings attempts to simulate this process (Morton Citation1984). Brockman et al. (Citation1970) reported a similar response in pasture production to applied K under sheep grazing and mowing with no clippings returned over five years. In contrast, Morton (Citation1981) measured similar pasture production responses to K regardless of whether clippings were returned or not but no response under sheep grazing where there was lateral movement of urine K through surface runoff on a poorly drained Podzol soil under high rainfall over four years. The results of these two experiments question the necessity of having to return clippings in small plot mowing trials to simulate the return of urine K, at least in the short term.
However, especially if soil K reserves decline in the longer term, the cycling efficiency of K would be assumed to be more critical. Urine containing 0.92% K (Doak Citation1952) has a lower K content than pasture clippings (about 2% K) but is returned at a higher rate ((209–825 kg K/ha per urine patch) (Morton Citation1984) than in clippings (200–300 kg K/ha/yr). Over a two-year duration in the soil, urine K has been estimated to affect 26–44% of the pasture at stocking rates of 9–18 SU/ha (Morton Citation1984) whereas pasture clippings are spread over the whole area. Taking all factors into account, it could be concluded that the return of nearly all of the clippings at a lower rate of K would be as least as efficient as the uneven return of urine at a higher rate of K. Although Shaw et al. (Citation1966) reported that the amount of N transferred per kg of clover N was approximately doubled in grazed compared with cut swards, all of the herbage was removed under cutting compared with the majority being returned in our trials.
Because the pasture clippings were not returned evenly to each of the plots that they originated from, the control and low K plots possibly received clippings with more K than was contained in their herbage. This method could have had the effect of dampening the response in pasture DM production to rate of K application. Nevertheless, even if this occurred, there was still significant increases in clover cover during spring and early summer at four of the five sites. Also herbage K content increased significantly with K rate.
Pasture production responses to nutrients have traditionally been measured by mowing one or two strips from small plots, weighing the wet herbage and drying a sample to measure DM content so that pasture DM production can be calculated. However this method is labour intensive and because it is atypical of grazing conditions, it still only provides a measure of relative pasture production. An alternative method of using a Rising Plate Meter is sometimes used. This technique measures pasture mass with a denser sward providing a higher reading than a more open one. Grasses being more upright in stature and containing more fibre would present more resistance to the downward force of the plate than more prostrate growing, thinner stemmed clovers. This difference is reflected in grasses having a higher DM content than clovers. The standard calibration equation used may have not been the most accurate one for the sward type but would have been consistent in terms of relative yield. Relative rather than absolute yield as more accurately assessed under grazing is all that can be achieved where animals are excluded so use of the Rising Plate Meter should satisfy this condition as well as the strip mowing technique. Calibration of the pasture mass levels from the Rising Plate Meter using quadrat cuts would have been ideal but would have required much extra work for little gain in accuracy of relative yield.
In the likely absence of any of these three factors limiting the measurable growth of clover and the transfer of N from clovers to grasses, the lack of response in pasture DM production to K may have been due to the substitution of grass growth for clover growth. Where clover cover was lower in the control and low K plots, more grass grew so that pasture production was similar to that in high K rate plots that had more clover and less grass. Clover was found to be less compatible with browntop, which was the dominant grass in the trial swards, than other grasses (Frame Citation1990) and the greater presence of this grass in the control and low K plots could have increased the overall pasture DM production. Browntop has a dense, mat-like growth habit (Harris Citation1973) which could have elevated the Rising Plate Meter readings of sward density where it was more abundant in the nil and low K plots.
These trial sites had an artificially high clover content compared to what is normally present in grazed sheep and beef pastures. For three years before the application of the K treatments, the pasture swards had been regularly mown (Gillingham et al. Citation2007) which favoured the growth of the more prostrate clovers. Grazed pastures, especially in hill country have an inherently low clover content because of selective grazing and competition from grasses. This lower clover content in grazed compared with fenced-off, mown pasture was observed adjacent to the trial plots. Lambert et al. (Citation1986) failed to maintain a significant clover content in sheep-grazed pasture at Ballantrae Research Station with the application of 375 kg superphosphate/ha/yr over nine years. This decline in clover content even occurred in the early stage of pasture development where grasses lack N and clovers are more competitive for light.
The relationship between soil QT K and relative pasture production for the many sites on each soil group is inherently weak to moderate because of variability between sites and years (Edmeades et al. Citation2010). Therefore it is dangerous to use data from a small number of sites to extrapolate to all sites that measure the same soil QT K level and expect the same pasture production response. For both Pallic and Pumice soils in the large national database of K trial results (Edmeades et al. Citation2010), at soil QT K levels of 3 and less, a reasonable proportion of the site x year data points corresponded to relative pasture yields below 90% where the application of fertiliser K would have given an economic pasture yield response. But at several sites in some years the opposite occurred, where just as in these trials, there was a high relative yield at low soil QT K levels (Edmeades et al. Citation2010). Also, even at soil QT K levels of 4 or more, significant increases in pasture DM production were measured in some trials (eg. Ledgard et al. Citation1997). On farms, it can be misleading to use soil QT K level alone to decide whether fertiliser K is required and other techniques such as clover analysis and visual assessment of clover vigour should be employed.
In the first and third year, herbage K contents for the control plots were generally in the deficient range of less than 2% (Morton and Roberts Citation2017) from which a growth response to applied K would be expected. The increase in herbage K content with increasing rate of K application also indicated that pasture uptake of K also increased. In these trials, this response was represented in clover growth on some occasions but not pasture DM production.
In any pasture, clovers apart from fixing N also provide feed to stock with higher protein and soluble carbohydrate content (Ulyatt et al. Citation1976). Therefore even in situations represented by these trials, where K application resulted in more clover on some occasions but not increases in total DM production, benefits could be achieved from improved stock performance.
It could be argued that these trial sites being atypical of established hill pasture because of their artificially high initial clover cover were not the most suitable for testing. However in their initial development phase, hill pastures are clover-dominant and these swards do represent that situation. In other research a response in pasture DM production to K has not occurred until the second year of the trial (eg Morton et al. Citation2014) and if these trials had been continued for longer, the pastures may have become more responsive.
There have been a large number of K response trials carried out in New Zealand (Edmeades et al. Citation2010) but because more K fertiliser is required on flatter land where more intensive dairying is carried out, only a small proportion of these trials have been on hill country pastures. On higher K weathering hill soils such as Brown soils, soil QT K test more accurately predicts pasture production responses because of less reserve K. Although these trials on Pumice soils were an exception, soils from volcanic origin also lack K reserves and responses in pasture production to fertiliser K are also more predictable from the soil QT K test. Also these soils occupy less hill country than sedimentary soils. Therefore any further K response research on hill pastures needs to be on Pallic soils in which non-exchangeable K contributes more to plant supply and requirements for K fertiliser are more difficult to assess.
Conclusions
There were significant increases in soil QT K level and herbage K content from a low to marginal initial status at each trial site and increases in clover cover on some occasions from increasing the rate of K fertiliser at five seasonally dry hill country sites. But except for one site in one year, there was no corresponding increase in total pasture DM production.
It was considered that this lack of response to K in pasture DM production was probably caused by compensation for less clover growth by more grass growth at the nil and low rates of K.
This lack of response in pasture production despite low soil QT K levels for control plots was similar to some of the results measured from a large number of K response trials.
Nevertheless, soil QT K levels should be carefully used as an indicator of a pasture requirement for K along with clover K analysis and visual assessment of clover vigour.
Acknowledgements
The authors wish to thank Ballance Agri-Nutrients and Agrow Australia Ltd for originally funding these trials and the Fertiliser Association of New Zealand for funding this publication. Thanks are also due to Dennis Munro, Roger Alexander and Nathan Walter for access to their properties and Fred Potter, AgResearch for assistance and guidance with statistical analysis.
Disclosure statement
No potential conflict of interest was reported by the authors.
References
- Brock JL, Hay MJM. 2001. White clover performance in sown pastures: a biological perspective. Proceedings of the New Zealand Grassland Association. 63:73–84.
- Brockman JS, Shaw PG, Wolton KM. 1970. The effect of phosphate and potash fertilisers on cut and grazed grassland. Journal of Agricultural Science, Cambridge. 74:397–407. doi: 10.1017/S0021859600023042
- Doak BW. 1952. Some chemical changes in the nitrogenous constituents of urine when voided on pasture. Journal of Agricultural Science, Cambridge. 42:162–171. doi: 10.1017/S0021859600058767
- Edmeades DC, Morton JD, Waller JE, Metherell AK, Roberts AHC, Carey P. 2010. The diagnosis and correction of potassium deficiency in New Zealand pastoral soils: a review. New Zealand Journal of Agricultural Research. 53:151–173. doi: 10.1080/00288233.2010.482954
- Frame J. 1990. Herbage productivity of a range of secondary grass species in association with white clover. Grass and Forage Science. 45:57–64. doi: 10.1111/j.1365-2494.1990.tb02182.x
- Gillingham AG, Morton JD, Gray MH. 2007. Pasture responses to phosphorus and nitrogen fertilisers on East Coast hill country: total production from easy slopes. New Zealand Journal of Agricultural Research. 50:307–320. doi: 10.1080/00288230709510299
- Gillingham AG, Morton JD, Gray MH. 2008. Pasture responses to phosphorus and nitrogen fertilisers on east coast hill country: 2. clover and grass production from easy slopes. New Zealand Journal of Agricultural Research. 51:85–97. doi: 10.1080/00288230809510438
- Harris W. 1973. Why browntop is bent on creeping. Proceedings of the New Zealand Grassland Association. 35(1):101–109.
- Hewitt AD. 2010. New Zealand soil classification third edition. landcare research science series no. 1. Lincoln: Manaaki Whenua Press.
- Lambert MG, Clark DA, Grant DA, Costall DA. 1986. Influence of fertiliser and grazing management on North Island moist hill country. 2. Pasture botanical composition. New Zealand Journal of Agricultural Research. 29:1–10. doi: 10.1080/00288233.1986.10417968
- Ledgard SF. 1991. Transfer of fixed nitrogen from white clover to associated grasses in swards grazed by dairy cows, estimated using 15N methods. Plant and Soil. 131:215–223. doi: 10.1007/BF00009451
- Ledgard SF, Roach CG. Sowry SR. 1997. N, P, K research at the Te Kuiti Research Station. Proceedings of the New Zealand Fertiliser Manufacturers’ Association Technical Conference. 24:160–170.
- McLaren RG, Cameron KC. 1994. Soil science. An introduction to the properties and management of New Zealand soils. Auckland: Oxford University Press. 294 pp.
- Metson AJ, Gibson EJ. 1974. Chemistry and physical chemistry of potassium. soil groups of New Zealand Part 1. yellow-brown pumice soils: 88–90. Wellington: Government Printer.
- Morton JD. 1981. The influence of trial management on pasture responses to potassium on a Pakihi soil. New Zealand Journal of Experimental Agriculture. 9:271–277. doi: 10.1080/03015521.1981.10425425
- Morton JD. 1984. Spatial distribution of excreta and its effect on soil fertility. MAF Internal Report. 35 pp.
- Morton JD, Stafford AD, Hawke M. 2014. Potassium requirements of pasture on pumice soils. Proceedings of the New Zealand Grassland Association. 76:193–196.
- Morton JD, Roberts AHC. 2017. Fertiliser use on New Zealand sheep and beef farms. Wellington: Fertiliser Association of New Zealand publication. 48 pp.
- O’Connor MB, Gray MH. 1984. Fertiliser requirements of Gisborne-East Coast hill country. Proceedings of the New Zealand Grassland Association. 45:83–91.
- Officer S., Tillman R., Palmer A., Kirkman J. 1997. Potassium requirements of hill country soils in North Island, New Zealand. Proceedings of the New Zealand Fertiliser Manufacturers’ Technical Conference. 24:77–86.
- Shaw PG, Brockman JS, Wolton KM. 1966. The effect of cutting and grazing on the response of grass/white clover swards to fertilizer nitrogen. Proceedings 10th International Grassland Congress. p. 240–244.
- Toxopeus MRJ, Gordon RB. 1971. Rational fertiliser practice on pumice soils. Proceedings of the New Zealand Grassland Association. 33:35–41.
- Ulyatt MJ, Lancashire JA, Jones JA. 1976. The nutritive value of legumes. Proceedings of the New Zealand Grassland Association. 38(1):107–118.