Seeds of eight nitrogen-fixing species (Acacia saligna H. Wendl., Acacia sophorae (Labill.) C. Martius. Astragalus cicer L., Chamaecytisus palmensis (Christ) Bisby et K. Nicholls, Doryenium hirsutum (L.) Ser., Lathyrus latifolius L., Lotus pedunculatus Schk., Teline stenopetala Webb et Berth.) which tolerate the coastal sand dune environment when introduced as healthy plants, were broadcast-sown in 1995 at two sites on sand dunes which had been partially stabilised by planting marram grass (Ammophila arenaria L.). Establishment rates of 0-1% of viable seed sown after 6 months and 0-9% after 10 months, were considered to be unsatisfactory for sand revegetation purposes. Other trials at the two sites commenced in 1996 and compared the effects of three sowing methods on establishment rates of Lathyrus latifolius and Lotus pedunculatus. Lupinus arboreus Sims, once successful in this habitat but now attacked by a fungal disease before reaching full maturity, was included as a benchmark species. Sand accretion resulted in establishment failure at one site but at the other site L. arboreus provided satisfactory vegetation cover within 6 months (84-99% where seed had been partially or completely buried; 31% where seed had been broadcast-sown). Plot coverage by the other two species did not exceed 4%. Results of these trials do not explain why establishment rates were low in species other than Lupinus arboreus, but indicate that direct seeding is unlikely to be a satisfactory technique for establishing most nitrogen-fixing species on sand partially stabilised by the planting of a sand-binding grass. Enhancement of the chronically low nitrogen status of coastal sand in order to promote effective vegetation cover continues to depend on either frequent fertiliser application or the introduction of nursery-raised, nitrogen-fixing plants.

The aim of this study was to model the effects on the New Zealand forest sector of tariff elimination under the ASEAN Free Trade Area - Closer Economic Relations (AFTA-CER) and P5 regional trade agreements, and of tariff reductions under the World Trade Organization (WTO) administered General Agreement on Tariffs and Trade (GATT 1994). These scenarios were compared to a base scenario of no trade liberalisation except for the Closer Economic Relations (CER) agreement between Australia and New Zealand. All trade agreements were compared under three scenarios regarding growth in New Zealand's roundwood supply. The projections were made with the Global Forest Products Model from 1998 to 2015. The Global Forest Products Model is a spatial equilibrium model, which gives projections of production, consumption, and trade for each of 180 countries and 14 forest commodity categories. Results suggested the regional trade agreements (P5 and AFTA-CER) would have a limited effect on New Zealand production of forest products under all supply scenarios. The P5 and AFTA-CER would result in greater paper and paperboard production and exports, with an associated increase in fibre consumption. Tariff reduction under GATT 1994 had the most significant overall impact, increasing the production of all major processed commodities. In particular, the production and exports of paper and paperboard, and wood-based panels, were projected to be higher, and those of roundwood and pulp lower as their domestic consumption increased. From a policy perspective, this study suggests that if New Zealand wishes to increase domestic utilisation of its future roundwood harvest, supporting the tariff reduction initiative of GATT 1994 would be of greater benefit than tariff removal under the P5 or AFTA-CER regional trade agreements. According to the Global Forest Products Model, New Zealand has a competitive advantage in pulp, paper, and fibre production, rather than sawlog and sawntimber production as suggested by previous New Zealand studies. A possible explanation for this disparity is that the Global Forest Products Model does not disaggregate industrial roundwood into pruned, saw, and pulp logs. The Global Forest Products Model, therefore, ignores New Zealand's relative abundance of material for producing sawnwood and possible differences in the trend of pruned log, sawlog, and pulp log prices.

There are increasing requirements for forest management agencies to estimate not only wood volume for timber production, but also biomass accumulation and carbon sequestration rates of their forests for environmental purposes. The common methods of biomass estimation have been to develop allometric equations to predict the biomass of individual trees from diameter or both diameter and height. The biomass equations are usually based on small samples, especially for large trees, due to the time-consuming nature of destructive biomass sampling. Consequently, the predictive performance of biomass equations has been seldom evaluated. Most forest management agencies do, however, have reliable volume estimates that are based on large samples. Converting stem or stand volume estimates, that are already available in forest inventory and growth and yield systems, to biomass seems to be the most convenient and reliable way to estimate forest biomass over a large management area. Adopting this approach, we developed a system of additive equations for converting stem volume into four biomass components (stemwood, bark, branches, foliage) and total above-ground tree biomass using data for two Australian tree species as an example. To correct log transformation bias and at the same time maintain additivity among the component equations, we proposed a regression-based bias correction factor and simultaneously estimated the biomass correction factors for the component equations. The distributional properties of the error in stem volume prediction were incorporated in stochastic simulations of the system of equations to determine the confidence bands of the biomass conversions. Such results would provide a clear indication whether the required precision of biomass prediction is met for a particular objective of investigation and, if not, where improvements can be made

The effect of harvesting logistics on the fuel characteristics of non-comminuted Picea abies (L.) Karsten (Norway spruce) logging residues from regeneration cuttings was studied in Southern Finland. Roundwood harvesting took place in the summer of 1999. Green residues were forwarded immediately to roadside storage areas for use during winter 1999-2000. Part of the material was left on the felling sites to dry for 4-8 weeks, after which the brown residues were also forwarded to the roadside for storage. Drying/ storage on the clearcut area continued into the fall. Altogether 12 roadside storage piles were made by mid-October 1999, half of them covered. Moisture, needles, and ash contents as well as element composition were analysed at the beginning and during the experiment.

Drying and storage conditions were exceptionally good during summer 1999. The cumulative evaporation of June through August was 150 mm, which was twice as high as the cumulative precipitation over the same period. As a result, the residues in roadside piles reached 22% moisture content in just 4 weeks. At the same time, the residues on the clearcut area in harvester-made heaps fell below the 20% mark, after which drying was much slower. With the approach of the fall and winter, the residue mass started to absorb moisture. As long as the weather was favourable, covering the piles did not seem to have much significance. However, with the increased precipitation, first rain and then snow, the uncovered roadside piles rewetted at a faster rate. This applied to both the green and the brown residues.

The fresh logging residues contained 19.1% needles. Seasoning the fuelstock lowered the percentage to 1.8% on the clearcut area and to 4.0% on the landing. This loss lowered the ash content from 2.1 % to 1.5% during the first month of storage. The concentrations of nutrient elements were correlated with the ash content. Seasoning helped to lower the amount of chlorine from 200 mg/kg to 50 mg/kg.

Experiments were carried out to determine the effect of quarantine heating conditions on temperature distribution inside logs of varying diameter. When 20-cm-diameter Pinus radiata D. Don logs were heated using a 3.5-hour ramp from 20°C to 65°C air temperature and this air temperature was maintained, internal temperatures at 10 cm depth inside the logs reached an average of 45°C 6 hours from start. This temperature had previously been shown to be lethal to huhu (Prionoplus reticularis White) larvae treated outside logs. Mortality of all huhu larvae heat-treated in logs was achieved using the 6-hour heat treatment. Huhu mortality in the insect-infested logs may have been further aided by the fact that internal temperatures continued to rise for more than 1 hour after treatment was completed, and that the average temperature remained above 40°C for more than 3.5 hours from completion of treatment.

As log diameter increased, longer treatment duration was required. The treatment time to reach 45 °C at 10 cm depth had to be extended from 6 hours for a 20-cm-diameter log to an average of 10.5 hours for a 30-cm log and 13.5 hours for a 40-cm log.

Heat treatments may be cost-effective compared to the currently used methyl bromide disinfestation treatments, and could also improve wood quality (reducing the incidence of surface checks, warping, or discoloration).

Laboratory and field studies were conducted to investigate the potential of dairy products and integrated systems, containing fungi and biocides, for controlling fungal invasion of Pinus radiata D. Don by wood-inhabiting fungi. In laboratory and field trials, branches and log billets of P. radiata were subjected to various experimental treatments prior to exposure to fungal infection. After 4 and 9 weeks of incubation, billets were visually assessed for internal discoloration. In the laboratory, satisfactory control against staining fungi was obtained with integrated systems containing Trichoderma viride Pers.: S.F.Gray plus sodium fluoride, T. viride plus sodium fluoride plus Sentry«, and a suspension of Camembert cheese rind in pasteurised milk. In the field, T. harzianum Rifai or T. viride plus sodium fluoride plus Sentry« controlled detrimental discoloration in log billets exposed to natural fungal invasion.

The location and size of branches in the tops (main stem diameter < 20 cm) of 16 Pinus radiata D. Don trees (26 years old) were recorded, and phyllotactic patterns in each branch cluster were described and analysed. Branch position and size within the cluster were examined in relation to ontogenetic sequence. Accurate three-dimensional branch locations were used to analyse branch azimuth and branch vertical position. Divergence angle, branch vertical position, and branch diameter within a cluster were all shown to be strongly related to position in the ontogenetic sequence. A mathematical model incorporating relationships between branch location, branch size, and phyllotactic pattern was constructed for the prediction of branch arrangement and development in P. radiata. The model has potential for forestry applications, particularly those involving log and timber grading, where reliable branch and knot size data are required.

Results from a trial at Rotorua, on a moist, moderately fertile site, suggested that tall fast-growing weed species reduced growth of Pinus radiata D. Don by restricting availability of light to tree crowns. Data from this experiment were used to model the height growth of the trees relative to neighbouring weeds to create a simple shade index. The model demonstrated that reductions in tree growth could be expressed as a function of this simple index of shading. However, the index was not independent of weed species or age, possibly because of the lack of a term in the model to describe weed abundance, and future studies will test this. The data showed that tree diameter growth was reduced by less than 10% when weeds were 50% tree height, and by about 30% when the weeds were 75% tree height. Thus, an appropriate time for management intervention to minimise tree growth losses would be when weeds exceed 50-75% tree height.

Models to predict herbicide spray deposition within plant canopies can help to define application characteristics which result in the desired distribution of herbicide deposits and minimum contamination of non-target surfaces. A modelling approach was used to explain the interception of large herbicide droplets in terms of foliage structure, using the canopies of bracken fern (Pteridium aquilinum L. Kuhn.) and greenleaf manzanita (Arctostaphylos patula Greene) as examples. Analysis demonstrated that an effective canopy interception model must account for both the distribution and the quantity of foliage so that the gap fraction can be quantified. Within a bracken canopy, predicted spray interception profiles were in close agreement with actual data. Predicted spray interception profiles in a greenleaf manzanita canopy over-estimated the observed rate of deposition, possibly because of low droplet retention.

The hypothesis that Pseudotsuga menziesii (Mirb.) Franco (Douglas-fir) is more windfirm than Pinus radiata D. Don was tested by examining the factors which affect tree stability and comparing available data for each species. A tree-winching study performed on Douglas-fir trees of different sizes growing on Orthic Pumice soils found that their maximum resistive bending moments were not significantly different from those of P. radiata (p > 0.05). A literature search was performed to determine the relative overturning bending moments of the two species. For each species, information was obtained on the drag coefficient, the crown frontal area, and the mass of the stem and crown. A mechanistic model for predicting the threshold hourly mean windspeed at which damage occurs was applied to a baseline stand of each species (mean top height 30 m, dbh 40 cm, and stocking 310 stems/ha). The Douglas-fir stand had a threshold windspeed of 24.3 m/s and the threshold windspeed for the P. radiata stand was 20.6 m/s. A recurrence function, fitted to extreme windspeed data from Rotorua airport, was applied to these critical windspeeds and showed that the risk of damage to P. radiata was 2.3 times greater than that to Douglas-fir. Over a period of 90 years, the analysis predicted that a P. radiata stand grown on a 28-year rotation was 3 times more likely to suffer catastrophic wind damage than a Douglas-fir stand grown on a 45-year rotation. The most critical factor behind these differences was the lower drag coefficient of Douglas-fir foliage.

The six fungi described in this paper have not previously been recorded in New Zealand. The fungi are: Caulicolous Mitosporic fungi, Coelomycetes: Cryptosporiopsis edgertonii nom.nov. on Acer davidii, Chamaecyparis lawsoniana, Eucalyptus regnans, Ilex sp., Liriodendron tulipifera, Nothofagus antartica, N. fusca, N. solandri, Podocarpus hallii; Foliicolous Mitosporic fungi, Coelomycetes: Kabatina thujae on Chamaecyparis lawsoniana, Thuja plicata; Pestalotiopsis stevensonii on Abies sp., Pinus edulis, P. jeffreyi, P. radiata; Phyllosticta concentrica on Corynocarpus laevigatus; P. spinarum on Chamaecyparis lawsoniana, Cryptomeria japonica, Cupressus arizonica, C. macrocarpa, Juniperus chinensis, Thuja plicata; Septoria alnifolia on Alnus rubra.

Land application of wastewater in New Zealand is becoming more common due to resource management legislation and the cultural preferences of the indigenous people. Impacts of municipal wastewater irrigation on some physical properties of two sandy soils under Pinus radiata D. Don forest were measured on a site of sand-dune origin (Waitarere) and one of volcanic origin (Whakarewarewa) which received primary-treated and tertiary-treated wastewater, respectively, at a rate of 60 mm/week. Intact soil cores were collected at 0-100 mm and 100-200 mm from wastewater-irrigated and non-irrigated control sites after 4 years of irrigation. At the Waitarere site, irrigation with primary-treated wastewater significantly decreased the bulk and particle densities and unsaturated hydraulic conductivity, but increased total porosity and macroporosity. However, there was no change in saturated hydraulic conductivity. At the Whakarewarewa site, irrigation with tertiary-treated wastewater did not change bulk and particle densities, total porosity, water retention, and hydraulic conductivity but significantly increased soil macroporosity. Wastewater quality and, possibly, soil origin appear to play important roles in the changes in soil physical properties observed under wastewater irrigation.

The long-term impacts of repeated prescribed-burning on surface (0-30 cm) soil chemical properties and forest floor nutrients in two contrasting native eucalypt forests in south-east Queensland, Australia were assessed. Most effects were confined to the topsoil (0-10 cm). At the dry sclerophyll site, more than 40 years of annual burning had not led to any loss in topsoil total nitrogen and organic carbon but had significantly increased topsoil acid-extractable phosphorus. At the wet sclerophyll site, biennial burning (10 burns) resulted in decreases in topsoil total nitrogen and potentially mineralisable nitrogen despite a shorter experimental period (22 years). The increase in acid-extractable phosphorus with burning on this latter site was not significant. The varying impacts of prescribed-burning on soil chemical properties at the two sites may be partly explained by differences in site quality and in the number of times the forests have been subjected to various burning regimes. These changes may have implications for tree growth in the long term. Repeated burning reduced the weight, organic matter, and nutrient contents of the forest floor at both sites, the decreases being directly proportional to the frequency of burning.

Nothofagus solandri var. cliffortioides (Hook.f.) Poole and N. menziesii (Hook.f.) Oerst. are both widely distributed in New Zealand, occurring naturally from sea level to the upper timberline. Despite the similarity in the distribution pattern between the two species, there are some areas where the natural stands of forests are either represented, or dominated by only one of the two species. To test the hypothesis that differences between N. solandri and N. menziesii in mineral nutrition may play a role in the localised distribution differentiation, we conducted two experiments examining the growth response of seedlings to the supply rates of five major nutrients, i.e., nitrogen, phosphorus, potassium, calcium, and magnesium, using the techniques of perlite and solution culture in a glasshouse. Results showed that N. solandri seedlings are less tolerant to supply limitations of phosphorus, magnesium, and calcium than are N. menziesii seedlings. These differences were related to the natural distribution of the two species.

Static bending tests were conducted on small clear timber specimens from 20 trees of each of four native provenances of Pinus radiata D. Don: An?o Nuevo, Monterey, and Cambria from mainland California, and Guadalupe Island. The specimens came from ?5 m up the tree, with c. 24 rings from the pith. Individual specimens were cut from five equidistant positions from pith to bark, giving 400 specimens in all. Density (unextracted), modulus of elasticity (MoE), and modulus of rupture (MoR) were recorded for each specimen. Density had the lowest coefficients of variation and MoE the highest. Provenances differed appreciably in density but not quite consistently across positions; while Guadalupe ranked top throughout, Cambria ranked lowest near the pith and An?o Nuevo lowest towards the bark. MoE and MoR showed the same sort of pattern, albeit less pronounced in MoE. MoE was strongly correlated with density near the bark, but towards the pith the correlations were weaker and regressions on density less steep. MoR was even more strongly correlated with density, and with little pith-to-bark trend in regression slopes. Pith-to-bark increases were strong in all variables, and were evident in MoE and MoR even when adjusted for covariance on density. MoE in particular showed some negative skewness about its regression on density, especially near the pith, and this was believed to reflect compression wood. After adjusting for covariance on density, provenance differences in MoE and MoR were negligible, but significant differences remained among individual trees across positions.

Comparisons of split-taper and no-taper sawing methods were based on the simulated sawing of 300 pruned logs of varying size, shape, and quality. Sawn timber conversions, Clears grade and Combined Clears grade conversions, and gross log value arising from each log were calculated for both methods and compared using paired t-test statistics. In general, significant gains (p < 0.01) were provided by the split-taper method. Sawn timber conversions averaged 64% (range 52-70%) for the split-taper method and 62% (range 48-70%) for the no-taper method. Conversion to Clears and Combined Clears grades was significantly greater with split-taper sawing of well-pruned logs (PLI ? 6). Overall, conversion to Combined Clears yielded an average conversion of 36% (ranging from 18 to 54%) for split-taper sawing and an average of 34% (ranging from 16 to 51%) for the no-taper method. Gross log value, calculated as timber value per cubic metre of round log volume, was also significantly increased with split-taper sawing. On average the split-taper method yielded $246/m3 whilst the no-taper method yielded $237/m3. This equates to an average increase in conversion value of 4% (when measured relative to no-taper sawing). Conversion value increases ranged from nearly - 14% to 27%, indicating that, although significant gains can be made with split-taper sawing, gains cannot be expected from all logs

An analysis of log and branch variables influencing timber grade outturn was performed using the AUTOSAW sawing simulator. Data from 115 unpruned Pinus radiata D. Don logs were used in the study; 85 logs were extracted from a plantation forest (Kaingaroa Forest) and the other 30 from a farm site (Tikitere Agroforestry Research Area). Each log had been intensively measured and reconstructed as a virtual 3D log model inclusive of all branches. The virtual logs were repeatedly sawn with the sawing simulator over a range of rotational settings and product types. Virtual boards arising from the simulations were graded according to clear cuttings and visual structural grade requirements. Simulated timber grades were correlated with indices reflecting log and branch characteristics. The sawn percentage of No. 2 cuttings and better grades (%2Cuts+) was correlated with combinations of traditional predictive variables and indices such as small-end diameter (SED), internode index (IIX), and mean internode length (MIL). New measures of internodal lengths, such as a fundamental internode length (FIL) that ignores whorls with three or fewer branches, were also trialled. FIL was developed on the premise that clearcuttings can be recovered that pass through whorls, especially when the branch count is low. Models for predicting the sawn percentage of No. 1 Framing and better grades (%1F+) were correlated with Branch Index (BIX) and other log variables. Newly developed indices based on the sum of branch diameters within a whorl, such as WGB that refers to that whorl with the greatest sum of branch diameters, and knot size ratio (KSR), were fitted to both linear and non-linear regression models. FIL gave higher correlations than IIX and MIL, when fitted using linear models, and in practice should be easier to calculate than MIL as whorls with three or fewer branches can be simply ignored. With non-linear regression models, the combination of IIX with SED gave the highest correlation. For predicting structural grade conversions, WGB gave higher correlations than BIX for both linear and nonlinear models. Almost as good a fit was achieved using WN4, which summed the branch diameters in the whorl with the most branches of 40 mm and over. In practice, WN4 is likely to be easier to use than the WGB. KSR was also effective in predicting sawn percentage of No. 1 Framing and better. These results suggest that there is one whorl that has a significant influence on sawn percentage of No. 1 Framing and better grades.

Breast height pith-to-bark 10-mm increment cores were collected from eight individual stems from each of 10 families of Pinus radiata D. Don (+ control seedlot) at seven sites after 22 years of growth. The sites ranged from Taringatura, in the deep south of New Zealand (lat. 46¦S; mean annual temperature 10.1¦C), to Woodhill (lat. 35¦S; mean annual temperature 14.6¦C), and covered an altitudinal range of 650 m. The cores were resin-extracted and scanned in a wood densitometer to yield a continuous density trace across all annual growth rings. Sample data were analysed by family and site to determine the influence of genotype and environment on growth ring density components. There were significant site effects for both growth rate and wood density, which broadly corresponded to expected patterns, but little correlation between the two variables. Average wood density at Woodhill reached "mature" levels (defined as 400 kg/ m3) well before age 10 years whereas the southernmost site, Taringatura, failed to reach this level even after 20 years. Percentage latewood proved to be the most sensitive variable in both juvenile and mature wood, and was reflected in the site values for average wood density. Site differences in earlywood and latewood density were comparatively small. Family ranks for density traits averaged across all sites maintained a high degree of consistency and moderate-to-high heritabilities, but analyses of variance also revealed a significant GxE component. The range in family density means of about 50 kg/m3 remained consistent from pith to bark across sites. There were large differences in the ability of individual families to reach "mature" levels of wood density. Families varied in the period required for mature wood initiation (8 to 20 years).

Statistical techniques for assessing the evidence of heritability (h2) of a highly non-normal random variable from limited-sized datasets were applied to internal checking in Pinus radiata D. Don wood. Bayesian hierarchical models for ordinal logistic regression with and without random family effects were fitted and compared using the technique of pseudo-priors to estimate the Bayes factor. Model parameters and Bayes factors were estimated using Gibbs sampling, and implemented using the computer program BUGS and importance sampling. Bayes factors of 203, 41.7 were obtained for the total number of checks and number of rings with checks, respectively. In the model assuming non-zero heritability, estimates and 95% credible intervals for h2 were 0.64 (0.15-0.996) and 0.60 (0.06-0.997) respectively. In contrast, non-Bayesian methods including ANOVA with transformed variables and a non-Bayesian ordinal logistic regression analysis, failed to detect any effects for p<0.05

The lymantriid forest defoliators, Lymantria monacha L. (nun moth) and Lymantria dispar L. (gypsy moth) are particularly severe pests in other countries in the world, but the ability of these moths to utilise and complete development on Pinus radiata D. Don had never been established. In laboratory trials, colonies of central European L. monacha and Russian far east (flight capable) L. dispar were fed on foliage from three mature P. radiata trees originating from three different advanced-selection families from New Zealand. The results showed that both moths were capable of completing their development from egg to adult on these families of P. radiata. However, P. radiata was a less suitable host for development of L. dispar than Quercus velutina Lam. (black oak), as evidenced by higher mortality and slower growth. Lymantria monacha developed faster and survived better on P. radiata than it did on mature foliage of Picea glauca (Moench) Voss. Neonate L. monacha larvae favoured male pine cones from Pinus strobus L. as a food source, but when these were absent did complete their development on P. radiata needles. There was no difference in larval development between those on the three P. radiata families tested. The study suggests an accidental introduction of L. monacha to New Zealand, even more so than L. dispar, could have a serious impact on P. radiata plantations.

Determining soil hydraulic parameters is necessary for accurate simulation of water movement. However, few data exist for key hydraulic parameters required to run such simulation models for volcanic soils. The aim of this study was to obtain hydraulic parameters for sandy volcanic soils (Vitric Orthic Allophanic under the New Zealand system) that are irrigated with wastewater from Rotorua, New Zealand. We measured drainage characteristics, soil water retention, and hydraulic conductivity at three plots and used a simulation model, SOIL. The drainage process was simulated for 20 days and model output was compared with measured water contents and total drainage. The model output was in excellent agreement with measured results from one of the three profiles (less than 1% error in the prediction of the cumulative drainage at 1 m depth). However, the agreement was poor for the other profiles (35% and 138% over-estimation of the drainage) when water retention curves determined from soil cores were used. We repeatedly ran the model using new values of the Brooks and Corey coefficients, until the best agreement between simulated and measured data was achieved for the two profiles with poor comparisons. Residual water content was found to be large (about 25%). Finally, we constructed a three-layer soil profile for the upper 1 m of these soils, with common soil hydraulic parameters in each layer. Simulation results indicated that the uppermost 1 -m layer of local volcanic soils may be adequately simulated as a profile with three layers for prediction of water fluxes and drainage. However, these parameters have to be tested on other soil profiles in order to evaluate their generality.

A magnesium (Mg) fertiliser trial was established in 1984 in a 5-year-old stand of Pinus radiata D. Don with magnesium deficiency in Tauhara Forest in the central North Island of New Zealand. The main trial consisted of rates of dolomite applied at 0, 20, 55, 150, and 400 kg Mg/ha. A range of magnesium fertilisers (kieserite, serpentine, coarseground calcined magnesite, fine-ground calcined magnesite, and Epsom salts) was also applied at 55 kg Mg/ha. Since establishment the trial has been regularly measured (height and dbh) and the foliage sampled, and the soil has been sampled periodically. The young trees were subjectively scored for visual symptoms of magnesium deficiency prior to fertiliser application in 1984 and a year later in 1985. In 1993, the trial was used to study the long-term effect of added magnesium on the incidence and severity of Upper Mid-Crown Yellowing (UMCY). Individual trees were scored for UMCY symptoms and needle retention, and measurements of height and dbh were taken. Foliage samples were taken from all plots and analysed for nitrogen, potassium, phosphorus, calcium, magnesium, and boron. Soil samples were collected from selected plots (at depths of 0-10 and 10-20 cm) and analysed for exchangeable cations (magnesium, potassium, calcium) and acid-extractable magnesium. The incidence and severity of UMCY symptoms were significantly reduced with magnesium additions of 150 and 400 kg/ha. There were no growth responses to magnesium addition in the trial but in an adjacent area of the stand, which had been operationally treated with magnesium (coarse-ground calcined magnesite at 200 kg Mg/ha) in 1984-85, there was a 5% response in dbh to magnesium addition. In the trial, strong positive correlations were found between tree growth measurements and needle retention, indicating that the needle-cast fungus Cyclaneusma minus (Butin) DiCosmo et al. may have been the most important influence on tree growth. Where dolomite had been added at 150 and 400 kg Mg/ha, and kieserite at 55 kg Mg/ha, foliar magnesium concentrations were still elevated nearly 10 years after addition. Dolomite added at 400 kg Mg/ha maintained high amounts of soil exchangeable magnesium in the top 20 cm and acid-extractable magnesium in the 0-10 cm depth while reducing the soil exchangeable potassium/magnesium ratio (both depths) after 10 years. High UMCY values were associated with low foliar and soil exchangeable magnesium and high foliar and soil exchangeable potassium/magnesium ratios. The latter relationships may be developed into useful diagnostics tools for the future identification of resistant genotypes and high-risk sites. It was concluded that the stand studied in Tauhara Forest was at "medium" risk for UMCY development by age 14 years and a future reassessment of the site may provide stronger relationships between UMCY and soil and foliar chemical characteristics.

Fertiliser requirements at establishment of pines and Pseudotsuga menziesii (mirb.) Franco (Douglas-fir) were examined in field trials on four soils in low to medium (600-to 900-mm) rainfall environments in the high-country of the South Island, New Zealand. A combination of boron, nutrient omission, and factorial nitrogen-phosphorus-sulphur trials was used to examine potential responses to nitrogen, phosphorus, sulphur, potassium, magnesium, and boron. Tree form and height responses to boron application were obtained in Pinus radiata D. Don breeds growing on free-draining Fork and Pukaki soils in a low-rainfall environment. In the same trials P. nigra Arn. subsp. laricio (Poiret) Maire (Corsican pine) and P. ponderosa P. Lawson et Lawson (ponderosa pine) showed no response to boron. The form response to boron varied amongst P. radiata breeds and depended on soil type. Douglas-fir showed no response to omission of boron from an otherwise-complete fertiliser mixture on Tekapo soils in moister (800-900 mm) environments. No responses were demonstrated to nitrogen, phosphorus, sulphur, potassium, or magnesium fertilisers by either pines or Douglas-fir.

Soil chemical properties under grassland were compared with those under adjoining first-rotation pine forest aged 20 and 25 years, at coastal and inland hill country sites in Canterbury, New Zealand. The pasture sites had been treated with fertiliser but not limed. Organic carbon, total nitrogen, sulphur, and phosphorus, and exchangeable potassium, calcium, and magnesium levels were lower in soil beneath forest at one or both of the sites. In contrast, available phosphorus and sulphur concentrations were marginally higher beneath forest at both sites despite fertiliser application to grassland. Mineralisable nitrogen also was higher beneath forest at the inland site, but not at the more agriculturally developed coastal site. Differences between sites for total sulphur and exchangeable magnesium were ascribed to greater atmospheric inputs of these elements to forest at the coastal site. At both sites, soils under forest were more acid and had higher exchangeable aluminum levels. Differences between forest and grassland for organic carbon and total nitrogen and phosphorus were confined to the upper soil layers (0-0.1 m), while differences in soil acidity progressed to a depth of 0.2 m, and differences in exchangeable cations were evident to 0.4 m, the greatest depth measured. Soil (< 2 mm) bulk density and nutrient pools were measured at the coastal site, and bulk density was similar beneath forest and grassland. Total nitrogen and exchangeable potassium and magnesium pools to 0.3 m depth were lower under forest than grassland, while the exchangeable aluminium pool was higher under forest. Organic carbon, total phosphorus and sulphur, and exchangeable calcium pools were similar under the two types of vegetation. Lower concentrations and pools of nutrients in soil beneath forest may have been due to uptake and sequestration of nutrients in forest biomass, or to fertiliser application to grassland, although the latter would have been counteracted to some extent by nutrient removals by grazing animals.