Reliability of Foliar Analyses of Norway Spruce Stands in a Nordic Gradient

Norway spruce stands at eleven sites in Finland, Norway and Sweden have been studied under various climates, atmospheric deposition of N and S and fertilisation regimes. Nitrogen was growth restricting at eight inland sites, while P was growth restricting at three coastal sites. Liming and N fertilisation caused serious B defi ciency on some of the inland sites. It is likely that liming affects uptake of B, whereas N fertilisation causes a dilution due to increased growth. Application of S combined with N probably caused K defi ciency at one of the sites. The reliability of foliar analyses as a method to diagnose nutrient status and the likely changes after nutrient input to spruce forests in the Nordic countries, are discussed. The CRand the DOP-method are evaluated for diagnostic purposes. Both methods seem to give reliable conclusions even if the CR-method often produces more specifi c results. Interpretation based on both current and one year old foliage improved the diagnostic prognoses. The accuracy of diagnosis also relies on knowledge and ability of the interpreter. Based on the results it is reason to be cautious about recommendations of single element fertilisations, e.g. with N alone, because the demand of other elements beyond available pools frequently occurs. Forest trees in the boreal region are probably well adapted to N defi ciency, which means that they can handle the physiological consequences rather well, while defi ciencies of other elements usually are more detrimental to growth vigour and stress related diseases.


Introduction
The general principles of plant nutrition are well documented (e.g.Epstein 1972, Marschner 1995).As proper nutrition is important for the health of trees, this theme has received much attention, especially the problem of developing reliable fi eld methods for diagnostic testing.Less expensive and more specifi c techniques than soil analyses are required to estimate the availability of nutrients.Good correlation between total N and P concentrations in humus and in foliage when nutrient supply is below optimal, encourages the use of foliar analyses techniques (e.g.Miller et al. 1977, Nohrstedt and Jacobson 1994, Braekke 1996).
Foliar analyses as a diagnostic method was introduced in the early 1930's.Lundegårdh (1951) stated that internal nutrient concentrations in a plant could be taken as an integration of soil nutrient availability and time related uptake fl ux.Tamm (1964) summarised important principles of foliar analyses, practical application, foliar sampling technique and gaps in knowledge.Evers (1986) has given revised standards of the foliar sampling technique.One major problem has been to establish the relation between nutrient concentrations in the foliar tissue and growth rate under fi eld conditions, i.e. response curves for individual elements.Critical optimum concentrations and ratios to nitrogen for different elements are available from laboratory experiments (Ingestad 1979, Ericsson et al. 1994).After corrections of concentrations and ratios these can be applied to fi eld conditions (Braekke 1994, Linder 1995).Braekke (1994) has coupled growth capacity or degree of defi ciencies and ranges of nutrient concentrations in current foliage of Norway spruce (Picea abies L. Karst.) and Scots pine (Pinus sylvestris L.).Four ranges representing different physiological stress levels were defi ned: optimum, pre-optimum, defi ciency, and strong deficiency (Table 1).The pre-optimum zone covers according to Ulrich and Hills (1967) the range from 80-100% of maximum growth.Two ranges were defi ned in the defi cient zone when actual growth is below 80% of maximum growth, deficiency (50-80% of maximum growth) and strong defi ciency (< 50% of maximum growth).The strong defi ciency range usually corresponds to visible defi ciency symptoms.Braekke (1994) also defi ned criteria of balanced nutrition as critical ratios to N, based on the principles given by Ingestad (1979).
A number of methods are available for diagnostic interpretation of foliar data e.g.i).Concentrations and ratios (the CR-method: Braekke 1994, 1996, Braekke et al. 1998, Linder 1995), Table 1.Ranges of concentrations and critical ratios of nutrients in current-year-needles (C) of Norway spruce and Scots pine at strong defi ciency (SD), defi ciency (D), pre-optimum (PO) and optimum (O).(From: Braekke (1994)  ii).Deviation from optimum percentage (the DOP-method: Montanes et al. 1993), iii).Diagnosis and recommendation integrated system (the DRIS-method: Beaufi ls 1973) and iv).Graphic vector analyses (the GVA-method: Kraus 1965, Timmer and Stone 1978, Valentine and Allen 1990, Braekke 1996).These methods differ with respect to type of data required, diagnostic reliability and fl exibility.The less fl exible method are the DRIS, which requires a large and specifi c database to establish its standard reference values, and the GVA-method which requires a specifi c fi eld plan for fertilisation and sampling.In addition a graphical method for displaying data on a relative scale is available, the RCC-method (relative concentration and content change: see e.g.Braekke et al. 1998).
The aims of this project were to 1) summarise and adjust existing knowledge about diagnostic foliar analyses of Norway spruce to common criteria to detect nutrient defi ciencies more precisely, 2) compare nutrient status of stands in a Nordic gradient of different climates under a wide N deposition range, 3) predict likely changes of stand nutrient status after input of different nutrients.

Materials and Methods
Eleven Norway spruce sites within the vegetation types, according to Kielland-Lund (1981), Eu-Piceetum myrtilletosum and Eu-Piceetum dryopteridetosum in Finland, Norway and Sweden were selected.Table 2 specifi es location, altitude, temperature sums (degree-days > 5°C), N-deposition, site index of control plots and experimental parameters.
The experiments in Finland were on a factorial design, as were the Norwegian experiment Lontjern and the Swedish ones at Åseda and Farabol.The experiments Vardal and Løten in Norway and Norråker in Sweden were traditional N fertiliser experiments with supplemental treatments including K and/or P. All treatments including N at Kemijärvi and Sodankylä were supplemented with B in 1985-86 and 1990-91.Birkenes and Marnardal in Norway were designed for other aims than fertilisation.Some of the essential elements in the foliage samples were not analysed for Åseda and Norråker.The data from these sites were therefore supplemented with data of biomass trees sampled some years later.Further information about fertiliser input, types and doses are described by Andersson et al. (1998) The foliar sampling was done on dominant and co-dominant trees according to specifi cations given in Table 3.The samples were kept The CR-method and the DOP-method were chosen for evaluation, as well as the RCC-method for graphical display.The fi rst step in the CRmethod uses the defi ned concentration ranges linked to nutrient stress levels and nutrient ratios as developed by Braekke (1994).This step sort out the key element which are the growth restricting one.Only nutrient elements with concentrations in the strongly defi cient and defi cient ranges were considered critical.The second step ranks the remaining critical elements after the key growth restricting element found in step one.When e.g.N was found growth restricting, the other critical elements were ranked after N according to their relative position in the range judged by their concentrations, starting in the strongly defi cient range.If another element than N was growth restricting, which means that the ratio to N for the element was below the critical limit, the other critical elements were ranked after this element as explained in the previous case.If more than one element had ratios at or below the critical ratio, the one with the lowest ratio relative to the limit was defi ned as the growth restricting one.The output is a predicted sequence of nutrient elements from the current restricting one to those, which potentially might be in short supply.If the actual defi ciency of an element is eliminated by nutrient input, its position should be taken by the next element in the sequence.
The DOP-method produces a sequence based on the element concentration of the sample in per cent of optimum.The RCC-method displays the data graphically on relative scales by combining foliage concentrations and contents of nutrients and needle weight in one graph.The graphs compare the control with each of the other treatments on the individual site.Such graphs illustrate the dynamic changes of nutrient concentrations and contents after dif- 1) When "upper 1/3" or "upper 1/4" is specifi ed then the sample is taken from the middle of that crown section.
ferent treatments.This is a supporting tool for the interpreter, especially to get a starting view of the data.Only a few of the produced graphs are presented in this report.

Results
A screening of the data identifi ed N, K, P, and B to be at concentrations below pre-optimums either generally or frequently.Only these elements were further processed by the CR-method, while elements at concentrations below optimums were processed by the DOP-method.

Finland
The control plots at all Finnish sites showed defi cient to strongly defi cient N concentrations.
Potassium was in pre-optimum range at Kemijärvi and P was in pre-optimum range at Sodankylä (Table 4a).Fertilisation with N alone or combined with other nutrients, raised the foliage concentration of this element considerably at Heinola, while changes were insignifi cant at the other sites.However, the doses of N at Heinola, 120-180 kg N ha -1 every fi fth year, were not suffi cient to hold N concentration permanently at the According to the CR-method (Table 4b) the specifi c growth restricting element at Heinola was N in the control and those treatments not given fertiliser nitrogen (Ca, KP, KPCa).In the remaining treatments B seems to be critical and growth restricting.Phosphorous was in defi cient range at treatment NK, whereas K levels were always at pre-optimum or optimum.The overall ranking of critical elements was: N>B>P.Nearly all treatments at Kemijärvi and Sodankylä showed strong nitrogen defi ciency.Potassium levels at Kemijärvi dipped into the defi ciency and strong defi ciency range at treatments NB, Ca and NCaB, while P levels dropped to defi ciency range at treatments NCaB and NKB.Boron concentration was critical at treatment Ca.The overall ranking of critical elements was: N>B>K>P.Sodankylä showed relative low supply of P and the concentration level dropped from pre-optimum to 1.2 at treatments NB and NKB, while K level dropped from optimum in control to defi cient range at treatments NB and NPB.The overall ranking was as follows: N>B>P>K.Limited but not critical supply of S is noted at the extreme northern sites at some fertiliser treatments.The DOP-ranking (Table 4b) deviated somewhat from the CR-ranking at certain treatments and was generally less specifi c.The main ranking, however, agreed well except that K was included at Heinola and S at Sodankylä and Kemijärvi.
The overall results from the RCC-diagrams are that N fertilisation at Heinola failed to cause signifi cant needle growth response despite increased foliage N concentrations (N applied in 1993 and needle sampled in 1994), whereas N in combination with P did (Fig. 1a).All treatments reduced B concentrations signifi cantly, least so at KP and rather much at NKPCa (Table 4a).The reduction in B concentration at NP treatment, which was accompanied by a signifi cant higher unit needle weight, may be explained by a dilution effect.In the other treatments, where only a slight change in needle weight was detected, the reduction in B concentrations could not have been a dilution effect.In the limed plots the concentration of Ca was moderately elevated, probably because 17 years has past since application.The only signifi cant response in terms of foliage weight at Kemijärvi was at treatment NB (Table 4a).
Here liming without B addition also reduced B concentration, while concentrations and contents of Ca increased signifi cantly (treatments Ca, KPCa, NCaB and NKPCaB).The more pronounced effect at this experiment than at Heinola, may refl ect that the lime had been applied more recently (8 years previous).The largest relative response (not signifi cant) on needle weight at Sodankylä was at treatment NPB (Fig. 1b).

Norway
Nitrogen concentrations (Table 5a) were in the defi cient and strongly defi cient ranges on control plots at Vardal, Löten, Birkenes and Marnardal.
Potassium was defi cient or sub-optimum at all sites other than Lontjern, a defi ciency that was exacerbated by N application.Similarly, at all sites P levels were inadequate, and remained so or were worse where N was applied without P. Boron levels were inadequate only when high doses of N was applied at Vardal, and there were no indications of problems with Mg and S.
The ratios indicate imbalanced nutrient uptake at treatments N and 3N at Vardal where probably potassium was growth restricting (Table 5b).The control plots at Birkenes, Marnardal, Lontjern and also the treatment N at Løten as well as treatment K at Lontjern had critical P/N-ratios.There are clear indications that C+1 needles refl ect the nutrient status more precisely than the C needles probably because of intensive re-circulation of critical elements.The overall ranking of critical nutrient elements by the CR-method was as follows: Vardal -N>K>P>B, Löten -N>P>K, Birkenes and Marnardal -P>K>N and Lontjern -P.Ranking of elements by the DOP was similar except for some deviation in sequence and details.
The changes in relative needle weight and relative concentrations and contents, shown by the RCC method at Vardal and Löten, clearly refl ect that nine years that had past since last fertilisation (Fig. 2).All treatments at Vardal and Löten had needle weights less or about equal to that of the control.Judged by these criteria, least stress was observed at treatment NP which also increased concentration and content of P, but reduced those of B. The stress, which was created by the 3N treatment at Vardal, signifi cantly lowered the concentrations of B, P and K.The Lontjern data illustrate responses on needle weights at all treatments, however, the only signifi cant one occurred at treatment N. The RCC results which indicate a general N limitation at Lontjern, are not con-fi rmed by the height growth responses 1992-95 (Table 5c).The signifi cant height growth response in 1994 at P treatment indicates a limitation of P.This result might have been due to the recovery process after the heavy drought in 1992, which were unevenly distributed between treatments at the drawing of plots.The trees subjected to both treatments K and P had longer shoots in 1992 and better growth up to 1994 (autocorrelation), compared to the other treatments.In 1995, however, there was a consistent height growth response at all plots treated with P supporting the conclusion that P was the actual growth restricting element at Lontjern.This conclusion was also predicted by using the CR-and the DOPmethods.

Sweden
The control plots of the Swedish sites had deficient to strongly defi cient levels of N, whereas concentrations of K and P were defi cient to optimum (Table 6a).As elsewhere on the inland sites, applications of N in the absence of K and P frequently lowered the concentration of these elements to critical levels.The ratio K/N = 33 in control plots at Farabol is a warning about temporary imbalanced uptake (Table 6b).However, the application of N seemed to have caused a deficiency of P, whereas NS treatment caused K limitation.At Åseda both treatment N and NS caused potassium defi ciency.The overall CR ranking at Farabol was: N>P>K, and at Åseda: N>K>P.Magnesium and S might become in short supply at Farabol if uptake of N, K and P were optimised.At Norråker where trees had serious N defi ciency, B came next in the ranking and was followed by P.There were some indications of short supply of K.The DOP-method gives similar ranking as the CR-method, although less detailed (Table 6b).
The RCC diagrams for Åseda, based on biomass tree-data from 1994, are presented in Fig. 3. Foliage unit dry mass weight was slightly changed at treatments N1S and reduced at all other treatments.Some patterns in concentration change should be noted.Application of N alone reduced P and K concentrations whereas Ca and Mg concentrations were increased.Application of S alone had an opposite effect.When N and S were combined, both concentrations and contents of all elements except P, increased.At treatment N1P2K2 the Mg concentration was reduced somewhat, although not critically.

Relationship Between N and P Concentrations and Foliar Unit Dry Mass
The control plots on sites where N was growth restricting had a weak, but signifi cant linear relationship between needle weights and N concentrations (Fig. 4).A stronger relationship was found for control plots on sites where P was growth restricting.Extrapolating these functions to optimum nutrition gives needle weights at about 6000 mgDM 1000 needles -1 .

Discussion and Conclusions
The DOP-method assumes that the physiological response to foliar concentrations of all nutrients is linear, starting from zero and ending at critical optimum.In principle optimum curves are S-shaped and do not start at zero.Despite this, the critical elements for individual sites were ranked satisfactory by the DOP when the CR-method was used as a standard.However, the rankings deviated in some cases at the plot or treatment level.The graphical RCC-method illustrates relative differences in concentrations, contents and specifi c unit needle weight between two samples representing different treatments, locations etc.This method cannot pinpoint growth-restricting elements unless the experiment is designed according to the specifi cation required for the GVA-method (Braekke 1996).Although the site Lontjern (973) in Norway had such specifi cations, no conclusive statements could be drawn from the RCC, probably because the site was subjected to a drought during the fi rst growing season after nutrient application.The GVA pointed out N as growth restricting element in the fi rst growing season, while the height growth response in the second growing season after fertilisation proved that P was the critical element.It is reasonable to conclude that the summer drought in 1994, had infl uenced N availability and that this element actually was growth restricting during the water stress.However, when water stress ended, P again became growth restricting.Such interactions must be considered seriously by the interpreter to avoid wrong conclusions (Braekke, 1996).The CRmethod turned out to be robust and pinpointed P as the growth restricting element, whereas the DOP added N as second element to the ranking sequence.Diagnostic foliar analysis by the CR-method (Tables 4b, 5b and 6b) certainly provided much information about nutrient status and cycling at the sites.These results explain reasonably well the stem volume growth at different treatments (Table 7).However, the unit needle dry mass weight and the volume growth response were usually not correlated, probably because needle weight represents one specifi c year, whereas volume growth was integrated over several years.Generally spoken, this implies that unit needle weight for one single year does not necessarily refl ect the total needle mass of a stand either for that particular year or the average for several years.
The experimental fi eld methods were not designed to optimise nutrient supply according to the principles given by Linder (1995).At all sites, except Åseda, nutrients were applied at intervals from 3-10 years, which means that the nutrient status of trees usually fl uctuated.This also implies that the sites never were brought to the maximum growth potential.The results demonstrate that if a primary defi ciency is overcome by fertiliser a secondary defi ciency of another element are encountered on most of the studied sites.Those located at the coast in southernmost Norway were exceptions.To overcome such problems a balanced fertiliser mixture of more elements are usually needed.Nitrogen and phosphorus should be key elements in such a mixture, but frequently also K and B are needed.This support the results from fertiliser optimising experiments reported by Tamm (1991).
It was diffi cult to fi nd the complete ranking by just analysing current foliage (C) of control plots.The Finnish and the Swedish sites, where only current needles were analysed, showed a truncated ranking sequence of elements at the control plots compared to the overall ranking.At the Norwegian sites, where current and previous year's foliage were analysed, the two rankings agreed satisfactory.None of the control plots had B in the range of critical concentrations, even if the supply turned out to be inadequate when stand growth was improved by adding other critical elements.This means that a proper diagnosis of forest stand nutrition by foliar analyses must also rely on the knowledge and ability of the interpreter.One example of needed knowledge is the regional variation of potential B defi ciency shown by Braekke (1979).All sites in Sweden and Norway where potential B defi ciency was indicated, are located inside the region proposed by Braekke (1983).Adverse effects of liming on B uptake are documented by Lehto and Mälkönen (1994).Such effects were demonstrated on Heinola, Kemijärvi and Farabol.
Application of S combined with N at Farabol seems to have rendered the stands into serious K defi ciency, but this effect was not supported by the results from Åseda.One likely explanation is that a critical amount of the available K-pool was leached out of the rhizospheric zone at Farabol because of higher annual S dose, which was 40 kg ha -1 yr -1 at Farabol whereas only 15 kg ha -1 yr -1 at Åseda).
Nitrogen was growth restricting on eight of eleven studied sites.The extreme northern ones, Norråker, Kemijärvi and Sodankylä, plus Löten in South Central Norway, had N concentrations in the strongly defi cient range.The other four, Heinola, Farabol, Åseda and Vardal, had concentrations in the defi cient range.This imply that the atmospheric input of N to these sites, which varied from 2 to10 kg N ha -1 yr -1 , had not changed the pattern of N limitation normally observed in boreal and boreonemoral forests.
The coastal sites Birkenes, Marnardal and Lontjern had a nutrient status different from the inland sites, as they suffered from P restrictions.High deposition rate of sea salts combined with increased deposition of S and N might explain this (Braekke 1996).Likely hypotheses are: i) N deposition has improved forest growth rate and caused a P demand beyond available supply, ii) soil acidifi cation has generated free Al, which has fi xed P chemically in soil solution and also turned the organic matter into complexes with decreased mineralisation rate, iii) N deposition and soil acidifi cation have caused changes in the mycorrhizal community and iv) N deposition has decreased decomposition rate of organic matter and increased the less available stores of organically fi xed P (Fog 1988, Berg 1986).

Acknowledgement
The Nordic Forest Research Cooperation Committee (SNS) and the Research Council of Norway made the Nordic cooperation, fi eld work and chemical analyses possible by their funding.Professor Eino Mälkönen and Dr. Leif Hallbäcken have supported urgent data.Professor Hugh Miller has suggested valuable corrections to the manuscript.We want to thank you all for contributions which made this report possible.

Fig. 1a .
Fig. 1a.Finland.Relative needle nutrient concentration and content changes (RCC) at Heinola site.The full drawn line and the open circle denote relative needle weight and relative concentration/content of nutrients in control plot.The dottted line with fi lled circles display relative changes due to the given treatment.Signifi cant differences (p<0.05) are given in the diagram by the signs: ^ signifi cantly different relative nutrient contents, > signifi cantly different relative nutrient concentrations.

Table 2 .
Description of sites by location, altitude, DD (temperature sums as degree-days >5°C 1961-1990), N deposition (wet + dry 1986-1990, dry deposition 30% of wet), site index as stem increment with bark on control plots and some experimental parameters.

Table 3 .
Specifi cations of foliar sampling technique, age of the stands used in the study and delay in sampling after last treatment.

Table 4a .
Finland . Mean nutrient concentrations and dry mass of 1000 current-year-needles (C).Elements with concentrations in pre-optimum range are in italics and those in defi cient range as well as needle weights signifi cantly higher than control (p < 0.05), are in bold face.

Table 4b .
Finland.Current and potential restricting elements predicted by concentrations and ratios (CR) and by deviation from optimum percentage (DOP)."The true ranking" of current and potentially restricting elements (in italics) are based on an overall evaluation of the factorial effects within each experimental site.

Table 5a .
Norway.Mean nutrient concentrations and dry mass of 1000 current-year-needles (C).Elements with concentrations in the pre-optimum range are in italics and those in the defi cient range as well as needle weight signifi cantly higher than control (p < 0.05), are in bold face.

Table 5b .
Norway.Current and potential restricting elements predicted by concentrations and ratios (CR) and by deviation from optimum percentage (DOP)."The true ranking" of current and potentially restricting elements (in italics) by the CR-method are based on C-and C+1-needles at different fertiliser treatments within the experimental site, while the ranking by the DOP-method is based on C-needles.
1) Critical ratios were not calculated when N concentrations are beyond critical optimum

Table 6a .
Sweden.Mean nutrient concentrations of current-year-needles (C).Elements with concentrations in the pre-optimum range are in italics and those in the defi cient range are in bold face.

Table 6b .
Sweden.Current and potential restricting elements predicted by concentrations and ratios (CR) and by deviation from optimum percentage (DOP)."The true ranking" of current and potentially restricting elements (in italics) are based on control as well as fertiliser treatments within the experimental site. )the data on elemental concentrations used in the CR ranking were supplemented for the missing elements in Table 6a by using data from the chemical analyses on biomass trees 1

Table 7 .
Growth response in stem volume over bark at different treatments.Treatments which are signifi cantly different from control have fi gures in bold face (p < 0.05).The Finnish sites are not tested statistically because data from only one block was available.