Comparison with published values

including micronutrients in CND score computations. However, it seems likely that including nutrients that contribute little to the total chemical composition of foliage may obscure the relationships between tree growth and multivariate nutrient ratios of macronutrients by decreasing the relative importance of the latter (personal communication L.-E. Parent, Univ. Laval).

2.2 Comparison with published values

demonstrating that optimum ranges for nutrient concentrations and CND scores derived from our boundary line approach are accurate for white spruce. Overall, the differences should be regarded as minor, given the important natural variation in foliar chemistry and the great distance between the two studies. Therefore, a relatively small number of trees from few sites can be appropriate to ob tain standards with the boundary line approach that can be applicable across the whole boreal forest. This is in agreement with results obtained for sugar maple with the boundary line approach (Vizcayno-Soto and Côté 2004).

Pub li shed optimum nutrient concentrations determined from field experiments were not found for black spruce and standards are limited to nitrogen for jack pine. The optimum N concentration (14 mg-g-¹) for jack pine determined from a long-term N fertilization trial by Weetman and Fournier (1984) is slightly higher than the optimum computed from the boundary line approach. This difference should be regarded as negligible, given that our N standard for jack pine was determined with data from a single sampling year and that annual variation of nutrient concentrations can be large (Lowry and Avard 1969; van den Driessche 1974). In addition, the computed critical and optimum N levels agree with the foliar N concentrations (10 to 13 mg-g-¹⁾ observed in a jack pine stand that failed to respond to N amendments in Ontario (Foster and Morrison 2002).

Manganese is the only nutrient for which no standard based on field data exists for the species under study. Manganese concentrations from several white spruce, black spruce and jack pine plantations across Quebec are however available and averaged 0.44, 0.70 and 0.24 mgog-^I, respectively (Sheedy and Thomassin 1994). These mean Mn concentrations are similar to our optimum for white spruce but lower for black spruce and jack pine. Our optimum Mn concentration for black spruce is also greater than the mean Mn concentration from six black spruce stands from northern Quebec (Morrison and Armson 1968). Similar to black sPruce, our Mn optimum for jack pi ne is higher than Mn published data found in several studies (Morris on 1972; Smirnoff and Bernier 1973;

Camiré and Bernier 1981; Brokheim and Leide 1991; Gagnon et al. 1994). These results suggest that the computed optimum Mn concentration for white spruce is accurate but that critical and optimum levels for black spruce and jack pine are overestimated.

Manganese nutrition has rarely been investigated in the latter species. Morrison and Annson (1968) observed Mn concentrations in black spruce and jack pine seediings that were above 4.0 mg-g-¹ and presented only minor growth decrease and toxicity symptoms, suggesting that the range of luxury consumption for this nutrient may be large.

The occurrence of unusually high soil moi sture, low soil pH or rapid organic matter decomposition at the Beardmore site during the year of leaf sampling could have contributed to increase the availability of Mn in the soil solution and in tree foliage.

Changes in tree growth in response to recent changes in Mn availability could have been small overall since growth was calculated over the two previous years. Moreover, it is likely that recent changes in Mn availability that would result in nutrient imbalances or toxicity would not be observed until the following year since tree growth of a given year is highly influenced by the tree previous-year nutrient status. It is also possible that another factor associated with Mn availability may have affected tree growth and induced significant re1ationships between Mn and growth. This shows that it should be mandatory to consider more than one site and spread sampling over several years to establish accurate nutritional standards.

2.2.2 Comparison with values for seedlings

In comparison with nutritional standards deve10ped for seedlings grown in hydroponic culture (Swan 1971), white spruce optimum nutrient concentration ranges derived from our boundary line mode1s were low for N and P, similar for K and high for Ca. For black spruce, the computed optimum range for Kwas lower than the published value (Swan 1970) for this nutrient. The optimum concentrations for N and Mg computed for jack pine were also lower than optimum leve1s for seedlings (Swan 1970).

Comparison is impossible for Mn, as no optimum concentration has been published for the three species under study. However, in contrast with a Mn criticallevel proposed by Ballard and Carter (1986) for many coniferous species of western Canada, the computed optimum Mn concentrations for the three species are ab ove deficiency leve1s by an order of magnitude. As mentioned above, our Mn standards for black spruce and jack pine are questionable and will require additional work to assess their validity.

With respect to computed optimum CNDs scores, comparison with nutrient log ratios derived from Swan's (1970, 1971) standards shows that nitrogen scores are lower in

trees than in seedlings whereas the reverse was observed for Ca scores. The computed optimum CND scores for P, K and Mg are closer to the scores derived from Swan (1970, 1971) values. Knecht and G6ransson (2004) have found a similar trend with bivariate nutrient ratios in many tree species growing in diverse ecosystems, suggesting that differences in N and Ca nutritional balance are constant for a broad range of tree species.

Results from white spruce show that despite the large differences in N and P optimum nutrient concentrations and CNDs scores computed for trees and from seedlings, the N:P ratio is nearly the same for trees (100: 16) and for seedlings (100: 15). Both values correspond closely to the optimum 100:15 ratio for conifers proposed by Ericsson (1994).

These similarities suggest that our optimum values for N and Pare accurate and that a precise chemical balance is maintained between the latter nutrients through the different life stages of trees.

2.2.3 Size of computed optimum ranges

Computed optimum ranges based on concentrations are rather small for many nutrients. Since soil nutrients are likely available in limited amounts in forest soils, it seems logical that at low nutrient concentrations, increasing tree growth rates observed with increasing foliar nutrient concentrations is indeed associated with nutrient deficiencies. However, our computed toxicity thresholds are often lower than proposed optimum values (except for Ca) and weIl under the range of "luxury to excess consumption" determined by Swan (1970, 1971). Trees growing in boreal ecosystems are often nearly nutrient limited and trees with excessive or toxic leaf nutrient levels should be uncommon. The validity of sorne of our toxicity levels is therefore questionable. The sampling of a small number of trees may have been insufficient to adequately cover the luxury consumption and toxicity range for many nutrients. With a lower number of observations at high nutrient concentrations, the probability of sampling trees with optimum growth rates decreases sharply. Hence, the additional precautionary steps developed in this thesis to reduce the risk of including trees with less than optimum growth for a specific interval of nutrition in the boundary-line models may not have been sufficient to prevent a possible underestimation of toxicity levels for the species under study. lndeed if the selected boundary points in high nutrient concentration intervals had sub-optimum growth, they would have induced an artificial decrease in the quadratic

models. Therefore, it is reasonable to conclude that the accuracy of the upper optimum limits of nutrients that are often considered growth-limiting in boreal forests, mainly N, P and K, could have been underestimated since sampling was inevitably biased towards trees with low foliar nutrient concentrations. This bias, however, increases the accuracy of the computed critical and optimum levels of most nutrients since these ranges were usually well represented in our samples.

CHAPTERV

FOLIAR NUTRIENT DIAGNOSIS IN BOREAL STANDS

1. Results