The influence of soil properties and nutrients on conifer forest growth in Sweden, and the first steps in developing a nutrient availability metric

Abstract

<p><strong>Abstract.</strong> The availability of nutrients is one of the factors that regulate terrestrial carbon cycling and modify ecosystem responses to environmental changes. Nonetheless, nutrient availability is often overlooked in climate–carbon cycle studies because it depends on the interplay of various soil factors that would ideally be comprised into metrics applicable at large spatial scales. Such metrics do not currently exist. Here, we use a Swedish forest inventory database that contains soil data and tree growth data for &amp;gt;<span class="thinspace"></span>2500 forests across Sweden to (i) test which combination of soil factors best explains variation in tree growth, (ii) evaluate an existing metric of constraints on nutrient availability, and (iii) adjust this metric for boreal forest data. With (iii), we thus aimed to provide an adjustable nutrient metric, applicable for Sweden and with potential for elaboration to other regions. While taking into account confounding factors such as climate, N deposition, and soil oxygen availability, our analyses revealed that the soil organic carbon concentration (SOC) and the ratio of soil carbon to nitrogen (C<span class="thinspace"></span>:<span class="thinspace"></span>N) were the most important factors explaining variation in “normalized” (climate-independent) productivity (mean annual volume increment – m<span class="inline-formula">³</span><span class="thinspace"></span>ha<span class="inline-formula">⁻¹</span><span class="thinspace"></span>yr<span class="inline-formula">⁻¹</span>) across Sweden. Normalized forest productivity was significantly negatively related to the soil C<span class="thinspace"></span>:<span class="thinspace"></span>N ratio (<span class="inline-formula"><i>R</i>²</span><span class="thinspace"></span><span class="inline-formula">=</span><span class="thinspace"></span>0.02–0.13), while SOC exhibited an empirical optimum (<span class="inline-formula"><i>R</i>²</span><span class="thinspace"></span><span class="inline-formula">=</span><span class="thinspace"></span>0.05–0.15). For the metric, we started from a (yet unvalidated) metric for constraints on nutrient availability that was previously developed by the International Institute for Applied Systems Analysis (IIASA – Laxenburg, Austria) for evaluating potential productivity of arable land. This IIASA metric requires information on soil properties that are indicative of nutrient availability (SOC, soil texture, total exchangeable bases – TEB, and pH) and is based on theoretical considerations that are also generally valid for nonagricultural ecosystems. However, the IIASA metric was unrelated to normalized forest productivity across Sweden (<span class="inline-formula"><i>R</i>²</span><span class="thinspace"></span><span class="inline-formula">=</span><span class="thinspace"></span>0.00–0.01) because the soil factors under consideration were not optimally implemented according to the Swedish data, and because the soil C<span class="thinspace"></span>:<span class="thinspace"></span>N ratio was not included. Using two methods (each one based on a different way of normalizing productivity for climate), we adjusted this metric by incorporating soil C<span class="thinspace"></span>:<span class="thinspace"></span>N and modifying the relationship between SOC and nutrient availability in view of the observed relationships across our database. In contrast to the IIASA metric, the adjusted metrics explained some variation in normalized productivity in the database (<span class="inline-formula"><i>R</i>²</span><span class="thinspace"></span><span class="inline-formula">=</span><span class="thinspace"></span>0.03–0.21; depending on the applied method). A test for five manually selected local fertility gradients in our database revealed a significant and stronger relationship between the adjusted metrics and productivity for each of the gradients (<span class="inline-formula"><i>R</i>²</span><span class="thinspace"></span><span class="inline-formula">=</span><span class="thinspace"></span>0.09–0.38). This study thus shows for the first time how nutrient availability metrics can be evaluated and adjusted for a particular ecosystem type, using a large-scale database.</p>