Trends in tree height and basal area growth

In Paper I, trends in the stand height growth of Norway spruce and Scots pine were examined on the basis of LTE data. For both species, a significant upward trend in the height growth was observed in the period 1986-2014 (Figure 6). Over the period, the relative height growth was estimated as 3.94

%/year for Scots pine and 0.89 %/year for Norway spruce. However, the increased growth trend appeared more pronounced after the millennium shift (year 2000), where the magnitude of the annual height growth was 16.92 % and 9.54 % higher than expected respectively, for Scots pine and Norway spruce. The increased height growth correlated positively with the rising temperature but not precipitation during the observation period.

Figure 6. Trends in annual height growth, temperature and precipitation during the period (1986-2014) for Scots pine (upper panels) and Norway spruce (lower panels).

The red lines show the equivalence of the expected and observed growth. Solid blue lines are LOESS smoothing. Shaded regions are 95 % confidence intervals.

Despite the increased growth on average in Sweden, substantial variations in the growth levels at the local (site) scale were observed. This was investigated further in Paper II, using extensive observations made on sample trees from the temporary plots of the Swedish NFI during the period 1983-2020. For height growth, a strong and significant positive (upward) trend was observed for both pine (R² = 88 %) and spruce (R² = 72 %), indicating that the average heights of the two species have increased during the observation period (Figure 7). The magnitude of the increased height in the 38-year period was 2.25 m for Scots pine and 2.12 m for Norway spruce.

The trends were similar for the different regions in Sweden.

Figure 7. Trends (dotted lines) in average height growth of pine and spruce in the period 1983-2020.

However, the trend in basal area growth was stable for both species, despite an observed longer and shorter periods of higher and lower growth in the period (Figure 8).

The increased height growth but stable basal area growth means trees have become taller and slender. The slenderness defined as the ratio of height to diameter (H-D ratio) was examined for dominating and co-dominating trees.

The variation in H-D ratio was studied as a linear function of total tree age, SIS and stand density. The trend of slenderness was studied by residual analysis over year. For a given height, trees in the period 1983-1987 were on

average 1 cm thicker than trees in the last 5 years (2016-2020), and the trend shows an increasing slenderness during the 38-year period (Figure 9).

Figure 8. Trends (dotted lines) in annual basal area growth of pine and spruce in the period 1983-2020.

Figure 9. Increasing (positive trends) slenderness (H-D ratio) of pine and spruce in the period 1983-2020.

There are important limitations on the results presented here. Those of highest interest are related to a change of the method for sample tree selection and successive changes of instruments for height measurement. The selection of plots and stands influence trends in growth through the effects of age-bias in relation to site index. Generally, highly productive forest stands are managed under short rotation whereas those kept longer are mostly

poor stands that require longer periods to produce certain amount of wood volume (Tegnhammar 1992). Despite the capabilities of LTEs in providing insights into the principles of forest production, their statistical requirements (“representativeness” especially for earlier LTEs) are usually questioned (Pretzsch et al. 2019). In Paper I, the similarities in the distribution of site indices for the reference and validation periods coupled with the wider ranges of site index suggested that a large portion of the fertility in Sweden were captured in the analyses. The age class distribution was quite different in both periods. This was attributable to the fewer number of sites for young stands in the validation period and they were located on fertile sites. Though these discrepancies could bias the parameters of the reference growth model, the wider range of top heights in the young stands presumes that the variability was captured in the growth trend analyses. The accelerated height growth was more or less consistent with the outcome in Paper II, which is based on the statistically representative sample of the Swedish NFI. Sample trees have always been selected in proportion to the trees’ basal areas but in the period 1983-2002 without respect to their spatial distribution and in the period 2003-2020 with 1-3 trees per plot (Fridman et al. 2014). Before 2003, the probability for a plot to be represented increased with the density and sizes of the trees on the plot. The number of sample trees per plot varied from 0-9. This resulted in larger sample trees in the 1983-2002 data than in the 2003-2020 data (DBH 23.6 and 21.5 cm). There was also a larger difference in the distribution of trees on the classes of dominant and co-dominant trees, but the background to this difference is hard to explain. In the models (Paper II), site fertility was registered by site index (SIS). The SIS was on average 22.8 and 22.1 m respectively, for the 1983-2002 and 2003-2020 data. There is also a possible shift in the classification of sites according to most suitable tree species. For example, the proportion of plots that were classified as spruce sites was 54 % in the 1983-2002 data and 47 % in the 2003-2020 data.

Another possible source of variation may be changes of instruments for tree height measurements. Until 1970, the Tiréns device was used and in the period 1970-1995 the Suunto hypsometer. After that, the Haglöf-HEC electronic clinometer was used even though the Suunto was used to some extent on the sample plots from 2002 to 2007. Since 2008, the Haglöf-Vertex hypsometer has been used. The accuracy of the measured heights was higher for the Vertex hypsometer (upward bias by 2 cm) than for

Haglöf-HEC (upward bias by 20 cm) and the Suunto (downward bias by 12 cm), in a controlled examination by the Swedish NFI (Fridman et al. 2019).

Despite these limitations, the search for the existence of growth trends is nonetheless important for sustainable forest management (Spiecker et al.

1996; Kahle et al. 2008). As is the case with many survey data, it is difficult to isolate the causes of changes in forest growth trends. Here, the trends are discussed under changes in forest management and environmental conditions in Sweden. Forest management affects growth trends largely through silvilcutural practices (Högberg et al. 2021b), for example, changes in harvesting systems and precision silviculture with improved site preparation, nitrogen fertilization, density-regulation and planting with improved genetic materials (Kahle et al. 2008). In the period 1953-1992, an increase in both the annual height and basal area growth was observed in Swedish forests, and the major contributing factors were attributed to the shift from selective cutting and thinning from above to clear felling, cleaning and thinning from below (Elfving and Tegnhammar 1996). In Papers I and II, the results obtained indicate that the height growth increase has continued until now in Swedish forests and corroborates well with the reported increase in other forest regions (Sharma et al. 2012; Kauppi et al. 2014; Henttonen et al. 2017;

Socha et al. 2021). However, the stable trend in basal area growth may imply that a peak growth was reached around 1992. It is difficult to attribute this effect to increasing stand density in the latter years, given that the mean basal area was about the same in the periods 1983-2002 and 2003-2020.

Variations in environmental conditions over time may strongly influence tree growth rates. The rising summer temperatures are mainly due to an increasing level of atmospheric CO2. Temperature sum is also a very strong variable at the determination of SIS. However, the stable level for basal area growth is very surprising and posed the question if the increased CO2 level itself has a direct impact on the height growth. The tallest trees are often found in humid climates. Givnish et al (2014) studied Eucalyptus in Australia over a humidity gradient and found that tree heights decreased from 60 to 10 meters from the humid to the dry sites. The water use efficiency means the balance between carbon intake and water loss through the leaf stomata and seems to have a strong impact on the height growth of trees. Increasing atmospheric CO2 levels means that the stomata can take in required carbon faster with less loss of water (Keenan et al. 2013). Increased nitrogen deposition has probably contributed to an accelerated tree height growth

(Etzold et al. 2020), but the atmospheric nitrogen deposition in Sweden has decreased by 30 % in the period 1983-2013 (Andersson et al. 2018).

It remains unclear for how long the increases in maximum tree height growth will continue in Swedish forests. It is possible that drought-induced hydraulic limitations (e.g. the 2018 and 2019 heatwaves in Europe) and the frequency and intensity of forest disturbance agents such as storms, fires, pests and diseases may constrain the maximum tree height growth under a changing climate (Silva et al. 2010; Girardin et al. 2014; Belyazid and Giuliana 2019; Forzieri et al. 2021).

The changes in tree growth rates demonstrate a changing growing condition in Swedish forests. This implies potential variations in tree allometry such as height to volume relationships and stand composition, especially for even-aged (monocultural) stands. The use of top height as an indicator of site productivity may not be valid anymore, site curves must be revised and include a time factor that mirror the ongoing transition of the growing conditions. Thus, new expressions for forest site productivity in contemporary decision support systems are needed to enhance accurate predictions of tree growth and sustainable forest management in a changing climate.