Pre-harvest Soil Acidification, Liming or N Fertilization Did Not Significantly Affect the Survival and Growth of Young Norway Spruce

Acidification, liming, and N fertilization affect a number of soil properties. Such changes may have an impact on forest regeneration and yield. The aim of this study was to investigate the survival and growth of Picea abies (L.) Karst. planted on plots that had been acidified (in 12 annual treatments totalling 600 or 1200 kg S ha–1 in the form of elemental sulphur), limed (12 × 500 = 6000 kg lime ha–1 in the form of CaCO3) or N-fertilized (3 × 200 = 600 kg N ha–1 in the form of urea) prior to harvest. Trees growing on plots treated with a combination of the N plus the lower S application were also tested. None of the treatments, applied in three replicate stands, significantly influenced either survival or growth of Picea abies trees during the first 11 growing seasons after planting.


Introduction
Several soil properties are affected by acidification (Tamm 1989, Abrahamsen et al. 1994), liming (Hallbäcken and Popovic 1985, Derome et al. 1986, Abrahamsen et al. 1994, Smallidge et al. 1993, Staaf et al. 1996), and by N fertilization (Bååth et al. 1981, Popovic 1985, Martikainen et al. 1989, Nohrstedt et al. 1989, Nohrstedt 1990, Nohrstedt 1992, Jacobsson and Nohrstedt 1993).Such changes to the soil may affect forest regeneration and forest yield either directly, for instance by influencing nutrient supply to the seedlings, or indirectly, by altering the development of field vegetation (Gerhardt andKellner 1986, Kellner 1993), causing changes in competition for light, water and nutrients.In addition, fine-root production (Persson et al. 1995) and the ectomycorrhizal communities (Finlay 1995, Arnebrant et al. 1996) associated with conifer trees may be affected by acidification, and by the addition of lime or N.However, published data based on field experiments have not confirmed any distinct short-term effects on the regeneration, i.e. survival and early growth of coniferous seedlings, as a result of preharvest soil acidification, liming or N fertilization (Högbom et al. 2001, Sikström 2001a, 2004).However, the long-term effects of these treatments are unclear.Sikström's (2001a) conclusion on the lack of short-term effects was based on a field experiment conducted at the Farabol-experimental site in SE Sweden.The original experiment at that site was designed to study the effects of experimental acidification, liming and nitrogen fertilization on subsequent tree-stem growth and soil properties in mature stands of Picea abies (L.) Karst.(Andersson et al. 1995).The study by Sikström (2001a) assessed the experimental plots after clearcutting.In the present paper, data from an additional assessment, conducted 11 years after planting, is reported.Thus, the aim of the study was to investigate the survival and growth of P. abies trees planted in plots that had been acidified, limed or N-fertilized prior to the final felling.To my knowledge, this is the only field experiment describing survival and growth of conifer seedlings on pre-harvest acidified and limed forest soils, and one of few on pre-harvest N fertilized soils.

Site Description
The experimental site is situated in SE Sweden (56°26´N, 14°35´E), 140-150 m a.s.l.The annual mean temperature ) is +6.4° and the precipitation 669 mm at the nearest climate station, Ryd, 10 km away (Alexandersson et al. 1991).The soils are mesic, sandy-silty tills, characterized as Haplic podzols according to the FAO soil classification system (1988).Site indices were estimated to be G30-G32 according to the Swedish classification system (H 100 , m; Hägglund 1973), corresponding to site quality classes of 10.1-11.3m 3 ha -1 yr -1 (Hägglund and Lundmark 1982).The stand and site characteristics, as well as deposition (bulk and throughfall) are described in detail by Andersson et al. (1995) and reviewed by Sikström (2001a).

Experimental Design and Treatments
A randomised block design was used, including three blocks and six treatments.Each experimental plot was 30 m × 30 m in size, except for the plots in block 3, which were 34 m × 34 m.The six treatments were: control (0), low sulphur (S), high sulphur (2S), lime (CA), nitrogen (N) and a combination of N and S (NS).The S treatment consisted of annual applications of 50 kg ha -1 and 2S of 100 kg ha -1 elemental sulphur powder (100% S) over the 12-year period 1976-1987.The CA treatment also involved annual applications, of 500 kg ha -1 limestone powder.The limestone consisted of 90% calcium carbonate (CaCO 3 ), with 0.3% Mg (the 12 annual treatments totalling 18 kg ha -1 yr -1 ), 0.2% K (12 kg ha -1 yr -1 ), 0.04% P (2.4 kg ha -1 yr -1 ) and 0.02% S (1.2 kg ha -1 yr -1 ).N was given in doses of 200 kg ha -1 as urea [(NH 2 ) 2 CO] on three occasions, in 1976, 1980 and 1985.Data collected in 1991, a year before the final felling and the start of the regeneration study, showed that many soil chemical properties had been affected by the treatments (Andersson et al. 1995;shown, in part, in Sikström 2001a).In treatment CA, for example, pH, BS and exchangeable Ca and Mg had increased significantly compared with the control.In contrast, pH had decreased significantly after the addition of sulphur powder, as had exchangeable K and Mg in the humus.The addition of N resulted in minor residual effects on soil chemical properties.

Regeneration, Seedling Survival and Growth
The regeneration study was based on 20 m × 20 m plots located in the centre of the original plots.
The former mature stands were whole-tree har- In autumn 2002, the survival of the trees, along with their total height and diameter at breast height, were recorded.The stem volume was estimated using equations published by Andersson (1954).
In the survival analysis, the original seedlings, i.e. those planted in May 1992, and replacement seedlings were analysed both separately and together.Calculations for all the growth variables (mean tree per treatment) were based on the data for the original seedlings that had not been seriously damaged during the experimental period.
However, a severe episode of browsing by roe deer in April 1993, resulting in damage to most of the seedlings, forced a major exception to this general rule.The differences between treatment means (proportion of browsed seedlings) were not statistically significant (for further details, see Sikström 2001a).The growth variables presented on an area basis (mean per plot) were based on all living trees in the plots in the autumn of the year 2002.The total numbers of seedlings and "undamaged" seedlings per plot used in the growth calculations for 2002 were 79-114 and 22-68, respectively (see also Table 1).

Statistical Analyses
Data on the proportion of survivors, mean tree size and mean height growth (for the period 1996-2002) of the original seedlings, and the mean stem basal area and stem volume per plot in the autumn of 2002, were analysed using a two-way analysis of variance.Two-way analyses of variance were also used to assess the frequency of replacement seedling survival.The GLM procedure of the SAS ® program (SAS Institute Inc. 1999) was used to calculate models of the form: where μ = the overall mean; α i = the fixed effect of pre-harvest treatment, i = 1,..., 6; b j = the random effect of block, IND( , ) 0 2 σ b , j = 1, 2, 3; e ij = the random error, IND( , ) 0 2 σ e .In addition, several covariates were tested in the model.The covariates, proportion of seedlings from the different planting occasions in the individual plots [(i) original seedlings, (ii) replacement seedlings planted in autumn 1992, and (iii) replacement seedlings planted in spring 1994] were tested individually to determine their influence on mean stem basal area and stem volume per plot.
Statistical analyses were also performed assuming a split-plot experimental design.In this case, pre-harvest treatment was regarded as the main plot and planting occasion as the subplot treatment.The following variables were tested: survival and tree size in autumn 2002, and height growth for the period 1996-2002.The MIXED procedure of the SAS ® program (SAS Institute where p ij = the random error of the main plot, IND ( , ) 0 2 σ p ; γ k = the fixed effect of planting occasion, k = 1, 2, 3 (S92, A92 and A94; see Table 1); (αγ) ik = the interaction between treatment and planting occasion; and d ijk = the random error, IND( , ) 0 2 σ d .For the other terms, see Eq. 1. Differences between the class means for treatment, block and planting occasion were evaluated using Tukey's significant differences (HSD) mean separation test.In addition, differences due to adding N [(N-0) + (NS-S)] and S [(S-0) + (NS-N)] were tested.

Results
There were no statistically significant treatment effects (p = 0.21-0.95) on the survival in 2002.This was the case for both the seedlings planted originally and for both of the two batches of replacement seedlings (Table 2, Fig. 1).For the original seedlings, the mean survival rate associated with most of the treatments was between 53% and 66%, except for the N treatment where it was 36%.When trees from all plantings were analysed together (the mixed model) there was no treatment effect, although the survival rate of seedlings planted in spring 1994 tended to be c.10% higher than those planted in spring or autumn 1992 (Table 3).
Variables describing the mean tree size and growth (mean height, mean stem basal area, mean stem volume and mean height growth [1996][1997][1998][1999][2000][2001][2002] showed no significant (p < 0.05) differences between the treatments (Table 4, Fig. 2).This was also the case for the mixed-model analyses (Table 3).However, the means in the plots were larger for trees planted in spring or autumn 1992 than in autumn 1994 (Table 3).
The growth variables on an area basis (mean basal area and mean volume per plot) also did not show any differences between treatments (Table 4).The proportion of replacement seedlings in the plots planted in spring 1994 influenced the mean stem basal area per plot (p covariate = 0.098) and the mean stem volume per plot (p covariate = 0.090).However, the p-value of the treatment was only marginally affected (p treat- ment = 0.22-0.24)compared with the analyses of variance where a covariate was not included in the model (Table 4).When a covariate was included, the relative least-square means per treatment (in relation to the control) were altered by less than 2 percent units (p.u.) for all treatments, except 2S (+7-8 p.u.) and N (+10-11 p.u.).

Discussion
The results, 11 years after planting, are in agreement with those reported for the site five years after planting (Sikström 2001a).
In the previous report (Sikström 2001a), a tendency towards reduced survival amongst the original seedlings following pre-harvest N addition was   2. Means and p-values for the contrasts: a) S92 (57%) and A92 (57%) < S94 (66%); p < 0.076 b) S92 (4.16 m) and A92 (4.07 m) > S94 (3.40 m); p < 0.001 c) S92 (3.31 m) and A92 (3.29 m) > S94 (2.75 m); p < 0.001 d) S92 (0.194 × 10 -3 m 2 ) and A92 (0.175 × 10 -3 m 2 ) > S94 (0.104 × 10 -3 m 2 ); p < 0.001 e) S92 (0.0058 m 3 ) and A92 (0.0052 m 3 ) > S94 (0.0030 m 3 ); p < 0.001 identified.In spite of this, it was concluded that pre-harvest N fertilization is unlikely to substantially affect survival of P. abies (Sikström 2001a) seedlings.This conclusion has been strengthened as a result of the latest monitoring, since the tendency reported earlier was reduced (Table 2), as found in similar studies (Högbom et al. 2001, Sikström 2004).The current survey also identified a somewhat higher survival rate for plants from the last planting occasion (spring 1994) than plants from the two other planting occasions (spring 1992 and autumn 1992).However, for the three seedling categories there were different lengths of the calculation periods, i.e. the time from planting to the last survival assessment.At the time of the latest survey, in autumn 2002, the seedlings from the last planting occasion were significantly smaller than those planted earlier (Table 3).Thus, the proportion of seedlings in the plots originating from the last planting occasion influences the standing tree volume on an area basis.This was confirmed in the analysis of variance, when a covariate (proportion of replacement seedlings planted in spring 1994) was included in the statistical model.Consequently, the most relevant comparison of treatment effects on growth should be the mean tree size of individuals planted at the same time.4.
Simply considering the mean values, the trees in the limed plots were somewhat larger than those in the other treatments, although this difference was not statistically significant (Table 4, Fig. 2).This was also the case five years after planting (Sikström 2001a, Fig. 2).In this experiment, of all treatments, liming was associated with the most marked changes in many of the documented soil chemical properties, both in the humus and the mineral soil, before final felling and regeneration (Andersson et al. 1995).However, these effects were not reflected in the tree growth of the previous mature stand (Andersson et al. 1995).Liming is known to have variable effects on growth (see Sikström 2001a and references therein), and its effect seems to be associated with soil fertility (Sikström 2001b).In the former mature stand, N was the only element applied (in treatments N and NS) that significantly increased tree growth (Andersson et al. 1995), indicating that growth of the previous stand at this site was limited by N availability.The cover of field-layer vegetation tended to be denser in the CA-treated plots than elsewhere, four years after planting (Sikström 2001a).One possibility is that the pre-harvest liming might have promoted somewhat better growth conditions after final felling, for example by increasing plant-N availability, than the other treatments.
Neither the survival nor growth (Table 2) of the P. abies seedlings was affected by the pre-harvest additions of sulphur powder alone, although the sulphur treatments (especially 2S) were associated with a significant reduction in the pH of the mineral soil, and in levels of exchangeable K and Mg in the humus, in the year before final felling and planting (Andersson et al. 1995).These results are consistent with growth data from other Nordic field experiments on soil acidification (Tamm and Popovic 1989, Tamm 1989, Abrahamsen et al. 1994), except when there was a very large input of acid, which resulted in decreased stem-wood growth (Abrahamsen et al. 1994).
In conclusion, allowing for both the advantages and the limitations of the Farabol experiment, as discussed by Sikström (2001a), the pre-harvest soil acidification, liming or N fertilization did not seem to significantly influence either seedling survival or growth of P. abies trees during the first decade after planting.

Fig. 1 .
Fig. 1.Survival (%) of Picea abies trees after planting in forest soils that were acidified (S, 2S), limed (CA) or nitrogen fertilized (N) prior to harvest (0 = control).Data refer to seedlings originally planted in spring 1992 (S92), replacement seedlings planted in autumn 1992 (A92) and in spring 1994 (S94) at each time of monitoring.Means of three blocks per treatment.For details of statistical analyses, seeTable 2.

Table 4 .
Growth of P. abies seedlings during the 11 growing seasons after planting in forest soils that were acidified (S, 2S), limed (CA) or nitrogen fertilized (N) prior to harvest (0 = control).Means of three blocks per treatment.Treatment effects from the analyses of variance (GLM) and least significant differences (HSD) between treatments are given.

Fig. 2 .
Fig. 2. Height development of Picea abies trees after planting in forest soils that were acidified (S, 2S), limed (CA) or nitrogen fertilized (N) prior to harvest (0 = control).Means of three blocks per treatment.Intervals for least significant differences from the analyses of variance are given for each time of inventory.For statistical analyses, see Table4.

Table 1 .
Mean number and proportion of living Picea

Table 2 .
Survival SikströmPre-harvest Soil Acidification, Liming or N Fertilization Did Not Significantly Affect the Survival and Growth …

Table 3 .
Fixed effects from the analyses of variance (mixed model) for survival and growth variables of Picea abies (means per tree) 11 growing seasons after planting in forest soils that were acidified (S, 2S), limed (CA) or nitrogen fertilized (N) prior to harvest (0 = control).N = nominator, D = denominator and df = degrees of freedom.Means and contrasts are mainly given for the classes of statistically significant (p < 0.05) effects.
Data refer to seedlings planted in spring 1992 (S92) that remained undamaged throughout the experimental period, except for a severe browsing episode by roe-deer in April 1993.b) Data refer to all seedlings in the plots in autumn 2002.