*Article* **The Effects of Soil Moisture on Harvesting Operations in** *Populus* **spp. Plantations: Specific Focus on Costs, Energy Balance and GHG Emissions**

**Farzam Tavankar 1, Mehrdad Nikooy 2, Francesco Latterini 3, Rachele Venanzi 4, Leonardo Bianchini <sup>4</sup> and Rodolfo Picchio 4,\***


**Abstract:** Background: Poplar tree plantations for wood production are part of a worldwide growing trend, especially in moist soil sites. Harvesting operations in moist sites such as poplar plantations require more study for detailed and increased knowledge on environmental and economic aspects and issues. Methods: In this study, the effects of soil moisture content (dry vs. moist) on productivity, cost, and emissions of greenhouse gases (GHG) caused by operations of different harvesting systems (chainsaw-skidder and harvester-forwarder) were evaluated in three poplar plantations (two in Italy and one in Iran). Results: The productivity (m<sup>3</sup> h−1) of both systems in the dry sites were significantly higher (20% to 30%) than those in the moist sites. Production costs (€ m−3) and GHG emissions (g m−3) of both systems in the dry sites were also significantly lower than those in the moist sites. The productivity of the harvester-forwarder system was about four times higher, and its production cost was 25% to 30% lower than that of the chainsaw-skidder system, but the calculated GHG emissions by harvester-forwarder system was 50–60% higher than by the chainsaw-skidder system. Conclusions: Logging operations are to be avoided where there are conditions of high soil moisture content (>20%). The result will be higher cost-effectiveness and a reduction in the emission of pollutants.

**Keywords:** skidding productivity; logging cost analysis; harvesting site conditions; sustainable forest operations

## **1. Introduction**

Poplar planting has occurred around the world for a very long time. The plantations in Iran and Italy provide an important source of wood supply. At present, in both countries there are over 100,000 ha of these monospecific plantations (50,000 ha in Iran and 66,000 ha in Italy) and they mainly consist of *Populus deltoides* and *P. euramericana* [1–3]. Although poplar plantations cannot be currently considered among the main sources of wood in both countries, their importance is rapidly increasing [4,5]. Poplar wood shows interesting features, such as uniform mechanical properties and a high percentage of juvenile wood. These make it possible to obtain several products from plantations, i.e., building and veneering material, paper pulp and wood chips for bioenergy [2,6,7]. Moreover, the poplars in both the Italian and Iranian conditions can reach a growth rate of approximately 10–30 m3 ha−<sup>1</sup> year<sup>−</sup>1, which is substantially higher than local tree species [8–10].

**Citation:** Tavankar, F.; Nikooy, M.; Latterini, F.; Venanzi, R.; Bianchini, L.; Picchio, R. The Effects of Soil Moisture on Harvesting Operations in *Populus* spp. Plantations: Specific Focus on Costs, Energy Balance and GHG Emissions. *Sustainability* **2021**, *13*, 4863. https://doi.org/10.3390/ su13094863

Academic Editor: António Dinis Ferreira

Received: 19 March 2021 Accepted: 24 April 2021 Published: 26 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Considering the growing importance of poplar plantations as a source of wood material, it is necessary to assess the technical and environmental characteristics of the harvesting operations in these plantations. Wood production from artificial stands is indeed a simplified multistage process as compared to forestry production, but proper planning of the logging operations is crucial when viewing the overall sustainability of the intervention [11]. The concept of sustainability in the forestry sector is strictly related to the paradigm of sustainable forest operations (SFO) [12]. SFO refers to the implementation of logging operations which are able to meet the requirements of all three pillars of sustainability (economy, environment and society) [13,14]. Machine productivity and operation costs are the two main factors in evaluating harvesting operations regarding the economic aspect of plantation management [15–18]. Having accurate information on the productivity of logging machines is therefore a key issue for the economic assessment of production. Work productivity evaluation is a complex issue, considering that the working performance of a given harvesting system is related to several factors, for instance, machine type, tree size, logging intensity, number of trees per hectare, terrain conditions, operator skills and planned treatment [19–23]. However, always considering the concept of SFO, assessing and optimizing work performance is not enough to obtain sustainable logging. The environmental aspect is also fundamental [24–27]. Along with soil impact and stand damage, greenhouse gases (GHG) emission related to mechanical operations is a major aspect to be evaluated so as to assess the environmental performance of a given harvesting system [28,29].

Considering the points listed above, it is a major challenge for forest managers to evaluate the manifold consequences of decisions and to estimate the economic and environmental performance of different alternatives before carrying out action.

These statements are valid for all forestry and agro-forestry interventions, but are even more important when dealing with poplar plantations, which show particular features. Poplar plantations are often located in plain or floodplain lands, which means that very often harvesting operations take place in soil conditions with high moisture content. Among all the variables, the soil moisture content during logging can significantly influence the degree of soil disturbance, with greater potential for higher soil compaction on wet/saturated soils than on dry ones [30]. Several studies have focused on the effects of different levels of soil moisture content during harvesting on soil disturbance and the physical properties of the soil [31,32]. On the other hand, the effects of different soil moisture content at the time of felling and skidding on the harvesting operations' performance has not yet been studied.

Considering these peculiarities of poplar plantations and their growing importance in both the Italian and Iranian forestry systems, the main aims of the present study were: (i) to provide a comparative analysis of different harvesting technologies for poplar plantations; (ii) to determine the influence of soil moisture on harvesting operations' performance both from an economic and an environmental point of view.

This study will make a detailed statement of what can be verified and a general statement of what cannot. More precisely, one can state that operations in specific soil conditions could produce the performance reported here under the specific conditions of this study. This knowledge is worth having, although poorly suited to any generalization, and can be used to estimate an expected forest operation's performance that may occur under different conditions with some level of approximation.

#### **2. Materials and Methods**

#### *2.1. Study Areas*

This study was carried out in three different geographical areas, two in Italy and one in Iran, investigating three different harvesting operations in poplar plantations.

The Iranian poplar plantation (IR) is located in the coastal area of the Caspian Sea in the Guilan province in northern Iran. The total area of the plantation is 60 ha, situated in flat terrain at an altitude range from 0 m to 20 m a.s.l. The average annual rainfall is from 1260 to 1340 mm and most of the precipitation occurs between the months of September and December. The average annual temperature is +15 ◦C, with the minimum during the winter at a few degrees below 0 ◦C, and the maximum at +25◦ during the summer. The soil type is clay loam with poor drainage. This plantation was divided into two areas of 30 ha each. The harvesting operation in 30 ha for the dry site was performed in the first half of September 2019 before the rainfall. Harvesting operations were carried out in the second half of September 2019 after rainfall in the 30 ha moist site. The soil moisture content was 14.4% in the dry site and 34.6% in the moist site. Trees were felled by chainsaw, and whole trees were extracted to roadside landings by wheeled skidder Timberjack 450C. Finally, processing operations were motor-manually performed at the landing site.

The first Italian poplar plantation (IT1) was located in the Lazio region in central Italy. The total area of the plantation is 20 ha and it is situated in flat terrain and with an altitude range from 90 m to 110 m a.s.l. The average annual rainfall is from 830 to 900 mm and most of the precipitation occurs from October to December. The average annual temperature is +14.9 ◦C with a minimum during the winter at 2.5 ◦C, and a maximum at +30.7 ◦C during the summer. The soil type is clay loam with a low level of organic matter, nitrogen and phosphorus and with poor drainage. This plantation was divided into two areas of 10 ha each. The harvesting operations in 10 ha on the dry site were performed in the second half of June 2018. Harvesting operations were carried out in the first half of April 2018 after rainfall in the 10 ha moist site. The soil moisture content was 12.1% in the dry site and 36.8% in the moist site. As in the Iranian site, trees were motor-manually felled by chainsaw, and whole tree extraction was carried out by a wheeled skidder Timberjack 450C. In this case too, motor-manual processing with a chainsaw was carried out at the landing site.

The second Italian poplar plantation (IT2) was located in the Veneto region in North Italy. The total area of the plantation is 20 ha and situated in flat terrain with an altitude range from 10 m to 30 m a.s.l. The average annual rainfall is from 730 to 850 mm and most of the precipitation occurs from September to November. The average annual temperature is +13.6 ◦C with a minimum during winter of 0.1◦C, and a maximum of +29.2 ◦C during the summer. The soil type is clay loam texture with a low level of organic matter and with poor drainage. This plantation is divided into two areas of 10 ha each. The harvesting operations in the 10 ha dry site were performed in the first half of July 2018. Harvesting operations were carried out in the second half of September 2018 after rainfall in the 10 ha moist site. The soil moisture content was 15.0% in the dry site and 35.7% in the moist site. Trees were mechanically felled and processed by a harvester and cut to length through an extraction system by a forwarder. Technical characteristics of the machinery used in the various harvesting sites are given in Table 1. Average dendrometric characteristics of the three plantations are shown in Table 2. A preliminary analysis for dendrometric characteristics of the three different stands was done by one-way ANOVA to check for differences among the average values of the three plantations. There were no significant differences of dendrometric characteristics between the Italian sites (IT1 and IT2). However, density, basal area and standing volume of trees in the Iranian site were higher than in the Italian sites.

## *2.2. Data Collection and Analysis*

Dendrometric data were obtained through systematic plot sampling. Grid dimension was 150 m × 150 m, the area of each circular plot was 1256 m<sup>2</sup> (20 m radius), and in total 20 plots in each area were established. Diameter at breast height (dbh) and height of tree species were measured by caliper and clinometer, respectively, in each plot. The volume of winched logs was calculated by Huber's formula (V = Am × L), where V is log volume (m3), Am is the middle point cross-sectional area of log (m2), and L is the length of log (m).

Soil samples were collected with a steel ring (inside diameter 5 cm, length 10 cm) and immediately put in hermetic plastic bags and labeled. The wet weight of all samples was measured before transfer to the laboratory (on the same sampling day). In the laboratory, soil samples were dried in an oven at 105 ◦C for 24 h until reaching a constant mass to determine the soil moisture content.


**Table 1.** Specification of the mechanization used in the three harvesting sites.

**Table 2.** Average dendrometric stand and main wood characteristics before harvesting in the three poplar plantations. The wood density values (±SD) showed refer to fresh matter and recorded during the harvesting operation.


A time-motion study was carried out to evaluate working productivity. Each working cycle was stop watched individually, separating productive time from delay time [33]. Calculated delay factor represents the quotient of delay time over net cycle time. Productivity was evaluated both on delay-free time and on actual total time, inclusive of all delays. Inclusion of delays was not capped on the basis of a maximum event duration. Scheduled Machine Hours (SMH) include all the time the machine is scheduled to work, whereas Productive Machine Hours (PMH) represent the time during which the machine actually performs work, excluding the time lost to both mechanical and non-mechanical delays.

The working group had 10 to 15 years of work experience with the machines and they were able to service and repair them.

The working cycles are reported, for the three areas, in Tables 3 and 4. Continuous time was recorded to the nearest second with a chronometer. The cycle times of the machines were divided into time elements (process steps) that were considered characteristic of this work.

**Table 3.** Description of felling and processing cycle elements of harvesting in the three yards.



**Table 4.** Description of extraction cycle elements of harvesting in the three yards.

Details of the harvested trees and volume in each treatment are given in Table 5.

**Table 5.** Harvested trees, volume and working cycles for work productivity analysis in each treatment (data reported to FU of 1 t of fresh mass showed in Table A1).


The system boundaries for the study area were set to those of the harvesting operations, from the felling to the landing site. The Functional Unit (FU) for the analyses was the cubic meter of round wood (m3); in Appendix A the data referring to another Functional Unit (FU) are shown (1 t of fresh mass, following the data shown in Table 2). This is important in order to compare these results more readily with other studies.

Operational costs were estimated according to the Miyata method [34] as previously explained in Spinelli et al. [35]. Economic evaluation of the different machines was carried out taking into consideration different periods of use. The skidder, harvester and forwarder depreciated by 1200 SMH per year [35,36] in a depreciation period of 10 years [37]. The chainsaw for the felling depreciated by 800 SMH per year in a depreciation period of 2 years [36]. Labor cost was set at € 15 SMH−<sup>1</sup> inclusive of indirect salary costs [38]. Lubricant consumption was calculated as reported by Picchio et al. [39].

Costs for insurance, repair and service were obtained by literature analysis [35], while the fuel and lubricant prices were taken by a market survey (second semester 2019) conducted upon three company products. The calculated operational cost, as reported in similar studies [35], was increased by 10% to account for overhead costs [40].

Focusing instead on the environmental aspects, an energy consumption analysis was performed, applying the Gross Energy Requirement (GER) method [41]. Indirect input (MJ kg−1) of harvesting machinery was evaluated taking into consideration the average energy value of the raw materials. This is related to several parameters, i.e., quantitative presence (%), total mass of the machine (kg), overall service life of the machine (h m<sup>−</sup>3) and use of the machine during harvesting. Energy consumption related to human manpower was evaluated according to what was reported in previous works [42–44], through the application of a standard value equal to 0.030 MJ min−<sup>1</sup> worker<sup>−</sup>1.

To calculate the energy balance, the energy value of poplar wood was determined as Higher Heating Value (HHV) (CEN/TS 14918), on 30 random samples, through Parr calorimeter, model 6200 [45].

Regarding pollutant emissions during logging operations, emissions related to fuel were evaluated as the sum of the emissions during combustion (Efc) and the emissions produced within the production and logistic process (Efp). For Efc assessment of fuel energy content, the emission factor of the engine and the thermal efficiency of the combustion were taken into consideration, as reported in Klvac et al. [46] and Athanassiadis [47].

Dealing with Efp assessment, fuel energy content and emission factors were obtained by [46], but HC emission factor was taken from [47].

#### *2.3. Statistical Analysis*

The first step in statistical analysis was checking for normality using the Kolmogorov-Smirnov test and for homogeneity of variance using the Levene test. Averages of dendrometric characteristics, skid trail network, and average extraction cycle time elements in each area between the two site conditions (moist and dry soil) were compared by independent t test. A regression analysis of time study data was used to check the model's capability of predicting productivity as a function of statistically significant independent variables such as distance and load size. If the data were not normally distributed, a non-parametric Spearman's rank coefficient was applied to analyze the correlation between the variables.

A major focus was placed on extraction operations investigating the relationships between time elements and dendrometric characteristics of extracted timber, and between time elements and bunching-extraction distance. This investigation was performed by nonlinear regression analysis, performed by SPSS 19.0 software (New York, NY, United States).

#### **3. Results**

Results of the *t*-test showed no statistically significant differences regarding both extraction distance and mean volume per working cycle between moist and dry sites in all the three plantations. Bunching distance was statistically lower in the IT1 dry site than in the moist one, while in IR and IT2 no statistically significant difference was found for this parameter (Table 6).

**Table 6.** Extraction trail and corridors' average features for the three areas in the two soil moisture conditions (mean ± SD), From the *t*-test for independent samples applied, statistically significant differences (*p* < 0.05) between the average values are highlighted (underlined text) (data reported to FU of 1 t of fresh mass showed in Table A2).


Data regarding working time analysis of felling and processing operations are given in Table 7 and Figure 1. In all three plantations felling operation time (motor-manual in IR and IT1 and mechanized in IT2) did not show any statistically significant differences between dry and moist sites. There were some differences found in processing and moving time. In every study area the most time-consuming operation was processing.

The soil moisture in IT2 did not affect working productivity, with no statistically significant difference among different soil moisture conditions for every phase, and consequently also for the overall working time. In IR1 and IT1, instead, both TET (Total Effective Time) and TGT (Total Gross Time) were significantly higher in moist conditions than in dry ones, with higher DT (Delay Time) also in the moist soil. However, as previously reported, such differences are related not to the felling operations, but only to moving (both IR and IT1) and processing (only IR).

Working time analysis data in bunching and extraction are reported in Table 8 and Figure 2. The only phase which did not show an effect of soil moisture on productivity was the landing operation (LO), while all the other phases were influenced by the moisture content of the soil. This led to a significant difference in overall working times (both TET and TGT), related to the different site conditions, in all the three yards. In particular, higher soil moisture negatively affected working productivity.

**Table 7.** Description statistics of felling-processing operations for the harvesting sites studied referring to a single tree. From the *t*-test for independent samples applied, statistically significant differences (*p* < 0.05) between the average values are highlighted (underlined text). Statistical comparisons are between columns.


M: moving, F: felling, P: processing, DT: delay time, TET: total effective time, TGT: total gross time.

**Figure 1.** Working time distribution for felling-processing operations in the harvesting sites studied.

**Table 8.** Description statistics of bunching-extraction operations for the harvesting sites studied referring to single cycle. From the *t*-test for independent samples applied, statistically significant differences (*p* < 0.05) between the average values are highlighted (underlined text). Statistical comparisons are between columns.


TUL, travel unloaded; B, bunching; TL, travel loaded; LO, landing operations; DT, delay times; TET, total effective time; TGT, total gross time.

**Figure 2.** Working time distribution for bunching-extraction operations in the harvesting sites studied.

The analysis of the various factors influencing extraction time (Table 9) revealed that the parameter with the highest impact on time consumption was bunching–extraction distance, with R2 values ranging from 0.6 to 0.8 for both dry and moist sites in all the three yards.

**Table 9.** Cycle time (Y) equations of bunching–extraction in studied sites. D: distance of bunching-extraction; and LV: load volume.


Increased bunching–extraction distance obviously led to increased bunching–extraction time; however, it is interesting to notice (Figure 3) how this effect is less evident in forwarding operations (IT2) than it is in winching operations (IR and IT1).

**Figure 3.** Graphical regression analysis referring to bunching–extraction time in relation to bunching–extraction distance in the studied areas (IR-d: Iranian dry site; IR-m: Iranian moist site; IT1-d: skidding Italian site with dry soil; IT1-m: skidding Italian site with moist soil; IT2-d: forwarding Italian site with dry soil; IT2-m: forwarding Italian site with moist soil).

Focusing on the overall harvesting system productivity (Figure 4), moist soil showed negative effects in all three plantations. In detail, SMH in dry soil conditions was 5.671 m3 h−<sup>1</sup> in IR, 6.403 m3 h−<sup>1</sup> in IT1 and 23.761 m3 h−<sup>1</sup> in IT2; while, respectively, were 12.53%, 18.68% and 16.27% lower in the moist soil. Moreover, the higher moisture content of soil also resulted in a higher percentage difference between PMH and SMH, i.e., 2.73% vs. 3.08% in IR; 3.39% vs. 4.40% in IT1; and 4.4% vs. 5.83% in IT2. Referring to the single operations, SMH in felling-processing was 7.541 vs. 7.101 m3 h−<sup>1</sup> in IR; 8.238 vs. 7.887 m3 h−<sup>1</sup> in IT1; and 37.941 vs. 36.481 m3 h−<sup>1</sup> in IT2. Bunching-extraction productivity was also negatively affected by higher soil moisture; specifically, SMH was 22.870 vs. 16.455 m3 h−<sup>1</sup> in IR; 28.746 vs. 15.325 m3 h−<sup>1</sup> in IT1 and 63.580 vs. 43.758 m3 h−<sup>1</sup> in IT2.

Focusing on harvesting costs, the results of the economic evaluation carried out within the present study are given in Tables 10 and 11.

The details of hourly costs reported in Table 10 show how the harvesting machinery applied in IT2 (harvester and forwarder) presents higher hourly costs, mostly related to the higher purchase price. However, the higher productivity of this fully mechanized harvesting system allowed IT2 to have a lower cost per m3 of timber produced (Table 11).

Regarding the influence of soil moisture conditions on harvesting costs, it is evident that the negative influence on working performance correlated to higher moisture also led to higher harvesting costs. In detail, this was about 16%, 26% and 16% higher in the moist site than in the dry one for IR, IT1 and IT2, respectively.

**Figure 4.** Average yard productivity (bars) and possible increase of performance (lines) from SMH to PMH for the six harvested sites (data reported to FU of 1 t of fresh mass showed in Figure A1).


**Table 10.** Summary cost assessment of mechanization used in the logging activities studied.


**Table 10.** *Cont.*

**Table 11.** Harvesting costs for one cubic meter of wood and percentage of costs at two main operations (felling–processing and bunching–extraction to landing) in the studied sites (data reported to FU of 1 t of fresh mass showed in Table A3).


Regarding environmental aspects, the results of the analysis of energy efficiency are given in Table 12. The highest energy input was reported for IT2, due to the complete mechanization of the overall harvesting operations. The effects of moisture on environmental performance can be observed in all of the three plantations where higher soil moisture led to lower energy efficiency, more exactly, 97.7% vs. 97.0% in IR; 98.1% vs. 96.9% in IT1 and 97.0% vs. 96.6% in IT2.

**Table 12.** Total energy inputs and balance in the studied harvesting yards (data reported to FU of 1 t of fresh mass showed in Table A4).


In IR and IT1 a major part of the energy input is related to bunching–extraction operations in both moist and dry soil conditions. Instead, in IT2, felling operations via harvester were the reason for the highest portion of energy input in this yard, in both soil conditions (Figure 5). Interestingly, moist soil led to an increased portion of energy input related to bunching–extraction in all three yards (77.2% vs. 81.6% in IR; 74.6% vs. 84.0% in IT1 and 25.8% vs. 32.7% in IT2), showing how this operation was most influenced by soil moisture when regarding environmental issues.

As shown in Table 13 and Figures 6 and 7, soil moisture also showed negative effects regarding pollutant emissions in the three different yards, with increasing emissions for all the investigated parameters in moist soil conditions. What is more, mechanized felling via harvester (IT2) led to higher emissions in comparison to motor-manual felling (IR and IT1).

**Figure 5.** Energy inputs percentage for each harvesting operation assessed in the studied sites.

**Table 13.** Total emission assessed in the studied harvesting yards (data reported to FU of 1 t of fresh mass showed in Table A5).


**Figure 6.** Percentage distribution of total PM10 emission in the harvesting sites studied, data shown for single operation (data reported to FU of 1 t of fresh mass showed in Figure A2).

**Figure 7.** Percentage distribution of GHG emission in the harvesting sites studied, data shown for single operation and reported in CO2 equivalent (data reported to FU of 1 t of fresh mass showed in Table A3).

#### **4. Discussion**

## *4.1. Comparison of Harvesting Systems Performance*

Several studies on work productivity and cost analysis in poplar plantations are available in the literature, even if most of these deal with Short Rotation Coppice (SRC) plants for bioenergy production [48]. Indeed, few studies have focused on productivity analysis in poplar plantations for timber production. In a poplar plantation located in Serbia, Danilovic et al. [49] reported a productivity for mechanized felling–processing via harvester of 30.3 to 34.7 m<sup>3</sup> h−1, depending on working method and stem dimension. In the same year, Spinelli et al. [50] carried out an extensive productivity and cost analysis in a 25 year old poplar plantation in Italy, reporting an average work productivity (SMH) for motor-manual felling and processing via chainsaw of 6.3 m3 h−<sup>1</sup> and a value of 21.1 m3 h−<sup>1</sup> for the same operation performed via harvester.

The productivity values found in the present study are higher than reported in the above cited studies, both for motor-manual and mechanized felling processing. This difference is more pronounced in comparison to Spinelli et al. [50]. Such a gap concerning mechanized felling–processing can be partially related to the lower average dbh of the stems in the previous study (around 30 cm vs. 40.4 cm). However, the major difference between the present studies and the literature, which can explain the higher productivity found in the present analysis, is the lower percentage of delay times. In particular, delay percentage ranges from 1.7% in IR to 2.9% in IT2, while the average delay for motor-manual felling-processing in Spinelli et al. [50] was 29.6%, decreasing to 13.0% in mechanized operations, while in Danilovic et al. [49] delay accounted for 28.5 of the working time.

Concerning bunching–extraction operations, no study on winching and forwarding in poplar plantations for timber production were found in the literature. However, there are several studies on bunching and extraction via TimberJack 450 cable skidder, which analyzed work productivity in different forest stands. Lotfalian et al. [51] found a winching productivity of 20.2 m3 h−<sup>1</sup> in beech high stand thinning with an average extraction distance of 289 m, while Mousavi [52] reported a bunching–extraction productivity of about 11 m<sup>3</sup> h−<sup>1</sup> in beech selection cutting with an average skidding distance of 439 m. Nikooy et al. [18] reported instead a lower value of 5.2 m<sup>3</sup> h−<sup>1</sup> productivity of Timberjack 450C in timber extraction of path cutting in a pine plantation. Such lower productivity is probably related to the lower dimension of trees considering that skidding distance was comparable to those in both IR and IT1 [18]. Therefore, as a general trend, bunchingextraction productivity in the present study showed higher values than previous works reported in the literature for the same machinery in different stands and silvicultural interventions. This difference is related to both the type of forest intervention (clear cut) and to the flat terrain of IR and IT1, which facilitated logging operations.

Regarding forwarding operations, also in this case it is possible to make a comparison regarding work productivity only between different stands, considering the lack of studies on poplar plantations for timber production. The forwarder is the most commonly applied machinery in CTL (Cut to Length) harvesting operations, and it has been widely applied in artificial plantations, mostly of softwood species [53]. This machinery can reach a very high working productivity [54,55], even if a proper planning of the intervention is needed to reduce the impact which can occur considering the average dimension of a forwarder [13]. Comparing the findings of the present study with other forestry interventions in artificial plantations, it is evident that the substantial average dimension of stems, the short extraction distance and the flat terrain features in IT2 led to higher work productivity. In detail, Puttock et al. [56] reported a SMH productivity of 11.2 m3 h−<sup>1</sup> in a poplar-dominated mixed-wood stand in Southern Ontario during thinning interventions; Eriksson and Lindroos [57] showed PMH productivity for forwarding in clear cutting in pine and spruce stands of 21.4 m<sup>3</sup> h−1. Another study performed in Romania reported a SMH productivity of 15.35 m3 h−<sup>1</sup> in a clear cut of spruce stand, with an average slope of 10% and an average extraction distance of 479 m [58]. In another study recently carried out

in Poland, Magagnotti et al. [11] reported a forwarding productivity of 24.4 m<sup>3</sup> h−<sup>1</sup> SMH for a poplar plantation.

Focusing on harvesting costs, it is possible to notice how felling and processing operations accounted for the major part of these in all the yards in both soil moisture conditions, as reported by previous literature [38,59]. Felling and processing costs, with values ranging from 4.22 € m−<sup>3</sup> (IT2 dry) to 5.77 € m−<sup>3</sup> (IR moist), were in line with the literature findings for several Italian poplar plantations for high value timber production, notwithstanding higher work productivity. Spinelli et al. [50] reported a unit cost of about € 5 m−<sup>3</sup> for both motor-manual and mechanized felling–processing. This can be explained by the higher purchase costs of the machinery applied in the studied plantations. Skidding and forwarding costs are also in line with the literature, for similar harvesting systems but in different kinds of stands. In the present study the lowest cost for extraction was shown by IT2 dry at € 2.46 m<sup>−</sup>3, while the highest cost was related to winching in IT1 moist (€ 5.28 m−3). Regarding winching operations through cable skidder, Jourgholami and Majnounian [33] reported 6.15 € m−<sup>3</sup> as the cost of Timberjack 450C in timber extraction on a pine plantation, while the findings of Lotfalian et al. [51] showed 5.15 USD m−3. Focusing on forwarding, extraction costs were assessed by Cabral et al. [60] at about 1.95 € m−3, while about 7.5 € m−<sup>3</sup> were reported by Kaleja et al. [61], and about 9.2 € m−<sup>3</sup> by Magagnotti et al. [11].

Concerning environmental impact, a comparison was carried out between the findings of the present study and other similar studies, regarding both high and medium level of mechanization. In both cases, energy inputs and energy balance in the investigated poplar plantations were substantially higher [37,62].

System efficiency values were high in all three yards in both the soil conditions (ranging from 96.6% to 98.1%), and thus were in line with previous literature findings in other forest interventions [16,62,63].

GHG emissions, mostly regarding CO2, were lower than in previous literature findings, which reported a range between 3 and 33 kg CO2eq, while the values of the present work ranged from 1.6 to 2.8 kg CO2eq [5,16,24,46,64].

## *4.2. Influence of Soil Moisture on Harvesting Performance*

There is a considerable amount of literature regarding working productivity evaluation, but a major part of the studies focused on the influence on working performance of parameters such as terrain features, working distance, age, species composition, labor skills, etc. However, not much attention has been directed towards soil moisture conditions and productivity. Studies were, however, conducted on soil impact related to logging activities [65–68].

In all of the three yards, higher soil moisture led to lower work productivity, thus to higher harvesting costs, in accordance to what was reported in the few studies dealing with this topic [69,70]. High soil moisture negatively affected both motor-manual and mechanized operations, resulting in higher working times, except for felling (both with chainsaw and harvester) and processing (only with harvester). Longer working time in higher soil moisture conditions for mechanized operations (bunching –extraction) is related to the lower driving speed which the machinery was able to achieve with moist soil. The high level of moisture of the terrain caused reduced tire grip and the operators had to reduce the working speed for safety reasons. Interestingly, this did not happen for felling–processing operations by harvester, which were not affected by soil moisture regarding working time. Soil moisture also negatively affected motor-manual operations, specifically, moving (IR and IT1) and processing (only IR). This is equally related to worker safety issues, with operators that had to be more cautious during the logging activities, due to the fact that high moisture in the soil made the trail slippery, and therefore prone to accidents.

High soil moisture showed negative effects on the environmental performance of logging activities in the studied poplar plantations. There were two reasons for these negative effects: the longer working time, which required more time in which motors were running, and the higher torque needed to move the machinery in moist soil, considering the lower grip and higher attrition. Thus, there were higher emissions of pollutants.

## **5. Conclusions**

Although poplar plantations are important sources of timber in both Iran and Italy, few studies have focused on work productivity evaluation under different aspects related to sustainability in such kinds of stand. Moreover, these plantations are often located in plain or floodplain lands, therefore harvesting operations can occur in soil conditions with a high moisture content.

The aims of this study were: (i) to evaluate different harvesting systems in poplar plantations and (ii) to evaluate the influence of soil moisture on economic and environmental performance of logging operations.

In order to assess different harvesting systems in poplar plantations, from what was analyzed it is possible to state that a fully mechanized harvesting system (harvesting forwarder) is the most productive and economically sustainable with respect to semimechanical harvesting/processing and skidding extraction. However, in terms of energy balance and emissions, it is possible to state exactly the opposite, that the best harvesting system was semi-mechanical harvesting/processing and skidding extraction. These are aspects to be carefully considered during operations planning, but they must also be analyzed in terms of greater efficiency of the mechanization used, seeking to bring mechanical technologies that are increasingly efficient also in environmental terms to the forestry sector.

In order to assess the influence of soil moisture on economic and environmental performance of the logging operations, the findings revealed that high moisture content led to lower work productivity in all of the three investigated plantations, with detrimental effects on harvesting costs, which were found to be higher in moist soil conditions in all three yards. Moreover, environmental features related to pollutant emissions were higher in moist soil conditions, as a consequence of the longer time and the major torque required for the machinery to perform the logging activities.

It can be concluded from these findings that it is advisable to avoid logging operations in conditions of high soil moisture (>20%), to decrease the impact on the soil, to create higher cost-effectiveness, and to reduce the emissions of pollutants.

**Author Contributions:** Conceptualization, F.T., M.N. and R.P.; Data curation, F.T., M.N., F.L., R.V. and R.P.; Formal analysis, F.T., M.N., R.V. and R.P.; Investigation, F.T. and R.P.; Methodology, F.T., M.N., R.V. and R.P.; Supervision, F.T., R.V. and R.P.; Validation, F.T., M.N., F.L., R.V., L.B. and R.P.; Writing—original draft, F.T., M.N., F.L., L.B. and R.P.; Writing—review & editing, F.T., F.L., R.V., L.B. and R.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study could be available on request from the corresponding author.

**Acknowledgments:** This research was in part carried out within the framework of the MIUR (Italian Ministry for Education, University and Research) initiative "Departments of Excellence" (Law 232/2016), WP3 and WP 4, which financed the Department of Agriculture and Forest Science at the University of Tuscia.

**Conflicts of Interest:** The authors declare no conflict of interest.

## **Appendix A**

The data assessed in the Appendix A are referred to another Functional Unit (FU) respect to how reported in the main text. The FU in this case was 1 t of fresh mass (following the data showed in Table 2). This was important in order to give more possibility to compare these results with other studies.

**Table A1.** Harvested trees, fresh mass and working cycles for work productivity analysis in each treatment.


**Table A2.** Extraction trail and corridors' average features for the three areas in the two soil moisture conditions (mean ± SD). From the *t*-test for independent samples applied, statistically significant differences (*p* < 0.05) between the average values are highlighted (underlined text).


**Table A3.** Harvesting costs for one wood, fresh tons and percentage of costs at two main operations (felling–processing and bunching–extraction to landing) in the studied sites.


**Table A4.** Total energy inputs and balance in the studied harvesting yards, referring to surface unit and to fresh mass.



**Table A5.** Total emission assessed in the studied harvesting yards, referring to fresh mass.

**Figure A1.** Average yard productivity (bars) and possible increase of performance (lines) from SMH to PMH for the six harvested sites.

**Figure A2.** Percentage distribution of total PM10 emission in the harvesting sites studied, data shown for single operation, referring to fresh mass.

**Figure A3.** Percentage distribution of GHG emission in the harvesting sites studied, data shown for single operation and reported in CO2 equivalent, data referring to fresh mass.

## **References**


**Ola Lindroos 1,\*, Malin Söderlind 2, Joel Jensen <sup>1</sup> and Joakim Hjältén <sup>3</sup>**


**Abstract:** Translocation of dead wood is a novel method for ecological compensation and restoration that could, potentially, provide a new important tool for biodiversity conservation. With this method, substrates that normally have long delivery times are instantly created in a compensation area, and ideally many of the associated dead wood dwelling organisms are translocated together with the substrates. However, to a large extent, there is a lack of knowledge about the cost efficiency of different methods of ecological compensation. Therefore, the costs for different parts of a translocation process and its dependency on some influencing factors were studied. The observed cost was 465 SEK per translocated log for the actual compensation measure, with an additional 349 SEK/log for work to enable evaluation of the translocation's ecological results. Based on time studies, models were developed to predict required work time and costs for different transportation distances and load sizes. Those models indicated that short extraction and insertion distances for logs should be prioritized over road transportation distances to minimize costs. They also highlighted a trade-off between costs and time until a given ecological value is reached in the compensation area. The methodology used can contribute to more cost-efficient operations and, by doing so, increase the use of ecological compensation and the benefits from a given input.

**Keywords:** restoration; no-net-loss; biodiversity conservation; wood living species; mining; forwarder; forest operations; cost-efficiency; boreal forest; Sweden

## **1. Introduction**

Anthropogenic disturbance has altered ecosystems worldwide, resulting in habitat loss and species extinctions over a wide range of biomes [1,2]. Forest ecosystems are no exception, and their exploitation has led to changes in ecosystem structures and processes, and to biodiversity loss [2,3]. Species associated with dead wood (saproxylic species) are especially vulnerable.

To conserve biodiversity and continue economic development simultaneously is a major challenge as human society depends on functional ecosystems in numerous ways [4]. Economic growth and biodiversity conservation are often perceived to be incompatible. There is continually increasing pressure on corporations by consumers and stakeholders to be environmentally conscious with greater focus being directed towards alternative approaches in adapting to this [5]. One such approach is the relatively recent concept of ecological compensation (biodiversity offsetting), which is based on the principle that those who destroy or damage natural values are to compensate for this by the creation or protection of natural values at a different/substitute location [6]. Thus, ecological compensation potentially provides an approach that links biodiversity conservation and human development associated with economic growth. Although legislation man-

**Citation:** Lindroos, O.; Söderlind, M.; Jensen, J.; Hjältén, J. Cost Analysis of a Novel Method for Ecological Compensation—A Study of the Translocation of Dead Wood. *Sustainability* **2021**, *13*, 6075. https://doi.org/10.3390/su13116075

Academic Editors: Rachele Venanzi, Janine Schweier and Rodolfo Picchio

Received: 16 April 2021 Accepted: 25 May 2021 Published: 28 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

dating biodiversity offsets exist in many countries at present, biodiversity offsets are still under development.

Biodiversity offsets are the last step in the mitigation hierarchy, with the overarching goal to achieve no net loss of biodiversity [6]. These actions can involve protection of areas that are otherwise at risk of exploitation, through ecological restoration or other positive management interventions and, in some circumstances, the re-creation of habitat that has been lost. Restoration of degraded habitat is often used for ecological compensation and, although our knowledge of the effects of different restoration methods on biodiversity has improved in recent years [7–10], it often suffers from the innate problem that even if substrates or habitat for species that we want to favor are provided, those species may not be able to migrate to the restored areas [7,9]. Furthermore, the delivery time of some types of substrates is long, e.g., large diameter dead wood takes a very long time to develop and reach late decompositions stages, and it could take centuries after restoration before these kind of substrates are available for specialist species. One way to circumvent this problem is the translocation of substrates and associated species from the impact area to compensation areas.

Translocation means that habitats and substrates with long delivery times are instantly created in a compensation area. In addition, in the best-case scenario, many of the organisms associated with these high-quality substrates are translocated together with the substrates rather than having to migrate there. Thus, fulfilment of the "fields of dreams" hypothesis would be unnecessary for successful compensation and/or restoration [11].

Due to the concept of ecological compensation being a novel one, there is little in the way of relevant research on methods and outcomes of ecological compensation. However, there have been assessments of the ecological functionality and methods for evaluation of the compensation measures (e.g., [12]), and also research into the social and economic effects on local societies (e.g., [13]). However, there has been limited focus on the costs of carrying out the actual compensation measures. When such costs have been investigated, it has often been in terms of the total costs for compensation projects carried out, with little focus on comparing alternatives in order to find and develop cost-efficient practices (e.g., [14]). Even when the ecological compensation constitutes a minor part of large-scale projects, such as the construction of roads and establishment of mines, cost-efficiency is, nevertheless, instrumental for increasing both the use of ecological compensation and increasing the benefits from a given economic input.

Evaluation of cost-efficiency is a central part of research in, for example, forest operations. Based on methods for cost assessment, such research focuses on evaluations of, and improvements to, work carried out, and on predicting the outcomes of planned operations. The aim is to estimate the cost of the work required to produce the desired outcome of the operation. To do so, the cost assessment focuses on two main parts: the cost per unit of time, and the time required per produced unit. The cost per time unit is based on the labor costs, operational costs and investment costs related to the operation [15]. The time required per produced unit is often called time consumption, and often there is a focus on the inverse of the time consumption—the productivity (produced units per time). The time consumption is dependent on the conditions in which the operations are carried out and is measured by dividing the work into distinct parts (work elements) in order to isolate influencing factors better. By doing so, relationships between time consumption and the influencing factors are easier to distinguish [16]. There is a wealth of research on time consumption and productivity of conventional forest operations, with well-known general relationships. For instance, the required time consumption per produced unit increases with transport distance and decreases with the number of units that can be handled at a given time (e.g., per load) [17–20]. Cost-based evaluations of conventional operations are common [21], and there have also been evaluations of how ecological considerations influence the costs of operations [22]. However, cost-based evaluations of ecological compensation are scarce.

The largest copper mine in Sweden is the Aitik mine, founded in 1968 by the mining company Boliden AB. The ore extraction process at the mine produces 50,000 tons of tailings daily. These are subsequently transported to a sand magazine. Recently, the Boliden AB was granted a permit to increase the sand storage area in the Aitik mine by 376 ha. The affected forest land had high to very-high natural value and The Land and Environmental Court of Appeal decided that offsets areas should be created to compensate for the impact. In addition, as part of the compensation measures it was decided that substantial amounts of dead wood and associated species of insect, wood fungi, lichens, bryophytes and lichens living in or on the dead wood should be translocated from the impact to a 397 ha compensation area 5 km (in a straight line) west-southwest from the affected area.

The aim of this study was to assess the cost of translocating dead wood from an affected area to the compensation area. More specifically, the focus was on assessing the costs for different parts of the translocation process and their dependency on some influential factors (i.e., transportation distances and load sizes). Moreover, being part of a scientific project to evaluate the ecological outcome of the translocation, the study also provided input on some of the costs associated with carrying out such an evaluation.

This project is unique in its usage of translocation of natural values as a primary means of ecological compensation in boreal forests. The results presented here are thus highly significant for future endeavors involving ecological compensation (in a boreal setting) and may be valuable to other companies facing similar challenges.

## **2. Materials and Methods**

The study was conducted during the creation of off-set areas required to compensate for the impact of increasing the sand storage area in the Aitik mine. Hence, it was an observational study (and not an experimental study), with only limited possibilities to interfere with the work in order to collect desired data.

## *2.1. Work Phases*

The translocation of dead wood consisted of seven phases, from the identification of substrates (phases 1 and 2) to the translocating work (phases 3–7) (Figure 1). All phases, except the log marking, were necessary for the actual translocation work. However, for the area identification, felling and the insertion phases, there was extra work carried out in order to facilitate the planned scientific evaluation of the translocation.

**Figure 1.** Progression of the work phases in the translocation project.

The machine operators who carried out the work in the phases Extraction, Road transport and Insertion were given instructions to drive as normal, but to handle the logs with additional care since they could be more fragile than newly harvested and fresh logs.

#### *2.2. Area Description*

The impact and compensation areas are located in the north boreal vegetation zone [23] (Ahti et al. 1968). The area had previously been under silvicultural management, predominantly subjected to selective felling, but had not been managed over recent decades. The forests were dominated by conifers (Scots pine (*Pinus sylvestris*) and Norway spruce (*Picea abies*)), and with scattered broadleaves (mainly Downy birch (*Betula pubescens*)).

The affected area covered 376 hectares, of which 167 hectares consisted of forests of high or very high conservation value as defined by Swedish Standards Institute [24]. The area included 21 red-listed species and forest structures important for biodiversity (e.g., very large old trees, snags and logs of unusual dimensions).

The compensation area covered 397 hectares, of which 192 hectares was productive forest (mean annual increment of at least 1 m3 per ha and year) with high conservation value, 113 hectares of productive forest with low conservation value but with no forest of very-high conservation value. The remaining areas were non-productive forest, mires or open water [25].

To allow for a scientific assessment of the outcome of the translocation of the dead wood for biodiversity, the whole compensation area was systematically inventoried prior to translocation of the dead wood. Thirty circular plots were randomly distributed in areas with productive forest. The translocated dead wood was placed in the plots to enable the planned scientific evaluation. The evaluation focused on assessing ecological effects due to the quality and quantity of dead wood. Hence, each plot received either 48, 16 or zero translocated logs, with 10 replications of each log concentration, giving a total of 30 experimental plots. For each concentration there were instructions on the number of logs of the different types to be placed in each plot. Plots were 50 m in diameter and separated from each other by at least 150 m. Detailed information on the scientific evaluation can be found in [26].

The inventory work and establishment of the plots were carried out by two nature value consultants, who were also involved in work phases 2, 4 and 7.

#### *2.3. Substrate Identification*

The impact area was systematically inventoried by the nature value consultants for standing and lying trees, and the meeting of criteria for any of the eight log assortments identified based on tree species (Scots pine or Norway spruce), posture (standing or downed) and decomposition stage (fresh, early or intermediate) as suitable for translocation (Table 1, examples in Figure 2). Trees meeting the criteria were marked.

**Table 1.** Log assortments that were translocated, and the mean log volume in each assortment, from sampling carried out for this study.


#### *2.4. Tree Felling and Log Marking*

The translocation work began on 2 October 2017, starting with felling of standing trees, cutting felled and lying trees to logs and marking the logs. The nature value consultants chose which of the marked trees to create logs from, and to which dimensions. The objective was to find logs with diameters greater than 15 cm and lengths between 3 and 5 m. Some trees marked in the previous phase were excluded from translocation as they were too decayed or too thin. Only dead wood substrates that were considered feasible to move without breaking were selected for the translocation. The number of relocation logs created per tree was not recorded, but the number of logs was just slightly higher than the number of trees marked during the substrate identification phase (Table 4).

The felling and cross-cutting work was carried out by two chainsaw operators, each with at least 25 years of experience in motor-manual forestry operations. Both used Husqvarna 550XPG chainsaws with 25.72 cm (18 inch) blades.

The logs were carefully handled so as not to reduce their value to the project e.g., by not disturbing the attached fruiting bodies of wood fungi. The logs were inventoried for red-listed or indicator species, tagged for dead wood assortment, photographed, and their upper side marked with paint so that they could be placed in the same position after

translocation. Finally, they were marked with a unique ID. A total of 671 logs were created. Due to a limited number of lying dead pine trees, with only 20 early and 66 intermediary decomposition logs found instead of the desired 80 in each class, they were supplemented with logs from the pine fresh wood and pine kelo style classes, respectively (see N in Table 1).

**Figure 2.** Examples of different classes of translocated logs. Downed intermediate decomposition logs of pine(**A**) and spruce (**B**), fresh standing pine (**C**) and standing pine of kelo type (**D**). Photo: Nordlund Konsult AB.

## *2.5. Extraction*

All 671 marked logs were extracted from the affected area to the roadside, using a conventional forwarder (Table 2). The size of the forwarder and its impact on the ground was not considered when the machine was being chosen, since the area was to be converted. Logs were arranged by assortment type at two separate landings. In order to protect the translocation logs, the bottom of the pile was made from a layer of conventional roundwood. The extraction was carried out by a machine operator with 18 years of experience of forwarding work.

### *2.6. Road Transport*

The extracted logs were transported using a conventional timber truck (Table 2) from the affected area to three separate landings at the compensation area. The first load (4 logs per assortments, 32 logs in total) was transported to the first landing, loads 2 and 3 were transported to the second landing (28 per log assortment, 224 logs in total) and loads 4, 5 and 6 were transported to the third landing (48 per log assortment, 384 logs in total). The logs were loaded in order of decreasing durability, with fresh wood being placed at the bottom, then kelo type trees, early lying dead wood and, finally, intermediate lying dead wood. In total, 640 of the 671 logs were transported, with the numbers per assortments reported in Table 1.

The road transport was carried out by two machine operators with at least 18 years of experience with self-loading timber trucks.


**Table 2.** Specifications for the machines used for the work phases Extraction, Road Transport and Insertion.

<sup>a</sup> metric tonnes (1 tonne = 1000 kg).

## *2.7. Insertion*

The insertion of logs to the compensation area was carried out using a small tracked forwarder (Table 2) under the supervision of nature value consultants. The size of the forwarder was chosen to minimize the impact on the ground, and so that it could navigate between the trees in the stand. A total of 640 logs were inserted in 20 plots, with 2 logs of each log assortment placed at each of 10 (16 logs per plot) of the 20 plots, and 6 logs of each log assortment at each of the remaining 10 plots (26 logs per plots). Within plots, log assortments were randomly distributed.

The insertion was carried out by two machine operators with 3 and 5 years of part-time experience, respectively, with small forwarders.

## *2.8. Cost Assessment*

Costs were collected in the national currency Swedish krona (SEK), which at the time of the study had an exchange rate of ca 10.0 SEK per Euro and 8.6 SEK per USD.

The total cost for the compensation project was calculated by totaling the total costs for each work phase. The total cost for each work phase was calculated by multiplying the times required to carry out the work involved in the phase by the hourly cost for the respective work.

Cost per translocated log was calculated for each work phase by dividing the total cost for the work phase by the number of translocated logs according to:

$$\mathbf{C}\_{\mathbf{x}} = \frac{\sum \left( T\_{\hat{i}} \times \mathbf{c}\_{\hat{i}} \right)}{O} \tag{1}$$

where *C* is the cost per translocated log for work phase *x*, *T* is the time required for work *i* in the work phase, and *c* is the hourly cost for work *i*. *O* is the number of translocated logs (irrespective of the number of objects handled in work phase *x*).

For the phases Extraction, Road Transport and Insertion, the work was assessed in more detail by use of time studies in order to enable the creation of models for cost estimations under various work conditions, i.e., with time requirements other than the observed total time requirements in this study.

## *2.9. Time Studies*

The work-time required for translocation, divided into the different work phases, was collected from the self-reported work-time of the operators. For the three last phases (Extraction, Road transport and Insertion) detailed time studies conducted on site were carried out on a number of loads.

For the studied loads, total values of load size, time consumption and distance driven were recorded. Driving speeds were derived from distance and time recordings. The time consumption per observed load in the phases was recorded, split over six work elements (Table 3) to isolate influential variables. Extraction was the first work phase to be time studied and suffered from initial technical challenges to record the driving distances. Hence, between 9–11 of the 26 loads were properly recorded with driving distances at work element level (see "*n*" under Transport distance in Table 6).

**Table 3.** Definition of work elements used in this study.


The work element Loading was observed at even higher resolution for the Road Transport and Insertion phases, by recording loading time per log assortment. In addition to time, distances driven, load sizes (number of logs) and number of crane cycles (loading/unloading) were also recorded.

During extraction, loading times per log assortment were not recorded, since the logs were loaded in the order they were found in the forest, which was not organized by log assortment. Furthermore, it was difficult to determine when the work element Loading was complete, due to the variation in load capacity usage and the indistinct loading area (i.e., loading both when driving from and to the landing).

All time studies were carried out by the same person. During Extraction and Road Transport, the person was located in the machine cabins and, for the Insertion, the person followed the forwarder on foot. The time consumption by the time studies work was recorded in units of a second.

## *2.10. Calculations*

Speed was calculated as the distance driven divided by the time consumption for the work element and/or load (round trips) of interest. Productivity, here defined as the output (in terms of logs or m3) per time unit, was calculated as the load size divided by the time consumption for the work element and/or load (round trips) of interest of the respective work element.

Between 14 and 55% of the logs in the eight translocated log assortments were sampled to give volume estimations (Table 1). Individual log volumes were calculated assuming a cylinder:

$$(V = \frac{\pi \ast d^2}{4} \times l) \tag{2}$$

where *V* is the volume, *d* is the log diameter over bark at half-length and *l* is the full length of the log. Hence, an even tapering of the logs was assumed. The volume unit used was solid cubic meter of wood over bark (m3).

Load sizes during Extraction, Road Transport and Insertion were calculated as the total volume of logs in the load, by multiplying the load's number of logs of each log assortment with the mean log volume for each log assortment (i.e., from Table 1). Utilized load capacity was calculated by dividing load size by the machine's load capacity (Table 2).

## *2.11. Statistical Analysis*

The impact of transport distance and load size during the work phases Extraction, Road transport and Insertion, on the time consumption of work elements Loading, Driving Full, Unloading and Driving empty, was assessed using linear regression. Data preparation was carried out using Microsoft Excel, whereas all the regression analyses were carried out using Minitab 17 (Mintab Inc., State College, PA, USA) with the critical significance level set to 5%.

#### **3. Results**

The results are here presented in three levels of detail. First, the work phase costs are reported, based on the costs invoiced. Second, detailed analysis of the work carried out with some work phases is reported, along with the analysis of the relationships between work time required and work conditions. Third, those relationships are used to model estimated costs under other conditions, in order to highlight cost-driving factors as well as to demonstrate the possibility of using the analysis to estimate costs for future projects.

## *3.1. Costs*

The total costs of the seven work phases involved in the compensation project was 520,800 SEK (Table 4). When distributing the total costs over the number of translocated logs, the cost per log was 813.8 SEK. The insertion of logs to the plots in the compensation area was the most expensive work phase, accounting for around 29% of the costs. About two thirds of the insertion cost was associated with the physical translocation of logs, while one third pertained to the work of the nature value consultants. The cheapest work phase was the Road Transport, which accounted for 5% of the costs.

**Table 4.** Time consumption, hourly cost, total cost, number of objects (logs/trees), cost per translocated log (*n* = 640) and the work phases' proportion of the cost.


The cost for work directly related to the planned scientific evaluation of the ecological compensation project was made up of all log marking, all of the nature value consultants' contribution at insertion, 20 h (33%) of the time consumption for area identification, 6 h (5%) of the time for insertion of logs (i.e., 20% of the unloading time, see time required below) and 40 h (20%) of the tree felling. When totaling those times and the related hourly cost, the cost related to carrying out the scientific evaluation was 348.8 SEK per log and the cost for the actual compensation work was 465.0 SEK/log. Hence, the included operational preparations for the scientific evaluation of the project required an additional 75% of financial resources, in addition to the resources needed for the compensation work.

## *3.2. Work Analysis*

On average, 18 logs were put on each load during extraction, which was almost twice as many as when inserting the logs (Table 5). However, in terms of utilization of the different vehicles' payload capacities, the conditions were reversed. In fact, only one third of the payload capacity was, on average, used during extraction, whereas more than two thirds were used during insertion. This was due to a much smaller payload capacity on the forwarder used for insertion (Table 2).


Logs were generally loaded and unloaded individually during the extraction and insertion, whereas slightly more than two logs were handled at a time when loading and unloading during road transport (Table 5).

On average, the total distance driven during Extraction was almost 1.6 km per load, 49 km during Road Transport, and 2.3 km during Insertion (Table 6). However, the distances varied between loads. For Extraction and Road Transport, the loading work required the highest amount of work time, whereas it was the driving with and without load, that required most work time for Insertion. The first three loads (to landings 1 and 2) of the Road transport loads were hauled to insertion landings 1 and 2, at about 13 km driving distance from the extraction landing. Loads 4–6 were hauled to landing 3, about 33 km from the extraction landing. Thus, over all loads, the distances reported had high variation (high SD in Table 6), but the speeds were rather consistent.

**Table 6.** Time consumption, transport distance and speed per load for the four work phases, distributed over work elements.


<sup>1</sup> For most studied loads, total values were recorded but not always at a work element level.

On average, the driving speed with and without a load was about 2 km/h during Extraction, whereas it was about twice as high during Insertion (Table 6). The average driving speed during Road Transport was almost 40 km/h. In all three work phases, the driving speed was slightly higher when driving without payload than with payload.

In total, the mean productivities were 27.6 extracted logs per hour (SD 21.9), 50.2 road transported logs per hour (SD 19.3) and 11.5 logs inserted per hour (SD 3.3) This corresponds to a volume-based productivity of 7.6 (SD 8.1), 9.2 (SD 3.9) and 2.8 (SD 0.6) m3 per hour.

The analysis of the observed work element relationships with external factors yielded significant relationships between distance and the time required for driving with and without load (Table 7). Moreover, the time required for loading as well as unloading depended significantly on the load size (i.e., number of logs per load). For Extraction, the distance driven during loading was found to be the strongest predictor of time consumption per load (Loading, time per load (h) = 0.1041 × distance (km), *n* = 10, *p* < 0.001, R2-adj = 89.7). However, given the features of the loading work, the loading is dependent on the number of logs loaded, and the distance driven during loading is directly dependent on the logs' dispersion. Indeed, the loading distance driven was strongly dependent on the load size (distance (km) = 0.0263 × load size (number of logs), *n* = 10, *p* < 0.001, R2-adj = 83.6). Following this logic, and due to the rather small sample size, load size was used for the models in Table 7.


**Table 7.** Models to predict time consumption per load as a function of transport distance or load size, based on regression analysis of observations in Table 6. In the case where no significant models could be found for a work element, the time consumption was predicted using the mean value from the observations.

(a) a = Transport distance in km, b = load size in number of logs. When no significant relationships with a or b were found, the mean values from Table 6 were used. \* The coefficient value for a corresponds to a speed, which can be calculated by dividing 1 by the coefficient (e.g., the model's speed for driving empty during extraction = 1/0.4431 = 2.3 km/h).

For Insertion, the loading time model yielded a relatively low level of explained variation. This was because two different work modes were used: one where several assortments (6 loads) were loaded, and one where fewer assortments (11 loads) were loaded. When analyzing just the load with fewer assortments (i.e., with less driving between log piles during loading), the model explained a much higher level of the variation (Loading time per load (h) = 0.0434 + 0.00832 × load size (number of logs), *n*= 11, *p* < 0.001, R2-adj = 85.7). However, both models resulted in very similar time predictions.

## *3.3. Modelling of Costs and Cost Sensitivity*

Using the models described in Table 7, it is possible to analyze the effect of distance and load size on the time consumption for Extraction, Road Transport and Insertion. By combining this with the hourly cost for the work, the effect on costs can also be analyzed.

For instance, the total time required per load for the translocation related Insertion work can be estimated by adding the six work element models thus:

Total insertion time = 0.2262a1 + 0.0612 + 0.00847b + 0.2770a2 +0.8 × 0.02830b + 0.11 + 0.12 (3)

where a1 and a2 are the distances (in km) expected to be driven when driving without and with load, respectively, and b is the number of logs expected in the load. Since 20% of the observed unloading time during Insertion was considered attributable to the work related to carrying out the scientific evaluation, the model for predicting the unloading time is multiplied by 0.8 (i.e., (100% − 20%)/100).

The cost per load is calculated by multiplying the hourly cost for insertion work (from Table 5) with Equation (3), and the cost per log is calculated by dividing the cost per load by *b*. It should be noted that this modeling assumes that a large number of logs are translocated and thus results in a large number of loads. The fewer the number of loads transported, the more influential the threshold effects of utilized load capacity will be. For instance, when considering translocating only a few logs, there will be a substantial difference in costs per log if a last trip with a single log in the load is required.

By applying this methodology and assuming that distances with, and without, a load are the same (for simplicity), the general effects of distance can be explored (Figure 3A). The distance-independent work per load is high for the Road transport, as indicated by the long time required for the distance of 0 km. However, when distributing the time based on load size, the Road Transport has the lowest time required for distance-independent work of all three work phases (Figure 3B). Insertion starts at the highest amount of time required per log, but extraction exceeds the time required at an expected driving distance of 1.5 km. However, since the insertion forwarder has a lower hourly cost than the extraction forwarder, the intersection is at 0.8 km when looking at the cost per log (Figure 3C). In this example, it costs 29.4, 16.7 and 45.2 SEK/log for the distance-independent work for the Extraction, Road Transport and Insertion phases, respectively. On top of this, there is an additional 44.2, 0.4 and 22.5 SEK/log in transportation costs for each kilometer.

As indicated by the models (Table 7), load size also affects the time consumption and thus costs, since it influences the loading and unloading work elements. Based on similar examples as those in Figure 3, it can be seen that the expected time per load increases most for each extra log in the load for Insertion (Figure 4A). When being analyzed as time per log, it can be seen that the load size effect is accentuated the smaller the loads are (Figure 4B). A similar pattern is also present for the cost per log (Figure 4C), and the effect is due to how load size-independent time/costs are distributed over fewer logs the smaller the load is. In contrast, the load size effect flattens out at a certain number of logs in the load, and this happens at a lower load size when the log-independent time requirements per load (and costs) are low. For 1 log per load, the time required is expected to be 0.37, 1.73 and 0.78 h/log for the Extraction, Road Transport and Insertion, respectively, under the assumptions shown in Figure 4. Correspondingly, the costs would be 336, 1559 and 581 SEK/log.

By applying different load sizes and hourly costs, the effects can be explored further than here, and the costs for possible alternatives can be evaluated.

**Figure 3.** Predicted time consumption (**A**) per load and (**B**) per log, and (**C**) predicted cost per log for the actual translocation work as a function of expected driving distance, when assuming that the distances driven with and without load are identical. Hence, 1 km in the figure gives a total distance of 2 km for driving with and without load. Load sizes are assumed to 18, 105 and 10 logs for the Extraction, Road Transport and Insertion, respectively. Correspondingly, the hourly costs are assumed to be 900, 900 and 750 SEK, respectively.

**Figure 4.** Predicted time consumption (**A**) per load and (**B**) per log, and (**C**) predicted cost per log as a function of expected load size. To mirror approximately the maximal load capacity of the study's vehicles, load sizes have been limited to 40 and 20 logs for Extraction and Insertion, respectively. Driving distances are assumed to be 0.3, 24 and 0.9 km for driving with and without load, respectively, for the Extraction, Road Transport and Insertion, respectively. Correspondingly, the hourly costs are assumed to be 900, 900 and 750 SEK, respectively.

## **4. Discussion**

Translocation of dead wood and associated organisms is a novel method for ecological compensation and restoration that could potentially provide a new important tool for biodiversity conservation. For example, by using this method, habitats/substrates with normally very long delivery times (e.g., large diameter dead wood in late stages of decomposition) are instantly created in a compensation area and, in the best-case scenario, many of the organisms are translocated together with the substrates. Hence, the time to achieve high ecological value in the compensation area is likely to be substantially shortened by the translocation. However, neither the outcome for biodiversity conservation nor the cost for translocation of dead wood have yet been properly evaluated. This study is, to the best of our knowledge, the first attempt to assess the cost of the translocation process. Our analyses revealed big differences in cost for different phases in the translocation process and thus identified phases where efforts should be made to reduce costs and improve efficiency.

The observed cost was 813.8 SEK per translocated log, of which 42% was for work related to facilitating the planned scientific evaluation of the translocation's ecological results. Thus, when omitting work related to the scientific evaluation and including only the work related to the actual translocation plan, the cost was 465 SEK/log.

Due to the different nature of translocation work compared to normal forest operations, it is difficult to compare the observed time consumptions and costs to previous research. However, it can be noted that the driving speeds are in line with previous research on extraction [18,27] and road transport [28,29]. There is a lack of research about the forwarder used for the insertion (Terri ATD), but it can be noted that it was driven at a considerably higher speed than during extraction with conventional forwarders. Hence, the small load volume of the insertion forwarder is, to some extent, compensated by its higher speed. In relation to the forwarder used in the extraction, the insertion forwarder's work becomes more competitive the longer the distances driven (Figure 3B,C).

In relation to costs, there is a limited amount of research to compare with. However, the cost of conventional logging can serve to put the costs in perspective. The average costs for harvesting trees and extracting the logs to roadside in Sweden are approximately 95 SEK/m3 for final felling, with average tree volumes of 0.23 m<sup>3</sup> under bark and with 44% of the costs being attributable to the extraction [30]. So, assuming the insertion corresponds to two extraction costs, conventional harvesting, extraction and insertion would be 137 SEK per m3. Conventional Swedish road transport costs are approximately 86 SEK per m3 in Northern Sweden [30]. So, in total, conventional felling, extraction, road transport and insertion could be estimated to cost 223 SEK per m3. With a mean log volume of 0.24 m3 over bark as in this study, this corresponds to approximately 54 SEK/log, whereas it was observed to be almost seven times higher (363 SEK/log) for the corresponding work phases of actual translocation work (i.e., with the costs of the scientific evaluation excluded). There are substantial shortcomings in the comparison, but it, nevertheless, clearly highlights the considerably higher costs related to translocation of ecologically valuable, and sensitive, logs compared to fresh logs for industrial uses.

In this observational study, total time consumptions and costs were based on selfreported data from the contractors executing the translocation work. The self-reported data was also provided to the customer for reimbursement purposes, and thereby motivating correctness in relation to agreements and to maintain the business relationship. Thus, the self-reported costs were the actual costs invoiced for executing the studied work. The conducted time studies of some work phases gave input on how the invoiced costs would change under different conditions. Additional time studies could provide similar information for also the other work phases. To estimate the actual time requirements and the actual hourly costs (and not invoiced) would naturally also be of interest, but would require a larger study set-up and also data for estimating hourly costs of contractors.

Although the full cost of the project to expand the sand magazine is not known, the compensation costs can be expected to be relatively small in comparison. Compensation costs are, nevertheless, one of many costs in projects that often require large investments, so cost efficiency can contribute to at least two beneficial effects. The first is by making ecological restoration more likely to be used when costs can be kept low. The second is by providing increased benefits for a given cost, when the most cost-efficient measures can be chosen.

There are several possible ways to influence the costs, by choices related to the planning and execution of the operations, and by the qualitative requirements of the result of the operation.

The results showed that there is an increased cost with increased driving distances for all three work phases in which logs are transported (Figure 3B). The cost increase was substantial for both extraction and insertion. Hence, the closer to roads that the logs to be extracted can be collected, the lower the costs will be. Correspondingly, insertion costs will be lower the closer to the road the logs can be placed in the compensation area. In contrast, road transport contributed very little to overall cost and the cost increased very little with distance. Thus, an increase in road transport distance will have relatively little effect on overall cost, so translocation over larger distances is possible without substantial impact on overall cost. In fact, a doubling of the road distance driven would only have resulted in a 4% increase in costs per translocated log (0.40 SEK/km and log × 50 km/456 SEK/log). For future translocation projects, this knowledge should be beneficial in allowing the prioritization of short extraction and insertion distances over road transportation distances when trying to minimize costs.

Another well-known time and cost driving factor that was not studied is the effect of log concentration. This can, briefly, be described as the more logs that are located in one place when being loaded or unloaded, the faster and thus cheaper the work [31]. Hence, if logs can be collected within a limited area during each round trip with the forwarder, the cheaper it will be. Similarly, it will be more efficient if the loading and unloading of trucks and the insertion forwarder can be carried out in concentrated areas. In fact, the organization of logs at the roadside landings was one important area of improvement highlighted by the operators in the study, in order to minimize the need for relocating the vehicles when loading/unloading different log types.

The operations described in this study were carried out for the first time by the people involved. It can be expected that repeating the operation will be quicker as these people gain more experience, and hence, the costs will decrease. A key way of improving extraction may be to use a different type of grapple than the conventional one used, preferably one that is small, straight and pointed to grip one log at a time without damaging the logs being loaded or the adjacent logs. Better planning and marking of log positions and log types, as well as clearer extraction trails, have been suggested by the operators to reduce the time for extraction, but it would also create extra work during extraction preparation. Furthermore, it would be desirable to only work with as many logs as are required for the relocation. In this study, 671 logs were produced during felling and subsequently extracted, but only 640 were road transported and inserted. The excess could be seen as a buffer to ensure that the desired number of logs of different qualities were available for the Road Transport and Insertion phases but, ideally, such an excess would not be necessary.

A rough estimation for the new improved planning and execution is that it could result in a total cost of 397 SEK/log, corresponding to a cost reduction of 15%. The estimation is the result of removing the time estimated to be associated with the 31 excess logs, a 30% decrease in time consumption during loading for both extraction and road transport (corresponding to better organization of roadside landings), and a subsequent 9% decrease in time consumption during extraction, road transport and insertion (corresponding to a general improvement in work efficiency due to experience).

The qualitative requirements of the operation greatly affected the cost. This was partly manifested by the high cost for the insertion, in part due to the choice of using a smaller forwarder to reduce the risk of damage to the ground in the compensation area. Thus, if ground damage had not been an issue, the use of a forwarder with a higher payload and thereby better cost-efficiency might have reduced costs. However, the quality of the translocated logs affected the cost even more, as indicated by the low load capacity utilizations (Table 5). If payload capacities were fully met, as with conventional logging, the costs would be substantially decreased. The challenge, however, is that the more decomposed and therefore fragile the logs are, the greater ecological value they might have as more rare- and threatened-species are associated with the late stages of decomposition [32,33]. Hence, there is a trade-off between costs and the qualitative properties of the translocated logs. Increasing payload usage by loading more logs with high ecological values carries with it an increased risk that log structures and species living on the logs would be severely damaged when piling them on top of each other. In fact, the most valuable logs in the impact area were not translocated since it was not considered feasible to move them due to their fragility [26]. Another way to utilize the payload capacity would be to translocate less ecologically valuable logs that can survive being transported in full loads. Yet another way would be to construct loads such that the most fragile logs were loaded on top with the most solid logs at the bottom of a load.

The level of decomposition of the logs could also be expected to influence the loading and unloading work. The more decomposed the logs are, the more carefully they need to be handled. The size of the logs is also a determining factor on the loading and unloading times. During road transport, the observed difference in mean loading times between the fastest log assortment (spruce logs, early decomposition) and the slowest (pine logs, intermediate decomposition) was 13.3 s per log. The same log categories were loaded fastest and slowest during insertion, with a mean difference of 23.4 s per log. It was not possible to evaluate whether the observed differences were statistically significant due to the aggregated data collection method, neither was it possible to evaluate possible reasons (such as differences in log sizes between classes).

When considering the choices related to qualitative results of the translocation, there is a trade-off between costs and time until a given ecological value is reached in the compensation area. If it is acceptable to wait for the ecological values to develop, economy of scale can be used in order to cut costs. Conversely, the time required to achieve high ecological values in the compensation area can be reduced by accepting higher costs.

To know the acceptable timeline for ecological values to develop, there is a need for knowledge about the speed at which such values develop under different treatments. The work carried out in this translocation project included facilitation of such scientific evaluation, which focused on evaluation of the effect of differences in concentration of dead wood. Hence, this study covered some of the costs of knowledge production and found that the operational work to carry out the scientific evaluation required an additional 75% of financial resources in addition to the resources needed for the actual compensation work. The operational work to lay out the experiment is only one part of an evaluation project. Thus, the costs related the researchers' design of the experiment, the inventories required to follow up the results of the translocation, the data analysis and publication of results should be totaled to give the full cost of the scientific evaluation. A modest estimation is that those costs not accounted for in this study are of a magnitude 10–100 times higher than the observed cost related to carrying out the scientific evaluation.

To the best of our knowledge, this study is unusual in that it added a cost-efficient dimension to ecological compensation. The data from a single observational study should naturally be handled with care when being used in other contexts, and there is a need for additional studies to verify and refine the findings. However, the general results are in line with similar studies from conventional forest operations, indicating that similar cost-driving aspects are in action. Hence, the methodology seems applicable to operations within the field of ecological compensation, and could be applied to enable cost-benefit evaluations of possible alternatives. This would enable efficient use of scarce resources, both when it comes to the strict delivery of the compensation (i.e., to spend as few resources as possible on a decided compensation action) as well as to the ecological quality of different actions (i.e., to choose the action that gives a desired benefit-to-cost level).

**Author Contributions:** M.S. and O.L. conceived and designed the study, with contributions from J.H.; M.S. collected the data; M.S. and O.L. analyzed the data, with contributions from J.J. and J.H., O.L., M.S., J.J. and J.H. wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by SLU, Boliden AB and the Swedish Research Council on Environment, Agricultural Sciences and Spatial Planning (Formas), grant number 2019-00923.

**Acknowledgments:** This paper was based on data from the Master's thesis of Söderlind [34]. The authors thank Boliden AB and Sveaskog AB for allowing access to the impact area and help with practicalities, Nordlund Konsult for invaluable help with planning and carrying out the translocations and the machine operators who volunteered to participate in the study.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

## **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Sustainability* Editorial Office E-mail: sustainability@mdpi.com www.mdpi.com/journal/sustainability

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel: +41 61 683 77 34

www.mdpi.com

ISBN 978-3-0365-4978-1